A recent blog posting of mine referred to Fuzzy Lookups and Fuzzy Groupings in SSIS and the ability to merge duplicate data based on inexact matches. Well, someone at Microsoft must be reading this blog because it has recently acquired a company called Zoomix that specializes in just this type of Fuzzy Logic. Looks like SSIS is going to receive a significant boost in functionality. According to the Zoomix web site, Zoomix’s Accelerator software “combines semantic and linguistic analysis with machine learning to classify, match and standardize complex corporate data”. Sounds like Fuzzy Logic to me. It even uses the word “de-duplicate” which seems to get through my spell checker just fine. A new word for me but it is self-describing, kind of like “normalcy”. I have not used the product yet, but looking at the System Architecture it appears to use Service Oriented Architecture (SOA) to provide inline data correction and matching with proprietary “Self Learning” technology. This sounds very promising as the hard work of Fuzzy Logic is to define the specifications or “rules” of what constitutes a match. It will be interesting to see exactly how it does this automatic de-duplication. Of course, the word “proprietary” means we will only be able to see the end results, but obviously Microsoft is sufficiently impressed to move forward.
Surely this technology has been acquired too late for SQL Server 2008, but we can look forward to this extra functionality, perhaps as an Add-In to SSIS, fairly soon, maybe at SP1. Hopefully, we won’t have to wait for SQL Server 2011, right? Philosophically, we can all look forward to a Fuzzy Future.
Cheers
Brian
Recent Posts:
6 things about SQL I wish Microsoft would change…
