Context Mining the Full-Text Index
By billy.cripe on Apr 11, 2008
An open question to the readers: is anyone doing anything in the way of text-mining the full-text indices of ECM systems out there?
For some time Biz Intel tools have been able to do text-mining to look for keywords, patterns and tallies of important words and phrases in semi-structured content (think a "comments" area in a Call Center log). But this is basically the application of traditional data mining and OLAP capabilities on different datatypes that still reside in (something that looks/functions very much like) a database.
Well, what about the full-text index that is maintained by ECM systems? The ability to automatically pull context out of content items in an ECM system, mash up those contexts with more traditional BI and OLAP queries and usage tracking metrics (e.g. everything from folksonomic ratings and tagging aggregates to click tracking and download counts) seems to me like a kind of holy grail of BI.
I'm not talking about improved search performance. So don't bother to reply with those kinds of answers. I am talking about merging capablities like THESE with traditional BI analysis and reporting.