Being careful about tag-combining

I'm of the opinion that tags should never be combined, but I know there are those out there who feel differently.

Nonetheless, there is such a thing as over-zealous tag combining. You really need to be sure that the two tags you are combining can't, won't, and don't mean two different things. I just uncombined the tags "manuscripts" and "manuscript," and honestly even just looking at the two different tag pages, it should be clear why they are different. People mostly "manuscripts" to tag books that are about manuscripts, while the tag "manuscript" is mostly used to indicate that a book is a manuscript.

I'm sure I'm preaching to the choir, but please be sure that your combination is warranted when it comes to tags!

We've had this discussion before and agreed that singular and plural tags for the same thing should not be combined.

Singular and plural is a real problem. It is something that should probably never, ever be combined--I have a real hard time thinking of a time it would be completely okay--but it is something that people ALWAYS want to combine because they seem so close to each other. I wish we could get some kind of notice displayed somewhere really prominent on the tag combination page, DO NOT COMBINE SINGULARS WITH PLURALS, LIKE, EVER, but since tag combination doesn't ever work quite right everyone is using the workaround URL anyway.

Related to this is something I've been debating on bringing up. Currently "lit" is combined with a bunch of other tags, as it's considered a short form for literature. I know this is generally accepted, but in this case .lit is also a well-used format for e-books and documents. Currently, someone who wants to tag their electronic files will have all their .lit titles thrown in with everything else tagged within "literature". Is this something where the vast majority is assumed to understand that "lit"means "literature" and so those who want to catalog file formats just have to suck it up?

lit should never have been combined with literature. I'd support separating it out.

It looks like lit .lit and literature are now separate. It does not appear that people are putting in lit when they mean .lit. But people are putting in lit when they mean literature.

At this point, I don't do any combining or separating of tags. However, I do put in reasonably synonymous tags on the same book, so I can see what other people are reading in my own field.
trolleys and streetcars
railroads and railways
public transit and public transport

I don't see why lit shouldn't be combined with literature. What else does it mean? .lit is a different tag entirely.

it means the same thing as pdf or cpd. It's a standard file format that some people will use for an e-book that happens to be commonly accepted shorthand for literature. However, that does not negate the completely different meaning and use. Given the argument about not combining plurals and other terms that may or may not be the same thing, I thought it pertinent to mention something like this. "lit" does have two possible and very different meanings. You *will* see, on here, that most people will likely be using "lit" to mean "literature", but that does not negate the fact that "lit" or ".lit" will be used by other people who wish to indicate a file format rather than a genre.

The difference between pdf and lit is that pdf doesn't have another, more common meaning. Looking at the list of people who use the tag "lit", it seems to be in the sense of "literature", based on the fact that they use other tags like "American lit" and "ref". I didn't look at all of the users, but about half of them.

I think it's one thing to avoid combining tags that actually have different usages, and another thing entirely to avoid combining tags that theoretically, possibly, might have different usages.

And kawika just gave a quite possible different usage. Here's some more:

LIT is a three-letter abbreviation that may refer to:

Limerick Institute of Technology, Ireland
Little Rock National Airport, Arkansas, United States
Liaoning Institute of Technology, China
Local income tax
Lost In Translation, a 2003 comedy drama film
LIT Ranch, Texas, United States
Long Island Iced Tea, a popular mixed drink

Not to mention what it might mean in another language.

So I wouldn't combine lit and literature for the same reason we don't combine SF and Science Fiction. Tim really means to invent a way for users to specify what they really mean by their tags, that is to tell LT "When I use lit I mean literature".

I can see why "lit" and "literature" should be left uncombined. But for those of you who think tags should never be combined... "saga of seven suns", "the saga of seven suns", and "the saga of the seven suns", all used exclusively for that one series of books? Would you really say they should be left uncombined?

a quite possible different usage

I think you're missing my point: it shouldn't be about whether a different usage is "quite possible", but whether it's actually used. If there was a significant group of users who really used "lit" to mean .lit, then by all means the tags should be separated. But there's not.

I think you're missing my point: it shouldn't be about whether a different usage is "quite possible", but whether it's actually used.

