社区问答网站中问题的刻面组织方法——以知乎网站为例

摘要

[Purpose/Significance]Community Question Answering (CQA) websites have become mainstream platforms for online knowledge sharing. Currently,most CQA websites adopt the topic-based organization to manage questions,which hardly satisfies users' need of fine-grained retrieval and positioning requirements.[Method/Process]This paper proposes a faceted organization method to man-age questions in CQA websites by applying facet information of topics into organization. The method consists of two processes,online process and offline process. The offline process generates facet structure of topics by using Contents of Wikipedia webpage and similarity of facets among related topics. While the online process constructs the mapping relations between topics and facets by using transfer learn-ing. We can organize the questions into hierarchical structure of topic-facet-question through the two processes.[Result/Conclusion]Ex-periments on questions of Zhihu website related to the knowledge topics of data structure were conducted,and the effectiveness of the proposed faceted organization method was verified.%[目的/意义]社区问答(Community Question Answering,CQA)已成为当前主流的在线知识共享方式.目前,大多数CQA网站都采用了基于主题的问题组织方式.这类组织方式很难满足用户细粒度的问题检索与定位需求.[方法/过程]将主题的刻面(Facet)信息引入问题组织,提出了一种针对CQA网站中问题的刻面组织方法.该方法包括离线与在线两个过程,离线过程利用Wikipedia中的目录信息与相关主题间的刻面相似性生成特定主题的刻面结构;在线过程则利用迁移学习技术建立主题与刻面的映射关系.通过这两个过程,将问题组织成"主题-刻面-问题"层次结构.[结果/结论]以知乎网站中"数据结构"领域相关知识主题的问题为对象进行了实验,验证了所提刻面组织方法的有效性.