http://curtis.ml.cmu.edu/w/courses/api.php?action=feedcontributions&user=Nqi&feedformat=atomCohen Courses - User contributions [en]2024-03-29T05:00:08ZUser contributionsMediaWiki 1.33.1http://curtis.ml.cmu.edu/w/courses/index.php?title=File:Niting_ppt.pdf&diff=5563File:Niting ppt.pdf2011-04-19T03:27:22Z<p>Nqi: </p>
<hr />
<div></div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=NqiPresentation.pptx&diff=5562NqiPresentation.pptx2011-04-19T03:26:47Z<p>Nqi: </p>
<hr />
<div>[[File:Niting_ppt.pdf]]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5520Dan Cosley, AAAI, 20102011-04-04T22:11:51Z<p>Nqi: /* Brief Description Of The Method */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]], then it defines some notation for various sets of tuples. At last, it gives a method about how to [[UsesMethod::Simulating Ordinal Time from Snapshots]]. <br />
<br />
Defination:<br />
<br />
* <math>B\left ( t_{1} \right )=\left \{ \left ( u,C,k_{1} \right )\right \}</math>: u joined C before t1 and u had k1 neighbors in C at t1.<br />
<br />
* <math>J\left ( t_{1},t_{2} \right )=\left \{ \left ( u,C,k_{1},k_{k2} \right )\right \}</math>: u had k1 neighbors in C at t1, u joined C between t1 and t2, u had k2 neighbors in C at t2.<br />
<br />
* <math>N\left (t_{2} \right )=\left \{ \left ( u,C,k_{k2} \right )\right \}</math>: u did not join C before t2 and u had k2 neighbors in C at t2.<br />
<br />
The sets B and N result in shifting no and do upwards with respect to <math>n_{s}</math> and <math>d_{s}</math>. Also, the set J results in stretching <math>n_{o}</math> and <math>d_{o}</math> when compared <math>n_{s}</math> and <math>d_{s}</math>. Finally, the sets J and N result in <math>d_{o}</math> becoming an accumulation or integration of <math>d_{s}</math>.<br />
<br />
== Experimental Result ==<br />
<br />
* For the English Wikipedia data, Figure 3(a) shows the plots of <math>n_{o}\left ( k \right )</math> and <math>d_{o}\left ( k \right )</math> on a log-log scale, along with the best linear fit. A linear model accounts relatively well for the data over a large range, suggesting a power law is a reasonable approximation for each of these quantities. This is not surprising, given that <math>d_{o}\left ( k \right )</math> can be viewed as a variation on the standard degree distribution: it measures the distribution of the “degree” (number of edges) of each node into each community. <br />
<br />
[[File:figure3.jpg]]<br />
<br />
* Using the Wikipedia ordinal time data the authors generated two snapshots, choosing November 1, 2005 and November 6, 2006 as two relatively arbitrary moments at which which they measure the full set of community memberships. It then computed p (k) using the snapshot method, shown in Figure 3(b).<br />
<br />
[[File:figure4.jpg]]<br />
<br />
* The approximation of ordinal time data from snapshot data depends on two factors: the number of snapshots used, and the amount of time between the snapshots. The authors begin by considering the effect of the number of snapshots. They show how the simulation of ordinal-time depends on the number of snapshots taken for English Wikipedia in figure below.<br />
<br />
[[File:figure5.jpg]]<br />
<br />
As one would expect, the approximation is becoming increasingly accurate with more snapshots. This is because as the number of snapshots increases the time between them goes to 0. Thus, in the limit, snapshot measurements converge to the ordinal-time measurements. This figure shows that empirically just a few snapshots produce good results for these datasets which means the convergence occurs fairly rapidly as the number of snapshots increases.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Simulating_Ordinal_Time_from_Snapshots&diff=5519Simulating Ordinal Time from Snapshots2011-04-04T22:11:01Z<p>Nqi: Created page with 'Roughly, the simulation works by hypothesizing the number of neighbors each node had at the moment it joined a community; choosing this number from among the possible values cons…'</p>
<hr />
<div>Roughly, the simulation works by hypothesizing the number of neighbors each node had at the moment it joined a community; choosing this number from among the possible values consistent with the snapshot observations. The authors exploit both the B(t1) and the J(t1,t2) sets. Recall the set B(t1) consists of triples (u,C,k1), where u joined C before t1, and u had k1 neighbors in C at t1. We choose an integer j uniformly at random in [0,k1] and assume that u had j neighbors in C at the time it joined C. Similarly, the set J(t1,t2) consists of tuples (u,C,k1,k2) where u joined C between t1 and t2, u had k1 neighbors in C at t1, and u had k2 neighbors in C at t2. Here we construct the approximation to ordinal-time by choosing an integer j uniformly at random from [k1,k2] and again assuming that that u had j neighbors in C at the time it joined C. Finally, we do not assume that u joins C for any tuple <math>\left ( u,C,k \right )\in N\left ( t_{2} \right )</math>.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5518Dan Cosley, AAAI, 20102011-04-04T22:06:38Z<p>Nqi: /* Brief Description Of The Method */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]], then it defines some notation for various sets of tuples. At last, it gives a method about how to [[UsesMethod::Simulating Ordinal Time from Snapshots]]. <br />
<br />
* <math>B\left ( t_{1} \right )=\left \{ \left ( u,C,k_{1} \right )\right \}</math>: u joined C before t1 and u had k1 neighbors in C at t1.<br />
<br />
* <math>J\left ( t_{1},t_{2} \right )=\left \{ \left ( u,C,k_{1},k_{k2} \right )\right \}</math>: u had k1 neighbors in C at t1, u joined C between t1 and t2, u had k2 neighbors in C at t2.<br />
<br />
* <math>N\left (t_{2} \right )=\left \{ \left ( u,C,k_{k2} \right )\right \}</math>: u did not join C before t2 and u had k2 neighbors in C at t2.<br />
<br />
The sets B and N result in shifting no and do upwards with respect to <math>n_{s}</math> and <math>d_{s}</math>. Also, the set J results in stretching <math>n_{o}</math> and <math>d_{o}</math> when compared <math>n_{s}</math> and <math>d_{s}</math>. Finally, the sets J and N result in <math>d_{o}</math> becoming an accumulation or integration of <math>d_{s}</math>.<br />
<br />
== Experimental Result ==<br />
<br />
* For the English Wikipedia data, Figure 3(a) shows the plots of <math>n_{o}\left ( k \right )</math> and <math>d_{o}\left ( k \right )</math> on a log-log scale, along with the best linear fit. A linear model accounts relatively well for the data over a large range, suggesting a power law is a reasonable approximation for each of these quantities. This is not surprising, given that <math>d_{o}\left ( k \right )</math> can be viewed as a variation on the standard degree distribution: it measures the distribution of the “degree” (number of edges) of each node into each community. <br />
<br />
[[File:figure3.jpg]]<br />
<br />
* Using the Wikipedia ordinal time data the authors generated two snapshots, choosing November 1, 2005 and November 6, 2006 as two relatively arbitrary moments at which which they measure the full set of community memberships. It then computed p (k) using the snapshot method, shown in Figure 3(b).<br />
<br />
[[File:figure4.jpg]]<br />
<br />
* The approximation of ordinal time data from snapshot data depends on two factors: the number of snapshots used, and the amount of time between the snapshots. The authors begin by considering the effect of the number of snapshots. They show how the simulation of ordinal-time depends on the number of snapshots taken for English Wikipedia in figure below.<br />
<br />
[[File:figure5.jpg]]<br />
<br />
As one would expect, the approximation is becoming increasingly accurate with more snapshots. This is because as the number of snapshots increases the time between them goes to 0. Thus, in the limit, snapshot measurements converge to the ordinal-time measurements. This figure shows that empirically just a few snapshots produce good results for these datasets which means the convergence occurs fairly rapidly as the number of snapshots increases.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=File:Figure5.jpg&diff=5517File:Figure5.jpg2011-04-04T22:01:06Z<p>Nqi: </p>
<hr />
<div></div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5516Dan Cosley, AAAI, 20102011-04-04T22:00:02Z<p>Nqi: /* Experimental Result */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]], then it defines some notation for various sets of tuples. <br />
<br />
* <math>B\left ( t_{1} \right )=\left \{ \left ( u,C,k_{1} \right )\right \}</math>: u joined C before t1 and u had k1 neighbors in C at t1.<br />
<br />
* <math>J\left ( t_{1},t_{2} \right )=\left \{ \left ( u,C,k_{1},k_{k2} \right )\right \}</math>: u had k1 neighbors in C at t1, u joined C between t1 and t2, u had k2 neighbors in C at t2.<br />
<br />
