Research title: Sentiment Analysis in Discussion Forums
Start date: February 2012
Sentiment analysis (SA) involves the extraction and assessment of subjectivity and opinion in text. Its main goal is to effectively determine, firstly, whether a piece of natural language text (word/phrase, sentence or document) expresses opinion, secondly, the sentiment orientation or polarity (i.e. positive or negative) of opinionated text, and thirdly, the strength of such polarity. Web discussion forums provide a means to share and express user views on different topics part of which are opinionated toward certain entities. Typically, a discussion thread is viewed based on a timeline. However, the user may want to view the thread differently (e.g. based on agreements, disagreements, positive opinions or negative opinions). In this research, we propose to explore lexicon-based methods of SA with a view to enabling different user views to a discussion thread for an effective search and browse. We focus on investigating contextual analyses relevant to the web discussion setting and effective integration of such analyses into sentiment aggregation process.
A.Muhammad, N.Wiratunga, R.Lothian, R.Glassey, (2013). Contextual Sentiment Analysis in Social Media Using High-Coverage Lexicon. To appear in proceedings of the Thirty-third SGAI International Conference on Artificial Intelligence (AI-2013). Springer.