Mindful is a web-based mental health interactive chatbot that performs a comprehensive 5W1H (What, Who, When, Where, Why, How) extraction on 1:1 conversations with its users.
Through a clean and simple design, Mindful provides a space for therapy dialogues tailored to individual needs. With a unique approach, it encourages users to explore their emotions in the moment, free from the burden of conversation history.
Mindful elevates user interactions through a sophisticated 5W1H analysis, unraveling the Who?, What?, When?, Where?, Why?, and How? of each conversation. This deep dive provides profound insights into the user's emotional landscape, fostering a personalized and empathetic experience.
To achieve this, we employed advanced natural language processing techniques, including custom Named Entity Recognition (NER), Sentiment Analysis, Topic Modeling (TF-IDF) & Association, Non-negative Matrix Factorization and Dependency Parsing. These intelligent methods were crucial for accurately dissecting complex, nuanced conversations and extracting relevant information.
Note: This repository openly documents the development of the 5W1H Extraction model used by Mindful. This is part of an ongoing collaborative project, whose source code is currently private.