Semantic Social Network Analysis with Text Corpora
We present the Document-Entity-Topic (DET) model for semantic social network analysis which tries to find out the interested entities through the topics we aim at, detect groups according to the entities which concern the similar topics, and rank the plentiful entities in a document to figure out the most valuable ones. DET model learns the topic distributions by the literal descriptions of entities. The model is similar to Author-Topic (AT) model, adding the key attribute that the distribution of entities in a document is not uniform but Dirichlet allocation. We experiment on the “Libya Event” data set which is collected from the Internet. DET model increases the precision on tasks of social network analysis and gives much lower perplexity than AT model.
KeywordsSemantic Social Network Analysis Topic Model Entity Modeling
Unable to display preview. Download preview PDF.
- 1.Blei, D.: Introduction to Probabilistic Topic Models. Communications of the ACM (2011)Google Scholar
- 2.Steyvers, M., Griffiths, T.: Probabilistic Topic Models. Handbook of Latent Semantic Analysis 427 (2007)Google Scholar
- 3.Blei, D., Carin, L., Dunson, D.: Topic Models. IEEE Signal Processing Magazine 27(6), 55–65 (2010)Google Scholar
- 6.Rosen-Zvi, M., Griffiths, T., Steyvers, M., Smyth, P.: The Author-Topic Model for Authors and Documents, pp. 478–494. AUAI Press (2004)Google Scholar
- 7.Shiozaki, H., Eguchi, K., Ohkawa, T.: Entity Network Prediction Using Multitype Topic Models. Springer (2008)Google Scholar
- 8.Newman, D., Chemudugunta, C., Smyth, P.: Statistical Entity-Topic Models, pp. 680–686 (2006)Google Scholar