First and foremost, I would like to emphasize that I personally have great faith in tagging as folksonomy has proven itself useful and efficient on numerous occasions. It's become such an intrinsic facet of the web architecture in recent years that imagining having to go without it is maddening as the amount of knowledge and time it would have taken an expert to built even the small fraction of the relations and collations is indeed unfathomable. Luckily enough, drawing significant amount of people to the task, while finding a sensible way to restrict the anarchy and ambiguity, can be extremely fruitful.
Elaborated on below are the cases when folksonomy either works extremely well, its efficiency is arguable or ambiguity reaches disastrous levels, which is in direct correlation with folksonomy's quality.
The Good
One outstanding example of folksonomy's success is definitely Last.fm artist page tag clouds. The beauty of tag clouds is that they allow visual representation of characteristic's weight which is far more true to life than one-level labeling. It also appears that search results are sorted by the quantity of people who'd deemed the tag appropriate for an artist as well as by weight of the tag in aritist's personal tag cloud. It must be noted that expertise of the masses does work extremely well in this case - for instance, one wouldn't be able to discover splendid, but yet obscure performers hadn't it been for tags (or similar artists - but that's a whole other story). Some tags ARE quite unspecific or irrelevant ("seen live", for once), but thanks to visual weight representation they do not seem to be distracting.
The Bad
The opinion to follow is perhaps slightly biased, but apparently and quite naturally amazon.com tagging in music department doesn't seem to match the eloquence and scale of Last.fm. This, ostensibly, derives from amazon.com dealing in multitude of wares and subject types, being continent or/and country specific and, essentially, very commercial - it's main objective is to Sell, whereas Last.fm's, if to put it the noble way, to educate, to broaden one's awareness - at least that's how it's marketed and that what it genuinely is doing (as many do not really have aversion to piracy and don't go and buy discovered artist's records). Therefore, the number of people who condescend to tagging on amazon.com isn't really enough to secure viable folksonomy, because in folksonomy quantity does indeed tend to turn into quality.
The Ugly
Just a few months ago it would have been extremely easy to name an internet instance where tagging was awfully incompetent and insufficient, but, apparently, things have improved for 500px.com since then. But even now, it's quite possible to find photos by very general tags, like "tiger", "rain", whereas more complex concepts, such as "withering", "matador", "burlesque" leave much to be desired. This might partially be caused by people not anticipating others wanting to search such specific entities and to some extent by a the user/picture ratio, which leads to the situation when there's a lot fewer users willing to contribute to folksonomy than photos to be classified. It also must be noted that due to result's display limit and deficient tagging prominent photos depicting flowers may not be found by searching for "flowers".
Now, the last and the least, comes the google image labeler folksonomy game - The Labeler Game- even though diversity and exactness is encouraged (by prohibiting proposition of words used to describe the picture very often, by granting more points for complex concepts), oftentimes the matching pair is indeed far to general to be of any real usefulness. Surprisingly enough, google image search does nevertheless prove to be quite efficient - probably mostly due to the fact that it doesn't heavily rely on the results of the aforementioned game-camouflaged folksonomy.
0 comments:
Post a Comment