Spatial + Time + Sentiment Analysis on Realtime Geo-located TweetsRedlands, CA

  • 12 Commits
  • 11 Pushes
  • 8 Deploys
Launch Site

Spatial + Time + Sentiment Analysis on Realtime Geo-located Tweets

By Here for Beer

Screencast
Quick Intro

Spatial + Time + Sentiment Analysis on Realtime geo-located tweets. App lets you Filter the tweets by specific terms or shows everything by default.

Description

Spatial + Time + Sentiment Analysis on Realtime tweets. Filter by terms or everything (default). Its fascinating to watch over time how people tweet about talk and think about specific things (e.g. game, politics, movies, issues etc..) over different time zones and around the world. This app/concept is just a sample, lot more functionality both spatial and analytics can be added for more fine grained targeting and understanding of the twitter-o-sphere.

Note: This is not a design demo. I do not have any designers/UI team members.

Highlights:

  • Realtime twitter stream (Free hose provides upto ~700 tweets per minute)
  • Realtime sentiment analysis. Positive and Negative sentiments as defined by a dictionary.
  • Realtime time normalization displayed via charts. So you can find how many tweets was tweeted at 9pm worldwide. This is done based on lat,long of the tweet to find the local timezone and then normalized for UTC and rounded to the nearest hour.
  • Solar Terminator layer for the map view, which shows the current day/night around the world. This affects to traffic per hour.
  • Map display data as points (color coded by sentiment) or Clusters or Heatmap.
  • Websockets delivery for each client for near realtime experience
  • No persistence as per Twitter API rules, all done in memory on Node.js
Judging Instructions

Please use a modern browser which support sockets and canvas for best results. Might not work well in mobile devices.

What they Used

Node.js, Twitter data, Esri Maps API, Mapbox Tiles, jQuery, Google charts and awesome NPM modules (Socket.io, Tuiter, Speakeasy NLP, Express, Moment-timezone, etc...)

Who

Votes

Your Vote

Voting is now closed.

Other Votes

  • (6)
  • contestant

    おもしろい。 

  • judge

    Bugsnag

    Hey Sathya,

    This looks really awesome., I like the idea of showing where things are happening and it's pretty powerful to see who's thinking what. I can imagine this being invaluable during elections etc. :).

    It's still a little bit hard to use, as I'm sure you're aware, but overall I think it's a great idea implemented really well.

  • judge

    Khan Academy

    Very impressive! Was fun watching different parts of the world and their activity based upon the time of the day. It'd be interesting to see, specifically, what was coming up in the sentiment analysis (like what people were happy/sad about - and where?). As it stands it's not too useful but it is an interesting display of the data.

  • judge

    innovative idea with good visualization.

  • judge

    Freelancer

    Interesting.

  • contestant

    MeteorHacks

    when we were planning , we considered this idea to develop. we chose ocb instead glad to see this implemented. good work. liked the visualisation.

  • contestant

    This is pretty sweet. I think adding a few more 'drilldowns' and maybe displaying some representative positive/negative tweets would have been cool too.

  • contestant

    Activimetrics LLC

    Nice map display. I wasn't sure what it was I was looking at though. Just numbers of tweets in different areas, but I wasn't sure how the sentiment analysis tied in to that, or why I'd want to do sentiment analysis in this way at all. But very impressive that you got as much done as you did.

  • contestant

    Editora Abril

  • contestant

    Cool concept - needs some polish on the user interaction part. It is hard to know what I am looking at. Specific search terms? All tweets everywhere?

  • contestant

    Not sure about the usefulness. The developer has done a good job though. The visualisation could have been better.

  • contestant

    There are many other apps like this out there; this one was well done. (a bit laggy on my mac though)

  • contestant

    I love the simple & easy to understand idea! Keep up the good work! What I would add next is a way to jump to the tweets and a way to filter them.

    • sathyaprasad
      contestant

      Thank you for voting. Appreciate your comments.

  • contestant
  • contestant

    Uva Wellassa University of Sri lanka

    Nice idea. Had some performance issues when a term is given.

  • contestant

    Cool interface with the map; could very well be a nice and robust product in the future.

    • sathyaprasad
      contestant

      Thank you for voting. Appreciate your comments. Hope so!

  • contestant

    e-conomic

  • contestant

    nice work!

  • contestant

    BrandExtract

    I had some intermittent problems when switching from clusters to points, but it eventually worked. I think a few times I didn't notice the "loading" in the upper right corner which led me to believe nothing was happening. I couldn't give a high score on design or innovation, thought I don't recall seeing real-time sentiment analysis before

    • sathyaprasad
      contestant

      I noticed it too. Too late to make any changes as per rules. Watch the video to get a comprehensive view on how the app should work.

  • contestant

    I struggled a bit with the map zoom and the overall graph design, though I like the general idea of your project. Good job in 48 hours!

  • contestant

    7digital

    Neat idea

  • contestant

    Square

    Points and heatmap don't seem to work. Interesting concept.

    • sathyaprasad
      contestant

      Thank you for voting. Appreciate your comments. Please watch the video to see the complete app.

IMPORTANT DATES

REGISTRATION
SEP 17
COMPETITION
NOV 9-11 UTC
JUDGING
NOV 11-17
WINNERS
NOV 18

Thank you to our Platinum Sponsors