For SF being "science fiction" vs. "San Francisco", the difference in usage is going to be obvious when looking at the tag page. (Though skimming the list, the top 200 uses are apparently all "science fiction", with nary a "San Francisco" among them.)

But how are you going to tell which of those "lit" tags are "literature", and which if any are the book format? Or if you're not familiar with the works in question, any other use of "lit"?

I was going by the other tags people used: so if they also had things like "American lit" and "ref", then they're using lit just as an abbreviation for literature.

> 7

What about bed ? Litouania ? throne (as in: lit de justice) ? liturgy/liturgical ? ...

>15 boekerij: I think this actually supports my point that we should only worry about meanings that are in use.

#16: Are you volunteering to keep watch on the tag usage of 8,350+ users (8181 for "literature", + 169 for "lit", = 8,350 if they were recombined—plus whoever actually uses the "lit" tag for the file format) to make sure that the other meanings don't come into use, then?

There's no need to look at the people using the "literature" tag. Looking at 169 users is manageable. Even when tags are combined, you can see only the members using one specific variant.

But really, it should be the people who actually use a minority meaning of a tag who separate the tags when needed and then complain about it so that no one combines it again. We don't need to anticipate these problems; they can be dealt with when they arise.

#20: There's no need to look at the people using the "literature" tag. Looking at 169 users is manageable. Even when tags are combined, you can see only the members using one specific variant.

Sorry—I wasn't aware of that. :)

But really, it should be the people who actually use a minority meaning of a tag who separate the tags when needed and then complain about it so that no one combines it again. We don't need to anticipate these problems; they can be dealt with when they arise.

I don't think I agree; but since I can't articulate why I feel that way, I'll just leave it at that.

I took a look at what people do with three tags:
Numbers in Zeitgeist on Tag page
Science fiction 335,069 294,637
Sci-fi 53,846 88,209
SF 138,137 114,209

Now in the weighted number, books like Neuromancer and Dune are near the top, but The Left hand of Darkness is #1 in SF, #12 in science fiction, and #31 in sci-fi. Ender's Game is #1 in sci-fi, and #4 in science fiction, but #32 in SF.

So, there is a certain overlapping randomness in how people tag.

> 16

Is it ?

You may take into account that "lit" is even a ISO 639-2 ("MARC") language code :
lit Lithuanian
(ISO 639-1 code for Lithuanian is "lt")

Thingamabrarians may use those language codes (ISO 639 family) as tags. As a matter of fact, I do.

I use them as tags e.g. to refine and specify and indicate the languages when the book language code is "mul" (Multilingual) or to indicate that apart from its main language(s), the book is dealing with different languages, too, or to indicate what languages a book is about, etc.

FYI : ISO 639 family language codes is quite some easy picking as well as user language neutral language indicator.

Some people may think that e.g. "Ελληνικά", "हिन्दी", "Hrvatski", "lietuvių", "русский" or "Српски" is the better way to indicate languages--and they may be right, while others may prefer "el", "hi", "hr", "lt", "ru", "sr" (ISO 639-1), either "ell"/"gre", "hin", "hrv"/"scr", "lit" "rus", "srp"/"scc" (ISO 639-2(/T)/(/B)) and think that this is quite some handier. Their choice. One can have them all.

FYI : I haven't added any books in Hindi, neither in Lithuanian--yet. Then again, tagging the latter i.a. "lt" and/or "lit" may be quite appropriate.

Is there a way of changing which tag ends up 'on top' after they have been combined? I have no major objection to my tag 'Science Fiction and Fantasy' being combined with 'sff' except that sff is what shows, and I find it about the least meaningful to the uninitiated of the many combined there. Even sf&f might just send my brain on the right track.

22: No, it's just based on how often the different aliases are used. Most popular wins.

> 18

We don't need to anticipate these problems; they can be dealt with when they arise.

I don't agree. It's much better to avoid the problem in the first place, rather than have to discover and fix it when it happens.

"An ounce of prevention is worth a pound of cure."

> 24

What's more : If ever a Thingamabrarians intends to use e.g. the tag "lit" as a (proper) language code, he will get bullied the message, loud and clear : "You {add your favourite insult here} ! You are wrong. 'lit' means 'literature'. Final."

