User Name Disambiguation in Community Question Answering

Proceedings of Recent Advances in Natural Language Processing, pages 707–713, Hissar, Bulgaria, Sep 7–9 2015. User Name Disambiguation in Community Question Answering Baoguo Yang Department of Computer Science University of York, UK by550@york.ac.uk Suresh Manandhar Department of Computer Science University of York, UK suresh@cs.york.ac.uk Abstract Community question answering sites pro- vide us convenient and interactive plat- forms for problem solving and knowl- edge sharing, which are attracting an in- creasing number of users. Accordingly, it will be very common that different people have the same user name. When a query question is given, some potential answer providers would be recommended to the asker in the form of user name. Howev- er, some user names are ambiguous and not unique in the community. To help question askers match the ambiguous user names with the right people, in this paper, we propose to disambiguate same-name users by ranking their tag-based relevance to a query question. Empirical studies on three community question answering datasets demonstrate that our method is ef- fective for disambiguating user names in community question answering. 1 Introduction In recent years, community-based question an- swering (CQA) sites like StackOverflow1, Quo- ra2 and Yahoo!Answers3, have achieved great suc- cess and attracted a huge number of users. It is not uncommon that some people in the CQA ser- vices share the same user names. Figure 1(a), Fig- ure 1(b) and Figure 1(c) show three lists of us- er names from three different CQA communities: Travel4, Webapps (Web Applications)5, and Cook- ing6, where each user name is shared by multiple 1http://www.stackoverflow.com/ 2https://www.quora.com/ 3http://answers.yahoo.com/ 4http://travel.stackexchange.com/ 5http://webapps.stackexchange.com/ 6http://cooking.stackexchange.com/ users. In Figure 1(b), “David” is the most common and ambiguous user name related to 57 users. In some cases, an off-line person asks people around a difficult question verbally, then he/she may be recommended by word of mouth to vis- it the CQA homepages of some potential answer providers. However, the links to their homepages are not provided sometimes, then the asker has to search them according to the provided user names. Some user names are unique, and they can easily access the historical QA records of these potential answer providers. However, some are very com- mon and ambiguous, accordingly, many users with the same user name will be displayed. Motivated by the above scenario, it is very nec- essary to help askers disambiguate these users, which can release them from wondering which us- er should be the right one. Moreover, if the user name is not clearly given, the askers will waste a lot of valuable time on searching and visiting ir- relevant users, which can cause misunderstanding and misleading. Then the asker will get puzzled. In CQA, given a new question, the related re- search studies mainly fall into three areas: 1) An- swer recommendation (Zhou et al., 2012b; Tian et al., 2013); 2) Similar question retrieval (Cao et al., 2010; Zhang et al., 2014b); 3) Expert user rec- ommendation (Pal and Konstan, 2010; Liu et al., 2011; Zhou et al., 2012a). As for …

Content truncated to 500 words