Saturday, May 06, 2006

Semantic Web Product Search?

Microsoft recently unveiled its product search engine interesting thing to note is the differences in the aproaches taken by Googles Froogle and Microsofts product search engine: Froogly relies on companies to send them structured data about the products they sell, Microsoft extracts information from vendor websites. Obviosly that means that Microsoft has potentially more data - because it does not need to wait until vendors decide to submit information to yet another product search engine. Google on the other hand eliminates the error prone information extraction step and hence has higher quality data.

This is a great example for a use case where the Semantic Web would make technical but not businness sense. If the vendors would just annotate their websites with their product data in a structured format, we could combine the advantages of both approaches. Vendors would no longer need to keep up to date with all the product search engines and submit their data in different formats. Anyone could just retrieve these files and have all the data for a successfull product search engine. The problem is that Google (or Yahoo or Microsoft for that matter) does not want it to be simple to create a competitor to its search business, hence vendors have to send the structured files for Froogle to Google instead of placing them on their websites.

I'm still optimistic that we'll see a "Semantic Web Product Search" soon - because it makes business sense for the vendors (they can stop worrying about different formats and search engines). I think Microsoft wasted a chance to change the rules of the game by not building a semantic web product search engine. By inviting more competition it would have risked to hurt Microsoft in this sector but would've surely hurt the market leader Google much more.