Inspiration
Recent hype around Cyptocurrency and NFTs can create potential for profit, considering the value of items is correlated heavily with the momentum of a given collection
What it does
--incomplete-- scrapes Twitter using its API to gather data on tweets over a recent timeframe. tweets filtered to only include those on the collections specified. Tweets are analysed and given a sentiment ranking, which is compared to the sentiment ranking of the collection recorded over a previous timeframe. An increase in sentiment over time, hence growing momentum, indicates a suggested buy, whereas a steady increase reccomends a sell of a collection.
How we built it
Backend using Python, with some HTML and Javascript. AI written heavily using the TensorFlow and numpy libraries, with data from Twitter scraped from its API.
Challenges we ran into
Surprisingly, scraping data from Twitter. Many unforeseen errors and rate-limiting meant consistently taking large volumes of data was more difficult than expected. Hence, while the AI structure was completed, no training nor test data could be run on it, leaving an unoptimized but proof-of-concept structure. Also, two first-year students with a project just too large. Backend is largely functional but Frontend incomplete due to lack of time (and fatigue)
Accomplishments that we're proud of
Creating a AI, using the TensorFlow library. Creating a proof-of-concept project which could work with a little more time and data scraping. Modelling a relational database and having a very organised and modular project structure, which makes code legible and maintainable.
What we learned
Don't bite off more you can chew. Get test data early. Get API keys ahead of time, plan ideas instead of thinking of a project idea on the spot. Get someone who actually likes Frontend.
What's next for NFT-Sentiment
Completing the frontend - adding support for scraping other social media, adding an NFT watchlist for users, better modelling and investment advice algorithms.
Log in or sign up for Devpost to join the conversation.