TweetSense

Inspiration

I was always really interested in AI and using it to analyze people, as I believe we are a lot more straightforward as humans than we may think. I wanted to design a tool that could apply the amazing technology of AI in a useful that was convenient, i.e. mobile devices.

What it does

It parses tweets from either a specific user or about a given topic and uses an artificial intelligence backend to analyze the poster of the tweets in several different areas. I chose to do both specific users and general topics because the code was pretty similar for both but the implications are very different.

When it comes to analyzing specific people, I was very interested in seeing their personality traits and sentiment through their 280 character tweets. I'm fascinated by the idea that these small tidbits of our life that we share online could be used by a machine to find out who exactly you are.

Using this same technology, I added the functionality of analyzing topics on twitter. A topic can literally be anything from a celebrity to something as simple as "dogs." The implications here are completely different than in the other use case, using this app to analyze topics can allow one to view how the general public feels about this thing and how they talk about it. What I find the most compelling about this app is just how fast it is, it takes under 30 seconds to parse thousands of tweets and analyze them to a high statistical significance.

How I built it

iOS app for the front end. Google Cloud for a web scraping api to get a user's timeline tweets (twitter's API is clunky when it comes to parsing one given user's timeline). Google Cloud for sentiment analysis. IBM's Watson for personality, tone and emotion analysis.

Challenges I ran into

I've never really done anything related to web development, so creating my own rest api and deploying it was a huge step for me. Although the api that parses tweets may not be the most compelling technological achievement of this project, it certainly was a difficult one for me to conquer.
Finishing this Devpost, I was coding all night until like 10 minutes ago.
A lot of the smaller UI features I wanted to implement took too long to do so I could not add as many animations as I would have liked to.