Cultivating Online: Question Routing in a Question and Answering Community for Agriculture

Abstract: Community-based Question and Answering (CQA) platforms are nowadays enlightening over a billion people with crowdsourced knowledge. A key design issue in CQA platforms is how to find the potential answerers and to provide the askers timely and suitable answers, i.e., the so-called \textit{question routing} problem. State-of-art approaches often rely on extracting topics from the question texts. In this work, we analyze the question routing problem in a CQA system named Farm-Doctor that is exclusive for agricultural knowledge. The major challenge is that its questions contain limited textual information. To this end, we conduct an extensive measurement and obtain the whole knowledge repository of Farm-Doctor that consists of over 690 thousand questions and over 3 million answers. To remedy the text deficiency, we model Farm-Doctor as a heterogeneous information network that incorporates rich side information and based on network representation learning models we accurately recommend for each question the users that are highly likely to answer it. With an average income of fewer than 6 dollars a day, over 300 thousands farmers in China seek online in Farm-Doctor for agricultural advices. Our method helps these less eloquent farmers with their cultivation and hopefully provides a way to improve their lives.