Keepr presentation at MIT CMSW

512 views
407 views

Published on

audio and live blog notes http://cmsw.mit.edu/liveblog-hong-qu-keepr/

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
512
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
5
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Keepr presentation at MIT CMSW

  1. 1. Keepr Algorithm for Extracting Entities, Eyewitnesses and Amplifiers Hong Qu September 19, 2013
  2. 2. http://twitpic.com/135xa https://twitter. com/jkrums/status/1121915133
  3. 3. http://www.telegraph.co.uk/technology/twitter/4269765/New-York-plane-crash-Twitter-breaks-the-news-again.html
  4. 4. http://www2.sims.berkeley.edu/courses/is290-2/f04/sched.html
  5. 5. https://twitter.com/hqu/status/1745171763
  6. 6. Humans vs Machines Pattern RecognitionValue Judgment
  7. 7. Computers count really, really fast!
  8. 8. What is a Tweet? 140 characters: ● words ● @user mentions ● #hashtags ● links
  9. 9. Hows does Keepr process tweets? 140 character * 100 tweets = 14,000 characters ❏ Parse it ❏ Count it ❏ Visualize it ❏ Zoom in ❏ Archive it
  10. 10. https://canvas.instructure.com/courses/812708/assignments/syllabus Natural Language Processing
  11. 11. Humans are way better at at making value judgements and telling stories
  12. 12. Social media and the Boston bombings: When citizens and journalists cover the same story
  13. 13. Keepr’s Algorithm ➔ Entity extraction ◆ Topics ➔ Media extraction ◆ images, videos ➔ Link expansion ◆ articles ➔ Conversation analysis ◆ @ mentions ◆ source discovery ◆ amplification velocity ➔ Source verification ◆ geo-location ◆ social media profiles
  14. 14. Topic Extraction by Term Frequency http://trimc-nlp.blogspot.com/2013/04/tfidf-with-google-n-grams-and-pos-tags.html
  15. 15. https://twitter.com/hqu/stawww.cs.cornell.edu/home/kleinber/bhs. pdftus/1745171763
  16. 16. Journalists want ❏ Source discovery and curation ❏ Passive monitoring and alerts ❏ Saving and archiving ❏ Visualizations ❏ Parity with TweetDeck user interface
  17. 17. People want “I want to catch up with a summary of key information about the breaking news story.” I want to get a list of Twitter accounts who are official organizations related to that story.
  18. 18. My Musing
  19. 19. What’s next for keepr? ➔ refine algorithm ➔ source classification ➔ conversation analysis and visualization ➔ archiving search results and tweets Rolling out a Beta program for newsrooms Sign up at www.keepr.com/beta
  20. 20. https://github.com/hqu/keepr
  21. 21. Verification Resources ● Verifying Social Media Content ● verificationjunkie.tumblr.com ● BBC processes for verifying social media content ● Storyful’s validation process ● InformaCam

×