July 9, 2005

Snippet on Getting Folksonomy Right

Today's summary on folksonomy... taxonomies and ontologies can help the many find information, but never help the whole of the people. The role of folksonomies is to fill in that gap to get far closer to the whole.

The failure that Google noted in other search companies in 1997 being happy with getting 80 to 85 percent of the correct answers for people, meant 15 to 20 percent of the people found the tools failed them (for me it seemed far higher than a 20 percent failure rate in 1998, which is why I switched to Google quite early). There are far too many that are complacent with their development of taxonomies and ontologies that are only helping the many and have no desire to change their practices to get to closer to the whole. It takes a diverse toolset to get the job down, which means including taxonomies and ontologies as well as other newer solutions.

So what is needed in a folksonomy? It must be broad to provide the best results. People must be tagging content or objects for their own purposes. The tags must be separated from the object so they are a point of reference. The person tagging must also be distinguishable from the objects to they are a point of reference. The objects must provide a point of aggregation to find common tags and common people and the matches on these three points. Tools like del.icio.us and CiteULike.com do this very well.

When we have these distinct elements we can begin filtering and aggregating, just as Jon Udell has been doing in his collaborative filtering.



Comments

Comments are closed.

This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike License.