h a l f b a k e r yTastes richer, less filling.
add, search, annotate, link, view, overview, recent, by name, best, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
People tag things. People tag things slightly differently. This is partly cultural and partly simply the way the coin was tossed given requisite variety of alternative representations in most people's lexicons. The same person will tag things differently, forgetting how they did it last time if it wasn't
visible to compare. Different people will exhibit an even wider variation, employing tags that the first person didn't even know were viable alternatives.
What the world needs now, is a tag synonym server* (as a deep web service, machine-to-machine, rather than a user-usable looking up resource like a thesaurus). People would strengthen the relationships by statistical means - alternate suggestions would be accepted by the host application presenting to the user, and this information is returned as a weight to the synonym server.
Initially, it might pre-provide a range of pluralisations so that when a person tags 'aarvark' it will register as the logical equivalent of 'aarvarks' and possibly even 'aardvarkii'. Pluralisation is something I'd treat as a special case, even if it's a majority and very common case.
Although I've left it out of the implementation considerations, I also think there needs to be some sort of analogue quantification of 'how much of a synonym' the suggestions are selected to be. This would allow weighted interrelationships of tags to be built, and then we'd sort out this whole folksonomy mess.
* I'm aware that the phrase 'synonym server' already exists, but it is exclusive to the realm of DNA sequencing and suchlike. This idea steals back that and re-generalises the phrase before re-eclecticising it again.
[link]
|
| |
[+] I think, subject to my having
understood it correctly. Would such a
server include elements of
disambiguation? |
|
| |
Nice. Have a tasty synonym bun! + |
|
| |
I have no idea what any of this means. Just thought I'd mention that. I'm sure it's very clever. |
|
| |
I'm concerned that the effectiveness of
this
might be subject to the semantic
context
in which the tags were authored. So, a
'Jaguar' tag on a car website has a
different
meaning from a 'Jaguar' tag on a large
primate conservation website. To
disambiguate homonyms, you might
need an additional layer of
disambiguation
which does something similar to search
engines which use analysis of how
search
terms form separable clusters of links
in the
search space. |
|
| |
dunt matter hippo, cruisin the strip in or on your jaguar would get attention. |
|
| |
Personally, I'd skip any attempt at
disambiguation in the early phases - a]
because it's a very difficult problem; b]
because it's probably not as necessary
as people think; 3] because given
enough diverse tags for the same
resource, it may prove possible to steer
in one direction or another through
inferred context of other tags; and d]
naughty marketing people will try and
break the system anyway by trying to
strengthen association of 'innocent'
tags with
brands and marques*. To the latter, I'd
say let 'em, by not even bothering with
disambiguation at all. Who kneeds it? |
|
| |
* (in a similar manner to
"googlebombing") |
|
| |