Modeling Time-Varying User Attitudes in Social Media

The underlying text classification setting generally used to model user attitudes in social media merely captures current attitudes whenever a text is generated. Under such constraints, the evolving historical attitudes are omitted and no anticipation of subsequent attitudes could be made. To alleviate this limitation, we propose a time-dependent approach that takes advantage of historical user comments to learn temporal descriptors that explain their evolving attitudes. We demonstrate that our approach could be used to predict user attitudes at subsequent times ahead. The experimental results on Tweets data, during the Covid-19 pandemic, corroborate our claim.