Sahacharza: Fostering Women’s Wellness via AI-Enhanced Bengali Conversations

Humayra Noureen, Mollika Deb, Enamul Hassan · 2023

Chatbots are intellectual computers being developed using Artificial Intelligence and Natural Language Processing algorithms. They are programmed to interpret and respond to user queries, making them increasingly popular in various industries. However, chatbots can also improve people's lives, including women's health. To address this issue, we have presented a Bengali women's healthcare chatbot that can respond to female reproductive organ disorders. In this study, the chatbot is capable of classifying women's reproductive diseases based on given symptoms, answering common questions about symptoms, and providing information about medical conditions. This system is implemented using Keras Sequential Model and Stochastic Gradient Descent. It is tailored to Bengali females and the ultimate goal is to promote healthy living for women in our community. To choose the most relevant response, the chatbot uses a retrieval-based approach and a number of techniques, including keyword matching, disease categorization using supervised algorithms, and deep learning. Specifically, we have utilized LSTM, a recurrent neural network, for the classification of the user-inputted message. We have also developed an unsupervised spelling correction technique based on word2vec for minimizing input errors to increase the accuracy of the chatbot. By providing a free-of-cost web application, women can access this chatbot wherever they are, making it a convenient tool for health-related queries.

Read the paper · More papers on PaperTik