Likemind

Inspiration

Think of any event we go to or any place we go to. There are hundreds and thousands of people. How can we find out who we should connect with or if there are any people worth connecting nearby? We all have that problem. Likemind solves that problem.

How it works

• Given your location, we find nearby tweets and the users who tweeted them
• We generate a user's personality using IBM's Personality Insight API on user tweets
• We generate a user's interests using Alchemy's keyword extraction API on user tweets and profile
• Then we match you against each of the nearby users on personality match and interest match scores

Challenges we ran into

• Speed is an issue, since we have to collect users nearby, then collect tweets for those users on real-time.
• Twitter also has very rigid restrictions on usage. We ended up using ~150 tweets per user for this demo to solve both the speed and twitter restriction problem.
• Automatically extracting user interests from twitter profile is also challenging. We tried a few of Alchemy API's services, but the problem was they were not very good for tweets. Then we used the keyword service, which returns some garbage, but we did some additional filtering and it ended up being very useful.
• iOS is a difficult platform to prototype on

Accomplishments that we're proud of

We successfully managed to overcome the challenges of collecting nearby user profiles and finding matches pretty well. We had many exciting feature ideas, we managed very well to define the MVP and execute it in time.

What we learned

We are still learning :-)

What's next for Likemind

We got very lucky ourself by finding like minded partners for this hackathon :-) We would see if we can continue working on this afterwards.

Built With

• alchemyapi
• bluemix
• ibm-personality-insight
• ibm-watson
• ios
• keyword-extraction
• python
• twitter