Joho the Blog
An Entry from the Archives

Nouns are often used for labels (tags). And, single word nouns in one language might have multiple word noun phrases in another. In languages like German and Spanish, those phrases might have articles or prepositions.

My German is too rusty, but my Spanish less so--so, I'll use a Spanish example:

What an English speaker might label as "Blue" could be labeled "El Azul" by a Spanish speaker.

Also, in the bigger history of tagging, people often put alternate titles and/or descriptive phrases in keyword / tag entry fields. This also could account for where the "stop words" come from that are showing up in the index.

Oh, and the MIT case is a good example of when the inclusiveness of a flat / aggregated namespace "doesn't work"--it doesn't work as well as an exclusive / hierarhical namespace might.

So, you don't want every MIT, you want education // MIT, or you want NOT ( german // mit ), etc.

Or, in the semantic web sense, you don't want to see everything with the tag (label) "mit," you want see everything with the tag (reference) "about mit.edu".

Jay, but if you were tagging in German and your tagging system did not allow tags to include spaces, tagging something as, say, "das" AND "Himmel" because you wanted to say "das Himmel" wouldn't do what you want. The "das" isn't staying connected to the "Himmel." So, if I'm understanding you, I don't think that can be the explanation. And if your tagging system allows spaces, then technorati should not be indexing those tags as consisting of two independent words, should it?

Yes, MIT is a good case of one language's substance being another language's stop word. But it's not necessarily the case that therefore a flat namespace can't work. It's plausible that a cluster analysis could cluster the MIT university references, just as Flickr clusters the Capri island references as well as the Capri car references. An exclusive, hierarchical namespace would work better, but you wouldn't get nearly as many items tagged into it.

. . . if your tagging system allows spaces, then technorati should not be indexing those tags as consisting of two independent words, should it?

I don't know what Technorati is really trying to achieve, so I don't know if what you're describing is intended and a more desirable result than not, or if it's something they would want to improve on.

But, generally, many indexes work at the "word" level, and break-up keyword phrases into separate index entries for each individual words. Some indexes preserve phrases all along. And, some can go both ways, e.g., "das Himmel" gets three entries: das Himmel, das, and Himmel.

Yes, MIT is a good case of one language's substance being another language's stop word. But it's not necessarily the case that therefore a flat namespace can't work. It's plausible that a cluster analysis could cluster the MIT university references, just as Flickr clusters the Capri island references as well as the Capri car references. An exclusive, hierarchical namespace would work better, but you wouldn't get nearly as many items tagged into it.

I totally agree that flat namespaces could work. We could also put aside the flat / hierarchical distinction and just think about this as an issue of using more precise terms, e.g. MIT-EDU-MA-USA could be a very distinct label in a flat namespace.

Though that still represents a hierarchy without using a dramatically hierarchical syntax, one could also create a precise term by spelling out Massasschuttes-Institute-of-Technology.

Also, let's not mix apples and picnics with getting into whether or not more people would tag given a flat space or a hierarchical one. That may be a data entry issue, or a context issue, or a lot of things that are independent of the resulting structure. For example, Wordpress lets me tag blog posts with tags that live in a hierarchical namespace as easily as tags in a flat namespace. (more on my blog's faceted CV / narrow folksonomy on the iawiki, fyi).

Young women at my alma mater started using LiveJournal back at its earliest days (1998) to find out about the schools they were planning to attend. LJ hosts many school-related groups, and these groups have, since '98, been used to help ease the transition into college. Potential students have "friends" before they even arrive.

The student blogs that are hosted by colleges are usually lifeless marketing tools that should just be phased out. Their messages are very spun--and you can tell when you read them.

Besides, it's also wicked-phony for the school to choose who its voices should be. Those faces are the ones parents want to see--not the realities of a school's life...

Nowadays, with MySpace, Facebook, (and probably, still, to a different degree) LiveJournal, young people have many ways of finding info about their schools through peers. Peer to peer is possibly the better marketing "tool." Schools should just let it happen.

My daughter, Jumper Girl, the high school senior, looked at a bunch of Facebook pages as a way of getting a feel for the colleges to which she had been admitted.

She also says that the high school juniors are using Facebook to help make decisions as to which colleges to apply to.

I suspect Technorati is invisible to JG & her cohort.