Computational Linguistic Research on the Sentiment Analysis of Chinese (and English) Media Texts

The same individual can be referred to as a “hero” or “terrorist”, or as being “stingy” or “thrifty”, and the writer’s choice of language reveals much about his/her subjective feeling towards the individual. Such sentiment analysis of the writer’s positive, negative, or neutral opinion by inferencing provides a basic summary of his/her stand, and can be important in today’s fast-paced society for politicians and public figures to monitor relevant opinion shift, for government departments to summarize critical public feedback, and for commercial enterprises to obtain bottom-line feedback on products. This new interdisciplinary area draws on linguistics and computational methods to implement computer understanding of non-obvious subjective opinions underlying textual material.This project aims to undertake fundamental theory-motivated research in sentiment analysis and to advance the state of the art through a linguistic investigation of subjective opinions. It builds on the team’s previous successful research on opinionated reporting on the 2004 US election in the Chinese news media to investigate deeper semantic relationship between the focussed subject and linguistic structures central to the deeper understanding of opinionated writing. It represents a linguistically motivated departure from the currently predominant bag-of-words approach and also has significant potential in many areas of language engineering.