PsyChatbot: A Psychological Counseling Agent Towards Depressed Chinese Population Based on Cognitive Behavioural Therapy

Abstract

Nowadays, depression has been widely concerned due to the growing depressed population. Depression is a global mental problem, the worst case of which can lead to suicide. However, factors such as high treatment costs and social stigma prevent people from obtaining effective treatments. Chatbot technology is one of the main attempts to solve the problem. But as far as we know, existing chatbot systems designed for depressed people are still sporadic, and most of them have some non-negligible limitations. Specifically, existing systems simply guide users to release their negative emotions or provide some general advice. They cannot offer personalized advice for users’ specific problems. In addition, most of them only support English speakers, despite the fact that depressed Chinese constitute a large population. Psychological counseling systems for the depressed Chinese population with improved responsiveness are temporarily lacking. As an attempt to fill in the research gap to some extent, we design a novel Chinese psychological chatbot system, namely PsyChatbot. First, we establish a counseling dialogue framework based on Cognitive Behavioral Therapy (CBT), which guides users to reflect on themselves and helps them discover their negative perceptions. Then, we propose a retrieval-based Q&A algorithm to provide suitable suggestions for users’ specific problems. Last but not least, we construct a large-scale Chinese counseling Q&A corpus, which contains nearly 89,000 psychological Q&A triples. Experimental results have demonstrated the effectiveness of PsyChatbot. The source code and data has been released at https://github.com/slptongji/PsyChatbot.