Can I search using tagmashes I create? If so, where?
neat
Don't get it. If I want X, I put that in and its enough to get everything else.
2> Just do a site-wide search (from the Search tab) on tags entering more than one separated by commas.
>2
Yes you still can. The same place as always, on the Search page in the Tag box (don't forget to use commas to separate the tags).
>4
As Tim said in the Blog post, it's like the tag mirror except for mashes rather than single tags. You see a list of mashes that apply to your books.
# 6 - Why ? One term does it all. Don't use Tag Mirror.
#7 - Don't you ever want to see the intersection of two tags? I do that quite often.
I suppose it depends on what kind of tags you use; if I want to find biographies of people who lived in the 16th century, a tagmash of "16th century, biography" would do the trick, but my books in that group would be tagged with both "16th century" and "biography". If you are tagging yours "16th century biography" you wouldn't need this tool, but would have far more tags, since you would need a separate one for "17th century biography", "18th century biography", etc.
Not only that, both the Tag Mirror and this tool lets you see what books you have by other peoples tags.
>8
Tags are great for personal use. That's why they are entered. But they have this emergent value when you have millions of them--they become something like a classification. So, the "chick lit" tag page isn't just your tags, but this wonderful assemblage of what thousands of other people think is chick lit. Tagmash extends that idea into multiple tags.
For more on tags, I've got a short talk up on YouTube:
http://www.youtube.com/watch?v=Ok7YinIhT...Tim
Can there be a way to exclude tagmashes with a particular tag, or, better, tagmashes that consist of only two tags, one of which is that particular tag? Basically, I've found that Young Adult tagmashes have about as much interest as Fiction tagmashes (i.e., almost none--I might as well be looking at a plain old list of tags), but they dominate my whole list (together with YA, children's, juvenile, etc.).
# 8, No I would tag "16C", "Biography", and so would just have to search my tags or the tags of others for what I want. If it was a permanent interest I would enter it into Tag Watch to find books that others have tagged similarly.
But, realistically the important single term is the one I would use. If it was more important to be Bio or from the 16C, that would drive my search.
# 9 - Why would I care ? My tags are for me, and their tags are for them. If they are similar then they would have been combined, and picked up by Tag Watch. If not, then its not something that I want to deal with.
Looking at the tag mash that is created from my books it combines terms that are basically the same: fantasy, horror, vampires - thats like saying black,black,black. They all seem to be the very similar, and the results always seem to have all the Harry Potters and/or Tolkiens.
It also seems to be confused by my tags. I often tag to indicate time period of the story. So I use historical and modern day. It has chopped it to modern and added it to fantasy and made modern fantasy: Harry Potter ! I am not interested in 'modern' fantasy (as a literary movement) in any event.
#10 - But that assumes I have an interest in what others do with their libraries.
I am quite happy with Tag Watch and the various book threads to let me know about new-to-me books.
Four, out of about 16k tags.
#13: It's at least as valid as Tim's infamous "No." review
12> It's true that the vast majority of my tags are some combination of fantasy and YA, and if I look at them they nearly all include the Potters and Tolkiens. I do have a goodly selection of non-fiction but none of that comes up, possibly because they're spread about in subjects?
>16
I think the overwhelming majority of tagmashes with big overlaps in my list are based on a handful of popular books I own (Harry Potter, Terry Pratchett, Patrick O'Brian). Obscure books, even if you have a lot of them, are obviously much less likely to appear in tagmashes, partly because they have fewer tags each, and partly because people are less likely to generate tagmashes in those areas in the first place.
The mash overlap is interesting, like it. (Coming from me, high praise ... but I'm not quite over on the dark side of social tagging yet ....)
>16,17
Yes, a lot of the absence of non-fiction must be the "no-one's ever done a search for this" effect - I made a new tagmash for "Germany, railways" this morning, and it has now popped up about a third of the way up my overlap list.
Presumably all the classics tagmashes at the top of Tim's list are from him testing the tagmash feature?
BTW - looking at Tim's overlap list, I notice one wonderfully misconceived tagmash: "1, war, world" - it returns all sorts of wonderful things, from Homer to Lolita, but I can only see one title in the list that matches what the creator of the search must have been looking for.
This message has been deleted by its author.
12> Why would I care ? I never said you should. If you have your catalog perfectly tagged to your needs and have no curiosity in how others think about tagging fine. Your posts just seem to have the feeling of "Why would ANYONE want this kind of information? How would this be useful?"
Probably a dumb question, but what (and where) is Tag Watch?
I made a new tagmash for "Germany, railways" this morning, and it has now popped up about a third of the way up my overlap list.You sir, have a weird library ;)
Presumably all the classics tagmashes at the top of Tim's list are from him testing the tagmash feature?Yes, probably. I've also spent a lot of time with the tags I use commonly in my talks—GLBT, chick lit, etc. Genres and interest areas are fertile ground for this feature.
Message edited by its author, Sep 16, 2009, 12:04pm.
>22
It's one of the modules on the Home page - "Tag watch provides a fun way to discover new books on topics you might be interested in. Edit this module to add a tag, or list of tags, and every time someone on LT applies that tag to a book, you'll see it here! " (
http://www.librarything.com/wiki/index.p... )
Again, "D&D, RPG, fantasy" is showing up on my tagmash list, and not only doesn't it link to a valid tagmash, I can't figure out how to produce it as a tagmash.
> You sir, have a weird library ;)
Thanks! :-)
Whatever its weaknesses, the tagmash overlap thing is a great tool for bringing that weirdness out. It's a shame about "elephants, zombies", though - that looked as though it was going to tell me something new and unexpected about my library, but it just seems to be the result of someone having an accident with Power Edit.
