Controlled Vocabularies – Why local search needs folksonomies.
I am working on a paper in the local search area for MindCommerce that will be published in a couple of months (if not sooner – it all depends on me – unfortunately) and today I had one of those “gotcha” moments. You know, when all of a sudden you are finally able to fix the concept that was always speeding around in your head, but had never slowed enough to “ideate” it.
Those of you who have read my last few blogs have gotten an earful about categorization and why I think it is the weakest link in local search. Tonight I found the words and concepts that help explain the problem with more clarity.
When you enter words and phrases in the search box while trying to find something using local search, what you are actually doing is creating a “tag”. The tag reflects the words that you would have used if you had stored the data.
In essence, local search has an input side that actually operates as a folksonomy (a user generated taxonomy). On the other side of the equation (the search platform) is an authoritative, controlled vocabulary that represents the “official indexing terms” used by the local search provider to enable the user to find business listings that fit the input terms. The mismatch between the use of a folksonomy on the input side and controlled vocabularies on the matching side is the most common reason for failed local search.
So what’s the solution?
Allow the users to add tags to the listing. In addition, consider allowing the use of negative tags that would help sharpen the results even more, for instance something like “tires, not rims” for a store that sold tires but not rims. By analyzing the IP of the users, you might be able to link people with geographic locations and assign some form of validity metric. You could even take the conservative stance and establish a MCT (minimum comment total) that would be necessary before you added a tag to the search vocabulary, which would, over time, become a folksonomy.
I think it would be useful to enter a local search site and ask to see businesses in your area that other users tagged with the terms you intend to search on. Hopefully you can see that this idea could play out in a number of interesting ways.
And yes, there are a number of difficult problems that spring from using user generated taxonomies. Take whatever stand you want, but don’t forget that the Web2.0 is about social networking and social searching is based on folksonomies. If that isn’t good enough, try this on for size – “Folksonomies are how you get every dime out of the “long tail”.
Hopefully someone in Local Search will take this advice and improve the relevancy of their responses to user searches.