* <math>N\left (t_{2} \right )=\left \{ \left ( u,C,k_{k2} \right )\right \}</math>: u did not join C before t2 and u had k2 neighbors in C at t2.<br />
<br />
The sets B and N result in shifting no and do upwards with respect to <math>n_{s}</math> and <math>d_{s}</math>. Also, the set J results in stretching <math>n_{o}</math> and <math>d_{o}</math> when compared <math>n_{s}</math> and <math>d_{s}</math>. Finally, the sets J and N result in <math>d_{o}</math> becoming an accumulation or integration of <math>d_{s}</math>.<br />
<br />
== Experimental Result ==<br />
<br />
* For the English Wikipedia data, Figure 3(a) shows the plots of <math>n_{o}\left ( k \right )</math> and <math>d_{o}\left ( k \right )</math> on a log-log scale, along with the best linear fit. A linear model accounts relatively well for the data over a large range, suggesting a power law is a reasonable approximation for each of these quantities. This is not surprising, given that <math>d_{o}\left ( k \right )</math> can be viewed as a variation on the standard degree distribution: it measures the distribution of the “degree” (number of edges) of each node into each community. <br />
<br />
[[File:figure3.jpg]]<br />
<br />
* Using the Wikipedia ordinal time data the authors generated two snapshots, choosing November 1, 2005 and November 6, 2006 as two relatively arbitrary moments at which which they measure the full set of community memberships. It then computed p (k) using the snapshot method, shown in Figure 3(b).<br />
<br />
[[File:figure4.jpg]]<br />
<br />
* The approximation of ordinal time data from snapshot data depends on two factors: the number of snapshots used, and the amount of time between the snapshots. The authors begin by considering the effect of the number of snapshots. They show how the simulation of ordinal-time depends on the number of snapshots taken for English Wikipedia in figure below.<br />
<br />
[[File:figure5.jpg]]<br />
<br />
As one would expect, the approximation is becoming increasingly accurate with more snapshots. This is because as the number of snapshots increases the time between them goes to 0. Thus, in the limit, snapshot measurements converge to the ordinal-time measurements. This figure shows that empirically just a few snapshots produce good results for these datasets which means the convergence occurs fairly rapidly as the number of snapshots increases.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5515Dan Cosley, AAAI, 20102011-04-04T21:02:52Z<p>Nqi: /* Experimental Result */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]], then it defines some notation for various sets of tuples. <br />
<br />
* <math>B\left ( t_{1} \right )=\left \{ \left ( u,C,k_{1} \right )\right \}</math>: u joined C before t1 and u had k1 neighbors in C at t1.<br />
<br />
* <math>J\left ( t_{1},t_{2} \right )=\left \{ \left ( u,C,k_{1},k_{k2} \right )\right \}</math>: u had k1 neighbors in C at t1, u joined C between t1 and t2, u had k2 neighbors in C at t2.<br />
<br />
* <math>N\left (t_{2} \right )=\left \{ \left ( u,C,k_{k2} \right )\right \}</math>: u did not join C before t2 and u had k2 neighbors in C at t2.<br />
<br />
The sets B and N result in shifting no and do upwards with respect to <math>n_{s}</math> and <math>d_{s}</math>. Also, the set J results in stretching <math>n_{o}</math> and <math>d_{o}</math> when compared <math>n_{s}</math> and <math>d_{s}</math>. Finally, the sets J and N result in <math>d_{o}</math> becoming an accumulation or integration of <math>d_{s}</math>.<br />
<br />
== Experimental Result ==<br />
<br />
* For the English Wikipedia data, Figure 3(a) shows the plots of <math>n_{o}\left ( k \right )</math> and <math>d_{o}\left ( k \right )</math> on a log-log scale, along with the best linear fit. A linear model accounts relatively well for the data over a large range, suggesting a power law is a reasonable approximation for each of these quantities. This is not surprising, given that <math>d_{o}\left ( k \right )</math> can be viewed as a variation on the standard degree distribution: it measures the distribution of the “degree” (number of edges) of each node into each community. <br />
<br />
[[File:figure3.jpg]]<br />
<br />
* Using the Wikipedia ordinal time data the authors generated two snapshots, choosing November 1, 2005 and November 6, 2006 as two relatively arbitrary moments at which which they measure the full set of community memberships. It then computed p (k) using the snapshot method, shown in Figure 3(b).<br />
<br />
[[File:figure4.jpg]]<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=File:Figure4.jpg&diff=5514File:Figure4.jpg2011-04-04T21:02:08Z<p>Nqi: </p>
<hr />
<div></div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5513Dan Cosley, AAAI, 20102011-04-04T21:01:57Z<p>Nqi: /* Experimental Result */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]], then it defines some notation for various sets of tuples. <br />
<br />
* <math>B\left ( t_{1} \right )=\left \{ \left ( u,C,k_{1} \right )\right \}</math>: u joined C before t1 and u had k1 neighbors in C at t1.<br />
<br />
* <math>J\left ( t_{1},t_{2} \right )=\left \{ \left ( u,C,k_{1},k_{k2} \right )\right \}</math>: u had k1 neighbors in C at t1, u joined C between t1 and t2, u had k2 neighbors in C at t2.<br />
<br />
* <math>N\left (t_{2} \right )=\left \{ \left ( u,C,k_{k2} \right )\right \}</math>: u did not join C before t2 and u had k2 neighbors in C at t2.<br />
<br />
The sets B and N result in shifting no and do upwards with respect to <math>n_{s}</math> and <math>d_{s}</math>. Also, the set J results in stretching <math>n_{o}</math> and <math>d_{o}</math> when compared <math>n_{s}</math> and <math>d_{s}</math>. Finally, the sets J and N result in <math>d_{o}</math> becoming an accumulation or integration of <math>d_{s}</math>.<br />
<br />
== Experimental Result ==<br />
<br />
* For the English Wikipedia data, Figure 3(a) shows the plots of <math>n_{o}\left ( k \right )</math> and <math>d_{o}\left ( k \right )</math> on a log-log scale, along with the best linear fit. A linear model accounts relatively well for the data over a large range, suggesting a power law is a reasonable approximation for each of these quantities. This is not surprising, given that <math>d_{o}\left ( k \right )</math> can be viewed as a variation on the standard degree distribution: it measures the distribution of the “degree” (number of edges) of each node into each community. <br />
<br />
[[File:figure3.jpg]]<br />
<br />
* Using the Wikipedia ordinal time data the authors generated two snapshots, choosing November 1, 2005 and November 6, 2006 as two relatively arbitrary moments at which which they measure the full set of community memberships. It then computed p (k) using the snapshot method, shown in Figure 3(b)<br />
<br />
[[File:figure4.jpg]]<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=File:Figure3.jpg&diff=5512File:Figure3.jpg2011-04-04T20:59:45Z<p>Nqi: </p>
<hr />
<div></div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5511Dan Cosley, AAAI, 20102011-04-04T20:58:01Z<p>Nqi: /* Experimental Result */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]], then it defines some notation for various sets of tuples. <br />
<br />
* <math>B\left ( t_{1} \right )=\left \{ \left ( u,C,k_{1} \right )\right \}</math>: u joined C before t1 and u had k1 neighbors in C at t1.<br />
<br />
* <math>J\left ( t_{1},t_{2} \right )=\left \{ \left ( u,C,k_{1},k_{k2} \right )\right \}</math>: u had k1 neighbors in C at t1, u joined C between t1 and t2, u had k2 neighbors in C at t2.<br />
<br />
* <math>N\left (t_{2} \right )=\left \{ \left ( u,C,k_{k2} \right )\right \}</math>: u did not join C before t2 and u had k2 neighbors in C at t2.<br />
<br />
The sets B and N result in shifting no and do upwards with respect to <math>n_{s}</math> and <math>d_{s}</math>. Also, the set J results in stretching <math>n_{o}</math> and <math>d_{o}</math> when compared <math>n_{s}</math> and <math>d_{s}</math>. Finally, the sets J and N result in <math>d_{o}</math> becoming an accumulation or integration of <math>d_{s}</math>.<br />
<br />
== Experimental Result ==<br />
<br />
For the English Wikipedia data, Figure 3(a) shows the plots of <math>n_{o}\left ( k \right )</math> and <math>d_{o}\left ( k \right )</math> on a log-log scale, along with the best linear fit. A linear model accounts relatively well for the data over a large range, suggesting a power law is a reasonable approximation for each of these quantities. This is not surprising, given that <math>d_{o}\left ( k \right )</math> can be viewed as a variation on the standard degree distribution: it measures the distribution of the “degree” (number of edges) of each node into each community. <br />
<br />
[[File:figure3.jpg]]<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Snapshot&diff=5510Snapshot2011-04-04T20:53:33Z<p>Nqi: </p>