That's why it is quite important that we have and keep combined "lit" and "literature" (for) now, for, hey, imagine, otherwise one may miss our message (see above).

Adding : The majority of Thingamabrarians is not dealing with e.g. Lithuanian anyway. Those who may do are poor suckers only, that's for sure. They 'd better understand whom is ruling and accept our message, loud and clear : "{F-word} ! Get lost !"

> "An ounce of prevention is worth a pound of cure."

Exactly. That's why we 'd better have "lit" and "literature" combined now.

It's much better to avoid the problem in the first place, rather than have to discover and fix it when it happens.

I have two objections to this:
1) The fix is extremely easy; this isn't like work combining
2) We lose valuable connections in the meantime by keeping them separate

*rolls eyes* I haven't tagged any books "lit" (yet), but I guess I should point out another usage for that tag, since it's the one I would be using: a shortened form of "literotica". Which could mean either literature-erotica, or the actual website Literotica. If/when I ever use this particular tag, it will be to tag books that fellow Literotica members have recommended to me. Therefore, I CERTAINLY wouldn't want that tag automatically combined with "literature".

Somebody has combined tag suomi with tag Finland, but suomi can be either the country Finland or the Finnish language. Some people don't use capitals in proper names so there is no way of knowing when suomi refers to the language and when it refers to the country.

Tag combining is tricky. Since recommendations are partly based on tags, I think that in most cases is better to leave tags uncombined. Especially tags in different languages (same word can mean different things in different languages) and tags that are or could be abbreviations.

Recently I noticed that somebody has combined tags like box01 and box1. Is there any point in that?

> 28

<preliminary>
---------------
W A R N I N G :

Part of this message is using ironic style. If you (*) are not sure whether you can deal with that kind of style, maybe you 'd better skip this message as a whole.

Worse yet : irony may be a very strong tool in order to get a message across. This may add to some people rather not wanting to read it, because the problem is not as much the style as is the message itself--i.e. : the content and therefore the expressed opinion--that makes them feel not at ease and thus that they cannot stand.

(*) Obviously, this "you" is NOT pointed at Anneli, neither at someone in particular. Then again, though I cannot know, it may suit... you indeed, whomever you are. I am afraid that I cannot help this.
---------------
</preliminary>
 
 
You 'd better applaud for some smart and avid combiner knew that "suomi" means "Finland" (or so he thought)--i.e. : that there is or may be some relationship between "suomi" and "Finland", even though the latter can be quite/too rough.

Fair enough, most probably most Thingamabrarians won't know the meaning of "suomi"--let alone the difference between "suomi" and "Suomi".

That's no problem, because they won't use either of those tags anyway and they couldn't care less. Therefore : why would they bother--and right so. Thus they ('d better) leave both of those tags alone--that's fine--and won't touch them in any way.

FYI: LT itself--TPTB and therefore the system, too--is mixing up languages and countries as well as both of their codes anyway.

FYI: By analogy, it may be just fine (if not preferable) to tag each and every book that has got anything whatever to do with English language and/or that is in English as "England", or still : to combine whatever "English" tags with/into the tag "England".

Fair enough, at least some Thingamabrarians will think that the latter is a most silly idea (and they may be right). Then again, IMO, this is exactly (analoguous to) what some smart and (most/overly) avid combiner has been doing vis-à-vis "suomi" now, by combining "suomi" and/with/into "Finland".

Though even trying and pretending the former--i.e. : "English (language)" == "England"--would be considered as plain madness, the latter--i.e. "Finnish (language)" == "Finland"--is just fine, is it ? Right.

Therefore, you 'd better conform to some (different) narrow-minded world view.

Have some examples of books (originally) in Finnish. Feel safe to tag them "Finland"--all of them, (even) including e.g. The Roman : the memoirs of Minutus Lausus Manilianus, who has won the insignia of a triumph, who has the rank of Consul, who is chairman of the Priests' Collegium of the God Vespasian and a member of the Roman Senate (link) (*), being a translation of Ihmiskunnan viholliset : Rooman senaattori Minutus Lausus Manilianuksen muistelmat ajalta 46-79 jKr, by Mika Waltari. Have you got a copy of the latter original edition? Don't even bother to try and tag it "suomi" (i.e. : Finnish language). It's no use anyway. Better tag it "Finland" from start.

