| Summary: | To provide better socio-feedback for the purpose of helping people conduct better conversation and human interaction activities, understanding the conversation context and topic has become a crucial task. This project concentrates on developing a Machine Learning system that tracks the context of conversation by speech recognition, natural language processing and text classification. Throughout the process, a Naïve Bayes Classification model is built and its performance is improved gradually through different methods in each stage. At the end, the classification model is able to classify the conversation into “Business meeting”, “Court”, “Sports Chatting” and “Restaurant” contexts with an overall accuracy of 96.3%.
|
|---|