<hr />
<div>Snapshot: Consider two snapshots of the network at different points in time. For each k, consider the set of all individuals who are [[k-exposed]] in the first snapshot. Let ps(k) be the fraction of individuals in this set who have become adopters by the time of the second snapshot.<br />
<br />
Figure 2 shows the shape of influence curves for the snapshot definitions using the Wikipedia data: <br />
<br />
[[File:snapshot.jpg]]<br />
<br />
As with ordinal time, it is now useful to define <math>p_{s}\left ( k \right )</math> as the ratio of two quantities, <math>p_{s}\left ( k \right )=n_{s}\left ( k \right )/d_{s}\left ( k \right )</math>, where <math>d_{s}\left ( k \right )</math> is the number of triples (u;C; k) for which u was k-exposed to C at time t1, and <math>n_{s}\left ( k \right )</math> is the number of triples (u,C,k) for which u was k-exposed to C at time t1, and then joined C between t1 and t2.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Ordinal-time&diff=5509Ordinal-time2011-04-04T20:49:25Z<p>Nqi: </p>
<hr />
<div>Ordinal-time: Consider a complete time sequence of an evolving social network that includes each time a new network link is formed and each time an individual adopts a new behavior. For each k, consider the set of all individuals who were ever [[k-exposed]] at any time, and define po(k) to be the fraction of this set that became adopters before acquiring a (k + 1)st neighbor who is an adopter.<br />
<br />
Figure 1 shows the shape of influence curves for the ordinal-time definitions using the Wikipedia data:<br />
<br />
[[File:ordinal-time.jpg]]<br />
<br />
<math>p_{o}\left ( k \right )</math> is the fraction of cases in which a node that is k-exposed to a community C proceeds to join C before acquiring a (k + 1)st neighbor in C. That is, we define <math>p_{o}\left ( k \right )</math> as the ratio of two quantities, <math>p_{o}\left ( k \right )=n_{o}\left ( k \right )/d_{o}\left ( k \right )</math>, where <math>d_{o}\left ( k \right )</math> is the number of triples (u;C; k) for which u was ever k-exposed to C, and <math>n_{o}\left ( k \right )</math> is the number of triples (u,C,k) for which u was k-exposed to C, and then joined C before acquiring a (k + 1)st neighbor in C.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5508Dan Cosley, AAAI, 20102011-04-04T20:44:32Z<p>Nqi: /* Brief Description Of The Method */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]], then it defines some notation for various sets of tuples. <br />
<br />
* <math>B\left ( t_{1} \right )=\left \{ \left ( u,C,k_{1} \right )\right \}</math>: u joined C before t1 and u had k1 neighbors in C at t1.<br />
<br />
* <math>J\left ( t_{1},t_{2} \right )=\left \{ \left ( u,C,k_{1},k_{k2} \right )\right \}</math>: u had k1 neighbors in C at t1, u joined C between t1 and t2, u had k2 neighbors in C at t2.<br />
<br />
* <math>N\left (t_{2} \right )=\left \{ \left ( u,C,k_{k2} \right )\right \}</math>: u did not join C before t2 and u had k2 neighbors in C at t2.<br />
<br />
The sets B and N result in shifting no and do upwards with respect to <math>n_{s}</math> and <math>d_{s}</math>. Also, the set J results in stretching <math>n_{o}</math> and <math>d_{o}</math> when compared <math>n_{s}</math> and <math>d_{s}</math>. Finally, the sets J and N result in <math>d_{o}</math> becoming an accumulation or integration of <math>d_{s}</math>.<br />
<br />
== Experimental Result ==<br />
<br />
This paper complements and extends the existing literature around influence in online communities.<br />
<br />
* Prior work has shown that how one’s friends influence the groups one joins online is quite similar across a variety of domains, content types, community goals, and ways of inferring ties. This paper shows that this type of social influence occurs in Wikipedia as well. <br />
* The demonstration of the relationship between snapshot and ordinal-time measurements may help researchers better understand social influence by allowing them to more easily compare data gathered with different sampling procedures.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5507Dan Cosley, AAAI, 20102011-04-04T20:40:50Z<p>Nqi: /* Brief Description Of The Method */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]], then it defines some notation for various sets of tuples. <br />
<br />
* <math>B\left ( t_{1} \right )=\left \{ \left ( u,C,k_{1} \right )\right \}</math>: u joined C before t1 and u had k1 neighbors in C at t1.<br />
<br />
* <math>J\left ( t_{1},t_{2} \right )=\left \{ \left ( u,C,k_{1},k_{k2} \right )\right \}</math>: u had k1 neighbors in C at t1, u joined C between t1 and t2, u had k2 neighbors in C at t2.<br />
<br />
* <math>N\left (t_{2} \right )=\left \{ \left ( u,C,k_{k2} \right )\right \}</math>: u did not join C before t2 and u had k2 neighbors in C at t2.<br />
<br />
== Experimental Result ==<br />
<br />
This paper complements and extends the existing literature around influence in online communities.<br />
<br />
* Prior work has shown that how one’s friends influence the groups one joins online is quite similar across a variety of domains, content types, community goals, and ways of inferring ties. This paper shows that this type of social influence occurs in Wikipedia as well. <br />
* The demonstration of the relationship between snapshot and ordinal-time measurements may help researchers better understand social influence by allowing them to more easily compare data gathered with different sampling procedures.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5506Dan Cosley, AAAI, 20102011-04-04T17:01:33Z<p>Nqi: /* Brief Description Of The Method */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence: [[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]]<br />
<br />
== Experimental Result ==<br />
<br />
This paper complements and extends the existing literature around influence in online communities.<br />
<br />
* Prior work has shown that how one’s friends influence the groups one joins online is quite similar across a variety of domains, content types, community goals, and ways of inferring ties. This paper shows that this type of social influence occurs in Wikipedia as well. <br />
* The demonstration of the relationship between snapshot and ordinal-time measurements may help researchers better understand social influence by allowing them to more easily compare data gathered with different sampling procedures.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=File:Snapshot.jpg&diff=5502File:Snapshot.jpg2011-04-04T16:14:12Z<p>Nqi: </p>
<hr />
<div></div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Snapshot&diff=5501Snapshot2011-04-04T16:13:36Z<p>Nqi: </p>
<hr />
<div>Snapshot: Consider two snapshots of the network at different points in time. For each k, consider the set of all individuals who are [[k-exposed]] in the first snapshot. Let ps(k) be the fraction of individuals in this set who have become adopters by the time of the second snapshot.<br />
<br />
Figure 2 shows the shape of influence curves for the snapshot definitions using the Wikipedia data: <br />
<br />
[[File:snapshot.jpg]]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=File:Ordinal-time.jpg&diff=5500File:Ordinal-time.jpg2011-04-04T16:12:18Z<p>Nqi: </p>
<hr />
<div></div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Ordinal-time&diff=5499Ordinal-time2011-04-04T16:10:44Z<p>Nqi: </p>
<hr />
<div>Ordinal-time: Consider a complete time sequence of an evolving social network that includes each time a new network link is formed and each time an individual adopts a new behavior. For each k, consider the set of all individuals who were ever [[k-exposed]] at any time, and define po(k) to be the fraction of this set that became adopters before acquiring a (k + 1)st neighbor who is an adopter.<br />
<br />
Figure 1 shows the shape of influence curves for the ordinal-time definitions using the Wikipedia data:<br />
<br />
[[File:ordinal-time.jpg]]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Ordinal-time&diff=5498Ordinal-time2011-04-04T16:10:03Z<p>Nqi: </p>
<hr />
<div>Ordinal-time: Consider a complete time sequence of an evolving social network that includes each time a new network link is formed and each time an individual adopts a new behavior. For each k, consider the set of all individuals who were ever [[k-exposed]] at any time, and define po(k) to be the fraction of this set that became adopters before acquiring a (k + 1)st neighbor who is an adopter.<br />
<br />
Figure 1 shows the shape of influence curves for the ordinal-time definitions:<br />
<br />
[[File:ordinal-time.jpg]]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Ordinal-time&diff=5497Ordinal-time2011-04-04T16:09:46Z<p>Nqi: </p>
<hr />