(*) Touchstones is defunct--yet another long standing issue.

Adding : Most probably, majority (whatever that means) of Thingamabrarians hardly know nor care about Finnish, about Finnish language neither about Finland anyway, so why would you care about ? You SHOULD learn and run with the pack.

What's more : you knedledge of English language turns out to be sufficient not to (want and) use Finnish language, or whatever different language any more--ever. The sooner all those different/silly/barbaric/inferior/retarded (add ad wish) languages are extinguished, the better it is--for everyone, really. They are a nuisance ! See ?

FYI : In French, "français" (avec minuscule) (en: French) relates to French (the language), whereas "Français" (avec majuscule) (en: French) relates to France (the country).

CONCLUDING :

UNLESS you know for sure each and every language and each of those's (finest) particularities--and does anyone, really?--better DON'T--i.e. : don't ever--combine whatever tags, because one/you don't and most probably cannot know what they mean--and therefore trying and combining them would be mean indeed.

OTOH, there is another, being perhaps, please forgive, a rather bullying point of view, too. In short, the latter view maintains that all those bastards and barbarians 'd better adapt to the one and single "right" language and (thus) unique and only and single "right" world view. French language has got a quite accurate and specific expression for the latter. It calls it (fr) "langue de bois". Alas, AFAIK, the latter is intranslatable, except for with some Slavic languages' (ru, hr, cs) expression "jednoumlje" that is covering quite the same meaning (and tone) indeed. (*)

(*) Hat tip: dr. Tomislav Sunić (LT: Tomislav Sunic). Warning: self-applauders (and being so proud) most probably won't like his Homo americanus: Child of the Postmodern Age (link).

Best tip yet : ban ideas, ban opinion, ban thought, and of course, ban different languages, too, for, hey, what purpose could any of those be, anyway ? Better play (and feel) safe (and soft) and be nice and cheer and support langue de bois and thus adhere to the Brave New World !

--
Thought crime does not entail death; thoughtcrime IS death.
-- Winston Smith, "Winston's Diary", in George Orwell, 1984.

>29 boekerij: What's the point of this? You've gone off on a ridiculous tangent that's far more insulting toward English speakers than whatever insult (perceived, or perhaps just invented by you) lies in the combination of suomi and Finland.

Message 28 is perfectly clear and wonderfully succinct. If your goal were simply to support the argument that tags in different languages shouldn't be combined, you'd have been better off just expressing agreement. But of course, the tag combination isn't really what you're concerned about, is it? It was a great opportunity to express your prejudiced view of English-speakers as stereotypically arrogant and ignorant, and you couldn't let that chance slip by. But it does get tiresome to read the same post again and again.

As for the actual discussion, I stand by my earlier remarks. I don't think there's any need to avoid combining abbreviated tags just because someone could conceivably use them to mean something else at some time in the future. Tags on LT can be a powerful tool in features like recommendations and tagmash, and their strength lies in their numbers. Since tag combinations can easily be undone, we might as well use them to get the most out of the data we have now.

>29 boekerij: boekerij, Would you please remove or edit those parts of your post that are not relevant to this thread and seem only calculated to insult and offend? Thanks.

Most, if not all, Thingamabrarians are thoughtful, well-read (often in several languages), and – this is the distinguishing characteristic – interested in broadening their horizons and learning more. Statements like "The sooner all those different/silly/barbaric/inferior/retard... (add ad wish) languages are extinguished, the better it is--for everyone, really. They are a nuisance ! See ?" do not reflect any stated opinions of those on this thread, and are blatantly offensive to most thoughtful readers. Cf. the popularity of the Reading Globally group.

Yes, the fact is that this is a primarily English-language site (due to its location), and that many U.S. English speakers have unfortunately not been forced to learn multiple second languages, as is the case in most parts of the world – where other languages are spoken in closer proximity. Most English speakers, particularly those who admire literature, will tell you that they regret that fact, and wish they had been afforded that opportunity. Criticize the educational system if you will, but please don't project bigoted views onto the users of this site simply because of the language they speak.

