If you start gathering data about whether a particular user would like a document then you might also be interested in the on-line learning algorithms that are currently being developed. These should be able to let the user say whether they wanted to read a document or not. The clustering could be used initially, but as interest information is collected, you can build a per user profile of interest.
On Fri, Jan 1, 2010 at 7:45 AM, Bogdan Vatkov <[email protected]>wrote: > > Don't people usually see if the new docs fit into an existing cluster and > > if they are a good fit, add them there, otherwise, maybe put them in the > > best match and kick off a new job. > > > > Actually this question goes back to the original attempt - to analyze > documents automatically by the machine, and not by people :). One of my > goals is to not read the new document but rather the system to tell me if I > should read it ;) - e.g. if it gets clustered/classified against given > cluster/topic which I am interested (not interested) in I could then take > more informed decision whether to read it (not to read it). -- Ted Dunning, CTO DeepDyve
