Abstract

Technology has opened up a new era of mental health care and information gathering, providing new means to access aid, connect to a peer counselor, and track progress. Chezlong is an Arabic text-based chatbot powered by the most modern NLP techniques. It aims to give guidance to those who need quick mental support. The development cycle is based on using the GPT-3.5 Large Language Model for Retrieval Augmented Generation (RAG). RAG is used on Arabic psychological articles, originally scrapped from the nafsy.net website. Cleaning, preprocessing, and topic modeling have been applied to prepare and explore the dataset. The chunked dataset is stored as vectors in Pinecone. Similarity search is done to retrieve query-related source knowledge from the dataset to answer a user query.