Herdict: Difference between revisions
(New page: First, we are looking for someone who can help us create a program that will scour a microblogging site, like twitter, and identify when someone might be saying that a site that is down or...) |
No edit summary |
||
| Line 1: | Line 1: | ||
Project 1: | |||
We are looking for someone who can help us create a program that | |||
will scour a microblogging site, like twitter, and identify when someone | will scour a microblogging site, like twitter, and identify when someone | ||
might be saying that a site that is down or blocked. So something that | might be saying that a site that is down or blocked. So something that | ||
| Line 8: | Line 10: | ||
language skills. | language skills. | ||
Project 2: | |||
We are looking to make some changes to our downloadable toolbar. | |||
In particular, we want to automate the detection of blocked sites. When | In particular, we want to automate the detection of blocked sites. When | ||
a browser throws a 504 error, we want the toolbar to catch it, and then | a browser throws a 504 error, we want the toolbar to catch it, and then | ||
Revision as of 11:13, 7 March 2012
Project 1:
We are looking for someone who can help us create a program that will scour a microblogging site, like twitter, and identify when someone might be saying that a site that is down or blocked. So something that would constantly look for things like URL within 3 words of "blocked" "censored" "inaccessible" "i can't get to" etc. We'd be looking for something north of 75% accuracy, but not perfection. Bonus points if we can do it for Sina Weibo instead of Twitter, but that will require some language skills.
Project 2:
We are looking to make some changes to our downloadable toolbar. In particular, we want to automate the detection of blocked sites. When a browser throws a 504 error, we want the toolbar to catch it, and then report it. In addition, we have known blocking patterns from ONI, and we want the toolbar to be able to identify a match with a known blocking pattern and report that as a inaccessible page as well.