Master Thesis Defence by Mr Radityo Eko Prasojo
Mr Radityo Eko Prasojo defended his master thesis on 'Entity and Aspect Extraction for Organizing News Comments'
Mr Radityo Eko Prasojo defended his master thesis on 'Entity and Aspect Extraction for Organizing News Comments' at unibz on 14 July 2015.
Abstract: Nowadays, social media platforms provide the latest breaking news on various topics about everyday events. Furthermore, they give the opportunity for Internet users to engage in discussions with each other on any published article, usually by writing comments and replies. Typically, these comments are unstructured making it hard to catch the flow of user debates and to understand their main points of agreements and disagreements. Thus, there is a need for organizing users’ comments around the main ideas they present and the sentiments they express. In this work, we address the above problem through the following contributions. First, we extend traditional Named-Entity Recognition and Classification tools, using coreference resolution and external knowledge bases, to detect all occurrences of entities in a given comment. Second, we propose a domain-independent approach for extracting explicit and implicit aspects around discussed entities. To handle aspect extraction, we exploit part-of-speech tag, dependency tag, and lexical databases. Third, we evaluate our entity and aspect extraction approach, on manually annotated data, showing that it highly increases precision and recall compared to baseline approaches. Fourth, we use the extracted entities and aspects to organize the comments through a visualization interface to have a better understanding on the flow of discussions around news articles.