Cleaning up groups 1: A simple classification system for groups

I've come late to a conclusion many others have come to—groups are out of control. You can't find anything. Interesting, even popular groups fall off the front page and stop getting attention (GenderQueer, SCA, Metafilter, Libertarian and Market Liberals, Wikipedia-Bibliothek).

I'm going to do a number of things to fix this. I'm going to put groups to "sleep" if they haven't seen any action in months. (By my current algorithm about half of LT's groups go to sleep; there will be a list of popular "sleeping" groups, so members can wake them up if they deserve it.) And I'm going to redesign the groups page to accommodate a classification system of some sort.

I've started a very basic ONE-LEVEL classification here:
http://www.librarything.com/wiki/index.php/Groups_classification

For now, I don't want to make a multi-level system. Nor do I want to try tags, although I'm obviously open to the idea.

Feel free to edit it. If there's disagreement, we can talk about it here. I'm going to make myself the final arbiter of this, but I certainly want membership involvement.

Starting with one level sounds good although a time may come when a particularly popular category needs more. At least it's many fewer group names to wade thru if you are looking for a particular kind of group.

Will the 'asleep' groups be like dormant threads? Easy to wake up?

Can I suggest something along the lines of 'Reading Groups' (which is probably not the clearest title but I'm lacking inspiration tonight) to encompass 'Group Reads' kind of groups which I think deserve a level classification of their own in some way. Particularly if you're thinking about ease of access for people who are not so familiar with Talk but may be interested in looking for group reading efforts going on (there are quite a few groups like this...)

There are the challenge groups as well. Maybe 'Reading and Challenge Groups'?

And, looking at the list, I was wondering about groups like 'The Green Dragon'.

First, let me say that I think the idea of cleaning up and categorizing is good in principle.

However, many of my groups wouldn't fit in those categories, to my mind. Maybe it would be good to have something for academic subjects?

My groups that don't fit the proposed categories well are:
Book History / Histoire du Livre, Books on Books, Ethical Theory, Feminist Theory, Graduate Students, History of the Book, Maps and Atlases, Philosophy and Theory, Philosophy of Science, Political Philosophy, theory.
I'm not sure it would make sense, for instance, to put political philosophy in the Politics section, and thus separate it from other academic subjects like sociology, philosophy of science, feminist theory, economics, social science, or whatever.
(Maybe book history & history of the book could go under history, but that would separate them from "books on books.")
Where do you put graduate students? Or Librarians who LibraryThing?

Looking at the wiki, I think the challenge groups are covered by 'Games and Challenges' and my understanding would be that groups like GD would be 'Online Communities' (could be wrong though...)

By the way, what I forgot to post before was 'Good Idea Tim!' ... let's not focus on the 'late to the party' bit, but instead just celebrate the fact that you've come at all ;)

Is the idea that once everything gets classified, new groups will be asked to pick a classification? Can group owners change a classification?

As one who participates in a challenge group but shuns games, I think it is misleading to include challenges with games. The challenge I'm part of, and others I've seen, involve in-depth discussions of books we've read. I think it would be more appropriate to include them with Books.

Good idea! But I agree with others that many groups won't fit into any of those classifications. Where *would* genderqueer go? Some could make an argument that any kind of sexual-orientation group could go under politics, but that's a big stretch. Maybe a "lifestyles" type classification? Would be good for the kink groups I'm in, too.

Where would the school-related groups go? I know there are a lot of groups for "so-and-so's class" and that kind of thing.

And the tv-show or movie-related groups? Harry Potter, Buffy, etc etc...

I think this is a *great* idea, but the classifications either need a lot of expanding or clear-cut definitions, so we know what is meant to go where.

I would hope Talk would follow suit with there being a way to see just the threads for the category (ies) you want (even if you aren't in the groups, or haven't posted to them) ?

I hate having to wade through the challenge junk, and other game stuff, so I would not want to see any of it in Books.

I think Challenges and Games sit well together. I guess that the 'Online Communities' and 'Local Communities' and 'Other Communities' are the catch-alls where you would find things like LibrariansonLT, genderqueer, and the so-and-so's class etc. but would guess that if that isn't immediately clear, then possibly these category titles might need tweaking a bit.

