Batch recommendation of experts to questions in community-based question-answering with a sailfish optimizer

Abstract

To facilitate question-answering in community-based question-answering (CQA), this paper proposes an approach for the batch recommendation of answerers by optimizing the utilization of expert resources. First, questions and experts are modeled with a biterm topic model (BTM). Next, the answered questions are clustered based on a novel discrete sailfish optimizer (SFO) with a genetic algorithm (GA), and the topic distribution is obtained. Then, experts are ranked in each cluster based on activeness, recency, and professionalism. Considering the limited number of experts, to ensure that core questions are answered and to avoid repeated answers to similar or duplicate questions, coverage, answerability and the consumption of expert resources are taken as objects to be optimized. This scenario is formulated as a multiobjective optimization problem and is addressed by the proposed novel binary multiobjective SFO (MOSFO) with a GA. The solution of the model includes not only the selected questions to be answered but also the matching between the questions and experts. The proposed approach is evaluated with a real dataset, and the experimental results show that the proposed approach is feasible and has superior performance to the question-priority method, the expert-priority method and other swarm intelligence (SI) methods. This study is the first to make batch recommendations, providing a new idea and extending research on expert recommendation. Additionally, the approach can be used practically to improve the satisfaction of the knowledge needs of users by improving the answerability of high-coverage questions.