View source for Chang, Annals of Applied Statistics(AoAS) 2010

== Citation ==

J. Chang and D. Blei, 2010. Hierarchical relational models for document networks. To appear in Annals of Applied Statistics, 2010

== Online version ==

[http://www.e-publications.org/ims/submission/index.php/AOAS/user/submissionFile/4451?confirm=6c1aa559 AoAS website]
[http://arxiv.org/abs/0909.4331 Arxiv]

== Summary ==

This [[Category::paper]] presents a [[UsesMethod::topic model]] that simultaneously models document text and the links (relations) between documents. The authors extend [http://en.wikipedia.org/wiki/Latent_Dirichlet_allocation Latent Dirichlet Allocation(LDA)], to enable links to be formed between documents. They call their model the [[UsesMethod::relational topic model]].

They ran their model on a subset of the [[UsesDataset::Cora_network|Cora network]], the [[UsesDataset::Web KB dataset]], [[UsesDataset:PNAS|PNAS citation data]], and a dataset of [[UsesDataset:Local News with Geographical Links| local news, with geographical links]].
They observed that their model is better than LDA (with logistic regression) and pairwise link-LDA in predicting held out words and held out links.

== Related papers ==
There has been a lot of previous work in modeling links and text, including
* The paper by [[RelatedPaper::Nallapati KDD2008]] models directed links by allowing links from citing documents to cited documents in a bipartite graph.
* The paper by [[RelatedPaper::Gruber UAI2008]] models directed links by assuming that a link from a word w to a document d depends on the frequency of the topic of w in d, and the in-degree of d. 
* The paper by [[RelatedPaper::Sinkkonen arxiv2008]] models the entire network as a bag of links using an interaction component model, which can scale to very large graphs with 670,000 nodes and 1.89 million links.