Single Topic Sentiment LDA: Analysis of Political Conversation Topics in Twitter

J. González Ortega, R. Naveiro Flores, M. Muñoz Aragón

Over the past decade, political debate has experienced significant transformation alongside the increased use of social networks which have emerged as dynamic public forums for discussions, prompting most politicians to establish multiple public profiles to advocate for their political views and policies. The impact of this new paradigm in politics has been extensively investigated, as examined by Tucker et al. (2018). Our contribution to this field is a novel Latent Dirichlet Allocation (LDA) model tailored for analyzing single-topic short texts with sentiment, thereby facilitating the identification of positive, negative and neutral conversation topics on social media platforms like Twitter. To substantiate our model's efficacy, we provide a model validation scheme and a case study involving an extensive database of tweets from Spanish parliament representatives, spanning from September 2021 to May 2022.

Keywords: Latent Dirichlet Allocation, Sentiment Analysis, Collapsed Gibbs Sampling, Political Debate

Bayesian methods II
November 10, 2023  12:00 PM
HC2: Canónigos Room 2

C. Ausin, M. Kalli

F. J. Girón González-Torre, C. del Castillo Vázquez

M. Morales Otero, V. Gómez Rubio, V. Núñez Antón