<div>Ordinal-time: Consider a complete time sequence of an evolving social network that includes each time a new network link is formed and each time an individual adopts a new behavior. For each k, consider the set of all individuals who were ever [[k-exposed]] at any time, and define po(k) to be the fraction of this set that became adopters before acquiring a (k + 1)st neighbor who is an adopter.<br />
Figure 1 shows the shape of influence curves for the ordinal-time definitions:<br />
[[File:ordinal-time.jpg]]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Snapshot&diff=5496Snapshot2011-04-04T05:24:55Z<p>Nqi: Created page with 'Snapshot: Consider two snapshots of the network at different points in time. For each k, consider the set of all individuals who are k-exposed in the first snapshot. Let ps(…'</p>
<hr />
<div>Snapshot: Consider two snapshots of the network at different points in time. For each k, consider the set of all individuals who are [[k-exposed]] in the first snapshot. Let ps(k) be the fraction of individuals in this set who have become adopters by the time of the second snapshot.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=K-exposed&diff=5495K-exposed2011-04-04T05:23:39Z<p>Nqi: Created page with 'An individual is k-exposed to the behavior at a particular point in time t if they are a non-adopter at time t, but they have exactly k neighbors in the network who are adopters …'</p>
<hr />
<div>An individual is k-exposed to the behavior at a particular point in time t if they are a non-adopter at time t, but they have exactly k neighbors in the network who are adopters at time t.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Ordinal-time&diff=5494Ordinal-time2011-04-04T05:22:46Z<p>Nqi: Created page with 'Ordinal-time: Consider a complete time sequence of an evolving social network that includes each time a new network link is formed and each time an individual adopts a new behavi…'</p>
<hr />
<div>Ordinal-time: Consider a complete time sequence of an evolving social network that includes each time a new network link is formed and each time an individual adopts a new behavior. For each k, consider the set of all individuals who were ever [[k-exposed]] at any time, and define po(k) to be the fraction of this set that became adopters before acquiring a (k + 1)st neighbor who is an adopter.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5493Dan Cosley, AAAI, 20102011-04-04T05:21:09Z<p>Nqi: /* Brief Description Of The Method */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
In order to study a method to measure influence, this paper gives two definitions of social influence:[[UsesMethod::Ordinal-time]] and [[UsesMethod::Snapshot]]<br />
<br />
== Experimental Result ==<br />
<br />
This paper complements and extends the existing literature around influence in online communities.<br />
<br />
* Prior work has shown that how one’s friends influence the groups one joins online is quite similar across a variety of domains, content types, community goals, and ways of inferring ties. This paper shows that this type of social influence occurs in Wikipedia as well. <br />
* The demonstration of the relationship between snapshot and ordinal-time measurements may help researchers better understand social influence by allowing them to more easily compare data gathered with different sampling procedures.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5492Dan Cosley, AAAI, 20102011-04-04T05:15:21Z<p>Nqi: /* Summary */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Brief Description Of The Method ==<br />
<br />
== Experimental Result ==<br />
<br />
This paper complements and extends the existing literature around influence in online communities.<br />
<br />
* Prior work has shown that how one’s friends influence the groups one joins online is quite similar across a variety of domains, content types, community goals, and ways of inferring ties. This paper shows that this type of social influence occurs in Wikipedia as well. <br />
* The demonstration of the relationship between snapshot and ordinal-time measurements may help researchers better understand social influence by allowing them to more easily compare data gathered with different sampling procedures.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5491Dan Cosley, AAAI, 20102011-04-04T05:15:10Z<p>Nqi: /* Summary */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
<br />
== Brief Description Of The Method ==<br />
<br />
== Experimental Result ==<br />
<br />
This paper complements and extends the existing literature around influence in online communities.<br />
<br />
* Prior work has shown that how one’s friends influence the groups one joins online is quite similar across a variety of domains, content types, community goals, and ways of inferring ties. This paper shows that this type of social influence occurs in Wikipedia as well. <br />
* The demonstration of the relationship between snapshot and ordinal-time measurements may help researchers better understand social influence by allowing them to more easily compare data gathered with different sampling procedures.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=User_talk:Nqi&diff=5490User talk:Nqi2011-04-04T05:14:18Z<p>Nqi: /* Paper */</p>
<hr />
<div>This is one of the [[Category::Presentation|presentations]] given in the course [[Social Media Analysis 10-802 in Spring 2011]].<br />
<br />
* Presented by: [[Speaker::User:Nqi]]<br />
* Slides : [[NqiPresentation.pptx]]<br />
<br />
== Paper ==<br />
<br />
Sequential Influence Models in Social Network:[http://malt.ml.cmu.edu/mw/index.php/Dan_Cosley,_AAAI,_2010]<br />
<br />
== Online Version ==<br />
<br />
Sequential Influence Models in Social Network [http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=User_talk:Nqi&diff=5489User talk:Nqi2011-04-04T05:14:04Z<p>Nqi: /* Online Version */</p>
<hr />
<div>This is one of the [[Category::Presentation|presentations]] given in the course [[Social Media Analysis 10-802 in Spring 2011]].<br />
<br />
* Presented by: [[Speaker::User:Nqi]]<br />
* Slides : [[NqiPresentation.pptx]]<br />
<br />
== Paper ==<br />
<br />
Sequential Influence Models in Social Network:[[http://malt.ml.cmu.edu/mw/index.php/Dan_Cosley,_AAAI,_2010]]<br />
<br />
== Online Version ==<br />
<br />
Sequential Influence Models in Social Network [http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=User_talk:Nqi&diff=5488User talk:Nqi2011-04-04T05:13:22Z<p>Nqi: /* Paper */</p>
<hr />
<div>This is one of the [[Category::Presentation|presentations]] given in the course [[Social Media Analysis 10-802 in Spring 2011]].<br />
<br />
* Presented by: [[Speaker::User:Nqi]]<br />
* Slides : [[NqiPresentation.pptx]]<br />
<br />
== Paper ==<br />
<br />
Sequential Influence Models in Social Network:[[http://malt.ml.cmu.edu/mw/index.php/Dan_Cosley,_AAAI,_2010]]<br />
<br />
== Online Version ==<br />
<br />
Discovering Overlapping Groups in Social Media [http://dmml.asu.edu/users/xufei/Papers/ICDM2010.pdf]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5487Dan Cosley, AAAI, 20102011-04-04T05:02:11Z<p>Nqi: /* Experimental Result */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Experimental Result ==<br />
<br />
This paper complements and extends the existing literature around influence in online communities.<br />
<br />
* Prior work has shown that how one’s friends influence the groups one joins online is quite similar across a variety of domains, content types, community goals, and ways of inferring ties. This paper shows that this type of social influence occurs in Wikipedia as well. <br />
* The demonstration of the relationship between snapshot and ordinal-time measurements may help researchers better understand social influence by allowing them to more easily compare data gathered with different sampling procedures.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5486Dan Cosley, AAAI, 20102011-04-04T05:01:58Z<p>Nqi: /* Experimental Result */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Experimental Result ==<br />
<br />
This paper complements and extends the existing literature around influence in online communities.<br />
<br />
* Prior work has shown that how one’s friends influence the groups one joins online is quite similar across a variety of domains, content types, community goals, and ways of inferring ties. This paper shows that this type of social influence<br />
occurs in Wikipedia as well. <br />
* The demonstration of the relationship between snapshot and ordinal-time measurements may help researchers better understand social influence by allowing them to more easily compare data gathered with different sampling procedures.<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5485Dan Cosley, AAAI, 20102011-04-04T04:58:49Z<p>Nqi: /* Summary */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Experimental Result ==<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Xufei_Wang,_ICDM,_2010&diff=5484Xufei Wang, ICDM, 20102011-04-04T04:28:57Z<p>Nqi: /* Brief Description Of The Method */</p>
