This is a Bachelor Project Thesis done at the Barcelona School of Informatics (FIB) for the Computer Science Degree with a specialization in Computing.
This project attempts to bridge the huge gap between people who struggle with mental health and people who actually get treated with the design and implementation of Kai, an AI-powered Chatbot that supports Cognitive Behavioral Therapy (CBT).
CBT is based on identifying cognitive distortions (negative thoughts) and challenging them to improve mood and overall mental health. This was done by using Text Classification, a Natural Language Processing (NLP) technique to identify potential cognitive distortions in text.
During the project, a model experimentation was done to compare different super- vised machine learning models in order to choose the best one for the text classification. The dataset needed to train the models was generated by manually giving labelled ex- amples of the 15 major cognitive distortions. The model experimentation was entirely done in python.
Furthermore, the design of the dialogue flow of the chatbot was done following the CBT’s guidelines and the implementation of the chatbot was done in Python using the TKinter in Framework for the interface. Finally, two test were made to check for the correct functioning of the chatbot.
To run the code, execute chatbot.ipynb. To test the chatbot you can use the script.txt.
- Creative Commons - Used to choose the license
- Mariama C. Djalo D.
This project is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License