I would like a way to follow twitter streams by keyword analysis.
No, I don’t mean… give me all tweets with the keyword php.
I mean… give me all tweets by the top 5% of people who talk about php the most. i.e. I want to see ALL the tweets by the top 5% of php tweeters.
Here’s how it could be done!
- Tokenize the twitter dataset and dump it into a table with two columns
- Put the tokenised word in column A put the tweeter in column B
- Group by column A,B and order by count desc
- Now show a twitter stream of the top 5% of tweeters in the results.
Simple as that!
If you make this, please give me a shout because I want to use it.
Over & out.
Advertisement
Matthew said
Have you given up posting new ideas? Or have you run out?
jv2222 said
haha, I’ve just been focusing all my spare creative attention on http://techzinglive.com I will be posting new ideas here soon!
Edward Seager said
Hey, thanks for the link from Techzing. I have a few thoughts about this – I would have preferred to have mailed you but I can’t find your address
First, I think this is a great approach to be able to find a list of people on Twitter who share similar interests to you, to lessen the amount of crap to wade through!
The practicalities worry me, like how would you obtain the Twitter dataset? I was thinking you could try doing lots of searches to obtain an approximate solution, but this may be tricky as you can only obtain 1500 tweets from each search. See http://apiwiki.twitter.com/Twitter-Search-API-Method%3A-search
There is also limit (unreleased) on the number of searches you can perform from a user or IP address per hour.
Final thing, I’m having no luck at searching for “PHP” and filtering out all the links to .php webpages with their advanced search http://search.twitter.com/advanced
(talk about falling at the first hurdle)
I hope I haven’t criticised your idea too much, I would really like to make it work! If you have any thoughts, feel free to e-mail me at my firstname.surname@googlemail.com