Click here to view the project in GitHub

Project Overview

This project involved the development of a sophisticated tool capable of reading and responding to queries with answers directly extracted from GitHub's FAQ section. This was achieved through the utilization of Beautiful Soup, a Python library, which was used to scrape a number of question and answer examples from GitHub's FAQ Page. The project further involved engineering new variations of questions using both context and keywords found within the questions and answers. A three-layer linear Neural Network was then implemented to construct a model capable of interpreting and responding to similar questions accurately. The final piece of this project involved developing a client-facing application using Flask. This application was seamlessly integrated with HTML, CSS, and JavaScript to create a user-friendly interface that effectively showcases the functionalities of the tool.

Languages Used

  • Python
  • HTML
  • CSS
  • JavaScript

Tools/Packages Used

  • BeautifulSoup
  • Flask
  • Pytorch
  • NLTK

A Little Reflection

This was my first project outside of a computer science course during undergrad that I really poured hours into. The inspiration for this project came from my final project in my neural networks course, which was meant to be a chatbot that could differentiate between emotions based on keyword context. For this interpretation, I decided to create a more business centered approach with a helpful virtual assistant tool. This was my first time using so many powerful packages and I am really happy with the final outcome because I know that with a few tweaks, it can become something truly useful. I got the opportunity to explore web scraping and data wrangling for the first time, so I have great exposure to all parts of the data pipeline. I also got to host an application with Flask for the first time, as well as create a frontend with HTML, CSS, and JavaScript. It was so much fun to go from scraping raw data off of a static webpage to a legitamate, working final product. Feel free to check out the project on GitHub, follow the download and run instruction in the ReadME file to try it out for yourself!

Client Preview