Finally – and this is just offered in the spirit of clarification – your statement the difference between Français and français is not entirely correct. Français (capital F) does not quite refer to the country, but to the nationality. A trivial difference, perhaps, but one which could make a difference in tagging. Perhaps just another example of why we shouldn't make assumptions about subtle differentiations in meaning in languages of which we are not native speakers.

>26 _Zoe_:, 30
I agree, it seems to me that the rules for tag combining should be something like this:
1. Avoid combining tags, unless:
(a) the two tags are so similar in usage, users, and meaning as to be identical (ideally, the tag pages would look very nearly the same)
(b)combining the tags would generate a wealth of useful information that otherwise would be lost
(c)there is no reasonable possibility of someone (currently or in the future) using either one or both of the tags to refer to some other meaning which should not be combined

2. If the tags combined are different enough that a user notices the change and uncombines them, they probably should not have been combined in the first place. No combining/uncombining wars, please. If the tag is unique to somebody, whoever they are and however 'wrong' we believe them to be, that tag is still unique.

edited for formatting

Would it be possible to have some sort of warning put on the tag page about not combining unless people are sure? Right now the notice says:

Are there synonymous tags in the "related tags" section? Go ahead and combine tags.

Nothing about possible problems. No way to put a note that says, "Please do not combine 'prayer' with 'prayers', as people are using them to differentiate books."

>30 _Zoe_:

Zoe, I wish that I didn't write my message at all. Now I feel that it was my fault that my message induced such an outburst (= message 29).

I usually tag my books with both English and Finnish tags, but I would not dare to combine them although they are the same in my vocabulary (or I try to choose tags that mean the same thing in English and in Finnish). English is foreign language to me - not inferior (how could any language be inferior to another language?), but not my mother tongue. I am not always sure about Finnish words, either. How can anybody be so sure what tags are synonyms?

It would be useful to know how exactly the tags and recommendations are connected. Could some of the odd recommendations be caused by incorrect tag combinations? Isn't it possible that the tag combinations cause cumulative error or something?

>32 MarthaJeanne: Yes, I agree that the tag page explanation is very weakly worded right now. I strongly feel that – unlike work combinations, which tend to strengthen the data pool – tag combinations tend to weaken the data by stripping it of the subtleties and nuances that are tagging's big advantage over categories/traditional classification. Tag combining should absolutely be a last resort (in my opinion, that is.)

>33 Anneli: Anneli, your comment in #29 was well-worded and entirely relevant to the conversation. Please don't feel bad about it.
I, too, have wondered why anyone would combine box1 and box01...I can't imagine anyone searching site-wide to find books tagged box1 or box01, actually. They seem like they would only mean something to the individual using them.
With the way the tag page is worded, though, I can see how people might get carried away and feel like they were being productive by combining as many similar tags as they can.

edited to add response to Anneli

>30 _Zoe_: / 31 / 33

#29 Boekerij is a Flemish user, and many Flemish people are very touchy about language and their own language in particular.

However, if he means that it's better not to meddle with tags in or about languages one doesn't speak, I think he has a good point.

34> Perhaps a user had accidentally used both the box1 and box01 tags and didn't know about the new functionality to easily change tags.

I don't think it is fair to Flemish people to generalize on the basis of one person who has a habit of writting over long messages that offend other people and make them want to disagree even with whatever good points he may make.

Anneli, your comment was well made, and useful for all of us. Please don't stop posting because some people may react unreasonably.

>38 MarthaJeanne:

I won't stop posting - I'm too loquacious - or voluble or talkative, maybe even garrulous ;-)

>38 MarthaJeanne: Agreed. Isn't it just as unfair to lump all Flemish-speakers together as it is to lump all English-speakers together? Besides, I'm not sure how being Flemish (which might justifiably lead to touchiness regarding the relationship of Flemish and Dutch) gives one any right to be touchy insulting re: English, Finnish, French, etc.

