Sentiment Analysis in Social Networks through Topic modeling

Sentiment Analysis in Social Networks through Topic Modeling Debashis Naskar 1, Sidahmed Mokaddem 1,2, Miguel Rebollo 1, Eva Onaindia 1 1 Universitat Politecnica de Valencia, Camino de Vera s/n Valencia Spain 2 University of Oran, BP 1524 El Menouar street 31000 Oran Algeria (DZ) debashis@drtc.isibang.ac.in, sid wise@hotmail.com, mrebollo@dsic.upv.es, onaindia@dsic.upv.es Abstract In this paper, we analyze the sentiments derived from the conversations that occur in social networks. Our goal is to identify the sentiments of the users in the social network through their conversations. We conduct a study to determine whether users of social networks (twitter in particular) tend to gather together according to the likeness of their sentiments. In our proposed framework, (1) we use ANEW, a lexical dictionary to identify affective emotional feelings associated to a message according to the Russell’s model of affection; (2) we design a topic modeling mechanism called Sent LDA, based on the Latent Dirichlet Allocation (LDA) generative model, which allows us to find the topic distribution in a general conversation and we associate topics with emotions; (3) we detect communities in the network according to the density and frequency of the messages among the users; and (4) we compare the sentiments of the communities by using the Russell’s model of affect versus polarity and we measure the extent to which topic distribution strengthen likeness in the sentiments of the users of a community. This works contributes with a topic modeling methodology to analyze the sentiments in conversations that take place in social networks. Keywords: sentiment analysis, topic model, emotion identification, multilayer network 1. Introduction Despite the amount of research done in sentiment analy- sis in social networks, the study of dissemination patterns of the emotions is limited. It is well known that social net- works exhibit some kind of positive correlation in the polar- ity of the sentiments associated to sequential pairs of mes- sages (Hillmann and Trier, 2012), and that the analysis of diffusion mechanisms is clue to determine the dynamic of the evolution of a particular emotion. Ziegler and Lausen analyzed propagation of trust and dis- trust on social networks, what can be considered the first paper in which sentiment propagation was studied (Ziegler and Lausen, 2005). Interesting conclusions, like that pos- itive and negative sentiments follow a different propaga- tion pattern (Hillmann and Trier, 2012), have been drawn from the various investigations on sentiments in social net- works. Other works studied the correlation between emo- tions and information diffusion, finding that those messages emotionally charged were re-tweeted more often (Stieglitz and Dang-Xuan, 2013), or investigated if the topic and the opinion of the user’s contacts affect the own user’s opinion (Tang and Fong, 2013). The ultimate objective of analyzing sentiments in social networks is to be able to predict the attitude of people and infer behaviour patterns like, for example, reactions against negative opinions. In this line, Nguyen et al. studied changes in collective sentiment and predicted the dynam- ics with statistical models that contemplate the complete network (Nguyen et al., …

Content truncated to 500 words