The Need For A Tagging Standard

John Carmichael writes "Tags are everywhere now. Not just blogs, but famous news sites, corporate press bulletins, forums, and even Slashdot. That's why it's such a shame that they're rendered almost entirely useless by the lack of a tagging standard with which tags from various sites and tag aggregators like Technorati and Del.icio.us can compare and relate tags to one another. Depending on where you go and who you ask, tags are implemented differently, and even defined in their own unique way. Even more importantly, tags were meant to be universal and compatible: a medium of sharing and conveying info across the blogosphere — the very embodiment of a semantic web. Unfortunately, they're not. Far from it, tags create more discord and confusion than they do minimize it. I have to say, it would be nice to just learn one way of tagging content and using it everywhere.""

Comments

Don't agree

[pubjames]

Isn't the power of tags that you can tag stuff however you want? To me a standard for tagging would be a negative thing.

I don't thing the problem is a standard for tagging, the problem is having a standard for sharing tags between applications. But that's another problem and it doesn't need to be solved to implement tagging itself.

Re:Don't agree

[lousehr]

Your analysis of "the problem" is exactly the point of TFA. The stated concern is not that the content of the tags has no standard, but that the format of the tags has no standard. If a single tag contains multiple words, should the words be separated by spaces or underscores, or should we use StudlyCaps?

Re:Don't agree

[radtea]

The stated concern is not that the content of the tags has no standard, but that the format of the tags has no standard.

The medium, as Marshall Maclluhan said, is the message. As soon as you standardize the format of the tags you will restrict the kind of information people can convey with them. That may be an acceptable limitation to you, but not to others, and they will find workarounds that effectively break the standard.

For example, if tags were standardized on underscores to separate words you would have to forbid spaces and caps to enforce that standard. And then we would have no way of distinguishing between Polish and polish, which would be bad if you were looking for things to do with Eastern European culture or furniture care products. People would then start doing things like expressing capitalization by some other syntactical hack which would be inconsistently applied and a greater mess would ensue.

Alternatively, tags could be represented as more complex markup:
<tag>
<word order="1">really</word>
<word order="2">stupid</word>
</tag>

But because words and concepts have no general one-to-one correspondence (many words do not convey a unique concept or a concept at all, and many concepts cannot be conveyed in one word) this would be inadequate, and in any case even if the content model of the "word" tag forbade spaces, caps and underscores, people would still create tags that looked like:

<tag><word>reallystupid</word></tag>

The basic idea of "semantic markup" is wrong. From the summary:

the very embodiment of a semantic web. Unfortunately, they're not. Far from it, tags create more discord and confusion than they do minimize it. I have to say, it would be nice to just learn one way of tagging content and using it everywhere.

Actually, tags as they stand are the very embodiment of the semantic web. The only function of the semantic web is to create confusion and discord, because confusion and discord is the essence of the human epistemological condition. And the call for "one way of doing X" has a nice religious ring to it, history shows that attempts to standardize things relating to human thought are very much misguided.

Automatic tagging

[drcoppersmith]

I'm inclined to disagree that 'tags' are the answer here. I wrote my masters thesis on a method automatically generating semantic webs from plaintext. It's a huge problem with about a dozen different stages, but I had backing in all of my research from the psycholinguistics and computer-science field.

Herein lies the rub: You're never going to get everyone to agree on a set of appropriate tags. Even if you do, you'll never have them uniformly applied (well I find that humorous but you have it tagged as inappropriate).

There are other solutions here, such as automatic semantic generation. Hey, I never said it was an easy solution, but it's one that I'm certain can be accomplished. Flame away ;-)

Re:Automatic tagging

[maxume]

The article is mostly talking about standardizing the envelope, not the message, which is to say, how do you share/create a two word tag, and how to you specify exactly what is supposed to be described by that tag, and how do you share that in a useful way.

The fact that someone thinks something is funny and someone else thinks it is inappropriate is useful information to gather, if you get 5000 funny and 5 inappropriate, you have a lot more information than if you have nothing at all, but even in you get 10 and 10 you still have more information, which is probably a good thing.

One Key Point

[Azarael]

How do you standardize something that has not been widely implemented before? It's great to say that it would be good idea to have one standard practice for tagging, but which one? There's no reason to make a huge fuss about this until it a least one clear contender for standardization emerges (which will probably happen on its own).

Re:One Key Point

[T-Ranger]

Well, not quite. Reading the blog post the problem lies with two areas: technology and linguistics.

For technology, as an example, how do you quote things? How do you separate tokens? Do you use StudlyCaps and spaces? "Quoted words", and commas? If the later, what about nested quotes?