The "Sorting Groups" thread discussed this issue earlier, if folks want to browse through ....

great. Late, but welcome thanks (is too slowing down the implimentation of collections? :-) )

I'd say that series/book groups can be part of the authors categry

there should be a sport category
"" food
and that politics should be subsumed into a wider society category (religion too?)

I think these would then capture most of the 140 odd groups I'm a member of.

But there is a danger of each category becoming so huge that groups are still lost in the wilderness.

a category talk would be great.

I really twitch at the "lifestyle" suggestion. I'd rather see a "Communities" sort of classification, for groups that are organized around the people rather than the topics. This could encompass GLBT groups, age-based groups, geographically-oriented groups, as well as the very generic groups like the Green Dragon, without judgmentally singling out one particular group of people as being "different".

Interesting that Tim mentions the genderqueer group as being "interesting", when he so dogmatically defends the gender binary imposition on Common Knowledge. Maybe there's hope.

14: I agree, but like I said, I think the catagory "names" are very vague and so I have no idea what each would encompass. If genderqueer and such would go in a "Communities" catagory, I think that needs to be made clearer somehow.... I for one would never think to look under, say, "online communities" for a glbt group, I would think that would be more like groups about MySpace and such.

13: What about an "Entertainment" catagory? It would encompass movies/tv, sports, etc....

15>

I didn't say "online communities", I said "communities".

The idea of classifying groups sounds good. Can I suggest that Private Groups get dealt with? Either highlighted in some fashion or seperated into a category?

I would have thought The Green Dragon would go under Genre as would Horwart's Express. Aren't most groups communities?

The classifications suggested are a bit arts biased! Could I suggest a heading for 'the sciences' or at least 'science and technology'.

As lquilter has identified, there has been a lot of discussion about this before, and countrylife has done some very good work.

However, Tim has put this up on WikiThing, so the idea is that you edit that wiki page to put up your suggestions under the 'Contested categories' heading. To sign any comments you put there, use ~~~~ (four tilde characters - above the hash on a UK keyboard), which will put your name and date to any comment.

groups are out of control. You can't find anything

Classifying groups is good, but I think a big part of the solution has to be improving the search.

there will be a list of popular "sleeping" groups

I'd like a list of ALL sleeping groups, not just popular ones.

My big concern about these classifications is, who decides where a given group goes? Earlier discussions often proposed that it would be up to the owner, which leads to problems in situations where the owner is no longer around.

For the record, I'm still in favour of group tags.

It seems to me that we are locking the barn door after the horse has gone.
I think there are 2 problems with groups and they are related.

1. You can't find anything, which leads to duplicates being made,

2. People have different ideas and approaches so even if you could search you might never connect a group to what its about. I think its not just categories that you need but subjects.

For example there is a group about Christopher Moore, but its called Lust Lizards. That is one of his books, but if you haven't read it, why would you think its about CM, not to mention that there is no reason to pick that book as the definitive way to identify him. Why not other books or phrases ? So you would have to think about searching for CM in the Ls to find it and perhaps avoid adding a duplicate group.

If you had to pick a subject when you created a group, and people could search the subject then it might keep down the duplicates and make it easier for people to find existing groups.

An idea:

Why not *temporarily* add tags to groups and see what categories the groups assort themselves into?

So you would have to think about searching for CM in the Ls to find it and perhaps avoid adding a duplicate group.

"Searching" shouldn't require looking in a particular place; the search engine should find groups with Christopher Moore in the description.

"but I think a big part of the solution has to be improving the search"

Group search works fine.

What needs improving is the descriptions of each group. If you search for a word the results will include all th e groups who use precisely that word in there description - that many don't have sensible descriptions isn't LTs fault. It doesn't help that you have to consider all possible derivatives making wrting ome good descripios tricky.

Group search works fine.

I really have to disagree.

