A latent representation model for sentiment analysis in heterogeneous social networks

Debora Nozza (1) , Daniele Maccagnola (1) , Vincent Guigue (2) , Enza Messina (1) , Patrick Gallinari (2)

1 DISCo - Dipartimento di Informatica Sistemistica e Comunicazione
2 MLIA - Machine Learning and Information Access

The growing availability of social media platforms, in particular microblogs such as Twitter, opened new way to people for expressing their opinions. Sentiment Analysis aims at inferring the polarity of these opinions, but most of the existing approaches are based only on text, disregarding information that comes from the relationships among users and posts. In this paper we consider microblogs as heterogeneous networks and we use an approach based on latent representation of nodes to infer, given a specific topic, the sentiment polarity of posts and users at the same time. The experimental investigation show that our approach, by taking into account both content and relationship information, outperforms supervised classifiers based only on textual content.

Domaines: Intelligence artificielle [cs.AI]

Debora Nozza, Daniele Maccagnola, Vincent Guigue, Enza Messina, Patrick Gallinari. A latent representation model for sentiment analysis in heterogeneous social networks. International Conference on Software Engineering and Formal Methods, 2014, Grenoble, France. pp.201-213, ⟨10.1007/978-3-319-15201-1_13⟩. ⟨hal-02503482⟩