View source for Yano et al ICWSM 2010. What’s Worthy of Comment? Content and Comment Volume in Political Blogs

== Citation ==
Tae Yano and Noah A. Smith. What’s Worthy of Comment? Content and Comment Volume in Political Blogs. In Proc of ICWSM 2010.

== Online Version ==
[http://www.cs.cmu.edu/~nasmith/papers/yano+smith.icwsm10.pdf What’s  Worthy of Comment? Content and Comment Volume in Political Blogs].

== Summary ==

This [[Category::Paper]] describes a  [[UsesMethod::topic model]]  based approach in modeling the relationship between the text content of a political blog post and the comment volume (i.e. the total amount of response) that a post will receive.

== Brief description of the method ==

The author's propose a generative model, called the Topic-Poisson model, which proceeds as follows. The number of topic <math>K</math> is fixed in advance.

[[File:yano-icwsm-tp.png|400px]]

Apart from step 2(c), the model is identical to a (smoothed) LDA. <math>v_d</math> is chosen from a mixture of distributions, where <math>m_{d,k}</math> are the mixture weights and <math>p</math> are Poisson mixture components.

The authors also present a variation of this model called the Topic Negative Binomial model, where the mixture of Poissons is replaced by a mixture of negative binomials.

== Experimental Result ==

'''Task''': Predict whether a blog post will have higher volume than the average seen in training data
(Note that they are NOT predicting the absolute number of words in the comments)

The authors use a subset of the [[UsesDataset::Yano & Smith blog dataset|Yano & Smith blog dataset]]; data from 2 blogs, [http://matthewyglesias.theatlantic.com Matthew Yglesias] (denoted MY) and [http://www.redstate.com/ Red State] (denoted RS) were used.

The compared models were:
* Naive Bayes (NB)
* Regression (Reg): linear regression with elastic net regularization
* Topic Poisson (T-Pois)
* Topic Negative Binomial (T-NBin)
* CommentLDA (C-LDA): refer to [[RelatedPaper::Yano et al NAACL 2009]]

'''Results''': 

[[File:yano-icwsm-result.png|400px]]

Word counts and comment counts were used to measure volume.
Unless noted, the topic models have their parameters set to <math>K = 15 </math>, <math>\alpha = 0.1</math>, and <math>\beta = 0.1</math>

The Topic-Poisson model improves recall substantially over Naive-Bayes, on both # words and # comments, with a slight loss in precision. Its precision lags behind the regression model.

The authors also show the topics discovered in MY by the word-volume Topic-Poisson model. Topics are ranked by <math>\lambda_k</math>.

[[File:yano-icwsm-my-topics.png|800px]]

== Discussion ==

Modeling topics can improve recall when predicting high volume posts.

== Related Papers ==

The Topic-Poisson model is essentially a type of supervised or annotated LDA as defined in [[RelatedPaper::Blei and McAuliffe 2008 Supervised topic models|Blei and McAuliffe (2008)]] and  [[RelatedPaper::Ramage 2009 Labeled LDA: A supervised topic model for credit attribution in multi-labeled corpora|Ramage et al. (2009)]].

== Study Plan ==