The challenge of understanding the flow of sentiments in social media documents

Abstract

This talk is focused on a key task in the area of Opinion Mining and Sentiment Analysis: polarity classification of social media documents (e.g. blog posts). Estimating polarity is much more demanding than estimating topicality. As a matter of fact, the effectiveness of polarity classification is still modest and does not compare with the effectiveness of standard retrieval tasks. Polarity estimation is severely affected by parts of the text that are off-topic or that simply do not express any opinion. In fact, the key sentiments in a document often appear in specific locations of the text. Furthermore, there are usually conflicting opinions in a given document and this mixed set of opinions harms the performance of automatic methods designed to estimate the overall orientation of the text. In this talk, I will argue that understanding the flow of sentiments in a text is a major challenge for effectively predicting the document's orientation towards a given topic. I will briefly outline some possible avenues to address this challenging issue and review some recent papers that take steps in this direction.