Let's say I'm looking for the group about Canada Reads. I search for "Canada Reads" in quotes, expecting that it will be one of the first few results. The first problem is that the quotes don't seem to mean a thing. It seems to return all the results that have either Canada or reads in the description. Why is the Green Dragon the second hit when it doesn't even have the word Canada anywhere in the name or description?

Likewise, "Read YA Lit" returns Biofiction and a Portuguese group as the top two results. Even if you know the exact name of a group, you can't find it in the search! (Okay, so you can find it if you use ctrl-f, but that really shouldn't be a part of the process.)

The search needs to deal with more complex terms, like phrases in quotes, and it needs to rank the results by relevancy. If it's currently aiming for relevancy ranking, it's doing a really bad job. Groups that contain both of the search terms should appear above groups that contain only one of them.

Sorry but I've got to agree with _Zoe_ in #28. For example, I did a search for "group reads" and it turned up 1615 groups. ctl-F turned up only 2 groups with "group reads" in their title and 2 with it in their description. 'Group Reads - Literature' and 'Group Reads - Sci-Fi' are the most relevant, with the phrase actually in the group title and yet they were almost at the bottom of the list (relatively). I imagine if I were a new user trying to find a group read to join in with, I would have given up at the 1615 groups as none of them seemed relevant.

I am looking at the list of groups that are on the wiki and I have to say, they are not going to pull everything into a group.

Librarything Railroad (which is the group for discussing all things train and rail related) would be left out, unless someone wanted to drop it in genre. But that doesn't really fit either. (Though it may also be a dormant group.)

I don't think a one level organization is going to cover the present NUMBER of groups adequately.

#26 Zoe

"Searching" shouldn't require looking in a particular place; the search engine should find groups with Christopher Moore in the description.

Search will find anything that has Christopher, and anything that has Moore, not just those 2 together. It also finds groups that have neither Christopher or Moore in the title/description.

1. Search often returns hundreds if not thousands of entries (not in alphabetical order either) if you have a very common search term. Searching in a letter helps to cut down the number of items returned.

2. There are groups with no descriptions, in fact at the start they didn't even have to have group names. I think they do all have names now, but they don't all have descriptions.

Search will find anything that has Christopher, and anything that has Moore, not just those 2 together. It also finds groups that have neither Christopher or Moore in the title/description.

My point wasn't about what search currently does, but about what it should do. Unfortunately the two are not the same right now.

Yes, there may be current workarounds, like searching in a particular letter. But the best solution would be to improve the search engine so that, for example, it could deal with phrases in quotes and rank the results in a more meaningful way. No matter how much we classify the groups, that can't make up for a decent search engine.

No argument. I agree with Zoe on the group search.

I want to add a better browse-layer to search—whether classification or tagging or both. And I think that gets "mixed into" the search.

But there's no question, the search needs to weight things better, and use the quotes, if it's not doing that now.

Even without the quotes it should rank groups with both Christopher and Moore before groups that have one or the other. And if it is feasible it would be even better to rank groups with Christopher and Moore next to each other and in that order the highest.

Oh, good. I really am in favour of browsing too, just not as a substitute for search.

So, I'll return to my other question: who decides where the groups go?

I'm not going to come up with any brilliant ideas (though I'll be thinking about it) - just wanted to say thanks, Tim, for tackling this (and to countrylife for taking on THAT mammoth task). Anything you come up with is bound to be an improvement on the current system (surely a case of LT being a victim of its own success) which has got completely out of hand. Looking forward to the new system.

You don't think 'disciplines' as a heading might attract a lot of people who are looking for something very different? ;-)

Heh, heh, heh - you can check out any time you like but you can never leave.

I like the broad definitions that countrylife offers. They allow for infinite growth, but are still suscinct enough that it makes sense where things fall.

Archery, Railroad go under subjects.

Hogwarts Express goes under books.

GBLT falls under People.

Definitely a direction I would lean for the one level organization.
Then we fix search and things are good.

Tim, as a group many of us have had this conversation several times (with or without you) and I think most people agreed that what will most likely be the best solution would be tags limited by the group members, meaning that you have to be part of the group to tag it. any other grouping is merely postponing the problem and as I am sure you have noticed, there are always groups that fall on the line or even completely out side of them.

