A Latent Dirichlet Allocation Approach using Mixed Graph of Terms for Sentiment Analysis

Casillo, Mario
Clarizia, Fabio
Colace, Francesco
De Santo, Massimo
Lombardi, Marco
Pascale, Francesco
Journal Title
Journal ISSN
Volume Title
The spread of generic (as Twitter, Facebook orGoogle+) or specialized (as LinkedIn or Viadeo) social networks allows to millions of users to share opinions on different aspects of life every day. Therefore this information is a rich source of data for opinion mining and sentiment analysis. This paper presents a novel approach to the sentiment analysis based on the Latent Dirichlet Allocation (LDA) approach. The proposed methodology aims to identify a word-based graphical model (we call it a mixed graph of terms) for depicting a positive or negative attitude towards a topic. By the use of this model it will be possible to automatically mine from documents positive and negative sentiments.Experimental evaluation, on standard and real datasets, shows that the proposed approach is effective and furnishes good and reliable results.
Data Analytics, Data Mining and Machine Learning for Social Media, Digital and Social Media, Information Extraction, Latent Dirichlet Allocation, Mixed Graph of Terms, Relations Learning, Sentiment Analysis, Structure Learning.
Access Rights
Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.