Inspiration

As a student, the concept of 'notes' is still a big part of our life even in this digital age. The number of books, digital documents is just increasing day by day. I always want to read notes/books, but they are just too big :( Last minute revisions are hectic just because of the number of pages and notes spread across physical and digital realm.

That's where our idea comes to rescue!
3 words - Summarize, Find and Revise through chatbot!

What it does

Our system does 3 things:

1. Grabs documents on your smartphones (Images/PDFs/Docs/PPTs) - User uploads documents or sets an automatic service to all capture of all documents (Database, user management on Firebase)
2. Extract raw text from these documents - Google OCR, NLP APIs for extracting raw data and getting text summary
3. NLP on the text to extract summary and create questions - Pass the NLP raw data gathered from APIs through our algorithm to generate questions and quizzes
4. Chatbot interface to help you learn - Ask questions to Alexa related to content of your books/notes, get summary about these books from Alexa, Ask Alexa to Quiz you based on content of the documents.

How we built it

  1. Android app, Firebase Storage, Firebase Database, Firebase Authentication for user and documents management.
  2. Google Cloud platform [http://hackny-218012.appspot.com/data/:docId], Google NLP APIs, Google Vision APIs to extract/translate/clean text in documents
  3. Javascript based algorithm implementation for generating questions from text
  4. Alexa interface for accessing the data points generated -> Alexa can find summary, search for queries (Search content of documents), Ask questions in form of Quizes

Challenges we ran into

  1. The hardest part was figuring out integration with Alexa (As we dont have lot of experience in Alexa development). The nature of Alexa platform made it difficult to host our quiz on Alexa. The quiz part of Alexa still needs works and fixes.
  2. Creating a simple but effective NLPish way to generate questions from text, given the time limit of 22 hours for the hackathon

Accomplishments that we're proud of

  1. End to end working prototype (though there are bugs)
  2. Working as a team of 2 yet working on a broad range of tools and platforms

What's next for docX

  1. Making the Alexa interface better
  2. Adding a sturdy NLP engine for generating questions

Built With

Share this project:

Updates