Bullshit question. The question is solved. Use XML. (Yeah, well, it is the web). We don't need Yet Another CSV "standard". Tags may be presented as lists, in spans, or WTF ever. But if you are talking about storage and transmission, then store the tokens separately, and transmit them in an unambiguous format; in 2007, on the web, the solutions are implementation-specific and XML, respectively.

For linguistics, thats harder. Nouns or verbs? Talk to a librarian, Im sure there are volumes of information on the right way. But I don't care, as I'm still disgusted that the technology problem even exists.

Right now it seems there is little discussion on the problem. Right now, if implementations are trying to reinvent data encoding schemes either the implementations are totally brain dead (and need a kick in the ass from an outside force), or are completely oblivious to the problems they are encoding into there core features (and thus still need a kick in the ass). This is so bad, its worse then wrong. You have to try to get to the point of being wrong.

Of course, I don't care because tags are stupid. OTOH, perhaps I would care if they at least were implemented in a potentially useful way.

Re:One Key Point

[robotninja]

I can't believe I'm reading this -- it's a sad day for information science (I'm a librarian) when many otherwise knowledgeable, tech-savvy people are blinded by Web-2.0-speak. Let me reiterate another poster's comment:

Tags are keywords. More specifically, they are subject keywords.

If you can wrap your head around this idea, then you might realize what the author is talking about is a list of 'standardized' subject headings. You may know this by its common street name: a thesaurus (although some people prefer to use the terms ontology, taxonomy, or controlled vocabulary).

There are plenty of well-established, long-standing, open thesauri out there - a few examples:
MeSH
LCSH (Library of Congress)
CSH (National Library of Canada)
... and hundreds more, in nearly every language.

There are even ANSI and ISO guideline standards for how to develop monolingual and multilingual thesauri for specific subject areas. This practice has existed well before the advent of the Internet, since before the first libraries even. Over the past hundred years or so, the practice has become highly refined in order to facilitate the practice of indexing.

That's right, when you "tag" a page, you're actually indexing it. But you can call it "tagging" if that makes you feel cooler.

Didn't they have this problem in Babel?

[zappepcs]

How to share and categorize information is an ages old problem. One man's trash is another man's treasure, likewise, one man's bread is another man's dietary problem.

I'm not sure, but haven't we already figured out that tagging would require more tags than the actual information being tagged to accomplish what the original poster was asking for?

People actually pay that much attention to tags?

[east coast]

I don't feel that tags have enough significance behind them to merit a standard. I'd be more concerned with truth in journalism first, for my part.

Hopeless

[bigmouth_strikes]

Trying to standardize tags in the context of standardizing what they are, is hopeless. It'll be like the Unicode standard; too complex to use in its entirety.

But to standardize the format of tags and to standardize how to exchange tags between systems, is a great idea.

I'm fired, aren't I?

[elzahir]

He said "blogosphere." Instantly, I don't care.

Only thing worse would be something like, I dunno, "tags should be a Web 2.0 standard" or somesuch.

Excuse me, but "proactive" and "paradigm"? Aren't these just buzzwords that dumb people use to sound important?

Tags are for things that AREN'T standardized

[vidarh]

Tags are human assigned labels for something that we don't have better meta-data for, or where we don't want to be bothered with formalism. If you want something formal, go use a proper taxonomy/ontology and put bucketloads of OWL or RDF-schema data on your site to define relationships, or use format with well defined semantics to add information. Noone is stopping you, and there are cases where formally defining relationships is worthwhile, such as when you want software agents to be able to infer stuff about the data. But that's not what tagging is used for. Tagging is used for ad-hoc manual classification in situations where it is good enough

Better hurry...

[supabeast!]

If someone gets started on a tagging standard right now, it might see a little use before the whole silly idea goes out of style next year.

They're not that useful anyway

[AdamHaun]

The only tags I like are my own. The real use of other people's tags is to show how they organize information, not to help me find something. The problems the article brings up are only the beginning -- the natural tendency of a global tagging system is for the number of tags applied to an object to increase without bound. If I'm doing a master's thesis on, say, web design, I might tag any number of sites "thesis". Is that useful to anyone else? Probably not. But it will interfere with someone who's searching for sites about writing theses.

MOD PARENT UP

[metamatic]

We've got a standard for keywords in HTML documents. There's no problem there.

The only issue is what to do when there are multiple sub-documents on a single page, like if Slashdot allowed individual replies to be tagged.

Re:Hyphens.

[Chryana]

I would add to this that slashdot tags tend to be not very useful.

Most of the time, the tags have little to do with the actual article (eg. yes, no, maybe, fud, notfud, flamebait). I thought the purpose of tags was to be able to find an article easily later on when it has been archived, and the usefulness of the tags I just mentioned for this purpose is dubious at best. I do not pretend to have a solution to this problem, but I think the situation would be improved if the editors or maybe the /.ers who wrote the top rated comment where the only people allowed to set the tags.