Now is this a nice intermediate step to help clean things up a little but I would hope it is only to give you more time until you can implement some kind of tagging. For this kind of 'organizational' problem I find it hard to believe you don't want to go straight for the gold.

I am not sure what Tim's objections to tags are, and at the risk of being in the minority, I am a bit leery of tags.

My concern would be that if everyone was able to add tags, you would get a tag that represents something that is only mentioned once. So looking at the tag on the group you would think it would meet your needs (like Tag watch). Only to be very frustrated combing through various threads trying to find the subject the tag represented.

I don't know if that would be very rare, or if it would happen a lot with prolific taggers.

>44 FicusFan:
Maybe we should restrict who can tag groups, or who can tag prolifically, or something along those lines.

The funny thing about Tim's "objections to tagging" for groups is that he's way for it for books. How is it great for one but not the other? The problems for one are the exact same problems I've pointed out before (a long time ago, I'll admit) for the other.

Where did Tim say he was against tagging groups? Did I miss that?

Tags for groups? No please, that will not solve the chaos. Categories will do a better job of ordering groups.

49 > yes Categories will make things nice and neat but I don't think it will be accurate, unless a group is allowed into more than one Category (and this only makes it a little more accurate) which is half way to tags.

People discussing books by Isaac Asimov could be in Authors, Genre, or Series and Books? Which one to look in? However it could be tagged Isaac Asimov, Sci_Fi, Robot Series, Foundation Series, etc.

ETA: If we have the one level categories as a primary sorting and then tags, but still searchable across all categories, it might be the best of both worlds.

>51 countrylife:

Ditto. Heck, I'll go so far as to say that for our groups, categories will only be truly effective if paired with tags.

Why? Because categories would have to be vague, or else very little would fit in them -- which means that most likely, very much will fit in them. And more than a few levels of sub-categorization would be a bit much to navigate, and will inevitably conflict (violently, no doubt) with how someone's world is ordered.

A combined approach (perhaps with tags as "related tags" à la book-tags and so on) will allow categories to be reasonably populated and groups reasonably findable.

Not to mention that a combined approach has the best shot at keeping browsers and seatchers happy with one (joint) system.

I think Tim was against tags because they only work well if people are tagging for themselves (contrast amazon tags with LT tags).

I argued that this would be accomplished if we could use tags to organize our own groups (e.g., to view Talk restricted to the groups that we've tagged "Reading Challenges"), but he wasn't convinced.

Perhaps people could be encouraged to "tag for themselves" then. Maybe there could even be an official tagfest when group tagging is first rolled out. With any luck that will generate a decently-sized initial tag cloud.

54>

I think the trouble is drive-by tagging; Tim argues that the requirement that you have a book in your library deters drive-by book tagging, but I can certainly imagine hostile tagging of, say, Happy Heathens.

Yeah, the question is: whose tags? Will there be a limited or pre-set number? Would LT have to implement another level of database structure to allow tag combining, etc.? Otherwise I think you'd get a lot of disparate, unorganized tags that would just mirror the chaos of groups.

55 > that is why I think you would be required to be a member of the group to tag, if you leave the group your tags leave as well.

Group tagging should be limited to the group's creator, just as with other group functions. Even with readafew's suggestion (#57), less high-minded types could join a group just long enough to cause trouble with the tags that would outlast their brief tenure. Sure, penalties could be put in place for hostile tagging, but why create the opportunity?

I think >57 readafew: is a good idea. If tagging is limited to group creators, what about all those abandoned groups, or groups which are active but the creator is not around much? I think the risk of transient tags would be small compared with the nuisance of active users being unable to tag their group - but that's just my opinion.

(edited to remove repeated words)

um, as I said I would assume that as with a book, if you if you delete the book (or leave the group) any tags you added will be deleted as well. I would want my groups to reflect the ideas and attitudes of the current members.

I say this partly because some groups were created by someone who is no longer active. And if only the creator can tag it is is not much of a step above Categories, at least in categories all the groups will be placed SOMEwhere.