<hr />
<div>== Citation ==<br />
Xufei Wang. 2010. Discovering Overlapping Groups in Social Media, the 10th IEEE International Conference on Data Mining (ICDM 2010).<br />
<br />
== Online Version ==<br />
http://dmml.asu.edu/users/xufei/Papers/ICDM2010.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|BlogCatalog]] [http://www.blogcatalog.com/]<br />
<br />
[[Property:UsesDataset|Delicious]] [http://www.delicious.com/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors propose a novel co-clustering framework, which takes advantage of networking information between users and tags in social media, to discover these [[AddressesProblem::overlapping communities]]. The basic ideas are:<br />
<br />
* To discover overlapping communities in social media. Diverse interests and interactions that human beings can have in online social life suggest that one person often belongs more than one community.<br />
<br />
* To use user-tag subscription information instead of user-user links. Metadata such as tags become an important source in measuring the user-user similarity. The paper shows that more accurate community structures can be obtained by scrutinizing tag information.<br />
<br />
* To obtain clusters containing users and tags simultaneously. Existing co-clustering methods cluster users/tags separately. Thus, it is not clear which user cluster corresponds to which tag cluster. But the proposed method is able to find out user/tag group structure and their correspondence<br />
<br />
== Problem Statement ==<br />
<br />
In this paper, the concept of community is generalized to include both users and tags. Tags of a community imply the major concern of people within it.<br />
<br />
Let <math>\mu = \left ( \mu _{1},\mu _{2},...,\mu _{m} \right )</math> denote the user set, <math>\tau = \left ( \tau _{1},\tau _{2},...,\tau _{n} \right )</math> the tay set. A community <math>C_{i}\left ( 1\leq i\leq k \right )</math> is a subset of user and tags, where k is the number of communities. As mentioned above, communities usually overlap, i.e., <math>C_{i}\bigcap C_{j}\neq \O \left ( 1\leq i,j\leq k \right )</math>.On the other hand, users and their subscribed tags form a user-tag matrix M, in which each entry <math>M_{ij}\in \left \{ 0,1 \right \}</math> indicates whether user <math>u_{i}</math> subscribes to tag <math>t_{j}</math>. So it is reasonable to view a user as a sparse vector of tags, and each tag as a sparse vector of users.<br />
<br />
Given notations above, the overlapping co-clustering problem can be stated formally as follows:<br />
<br />
'''Input:'''<br />
* A user-tag subscription matrix <math>M_{N_{\mu }\times N_{t}}</math>, where <math>N_{\mu }</math> and <math>N_{t}</math> are the numbers of users and tags.<br />
<br />
* The number of communities k.<br />
<br />
'''Output:'''<br />
* k overlapping communities which consist of both users and tags.<br />
<br />
== Brief Description Of The Method ==<br />
<br />
Communities that aggregate similar users and tags together can be detected by maximizing intra-cluster similarity, which is shown below:<br />
<math>arg max\frac{1}{k}\sum_{i=1}^{k}\sum_{x_{j}\in C_{i}}^{}S_{C}\left ( x_{j},c_{i} \right )</math><br />
where k is the number of communities, x is the edges and c is the centroid of community. This formulation can be solved by a [[UsesMethod::k-means]] variant.<br />
<br />
This paper uses different methods to solve the problem of overlapping communities:<br />
<br />
'''A. [[UsesMethod::Independent Learning]]'''<br />
<br />
'''B. [[UsesMethod::Normalized Learning]]'''<br />
<br />
'''C. [[UsesMethod::Correlational Learning]]'''<br />
<br />
== Experimental Result ==<br />
<br />
The authors use two kinds of datasets: one is a synthetic data and the other kind is real data from [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]]<br />
<br />
'''A. Synthetic Data'''<br />
<br />
Synthetic data, which is controlled by various parameters, facilitates a comparative study between the uncovered and actual clusters. It has 1,000 users and 1,000 tags and with different number of clusters which range from 5 to 50.<br />
<br />
[[File:figure1.jpg]]<br />
<br />
From the experiment result, we see that correlational Learning is more effective than the other two methods in recovering overlapping clusters. It works well even when the intra-cluster link density is low. Co-clustering performs poorly because it only finds non-overlapping clusters.<br />
<br />
'''B. Social Media Data'''<br />
<br />
[[File:Social Media experiment.jpg]]<br />
<br />
From the experiment with [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]], the paper show us that:<br />
<br />
* The probability of a link between two users increases with respect to the number of tags they share.<br />
<br />
* Correlational Learning consistently performs better, especially when the training set is small.<br />
<br />
* Higher co-occurrence frequency suggests that two users are more similar. Similar patterns are observed in the three methods.<br />
<br />
== Related papers ==<br />
<br />
The author uses Co-Clustering method in [[RelatedPaper::Co-clustering documents and words using bipartite spectral graph partitioning]] as a comparison to above methods.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Feedback_effects_between_similarity_and_social_influence_in_online_communities&diff=5483Feedback effects between similarity and social influence in online communities2011-04-04T04:27:09Z<p>Nqi: Created page with '== Citation == Crandall, D.; Cosley, D.; Huttenlocher, D.; Kleinberg, J.; and Suri, S. 2008. Feedback effects between similarity and social influence in online communities. Proc…'</p>
<hr />
<div>== Citation ==<br />
Crandall, D.; Cosley, D.; Huttenlocher, D.; Kleinberg, J.; and Suri, S. 2008. Feedback effects between similarity and social influence in online communities. Proc. 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/kdd08-sim.pdf<br />
<br />
== Summary ==<br />
This [[Category::paper]] develop techniques for identifying and modeling the [[AddressesProblem::interactions between social influence]] and selection, using data from online communities where both social interaction and changes in behavior over time can be measured. It finds clear feedback effects between the two factors, with rising similarity between two individuals serving, in aggregate, as an indicator of future interaction — but with similarity then continuing to increase steadily, although at a slower rate, for long periods after initial interactions. The authors also consider the relative value of similarity and social influence in modeling future behavior. For instance, to predict the activities that an individual is likely to do next, is it more useful to know the current activities of their friends, or of the people most similar to them?</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Scale-free_networks&diff=5482Scale-free networks2011-04-04T04:20:42Z<p>Nqi: </p>
<hr />
<div>A '''scale-free network''' is a network whose degree distribution follows a power law, at least asymptotically. That is, the fraction ''P''(''k'') of nodes in the network having ''k'' connections to other nodes goes for large values of ''k'' as<br />
<br />
<math><br />
P(k) \ \sim \ ck^\boldsymbol{-\gamma}<br />
</math><br />
<br />
where <math> c </math> is a normalization constant and <math>\gamma</math> is a parameter whose value is typically in the range 2 < <math>\gamma</math> < 3, although occasionally it may lie outside these bounds.<br />
<br />
Scale-free networks are noteworthy because many empirically observed networks appear to be scale-free, including the World Wide Web, citation networks, biological networks, airline networks and some social networks.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Scale-free_networks&diff=5481Scale-free networks2011-04-04T04:20:08Z<p>Nqi: Created page with 'A '''scale-free network''' is a network whose degree distribution follows a power law, at least asymptotically. That is, the fraction ''P''(''k'') of…'</p>
<hr />
<div>A '''scale-free network''' is a [[complex network|network]] whose [[degree distribution]] follows a [[power law]], at least asymptotically. That is, the fraction ''P''(''k'') of nodes in the network having ''k'' connections to other nodes goes for large values of ''k'' as<br />
<br />
<math><br />
P(k) \ \sim \ ck^\boldsymbol{-\gamma}<br />
</math><br />
<br />
where <math> c </math> is a normalization constant and <math>\gamma</math> is a parameter whose value is typically in the range 2 < <math>\gamma</math> < 3, although occasionally it may lie outside these bounds.<br />
<br />
