A Seq2seq based Chatbot for Dyslexic Patients

About the project:

Dyslexia is a disorder in the nervous system of the brain that affects a person’s learning abilities. The person suffering from dyslexia has difficulties in reading, writing, and spelling.

The person faces problems in identifying speech sounds and how to decode those sounds, not because of a lack of hearing, intelligence, or vision. It is estimated that dyslexia is present in nearly 7% of school-aged children. Children with dyslexia face difficulty learning and an inability to perform academically, which might culminate in frustration and social anxiety. After analysing all of these issues, we have devised a solution that involves presenting people with a chatbot so that they can interact with it and learn more about it. Our objective here is to build a chatbot for dyslexic patients and their family members and provide them with adequate information. We have created some of our datasets by conducting extensive online research and using information from websites like Chat-GPT and other reliable sites. In this research, we have focused on a generative model called the sequence-to-sequence model. The main focus of creating a chatbot is to stimulate the conversion between the AI and the user. Our model performed extremely well on both the training and validation sets, with training accuracy of 98.12% and validation accuracy of 98.25%.

Group Members:

  1. Kukil Bharadwaj – 190310007024
  2. Manash Kar – 190310007027
  3. Sazzadur Rahman – 190310007043

Project Guide:

  • Mr. Adarsh Pradhan, Assistant Professor, CSE

Screenshot of the UI:

GDPR