Hi Tony,


What do you mean by buckets?


The matching between social network messages and the web pages is open-ended.   We use 2 algorithms in our tool to do the matching, and 1 is trained on wikipedia and recognizes the kinds of things talked about in wikipedia pretty smartly; the other technique is a context-free matching algorithm on n-grams.  The former gives better results and we rank those matches higher.  But neither technique is limited to consumer scenarios.


I think the trick to getting this to work for specialized scenarios will be making sure that you have the data you care about available in the social network.