An CNN-LSTM Attention Approach to Understanding User Query Intent from Online Health Communities

Understanding user query intent is a crucial task to Question-Answering area. With the development of online health services, online health communities generate huge amount of valuable medical Question-Answering data, where user intention can be mined. However, the queries posted by common users have many domain concepts and colloquial expressions, which make the understanding of user intents very difficult. In this paper, we try to find and predict user intent from the realistic medical text queries. A CNN-LSTM attention model is proposed to predict user intents, and an unsupervised clustering method is applied to mine user intent taxonomy. The CNN-LSTM attention model has a CNN encoders and a Bi-LSTM attention encoder. The two encoder can capture both of global semantic expression and local phrase-level information from an original medical text query, which helps the intent prediction. We also utilize extra knowledge like part-of-speech tags and named entity tags to enrich feature information. Based on the experiments on a health community query intent(HCQI) dataset, we compare our model with baseline models and experiment results demonstrate the effectiveness of our model.