情报科学

• • 上一篇    下一篇

基于语义网络的社会化问答社区答案聚合与排序研究


  

  1. 1.吉林大学管理学院,吉林长春130022;2. 吉林大学信息资源研究中心,吉林长春  130022

Research on Answer Aggregation and Sorting of Social Q&A Community Based on Semantic Network

#br#   

  1. 1.School of ManagementJilin UniversityChangchun 130022,China2.Research Center of Information Resources of Jilin UniversityChangchun 130022,China

摘要:

【目的/意义】旨在将社会化问答社区中碎片化的答案关联起来,并为用户提供不同主题的高质量答案和更好的知识服务。【方法/过程】首先,本研究利用Doc2vec算法计算答案之间的语义相似度,并构建答案语义网络。其次,利用Louvain算法对答案语义网络进行社区划分,并用TextRank算法抽取各个主题下文档的关键词,使用词云对每个主题进行可视化展示。最后,利用PageRank算法对聚类后的答案语义网络进行排序,从而实现答案文档的主题聚合和排序。【结果/结论】本研究使用“知乎”上的问答数据进行了实证研究。结果表明,所提出的答案聚合和排序方法不仅能够向用户直观地展示答案之间的关联强度和各个主题答案的主要内容,还能够为用户提供分主题的答案排序结果,自动为用户筛选高质量的答案。【创新/局限】创新性地提出了答案语义网络,并基于与答案语义网络,提出了一种集聚合、主题可视化和排序于一体的答案知识组织方法。

关键词:

社会化问答社区, 答案聚合, 答案排序, 语义网络, Doc2vec算法

Abstract:

Purpose/significanceAims to connect the fragmented answers in the social Q&A community and provide users with high-quality answers on different topics and better knowledge services.【Method/process】First, we uses the Doc2vec algorithm to calculate the semantic similarity between answers and constructs the answer semantic network. Secondly, the Louvain algorithm is used to divide the answer semantic network, and the TextRank algorithm is used to extract the keywords of the documents under each topic, and the word cloud is used to visually display each topic. Finally, the PageRank algorithm is used to sort the answer semantic network after clustering, so as to realize the topic aggregation and sorting of answer documents.【Results/ConclusionsWe  conducted an empirical study using the question and answer data on “Zhihu”. The results show that the proposed answer aggregation and ranking method can not only visually show users the correlation strength between answers and the main content of each topic answer, but also provide users with topic-based answer ranking results, and automatically filter high-quality answers for users.【Innovation/limitations】An innovative answer semantic network is proposed, and based on the answer semantic network, we propose an answer knowledge organization method integrating aggregation, topic visualization and ranking.

Key words:

text-indent:0.0000pt, "> social Q&, A , community, answer aggregation, answer ranking, semantic network, Doc2vec algorithm