Skip to content

PandiaJason/AI-Chatbot-using-Deep-Learning-based-NLP

Repository files navigation

AI Chatbot using Deep Learning based NLP

  • 🌐 Pioneered the concept of Gen AI with BERT and Transformers in 2019.
  • 🤖 Built both open-domain and closed-domain chatbots utilizing DuckDuckGo API, Wikipedia API, and Custom dataset.
  • 📊 Organized a 70-year NLP survey up to 2020, breathing life into the concept of chatbots with Python and NgRok-based web app using Google Colab with DNN models.
  • 🌐 Tested it with a wide range of chatbots like OK Google, Amazon Assistant, and others, as detailed in my full thesis.
  • 🏆 Awarded first place for this groundbreaking project.

image

Overall Concept

  • The proposed system aims to provide precise and selective answers from the web, eliminating the need for users to filter or parse through extensive search results.
  • The system is capable of handling open-domain questions, making it predict answers almost every time. Privacy is a significant focus in open-domain search, utilizing DuckDuckGo and Wikipedia APIs for search.
  • It is easily implementable and customizable for specific tasks within a particular application domain. The system is cost-effective and well-suited for use in any browser. Additionally, the system can provide chit-chat answers, reducing traditional knowledge gathering/learning time.
  • Employing Deep Learning (DL)-based NLP technology helps minimize bot misunderstandings, resulting in more specific answers compared to traditional systems.

Notice: This project was conceived before the Gen AI era. I have made the code and documentation completely available under the GPL license. Some of the DL and NLP modules may have been updated in the past 3-4 years, and as a result, the AI Chatbot might not work. However, this project marked the beginning of conceptualizing BERT-like transformers for Gen AI tasks. I am proud to have pioneered this concept and was awarded first place for my B.E. thesis.

For More information Check the thesis

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published