Understanding and Summarizing Answers in Community-Based Question Answering Services

Proceedings of the 22nd International Conference on Computational Linguistics (Coling 2008), pages 497–504 Manchester, August 2008 Understanding and Summarizing Answers in Community-Based Question Answering Services Yuanjie Liu1, Shasha Li2, Yunbo Cao1,3, Chin-Yew Lin3, Dingyi Han1, Yong Yu1 1Shanghai Jiao Tong University, Shanghai, China, 200240 {lyjgeorge,handy,yyu} @apex.sjtu.edu.cn 2National University of Defense Technology, Changsha, China, 410074 Shashali @nudt.edu.cn 3Microsoft Research Asia, Beijing, China, 100080 {yunbo.cao,cyl} @microsoft.com Abstract Community-based question answering (cQA) services have accumulated millions of questions and their answers over time. In the process of accumulation, cQA ser- vices assume that questions always have unique best answers. However, with an in- depth analysis of questions and answers on cQA services, we find that the assump- tion cannot be true. According to the anal- ysis, at least 78% of the cQA best answers are reusable when similar questions are asked again, but no more than 48% of them are indeed the unique best answers. We conduct the analysis by proposing taxonomies for cQA questions and an- swers. To better reuse the cQA content, we also propose applying automatic sum- marization techniques to summarize an- swers. Our results show that question-type oriented summarization techniques can improve cQA answer quality significantly. 1 Introduction Community-based question and answering (cQA) service is becoming a popular type of search re- lated activity. Major search engines around the world have rolled out their own versions of cQA service. Yahoo! Answers, Baidu Zhidao, and Naver Ji-Sik-In1 are some examples. In general, a cQA service has the following workflow. First, a question is posted by the asker in a cQA service and then people in the commu- nity can answer the question. After enough num- ber of answers are collected, a best answer can © 2008. Licensed under the Creative Commons Attri- bution- Noncommercial-Share Alike 3.0 Unported license (http://creativecommons.org/licenses/by-nc- sa/3.0/). Some rights reserved. be chosen by the asker or voted by the communi- ty. The resulting question and answer archives are large knowledge repositories and can be used to complement online search. For example, Nav- er’s Ji-Sik-In (Knowledge iN) has accumulated about 70 million entries2. In an ideal scenario, a search engine can serve similar questions or use best answers as search result snippets when similar queries are submit- ted. To support such applications, we have to assume the best answers from cQA services are good and relevant answers for their pairing ques- tions. However, the assumption might not be true as exemplified by the following examples. Question Title Which actress has the most seductive voice?..could range from a giggly goldie hawn..to a sultry anne bancroft? Question Description or any other type of voice that you find allur- ing. .. Best Answer (Polls & Surveys) Fenella Fielding, wow!!!! Best Answer (Movies) i think joanna lumlley has a really sexy voice Table 1. Same Question / Different Best Answers Question Title Does anyone know of any birthdays coming up soon? Question Description Celerities, people you know, you? Anyway I need the name and the date. If you want to know it …

Content truncated to 500 words