Answer Quality Evaluation of Social Q&A Platforms Based on Feature Filtering
Weiyi Ye, Yizhou Chen, Gangfeng Ma, Xu-Hua Yang · 2021
At present, social Q&A platforms mainly evaluate and rank the quality of answers based on the user voting mechanism, and lack an objective answer quality evaluation mechanism. This leads to the behavior of some users can let some low-quality answers be ranked higher, which affects users' access to relevant knowledge. Based on this phenomenon, our study proposes an answer quality evaluation model based on a text feature filtering mechanism. It distinguishes high-quality answers from general answers by quantifying various text features, ranking the importance of those text features, and assigning weights to them. The experimental results from 49758 answers to the Zhihu website demonstrate that the proposed model can be more accurately filter and predict high-quality answers to users than the state-of-the-art model.