Building a learning engine to understand what tweets are relevant.

[svigraham]

Basically, using AI techniques to classify what tweets are relevant to me. Building this in rails.

[irfan]

I am building a similar system to filter content based upon its popularity among a specific group of people (rest of the content is noise for this group). I have already developed an engine to make groups based upon their geological location. So, in essence the system would provide localized popular content. Would love to see your implementation :-)

[svigraham]

Thats nice. Are the groups are based on the content's location or the person's original location? That might be an interesting way to group too. Irrespective of where I am from, can my content automatically get grouped based on context, which in this case is tied to location?

[irfan]

Right now they are based on person's location but we can extract the location from content only if content has this information. In case of tweets, not all tweets contain this information.

[svigraham]

Yeah, you are right. But, you can use foursquare/gowalla checkins and pair it up with content. But, it might not be worth doing.