Limiting it to the group's creator would sort of defeat the purpose of tags rather than categories. The value of tags comes when they're aggregated.

If tags disappeared when someone left the group, I think that would be enough (though I would also argue that we should be able to tag groups we're just watching, at least for our personal use). Other measures could be implemented if necessary, but I don't think there's been a problem with malicious book-tagging, so I don't see why groups would be different.

Yes, tagging by members/watchers of the group would be ideal, and would presumably work as well as tagging of a book only by people who have cataloged it.

>57 readafew:
I think that's a good idea. I don't know much about programming, but I imagine that such a requirement would mean that more of the code could be recycled from books.




>58 ExVivre:
I don't think that's a good idea at present. One person's idea of how a group should be tagged may end up being, well, not what most other people would tag it (see also, post 53). Really, the only way it would work and yet be restricted to so few people is if the group's creator could name admins/moderators/whatever who were then given tagging privileges.

And as for the "less high-minded types", there's already spam on LT, and casting a wide net of who gets to tag a group won't really make it significantly worse. Or, if it does, we have bigger problems than entry requirements for tagging.

A group suggester could also help the smaller groups get more traffic . . .

>60 readafew: Yes, I understood that the tags would leave - it's the issues created by malicious tags that would last.

>61 _Zoe_: I don't think there's been a problem with malicious book-tagging, so I don't see why groups would be different.

Check the various editions of the Bible, the Book of Mormon, the Koran, etc. Among tags for the KJV: fiction, myth, mythology, and auctorial wankery. Tags for the Book of Mormon include: apostasy, apostasy/heresy/poison, blasphemy, bullshit, crackpot, crank, crap, cult books, cult studies, cults, cults and sects, false religions, false teachings, fiction, fraud, heresy, make-believe, satanic, soporific, speculative fiction, spurious, tree died for this?!?!, very weird, wacky, and weird. And that's just among people who have catalogued those books.

I can see a lot of the political and religious groups going private really fast, and that's where a hefty chunk of Talk activity takes place.

>65 Katya0133: Best idea I've seen, yet, inclusive of my own.

66 Exvivre

Yes, there are a lot of bad tags on books, as well as a lot of personal tags that don't relate to anyone else (location, etc). Those are dealt with by the...I forget exactly what it's called, the long tail or whatever, because only one or two people tag them that vs many people tagging a book 'fiction' or whatever.

If all members could tag a group, my assumption is that they would be encouraged to tag it with everything they think is appropriate (including repeating things that other people have already tagged it)? That's how I always imagined it, anyway, so that we'd end up with a tag cloud of bigger and smaller tags, like books have, and maybe there could be automatic group recommendation based on tags, like there is with books (that is based on tags, isn't it?).

Those are just the sort of vague assumptions I had when this idea was floated. I also like the idea from msg 65.

>66 ExVivre:: I wouldn't necessarily label those tags as "malicious". Any hard core sceptic would perceive of those books as "mythology" - having (read) those books doesn't mean you believe them. And various christian denominations do consider mormonism as heretical. You may not like it, but it's within their rights.

I would never use tagging for evaluation myself - and I agree that some of them are not polite -, but even that is not necessarily "malicious" in itself.

>68 Nicole_VanK: I don't consider the book tags necessarily malicious because all is fair when discussing books. My point, poorly stated, is that those tags will be just as likely applied to the groups who hold the views, and all is not fair when labeling people. I know this could devolve into some argument between the dictates of polite society and the perceived rights of individuals to do whatever the hell they please - the same argument has been done over a thousand times on LT - but the cold reality is that users can brush off tags applied to mere objects much more easily than tags applied to their communities. I'd rather not see LT descend into a flaming tag war of heretics, faggots, facists, wingnuts, wankers, cultists, communists, blasphemers, and abominations.

The person best situated to moderate the tagging of a group is the group's creator, who picked the name, wrote the group description, and presumably has a vested interest. If the creator is inactive, let the members choose a new moderator - that should satisfy the more democratic leanings of some.

>70 Nicole_VanK:: Put it like that and I agree.

>69 ExVivre: But how would this not be prevented by restricting tagging to group members?