Tagthe.net
I don’t know if people have seen Tagthe.net, but it’s pretty interesting.
It’s another site that tries to automatically generate tags based on a URL or a block of text, but this one also tries to group these tags into categories, “topic,” “content type,” “person,” “title,” “location,” and “language.”
Like most auto-tagging sites, sometimes it works and sometimes it doesn’t. Here’s this blog’s generated tags:
You wanted some tags? Here you are!
http://blogs.ischool.berkeley.edu/i202f08
topic
talk Knowledge analysis Gruber book Web Permalink google information content
content type
text/html
person
Bill Schilit Nathaniel Wharton Bob Nick Doty
title
INFO 202 Fall 08 Blog
language
english
Nathaniel Wharton Said,
October 19, 2008 @ 8:57 am
Cool. People sometimes want to think inside their own vocabulary space, and it seems like this tool could auto-map personal vocabularies.
In lecture 7, Bob said, “Searches made using an uncontrolled vocabulary will not have high recall because they will fail to match documents indexed using “good terms”
Made me wonder… could this change that game?
Thinking about it, I wonder, could this personal vocabulary be connected to a hub vocabulary… And from there could people map personal semantic content across language barriers?! That would be pretty cool.
I looked at their FAQ: “This is just a sneak preview into a product we are currently working on.” .. Curious what market they’re going after.
Two guesses:
1. They’re going build a tool to allow people to autotag email. (uncontrolled vocabulary area, so I’m thinking a good candidate)
2. bookmark autotagging as a firefox plugin. (tagging my bookmarks is a pita) – I bet they mine delic.io.us for this.
I could imagine A RESTful interface for the “low-budget” wikipedia autotagger (with a controlled vocabulary) I blogged earlier wouldn’t be that difficult to create.. and it could also do #1 or #2 above… of course that method would create controlled, not personal terms..