———- Forwarded message ———-
From: Alan Ritter <aritter@cs.washington.edu>
Date: Mon, Aug 22, 2011 at 10:30 AM
Subject: Twitter Calendar
To: cs-grads@cs.washington.edu, faculty@cs.washington.edu, cs-ugrads@cs.washington.edu, cs-staff@cs.washington.edu
Cc: Oren Etzioni <etzioni@cs.washington.edu>
Hi All,
We have been working on automatically extracting a calendar of upcoming events from Twitter status messages, and have a prototype system available here:
http://statuscalendar.com/
You can click on the individual calendar entries to drill down and get more details.
A couple of details about what is happening here:
-We are continuously gathering a random sample of Tweets.
-We extract Named Entities (people, locations, products, movies, etc…) from the Text of the tweets.
-We extract and resolve temporal expressions (for example, we can figure out which calendar day “Next Friday” refers to based on the timestamp of the Tweet).
-We count the number of times each entity co-occurs with a reference to each date, and plot the highest-ranking entities on the calendar.
The idea is to provide a summary of the most popular/important events occurring in the near future. For example you can see many people are talking about the rumored iphone 5 announcement on September 7th.
Of course there are a few errors here and there, some text processing tasks are more difficult in Twitter text due to it’s noisiness (misspellings/abbreviations, unreliable capitalization, etc…).
Anyway, if you have time for a quick look, we would love to get any comments or suggestions for improvement that you might have.
Thanks!
-Alan