Question Analysis and Answer Passage Retrieval for Opinion Question Answering Systems

Question Analysis and Answer Passage Retrieval for Opinion Question Answering Systems Lun-Wei Ku, Yu-Ting Liang and Hsin-Hsi Chen Department of Computer Science and Information Engineering National Taiwan University {lwku, eagan}@nlg.csie.ntu.edu.tw; hhchen@csie.ntu.edu.tw Abstract Question answering systems provide an elegant way for people to access an underlying knowledge base. Humans are not only interested in factual questions but also interested in opinions. This paper deals with question analysis and answer passage retrieval in opinion QA systems. For question analysis, six opinion question types are defined. A two-layered framework utilizing two question type classifiers is proposed. Algorithms for these two classifiers are described. The performance achieves 87.8% in general question classification and 92.5% in opinion question classification. The question focus is detected to form a query for the information retrieval system and the question polarity is detected to retain relevant sentences which have the same polarity as the question. For answer passage retrieval, three components are introduced. Relevant sentences retrieved are further identified whether the focus (Focus Detection) is in a scope of opinion (Opinion Scope Identification) or not, and if yes, whether the polarity of the scope matches with the polarity of the question (Polarity Detection). The best model achieves an F-measure of 40.59% using partial match at the level of meaningful unit. With relevance issues removed, the F-measure of the best model boosts up to 84.96%. 1 Introduction Most of the state-of-the-art Question Answering (QA) systems serve the needs of answering factual questions such as “When was James Dean born?” and “Who won the Nobel Peace Prize in 1991?”. In addition to facts, people would also like to know about others’ opinions, thoughts, and feelings toward some specific topics, groups, and events. Opinion questions (e.g. “How do Americans consider the US-Iraq war?” and “What are the public’s opinions on human cloning?”) revealing answers about people’s opinions have long as well as complex answers which tend to scatter across different documents. Traditional QA approaches are not effective enough to retrieve answers for opinion questions as they have been for factual questions (Stoyanov et al., 2005). Hence, an opinion QA system is essential and urgent. Most of the research on QA systems has been developed for factual questions, and the association of subjective information with question answering has not yet been much studied. As for subjective information, Wiebe (2000) proposed a method to identify strong clues of subjectivity on adjectives. Riloff et al. (2003) presented a subjectivity classifier using lists of subjective nouns learned by bootstrapping algorithms. Riloff and Wiebe (2003) proposed a bootstrapping process to learn linguistically rich extraction patterns for subjective expressions. Kim and Hovy (2004) presented a system to determine word sentiments and combined sentiments within a sentence. Pang, Lee, and Vaithyanathan (2002) classified documents not by the topic, but by the overall sentiment, and then determined the polarity of a review. Wiebe et al. (2002) proposed a method for opinion summarization. Wilson et al. (2005) presented a phrase-level sentiment analysis to automatically identify the contextual polarity. Ku et al. …

Content truncated to 500 words