Twitter-Network Topic Model: A Full Bayesian Treatment for Social Network and Text Modeling

TL;DR

This paper introduces a Bayesian nonparametric model that jointly analyzes Twitter text and social network data, improving topic modeling accuracy and enabling various applications like author recommendation and hashtag suggestion.

Contribution

It presents the Twitter-Network (TN) topic model combining hierarchical Poisson-Dirichlet processes and Gaussian processes for social network and text modeling, outperforming existing models.

Findings

TN model significantly outperforms existing nonparametric models

Enables inference of authors' interests and hashtag analysis

Supports applications like author recommendation and hashtag suggestion

Abstract

Twitter data is extremely noisy -- each tweet is short, unstructured and with informal language, a challenge for current topic modeling. On the other hand, tweets are accompanied by extra information such as authorship, hashtags and the user-follower network. Exploiting this additional information, we propose the Twitter-Network (TN) topic model to jointly model the text and the social network in a full Bayesian nonparametric way. The TN topic model employs the hierarchical Poisson-Dirichlet processes (PDP) for text modeling and a Gaussian process random function model for social network modeling. We show that the TN topic model significantly outperforms several existing nonparametric models due to its flexibility. Moreover, the TN topic model enables additional informative inference such as authors' interests, hashtag analysis, as well as leading to further applications such as author recommendation, automatic topic labeling and hashtag suggestion. Note our general inference framework can readily be applied to other topic models with embedded PDP nodes.