The goal of this collaborative exploratory research project (0842546, Clement Yu, UIC and 0842608 Weiyi Meng, SUNY Binghamton) is to investigate two important problems in opinion retrieval. The first is to determine whether a negation word flips the sentiment of an opinionated word and the second is to examine how the opinions about a topic change over time and what causes significant changes. This project aims to demonstrate within 18 months that within two domains (consumer products/services and medical/health), whether a negation word flips the polarity of an opinionated word or phrase can be accurately determined when there is a single occurrence of such negation word in a sentence, and the significant time points and words/phrases that characterize the changes of sentiments for a given topic can be accurately obtained. The research results from this project will lay the foundation in achieving longer-term goals that include determining the polarity of the sentiment of a feature with reference to a given query in any given domain, identifying the events that cause the significant change of sentiments, measuring the intensity of sentiments and its change over time, and studying the interactions between the negation and temporal aspects of opinion retrieval. It is expected that the proposed project can have a significant impact on future search engine technology, information retrieval, text mining, and blog retrieval and analysis. Research results will be incorporated into several courses the PIs teach and students will be recruited to participate in the research activities. Research results will be disseminated through published papers. The project Web site (www.cs.binghamton.edu/~meng/OpRetSGER.html) provides access to research results.