Scale-free networks are noteworthy because many empirically observed networks appear to be scale-free, including the World Wide Web, citation networks, biological networks, airline networks and some social networks.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Measuring_wikipedia&diff=5480Measuring wikipedia2011-04-04T04:18:27Z<p>Nqi: Created page with '== Citation == Voss, J. 2005. Measuring wikipedia. In Proc. International Conference of the International Society for Scientometrics and Informetrics. == Online Version == http:…'</p>
<hr />
<div>== Citation ==<br />
Voss, J. 2005. Measuring wikipedia. In Proc. International Conference of the International Society for Scientometrics and Informetrics.<br />
<br />
== Online Version ==<br />
http://eprints.rclis.org/handle/10760/6207<br />
<br />
== Summary ==<br />
Wikipedia, an international project that uses Wiki software to collaboratively create an encyclopaedia, is becoming more and more popular. Everyone can directly edit articles and every edit is recorded. The version history of all articles is freely available and allows a multitude of examinations. This [[Category::paper]] gives an overview on Wikipedia research. Wikipedia’s fundamental components, i.e. articles, authors, edits, and links, as well as content and quality are analysed. Possibilities of research are explored including examples and first results. Several characteristics that are found in Wikipedia, such as [[AddressesProblem::exponential growth]] and [[AddressesProblem::scale-free networks]] are already known in other context. However the Wiki architecture also possesses some intrinsic specialities. General trends are measured that are typical for all Wikipedias but vary between languages in detail.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5479Dan Cosley, AAAI, 20102011-04-04T04:12:59Z<p>Nqi: /* Related papers */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Related papers ==<br />
<br />
This paper uses conclusions in [[RelatedPaper::Measuring wikipedia]] to compare with its experimental result. <br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5478Dan Cosley, AAAI, 20102011-04-04T04:10:28Z<p>Nqi: /* Summary */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].<br />
<br />
== Related papers ==<br />
<br />
[[RelatedPaper::Feedback effects between similarity and social influence in online communities]].<br />
<br />
[[RelatedPaper::Measuring wikipedia]]</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5477Dan Cosley, AAAI, 20102011-04-04T04:04:48Z<p>Nqi: /* Summary */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on [[Property:UsesDataset|Wikipedia]]; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5476Dan Cosley, AAAI, 20102011-04-04T04:03:04Z<p>Nqi: /* Summary */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors consider two of the most fundamental definitions of influence, one based on a small set of “snapshot” observations of a social network and the other based on detailed temporal dynamics. The former is particularly useful because large-scale social network data sets are often available only in snapshots or crawls. The latter however provides a more detailed process model of how influence spreads. The authors studied the relationship between these two ways of measuring influence, in particular establishing how to infer the more detailed temporal measure from the more readily observable snapshot measure. It validates the analysis using the history of social interactions on Wikipedia; the result is the first large-scale study to exhibit a direct relationship between snapshot and temporal models of [[AddressesProblem::social influence]].</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5475Dan Cosley, AAAI, 20102011-04-04T03:53:46Z<p>Nqi: /* Databases */</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors propose a novel co-clustering framework, which takes advantage of networking information between users and tags in social media, to discover these [[AddressesProblem::overlapping communities]]. The basic ideas are:</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Xufei_Wang,_ICDM,_2010&diff=5474Xufei Wang, ICDM, 20102011-04-04T03:51:43Z<p>Nqi: /* Databases */</p>
<hr />
<div>== Citation ==<br />
Xufei Wang. 2010. Discovering Overlapping Groups in Social Media, the 10th IEEE International Conference on Data Mining (ICDM 2010).<br />
<br />
== Online Version ==<br />
http://dmml.asu.edu/users/xufei/Papers/ICDM2010.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|BlogCatalog]] [http://www.blogcatalog.com/]<br />
<br />
[[Property:UsesDataset|Delicious]] [http://www.delicious.com/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors propose a novel co-clustering framework, which takes advantage of networking information between users and tags in social media, to discover these [[AddressesProblem::overlapping communities]]. The basic ideas are:<br />
<br />
* To discover overlapping communities in social media. Diverse interests and interactions that human beings can have in online social life suggest that one person often belongs more than one community.<br />
<br />
* To use user-tag subscription information instead of user-user links. Metadata such as tags become an important source in measuring the user-user similarity. The paper shows that more accurate community structures can be obtained by scrutinizing tag information.<br />
<br />
* To obtain clusters containing users and tags simultaneously. Existing co-clustering methods cluster users/tags separately. Thus, it is not clear which user cluster corresponds to which tag cluster. But the proposed method is able to find out user/tag group structure and their correspondence<br />
<br />
== Problem Statement ==<br />
<br />
In this paper, the concept of community is generalized to include both users and tags. Tags of a community imply the major concern of people within it.<br />
<br />
Let <math>\mu = \left ( \mu _{1},\mu _{2},...,\mu _{m} \right )</math> denote the user set, <math>\tau = \left ( \tau _{1},\tau _{2},...,\tau _{n} \right )</math> the tay set. A community <math>C_{i}\left ( 1\leq i\leq k \right )</math> is a subset of user and tags, where k is the number of communities. As mentioned above, communities usually overlap, i.e., <math>C_{i}\bigcap C_{j}\neq \O \left ( 1\leq i,j\leq k \right )</math>.On the other hand, users and their subscribed tags form a user-tag matrix M, in which each entry <math>M_{ij}\in \left \{ 0,1 \right \}</math> indicates whether user <math>u_{i}</math> subscribes to tag <math>t_{j}</math>. So it is reasonable to view a user as a sparse vector of tags, and each tag as a sparse vector of users.<br />
<br />
Given notations above, the overlapping co-clustering problem can be stated formally as follows:<br />
<br />
'''Input:'''<br />
* A user-tag subscription matrix <math>M_{N_{\mu }\times N_{t}}</math>, where <math>N_{\mu }</math> and <math>N_{t}</math> are the numbers of users and tags.<br />
<br />
* The number of communities k.<br />
<br />
'''Output:'''<br />
* k overlapping communities which consist of both users and tags.<br />
<br />
== Brief Description Of The Method ==<br />
<br />
Communities that aggregate similar users and tags together can be detected by maximizing intra-cluster similarity, which is shown below:<br />
<math>arg max\frac{1}{k}\sum_{i=1}^{k}\sum_{x_{j}\in C_{i}}^{}S_{C}\left ( x_{j},c_{i} \right )</math><br />
where k is the number of communities, x is the edges and c is the centroid of community. This formulation can be solved by a [[UsesMethod::k-means]] variant.<br />
<br />
This paper uses different methods to solve the problem of overlapping communities:<br />
<br />
<br />
'''A. [[UsesMethod::Independent Learning]]'''<br />
<br />
<br />
'''B. [[UsesMethod::Normalized Learning]]'''<br />
<br />
<br />
'''C. [[UsesMethod::Correlational Learning]]'''<br />
<br />
== Experimental Result ==<br />
<br />
The authors use two kinds of datasets: one is a synthetic data and the other kind is real data from [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]]<br />
<br />
'''A. Synthetic Data'''<br />
<br />
Synthetic data, which is controlled by various parameters, facilitates a comparative study between the uncovered and actual clusters. It has 1,000 users and 1,000 tags and with different number of clusters which range from 5 to 50.<br />
<br />
[[File:figure1.jpg]]<br />
<br />
From the experiment result, we see that correlational Learning is more effective than the other two methods in recovering overlapping clusters. It works well even when the intra-cluster link density is low. Co-clustering performs poorly because it only finds non-overlapping clusters.<br />
<br />
'''B. Social Media Data'''<br />
<br />
[[File:Social Media experiment.jpg]]<br />
<br />
From the experiment with [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]], the paper show us that:<br />
<br />
* The probability of a link between two users increases with respect to the number of tags they share.<br />
