An automated approach for temporal specificity classification of user-generated content

The temporal specificity of user-generated content (UGC) affects the information recommendation accuracy in online question answering (Q&A) community. Exploring the method for the temporal specificity classification of UGC is beneficial to optimize the content organization of online Q&A communities. In assessments of temporal specificity of UGC, it is relatively one-sided to judge only by the post time of the content. The relationship between the content’s topic category and temporal specificity is often ignored. To solve these problems, an automated approach for assessing the temporal specificity of user-generated content is proposed. The temporal specificity of UGC in the online Q&A community was divided into three types: high temporal specificity, medium temporal specificity, and low temporal specificity. The temporal specificity of UGC is automatically classified based on the Word2Vec-XGBoost algorithm, and question text for different topics from the Zhihu Q&A community is collected for experimental verification. The results show that the classification’s accuracy, recall, and F1 score with weighted temporal specificity are increased by 2.63%, 2.67%, and 2.69%, respectively, compared to those in the base case. The overall accuracy, recall, and F1 score reached 90.30%, 90.24%, and 90.12%, respectively, and the temporal specificity classification performance was good.