Statistical Machine Translation Improves Question Retrieval in Community Question Answering via Matrix Factorization

Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, pages 852–861, Sofia, Bulgaria, August 4-9 2013. c⃝2013 Association for Computational Linguistics Statistical Machine Translation Improves Question Retrieval in Community Question Answering via Matrix Factorization Guangyou Zhou, Fang Liu, Yang Liu, Shizhu He, and Jun Zhao National Laboratory of Pattern Recognition Institute of Automation, Chinese Academy of Sciences 95 Zhongguancun East Road, Beijing 100190, China {gyzhou,fliu,liuyang09,shizhu.he,jzhao}@nlpr.ia.ac.cn Abstract Community question answering (CQA) has become an increasingly popular re- search topic. In this paper, we focus on the problem of question retrieval. Question retrieval in CQA can automatically find the most relevant and recent questions that have been solved by other users. However, the word ambiguity and word mismatch problems bring about new challenges for question retrieval in CQA. State-of-the-art approaches address these issues by implic- itly expanding the queried questions with additional words or phrases using mono- lingual translation models. While use- ful, the effectiveness of these models is highly dependent on the availability of quality parallel monolingual corpora (e.g., question-answer pairs) in the absence of which they are troubled by noise issue. In this work, we propose an alternative way to address the word ambiguity and word mismatch problems by taking advan- tage of potentially rich semantic informa- tion drawn from other languages. Our pro- posed method employs statistical machine translation to improve question retrieval and enriches the question representation with the translated words from other lan- guages via matrix factorization. Experi- ments conducted on a real CQA data show that our proposed approach is promising. 1 Introduction With the development of Web 2.0, community question answering (CQA) services like Yahoo! Answers,1 Baidu Zhidao2 and WkiAnswers3 have attracted great attention from both academia and industry (Jeon et al., 2005; Xue et al., 2008; Adamic et al., 2008; Wang et al., 2009; Cao et al., 2010). In CQA, anyone can ask and answer ques- tions on any topic, and people seeking information are connected to those who know the answers. As answers are usually explicitly provided by human, they can be helpful in answering real world ques- tions. In this paper, we focus on the task of question retrieval. Question retrieval in CQA can automati- cally find the most relevant and recent questions (historical questions) that have been solved by other users, and then the best answers of these his- torical questions will be used to answer the users’ queried questions. However, question retrieval is challenging partly due to the word ambiguity and word mismatch between the queried questions and the historical questions in the archives. Word ambiguity often causes the retrieval models to re- trieve many historical questions that do not match the users’ intent. This problem is also amplified by the high diversity of questions and users. For example, depending on different users, the word “interest” may refer to “curiosity”, or “a charge for borrowing money”. Another challenge is word mismatch between the queried questions and the historical questions. The queried questions may contain words that are …

Content truncated to 500 words