Difference between revisions of "Compare Ramage Naaman"
| Line 27: | Line 27: | ||
== Algorithm == | == Algorithm == | ||
| − | |||
| − | + | The Ramage paper used LDA topic model to model each tweet, while the Naaman paper used various quantitative (statistical) methods, such as Pearson Chi-square, Kalensky’s analysis, Ward’s linkage cluster analysis. | |
== Dataset == | == Dataset == | ||
Revision as of 23:10, 5 November 2012
Two Papers
Problem
In the Ramage paper, the authors claims that the topics of all the tweets in Twitter can be classified into four categories:
- Substance Topics about events and ideas - Social Topics recognizing language used toward a social end - Status Topics denoting personal update - Style Topic that contains broader trends in language usage
They used the LDA topic model to model the latent topic information in each tweet.
In the Naaman paper, the authors also categorized the tweets from their underlying social meanings. However, they came up with 9 categories instead of 4:
- Information Sharing (IS) - Self Promotion (SP) - Opinions/Complaints (OC) - Statements and Random Thoughts (RT) - Me now (ME): talking about the user's own feelings - Question to followers (QF) - Presence Maintenance (PM) - Anecdote (me) (AM) - Anecdote (other) (AO)
Although the two categorizing methods are quite different, they both look into the latent meanings of the tweets and try to find the connection between the tweet's words and its social meaning.
Algorithm
The Ramage paper used LDA topic model to model each tweet, while the Naaman paper used various quantitative (statistical) methods, such as Pearson Chi-square, Kalensky’s analysis, Ward’s linkage cluster analysis.
Dataset
Again, because of the different perspective of the two papers, they used different dataset to evaluation their methods. In the ICDM paper, the authors used MSN Space blog data and randomly selected 10k users so that the major concern is how to measure the similarity between two users without worrying about the scalability too much.
The ICWSM paper used BlogCatalog and DBLP dataset. For both of datasets, the ICWSM paper used user metadata to define similarity and evaluated the proposed methods on a large scale.
Big Idea
The two papers were solving similar problem from different perspectives, However, those two perspective are not likely to be combined easily as both of the papers make some simplifying assumption about other perspectives so that they can just focus on one.
Questions
1. How much time did you spend reading the (new, non-wikified) paper you summarized?
35 min
2. How much time did you spend reading the old wikified paper?
20 min
3. How much time did you spend reading the summary of the old paper?
10 min
4. How much time did you spend reading background materiel?
15 min
5. Was there a study plan for the old paper?
if so, did you read any of the items suggested by the study plan? and how much time did you spend with reading them?
No study plan
6. Give us any additional feedback you might have about this assignment.
