MindIntent: A Chatbot Using SVM for Identifying Mental Health Intervention Intentions

Abstract

Mental health issues are increasingly acknowledged as vital, yet traditional support systems have limitations in availability, cost, and accessibility. Chatbots provide a scalable solution for real-time support, but their usefulness is dependent on correctly understanding user input. This work uses a Support Vector Machine (SVM) to improve chatbot performance in mental health applications by effectively dealing with high-dimensional data and discriminating between emotional states and intents. We created a chatbot by gathering and annotating text data, preprocessing it, and extracting features with tools like TF-IDF and word embeddings. The SVM model was trained on this data and implemented into the chatbot, allowing for real-time classification of user inputs and contextually relevant responses. Our results show that the SVM-powered chatbot excels at accurately categorizing users.