Support/TwitterMetricsProject: Difference between revisions
< Support
Jump to navigation
Jump to search
(Created page with ' * Scoutlabs doesn't do well with twitter * http://www.twitterfall.com/ is powerful for lots of searches at once and being pretty comprehensive but lacks historical data …') |
(category -> Support Archive) |
||
| Line 17: | Line 17: | ||
* Hard to build tools that other people (not me) can use since it'll all be command-line + MySQL queries | * Hard to build tools that other people (not me) can use since it'll all be command-line + MySQL queries | ||
* May be a good stopgap if the marketing tool actually happens | * May be a good stopgap if the marketing tool actually happens | ||
[[Category:Support Archive]] | |||
Latest revision as of 08:45, 14 July 2021
* Scoutlabs doesn't do well with twitter * http://www.twitterfall.com/ is powerful for lots of searches at once and being pretty comprehensive but lacks historical data * http://tweetcloud.com/search/term/firefox is probably my favorite for now since I can click on particular items and get posts so I can see what people are saying about "Crash" etc. * Cotweet is much less geared towards metrics and more towards acting on tweets but we're already using it for other things so this would be easy to build in. * Marketing may be working on a tool but timeline is unknown
- Searches have to be very carefully set up. Even quick searches show that at least 1/3rd of all tweets that have "Firefox problem" are "I have a problem with IE/Chrome/Safari, switching to Firefox"
- Twitter has an API to collect tweets that we can code off of
* If all we need is common/trending words as well as volume, this could work well.
* Can have the following features
o Archived list of all tweets
o Analysis by trending words and ability to segment the tweets by time, by words used, by user
o Can quickly locate words with simple DB searches
o Can annotate various tweets with data like tags
* Traffic is low enough that we won't have to worry about API limits as long as we don't try to do too many searches.
* Hard to build tools that other people (not me) can use since it'll all be command-line + MySQL queries
* May be a good stopgap if the marketing tool actually happens