<br />
* Correlational Learning consistently performs better, especially when the training set is small.<br />
<br />
* Higher co-occurrence frequency suggests that two users are more similar. Similar patterns are observed in the three methods.<br />
<br />
== Related papers ==<br />
<br />
The author uses Co-Clustering method in [[RelatedPaper::Co-clustering documents and words using bipartite spectral graph partitioning]] as a comparison to above methods.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Xufei_Wang,_ICDM,_2010&diff=5473Xufei Wang, ICDM, 20102011-04-04T03:51:19Z<p>Nqi: /* Databases */</p>
<hr />
<div>== Citation ==<br />
Xufei Wang. 2010. Discovering Overlapping Groups in Social Media, the 10th IEEE International Conference on Data Mining (ICDM 2010).<br />
<br />
== Online Version ==<br />
http://dmml.asu.edu/users/xufei/Papers/ICDM2010.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|BlogCatalog]] BlogCatalog [http://www.blogcatalog.com/]<br />
<br />
[[Property:UsesDataset|Delicious]] Delicious [http://www.delicious.com/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors propose a novel co-clustering framework, which takes advantage of networking information between users and tags in social media, to discover these [[AddressesProblem::overlapping communities]]. The basic ideas are:<br />
<br />
* To discover overlapping communities in social media. Diverse interests and interactions that human beings can have in online social life suggest that one person often belongs more than one community.<br />
<br />
* To use user-tag subscription information instead of user-user links. Metadata such as tags become an important source in measuring the user-user similarity. The paper shows that more accurate community structures can be obtained by scrutinizing tag information.<br />
<br />
* To obtain clusters containing users and tags simultaneously. Existing co-clustering methods cluster users/tags separately. Thus, it is not clear which user cluster corresponds to which tag cluster. But the proposed method is able to find out user/tag group structure and their correspondence<br />
<br />
== Problem Statement ==<br />
<br />
In this paper, the concept of community is generalized to include both users and tags. Tags of a community imply the major concern of people within it.<br />
<br />
Let <math>\mu = \left ( \mu _{1},\mu _{2},...,\mu _{m} \right )</math> denote the user set, <math>\tau = \left ( \tau _{1},\tau _{2},...,\tau _{n} \right )</math> the tay set. A community <math>C_{i}\left ( 1\leq i\leq k \right )</math> is a subset of user and tags, where k is the number of communities. As mentioned above, communities usually overlap, i.e., <math>C_{i}\bigcap C_{j}\neq \O \left ( 1\leq i,j\leq k \right )</math>.On the other hand, users and their subscribed tags form a user-tag matrix M, in which each entry <math>M_{ij}\in \left \{ 0,1 \right \}</math> indicates whether user <math>u_{i}</math> subscribes to tag <math>t_{j}</math>. So it is reasonable to view a user as a sparse vector of tags, and each tag as a sparse vector of users.<br />
<br />
Given notations above, the overlapping co-clustering problem can be stated formally as follows:<br />
<br />
'''Input:'''<br />
* A user-tag subscription matrix <math>M_{N_{\mu }\times N_{t}}</math>, where <math>N_{\mu }</math> and <math>N_{t}</math> are the numbers of users and tags.<br />
<br />
* The number of communities k.<br />
<br />
'''Output:'''<br />
* k overlapping communities which consist of both users and tags.<br />
<br />
== Brief Description Of The Method ==<br />
<br />
Communities that aggregate similar users and tags together can be detected by maximizing intra-cluster similarity, which is shown below:<br />
<math>arg max\frac{1}{k}\sum_{i=1}^{k}\sum_{x_{j}\in C_{i}}^{}S_{C}\left ( x_{j},c_{i} \right )</math><br />
where k is the number of communities, x is the edges and c is the centroid of community. This formulation can be solved by a [[UsesMethod::k-means]] variant.<br />
<br />
This paper uses different methods to solve the problem of overlapping communities:<br />
<br />
<br />
'''A. [[UsesMethod::Independent Learning]]'''<br />
<br />
<br />
'''B. [[UsesMethod::Normalized Learning]]'''<br />
<br />
<br />
'''C. [[UsesMethod::Correlational Learning]]'''<br />
<br />
== Experimental Result ==<br />
<br />
The authors use two kinds of datasets: one is a synthetic data and the other kind is real data from [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]]<br />
<br />
'''A. Synthetic Data'''<br />
<br />
Synthetic data, which is controlled by various parameters, facilitates a comparative study between the uncovered and actual clusters. It has 1,000 users and 1,000 tags and with different number of clusters which range from 5 to 50.<br />
<br />
[[File:figure1.jpg]]<br />
<br />
From the experiment result, we see that correlational Learning is more effective than the other two methods in recovering overlapping clusters. It works well even when the intra-cluster link density is low. Co-clustering performs poorly because it only finds non-overlapping clusters.<br />
<br />
'''B. Social Media Data'''<br />
<br />
[[File:Social Media experiment.jpg]]<br />
<br />
From the experiment with [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]], the paper show us that:<br />
<br />
* The probability of a link between two users increases with respect to the number of tags they share.<br />
<br />
* Correlational Learning consistently performs better, especially when the training set is small.<br />
<br />
* Higher co-occurrence frequency suggests that two users are more similar. Similar patterns are observed in the three methods.<br />
<br />
== Related papers ==<br />
<br />
The author uses Co-Clustering method in [[RelatedPaper::Co-clustering documents and words using bipartite spectral graph partitioning]] as a comparison to above methods.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Xufei_Wang,_ICDM,_2010&diff=5472Xufei Wang, ICDM, 20102011-04-04T03:50:33Z<p>Nqi: /* Databases */</p>
<hr />
<div>== Citation ==<br />
Xufei Wang. 2010. Discovering Overlapping Groups in Social Media, the 10th IEEE International Conference on Data Mining (ICDM 2010).<br />
<br />
== Online Version ==<br />
http://dmml.asu.edu/users/xufei/Papers/ICDM2010.pdf<br />
<br />
== Databases ==<br />
[[Property:UsesDataset|BlogCatalog]][[Category::Dataset]] BlogCatalog [http://www.blogcatalog.com/]<br />
<br />
[[Property:UsesDataset|Delicious]][[Category::Dataset]] Delicious [http://www.delicious.com/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors propose a novel co-clustering framework, which takes advantage of networking information between users and tags in social media, to discover these [[AddressesProblem::overlapping communities]]. The basic ideas are:<br />
<br />
* To discover overlapping communities in social media. Diverse interests and interactions that human beings can have in online social life suggest that one person often belongs more than one community.<br />
<br />
* To use user-tag subscription information instead of user-user links. Metadata such as tags become an important source in measuring the user-user similarity. The paper shows that more accurate community structures can be obtained by scrutinizing tag information.<br />
<br />
* To obtain clusters containing users and tags simultaneously. Existing co-clustering methods cluster users/tags separately. Thus, it is not clear which user cluster corresponds to which tag cluster. But the proposed method is able to find out user/tag group structure and their correspondence<br />
<br />
== Problem Statement ==<br />
<br />
In this paper, the concept of community is generalized to include both users and tags. Tags of a community imply the major concern of people within it.<br />
<br />
Let <math>\mu = \left ( \mu _{1},\mu _{2},...,\mu _{m} \right )</math> denote the user set, <math>\tau = \left ( \tau _{1},\tau _{2},...,\tau _{n} \right )</math> the tay set. A community <math>C_{i}\left ( 1\leq i\leq k \right )</math> is a subset of user and tags, where k is the number of communities. As mentioned above, communities usually overlap, i.e., <math>C_{i}\bigcap C_{j}\neq \O \left ( 1\leq i,j\leq k \right )</math>.On the other hand, users and their subscribed tags form a user-tag matrix M, in which each entry <math>M_{ij}\in \left \{ 0,1 \right \}</math> indicates whether user <math>u_{i}</math> subscribes to tag <math>t_{j}</math>. So it is reasonable to view a user as a sparse vector of tags, and each tag as a sparse vector of users.<br />
<br />
Given notations above, the overlapping co-clustering problem can be stated formally as follows:<br />
<br />
'''Input:'''<br />
* A user-tag subscription matrix <math>M_{N_{\mu }\times N_{t}}</math>, where <math>N_{\mu }</math> and <math>N_{t}</math> are the numbers of users and tags.<br />
