Brian Dear imagines a search engine where all user-generated tag information is searchable – an engine that confederates the various nations of Flickr, iTunes, etc. Neat idea, and certainly another step toward the semantic web vision.
. . . if more and more services in 2005 add user-generated tagging, will “federated tagging” be far behind? And if someone were to index all the tags from these various sites…. would the result be Taggle? Imagine: a service where you type in a keyword, and you get back all the hits that have that word as a tag. If Flickr, del.icio.us, and umpteen other sites cooperated, then an uber-tag-search service might just work . . .
Isn’t this just RSS feed search focused on a particular bit of metadata? We’re happy to work with someone on it if we’re as close as I think we are.
i am finding more and more often that searching del.icio.us tags gives me better results for many things I’m looking for.
I keep wondering when the spammers will start exploiting the usefulness of this sort of metadata, posting ads and phony links disguised as good info. I understand flickr has already had to fend off something like this.
Blogdigger indexes by tags/categories/subjects; we combine all the tag metadata and make it searchable with the subject: prefix. We also list the tags for each search result returned on any search. As an example, here’s a search for the tag “tags” – http://blogdigger.com/search.jsp?q=subject%3Atags&sortby=date
I posted more about this on the Blogdigger blog (http://www.blogdigger.com/blog/2005/01/04/1104819887000.html).
Also, the spam issue is real, has already hit free blog services such as Blogger quite heavily, and will only get worse as the spammers catch on. Any good aggregation tool is going to have to develop spam identification algorithms.
in my opinion user genarated metadata will ony be extremely usefull for retrieval purposes if we have a reputation managed community of users who annotate content with tags that refer to entries in an ontologie.
This way we fix 2 major problems of metadata; no spam (because annotations get valued by the reputations of the annotator), no ambiguity (because all annotations refer to a single meaning concept in the ontology).
Two major hills to take:
1-building the community,
2-standardising the used ontologie (ideally this is user maintained as well, perhaps the wikipedia community could help us here)