Resourceful

Take control of your future with Resourceful

Table of Contents

• About
  • Inspiration
  • What
  • How
  • Challenges
  • Accomplishments
  • Learnings
• What's Next

About Resourceful

Inspiration

• We were inspired to make this project after thinking about ways to level the playing field between those in poor and rich school districts

What it Does

• Takes in a list of keywords (interests and skills)
• Outputs a list of URLs and summaries of the webpages deemed most similar to the user’s interests
• Calculates Wu-Palmer Similarity and Levenshtein Distances to assess how similar a website is to the wants of the user
• Uses Beautiful Soup in order to get a summary of the activity (generally) as well as the time it takes to complete the course and who the course is offered by for Coursera

How We Built it

• We built the project using Python, Beautiful Soup, NLTK, HMNI, FuzzyWuzzy, and Flutter

Challenges we ran into

• We had trouble in implementing the code that used Levenshtein distances and Palmer similarity, leading to words that had nothing to do with each other being assigned high similarity values.
• We also had trouble with the UI as the version of TensorFlow that HMNI uses was incompatible with mac(the system that our frontend member was using)

Accomplishments that we're proud of

• Each member was working with tools they were not familiar with, yet we still completed the product
• We were able to overcome an issue where TensorFlow would not work with macOS
• The UI is intuitive and minimalistic

What we learned

• We learned a lot about web scraping, website interactions, and measuring abstract concepts like context-empowered text similarity

What's Next for Resourceful

• In Progress: Using multithreading to run Selenium search processes simultaneously
  • Subsequently running BeautifulSoup sub processes simultaneously
  • Effect: Greatly reduce runtime
• Use of Machine Learning to more accurately find the description of any activity no matter the website
• Use of Machine Learning to compare user tags for skills and interests through categorizing them by field or topic and then choosing the more specific one
  • ML model will be a “Bag-of-words” model → Train using words correlated to category
  • Ex. “python” in skills replaces “coding language” in interests for the combined skills-interests array
• Main Goal: Host Flask API online on services like AWS and deploy flutter app to iOS, Android, Web, and Windows