<br />
* The number of communities k.<br />
<br />
'''Output:'''<br />
* k overlapping communities which consist of both users and tags.<br />
<br />
== Brief Description Of The Method ==<br />
<br />
Communities that aggregate similar users and tags together can be detected by maximizing intra-cluster similarity, which is shown below:<br />
<math>arg max\frac{1}{k}\sum_{i=1}^{k}\sum_{x_{j}\in C_{i}}^{}S_{C}\left ( x_{j},c_{i} \right )</math><br />
where k is the number of communities, x is the edges and c is the centroid of community. This formulation can be solved by a [[UsesMethod::k-means]] variant.<br />
<br />
This paper uses different methods to solve the problem of overlapping communities:<br />
<br />
<br />
'''A. [[UsesMethod::Independent Learning]]'''<br />
<br />
<br />
'''B. [[UsesMethod::Normalized Learning]]'''<br />
<br />
<br />
'''C. [[UsesMethod::Correlational Learning]]'''<br />
<br />
== Experimental Result ==<br />
<br />
The authors use two kinds of datasets: one is a synthetic data and the other kind is real data from [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]]<br />
<br />
'''A. Synthetic Data'''<br />
<br />
Synthetic data, which is controlled by various parameters, facilitates a comparative study between the uncovered and actual clusters. It has 1,000 users and 1,000 tags and with different number of clusters which range from 5 to 50.<br />
<br />
[[File:figure1.jpg]]<br />
<br />
From the experiment result, we see that correlational Learning is more effective than the other two methods in recovering overlapping clusters. It works well even when the intra-cluster link density is low. Co-clustering performs poorly because it only finds non-overlapping clusters.<br />
<br />
'''B. Social Media Data'''<br />
<br />
[[File:Social Media experiment.jpg]]<br />
<br />
From the experiment with [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]], the paper show us that:<br />
<br />
* The probability of a link between two users increases with respect to the number of tags they share.<br />
<br />
* Correlational Learning consistently performs better, especially when the training set is small.<br />
<br />
* Higher co-occurrence frequency suggests that two users are more similar. Similar patterns are observed in the three methods.<br />
<br />
== Related papers ==<br />
<br />
The author uses Co-Clustering method in [[RelatedPaper::Co-clustering documents and words using bipartite spectral graph partitioning]] as a comparison to above methods.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Xufei_Wang,_ICDM,_2010&diff=5471Xufei Wang, ICDM, 20102011-04-04T03:47:51Z<p>Nqi: </p>
<hr />
<div>== Citation ==<br />
Xufei Wang. 2010. Discovering Overlapping Groups in Social Media, the 10th IEEE International Conference on Data Mining (ICDM 2010).<br />
<br />
== Online Version ==<br />
http://dmml.asu.edu/users/xufei/Papers/ICDM2010.pdf<br />
<br />
== Databases ==<br />
[[Category::Dataset]] BlogCatalog [http://www.blogcatalog.com/]<br />
<br />
[[Category::Dataset]] Delicious [http://www.delicious.com/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors propose a novel co-clustering framework, which takes advantage of networking information between users and tags in social media, to discover these [[AddressesProblem::overlapping communities]]. The basic ideas are:<br />
<br />
* To discover overlapping communities in social media. Diverse interests and interactions that human beings can have in online social life suggest that one person often belongs more than one community.<br />
<br />
* To use user-tag subscription information instead of user-user links. Metadata such as tags become an important source in measuring the user-user similarity. The paper shows that more accurate community structures can be obtained by scrutinizing tag information.<br />
<br />
* To obtain clusters containing users and tags simultaneously. Existing co-clustering methods cluster users/tags separately. Thus, it is not clear which user cluster corresponds to which tag cluster. But the proposed method is able to find out user/tag group structure and their correspondence<br />
<br />
== Problem Statement ==<br />
<br />
In this paper, the concept of community is generalized to include both users and tags. Tags of a community imply the major concern of people within it.<br />
<br />
Let <math>\mu = \left ( \mu _{1},\mu _{2},...,\mu _{m} \right )</math> denote the user set, <math>\tau = \left ( \tau _{1},\tau _{2},...,\tau _{n} \right )</math> the tay set. A community <math>C_{i}\left ( 1\leq i\leq k \right )</math> is a subset of user and tags, where k is the number of communities. As mentioned above, communities usually overlap, i.e., <math>C_{i}\bigcap C_{j}\neq \O \left ( 1\leq i,j\leq k \right )</math>.On the other hand, users and their subscribed tags form a user-tag matrix M, in which each entry <math>M_{ij}\in \left \{ 0,1 \right \}</math> indicates whether user <math>u_{i}</math> subscribes to tag <math>t_{j}</math>. So it is reasonable to view a user as a sparse vector of tags, and each tag as a sparse vector of users.<br />
<br />
Given notations above, the overlapping co-clustering problem can be stated formally as follows:<br />
<br />
'''Input:'''<br />
* A user-tag subscription matrix <math>M_{N_{\mu }\times N_{t}}</math>, where <math>N_{\mu }</math> and <math>N_{t}</math> are the numbers of users and tags.<br />
<br />
* The number of communities k.<br />
<br />
'''Output:'''<br />
* k overlapping communities which consist of both users and tags.<br />
<br />
== Brief Description Of The Method ==<br />
<br />
Communities that aggregate similar users and tags together can be detected by maximizing intra-cluster similarity, which is shown below:<br />
<math>arg max\frac{1}{k}\sum_{i=1}^{k}\sum_{x_{j}\in C_{i}}^{}S_{C}\left ( x_{j},c_{i} \right )</math><br />
where k is the number of communities, x is the edges and c is the centroid of community. This formulation can be solved by a [[UsesMethod::k-means]] variant.<br />
<br />
This paper uses different methods to solve the problem of overlapping communities:<br />
<br />
<br />
'''A. [[UsesMethod::Independent Learning]]'''<br />
<br />
<br />
'''B. [[UsesMethod::Normalized Learning]]'''<br />
<br />
<br />
'''C. [[UsesMethod::Correlational Learning]]'''<br />
<br />
== Experimental Result ==<br />
<br />
The authors use two kinds of datasets: one is a synthetic data and the other kind is real data from [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]]<br />
<br />
'''A. Synthetic Data'''<br />
<br />
Synthetic data, which is controlled by various parameters, facilitates a comparative study between the uncovered and actual clusters. It has 1,000 users and 1,000 tags and with different number of clusters which range from 5 to 50.<br />
<br />
[[File:figure1.jpg]]<br />
<br />
From the experiment result, we see that correlational Learning is more effective than the other two methods in recovering overlapping clusters. It works well even when the intra-cluster link density is low. Co-clustering performs poorly because it only finds non-overlapping clusters.<br />
<br />
'''B. Social Media Data'''<br />
<br />
[[File:Social Media experiment.jpg]]<br />
<br />
From the experiment with [[Category::Dataset|BlogCatalog]] and [[Category::Dataset|Delicious]], the paper show us that:<br />
<br />
* The probability of a link between two users increases with respect to the number of tags they share.<br />
<br />
* Correlational Learning consistently performs better, especially when the training set is small.<br />
<br />
* Higher co-occurrence frequency suggests that two users are more similar. Similar patterns are observed in the three methods.<br />
<br />
== Related papers ==<br />
<br />
The author uses Co-Clustering method in [[RelatedPaper::Co-clustering documents and words using bipartite spectral graph partitioning]] as a comparison to above methods.</div>Nqihttp://curtis.ml.cmu.edu/w/courses/index.php?title=Dan_Cosley,_AAAI,_2010&diff=5470Dan Cosley, AAAI, 20102011-04-04T03:42:53Z<p>Nqi: Created page with '== Citation == Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Art…'</p>
<hr />
<div>== Citation ==<br />
Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Xiangyang Lan, Siddarth Suri. Sequential Influence Models in Social Network. Association for the Advancement of Artificial Intelligence. 2010.<br />
<br />
== Online Version ==<br />
http://www.cs.cornell.edu/home/kleinber/icwsm10-seq.pdf<br />
<br />
== Databases ==<br />
[[Category::Wikipedia]] [http://www.wikipedia.org/]<br />
<br />
== Summary ==<br />
In this [[Category::paper]], the authors propose a novel co-clustering framework, which takes advantage of networking information between users and tags in social media, to discover these [[AddressesProblem::overlapping communities]]. The basic ideas are:</div>Nqi