But I digress. Regarding jjwilson's post, I think that's a good point. If the tag-page explanation is to be revised (and I think it should), it would be good to include a note like: "Combining tags on this page will affect data for all LibraryThing members. If you want to change your own tags, go to your tags."

I certainly didn't mean to be offensive towards Flemish people, and sincerely apologize if I gave that impression.

A thread where angels would fear to tread! ...but I just want to say, Matt, that I understood what you meant and I know there was no intention to offend anyone, least of all the Flemish.

>41 Nicole_VanK:, 42 Sorry...didn't mean to say that you were insulting Flemish users. I see now that that's the way it sounded. My bad.

Why wouldn't you combine box1 with box01? They're fundamentally the same tag, and one could argue it's a bit OCD, but it's a good combination according to the rules.

I don't think lit should be combined with literature, since how do you know they didn't mean literary? I suspect the abbreviation itself has as much different weight as sci-fi versus science fiction or humor versus humour.

>44 prosfilaes:

I agree that box1 and box01 appear to be the same tag, and that combining them violates no currently-stated rule of tag combining. As I stated in #31 (1.2), though, I feel that tags should only be combined when (in addition to the tags being synonymous) the combination would help generate better data. The reason I say that is that we can never quite know what someone intended in their tagging, and unless there's some 'greater good' argument to be made, we're better off letting sleeping dogs lie, if you will.
In this example, I don't know why someone would use box1 and box01 to mean two different things, but it's entirely possible that someone did (or will)...and since there's no really valuable data to be mined from that combination anyway, I wouldn't risk it.

That said, I realize this is just my personal opinion, and I'm not that passionate about it that I'm going to go out and uncombine things like box1/box01...but I'd like to know what others think.

45> Well, if there's no really valuable data to mined from that combination, it's because there's no really valuable data here at all, so there's also no really valuable data to be hurt by that combination. Even if someone uses both contrastively--which I find less likely than using WWII and WW2 contrastively--the only problem it will cause is that on books they've tagged box01, the group tags will include box1. Since the group tags aren't really yours anyway, I don't see the problem.

28> Because recommendations are partially based on tags, I think we should combined tags where ever possible. I don't think we should isolate by language; whatever Finnish mystery readers are reading and whatever Finnish mathematicians are teaching is indeed interesting, and should show up in recommendations. When I search for mystery, Russian literature in tag mash, I would like the tagging of the Finnish readers to contribute to the tagging of the English readers in finding all the Russian mysteries in LibraryThing.

>47 prosfilaes:

I haven't said that tags in different languages should not be combined, but when you are combining you should be sure that the tags are synonyms. It is not useful to combine names of countries and languages, e.g. Finland (=Suomi) and Finnish language (=suomi).

edited to add an example

47> I agree with you and I think that the same argument applies to combining humor and humour.

48> And Finnish and Finnish.

> 50

Is there intended to be a difference in your examples?

Yes, one refers to a language and the other to the Finnish people, just like Suomi and Suomi.

The lumper vs the splitter issue. LT probably has more of the latter.

Also, note that humor and humour, represent semantic dialect variations, but not necessarily phonlogical dialect variations.

51/52:
I actually thought of the difference between the language and just the adjective form of Finland: a book written by a Finnish author, which did not itself relate to the people or the language, could still be described as a Finnish book. (Although I don't know if anyone does actually tag that way.)

54> I tag that way.

I think we should Suomi this discussion.

;-))

> 52

I think asking the system to differentiate between tags that are spelled identically might be asking a little much. Keeping Finland/Finnish separate is totally understandable.

I think the point s that Suomi could be combined with either Finland or Finnish but to combine it with both would combine Finland and Finnish. So it's best to combine it with neither.

#56: it took me a few seconds to understand what you meant, lilithcat... Well done! You can smile.

>54 r.orrison:
When I think about it more closely, it seems to me that it really doesn't matter if suomi is combined with finnish and finland, because people are using tags like Finnish in three different meanings (at least): books in Finnish (original language or translated into Finnish) books about Finland or Finnish people and books written by Finnish authors.

I'm finished now with suomi tag discussion. Let's hope that nobody combines United States with United Kingdom - they are both English, aren't they ;-)