Evaluating and predicting answer quality in community QA

摘要: Question answering (QA) helps one go beyond traditional keywords-based querying and retrieve information in more precise form than given by a document or list of documents. Several community-based QA (CQA) services have emerged allowing seekers pose their need as questions receive answers from fellow users. A question may multiple users the asker community can choose best answer. While thus indicate if he was satisfied with received, there is no clear way evaluating quality that information. We present study to evaluate predict an answer CQA setting. chose Yahoo! Answers such service selected small set questions, each at least five answers. asked Amazon Mechanical Turk workers rate for based on 13 different criteria. Each rated workers. then matched assessments actual asker's rating show criteria we used faithfully match perception furthered our investigation extracting various features answers, who posted them, training number classifiers select using those features. demonstrate high predictability trained models along relative merits prediction. These support argument case CQA, contextual user's profile, be critical predicting content quality.