twivent

Inspiration

trying to figure out up-to-date info on the "mbta" by following the official mbta account yielded tweets about routes, lines, trains, or issues I wasn't interested in. I needed a way to get info on only the transportation stations or routes that pertained to me.

What it does

we aggregate tweets by points of interests using regex/nlp/ai vision technologies. we then let you subscribe to those points of interest on a web application to get notified via web push notification when something of significance happens at the location.

How we built it

just started hacking on my commute using my mobile device (android/tmux/emacs)

Challenges we ran into

• tweets about a location in the right context (polysemous; nouns used as verbs; etc)
• handling of deleted tweets
• what constitutes a tweet that should be highlighted (∴ send a push notifications to all subscribers to the point of interests tagged in the tweet)
• nsfw content

Accomplishments that we're proud of

we are in production with an appropriate domain (twivent.com) and we have engagement

What we learned

• anatomy of a tweet
• 1% of the firehose is good enough
• experience filtering nsfw content on twitter

What's next for twivent

capturing dates & time of upcomimg events in tweets and displaying them as an event calendar

Built With

• clojure
• emacs
• node.js