Hi all, I've worked on projects that have used Mahout for recommendations-generation, but I'm not terribly familiar with all it has to offer.
I've got a requirement to, for a large set of data: 1.. Cluster together similar entities (object with many textual properties) 2.. Within a cluster, perform fuzzy collapsing of each property of an entity (to find the most frequent valid value) 3.. Add the collapsed version of that entity to a searchable index (so when we get a new entity arrive, we can try and match against an existing one) Are there parts of the Mahout toolset that would help with the first two problems? I've been searching for fuzzy field collapsing libraries without much luck, but perhaps I'm not using the correct terminology. Item #3 sounds like a job for Solr/Lucene, so that's one thing that we don't need to worry about. Any advice would be most gratefully received. I'm trying to avoid reinventing the wheel, and I'm sure someone must have done this before! Regards, Deejay Proprietor [email protected] www.binarytweed.com Binary Tweed - New games that are a bit like old games, but better.