> 23
I wasn't getting this: so, the tagmashes that appear ~depend upon~ the specific tagmashes that have been entered previously, either (a) because someone tagged a specific book that way, or (b) someone did a search on that tagmash combination?
Why (apart from the inevitable processing power constraint)* wouldn't the algorithm simply find the set of all tags applied to a book, whether applied singly or in a -mash, and then generate the set of tagmashes from that?
Or perhaps I'm misunderstanding what was already said. 'Pologies if that's so.
* I fully recognise that this parenthetical qualifier is a doozy.
Edited for clarity (which removed the basis for Asimov's touchstone, heh).
Message edited by its author, Sep 16, 2009, 2:12pm.
Why (apart from the inevitable processing power constraint)* wouldn't the algorithm simply find the set of all tags applied to a book, whether applied singly or in a -mash, and then generate the set of tagmashes from that?Two answers:
1. The processing constraint is huge. It's basically every tag in the system times every tag it ever coincides with. We currently have 2,235,061 distinct tags.* I'd bet it's something like that times 100. Obviously some tags, like fiction, go with almost all of them.
2. Some tagmashes make sense and some don't. The ones that don't make sense "bathtubs, Isle of Man, dark" don't get made. That's a good thing--they won't clutter up any lists. Tagmashes are, in a sense, curated. They exist because someone thinks they're interesting.
*That's
combined tags, so fiction is also Fiction and fictino. Total distinct tags without combination is 3,475,155.
Message edited by its author, Sep 16, 2009, 2:30pm.
The "curated" aspect makes sense, and I see that goes back to the emergent character of tags you mentioned.
Curious to think about how to identify interesting tagmashes that have not yet been actualised, rather than simply waiting for someone to come along and tag a book or make a search. But I suppose that's part of what makes the emergent property of mass data so interesting: it does that for MY library, though I haven't used those tagmashes myself yet.
Hmmn. If I could hack into the LT of the Gods, we might be in business. That, or import the tagmash from LT 100 years from now. But I suppose there are processing constraints there, too.
Oooh, LT of the Gods!
Athena's library - now there's one I'd like to cruise!
27> I don't believe that you get tagmashes in this feature from (a) tags on books, but only from (b) tag mashes that have been searched for.
28> I searched for "bathtubs, Isle of Man, dark" but it didn't find any books. :(
But I have been entering tagmashes that might describe certain parts of my library. It's slow going but I hope it pays off. I'm suprised no one has mashed on history, germany, france before though.
A lot of people seem to think it's additive, eg. "sf, science fiction." Rather than yielding either, it yields only the intersection, which isn't actually helpful.
> 33
Does the feature's use of combined tags obviate the need for an additive search, such as the union of "sf" and "science fiction"?
If not, I definitely would want a way to do that, esp once my initial search hasn't returned something specific I'm looking for.
Obviate it? No, it doesn't exist.
I can see why it's Tim favourite feature - I hadn't even thought about using tagmashes as a way of finding books of possible interest. Loving it :-)
Now all I have to do is enter many years of Read But Unowned books to balance out the owned books already entered in my catalogue (mainly genre series - it is so annoying when one is missing at the library!). My tagmash page makes that too obvious!
34>
What he means is that tagmashes work on AND, but some people seem to think they work on OR. There is no OR search for tags.
>37.
Using De Morgan's law, there kinda is. If you're looking for not x AND not y, you're also looking for not (x OR y).
But, that's an implicit case...
I would like to see a way to favorite a tagmash. I've been looking at my tagmash page, and there are several really good ones that I would like to be able to quickly spot for re-visiting. Thoughts?
As "robots,sheep" only gives three works, and I have two of them already.... I fear I have no option but to get the third.... aaargh!
By the time I read that my search for "carnivores, mystic" had not been done before and that I would have to wait, it was reported that there were no matches.
Robert
But that didn't capture any of these:
* Don't Eat the Mystery Meat! by Tom B. Stone (2 copies)
* Inside Outside: The Meat Cove Mystery by Livingston, Neal (1 copy)
* The Mystery of the Mystery Meat (Pretty Freekin Scary) by Flesh, Chris P. (4 copies)
* DON'T EAT THE MYSTERY MEAT (Graveyard School) by Thacker, Nola (2 copies)
* Graveyard School #1: Don't Eat the Mystery Meat! by Stone, Tom B. (1 copy)
Fortunately I haven't had much to do with mystery meat since I got out of the Navy, but I think it is still a risk we should be able to look into.
Robert
Why would a tagmesh of "Western Australia, history" produce Great Expectations (Charles Dickens)?
Rob
>44
It's a shame you can't easily find out who the two people are who tagged it "Western Australia". Very odd, considering that Magwitch is always referred to as being in New South Wales, and there's only one other mention of Australia ("'Deep,' said Wemmick, 'as Australia.'"
You'd have thought that someone using the tag for physical location of their book would tag it "Australia" or "Perth", rather than with the state name.
Even odder, only one person each has tagged the book "Australia" or "newsouthwales" (sic.)
At least it wasn't "West Australia", which the
Sandgropers dislike.
Maybe those two users have the book stored in Western Australia? Tagmash probably shouldn't bring up works with such faint associations though.
The problem is, what does faint mean? Sometimes a faint association is someone noticing something obscure, but worthwhile.
Here's one from my list: England, boy, juvenile literature, magic, orphans. So, if they wanted to find the Harry Potter books why didn't they just do a title search for Harry Potter?
48? I would have thought that 2 out of however many tags Great Expectations has would be faint. But I take you point about obscure but worthwhile. It would be boring if this were limited to just the most common tags.
> 49: Hehehe. But somebody possibly checked if there were other books with these very same ingredients.
Message edited by its author, Sep 17, 2009, 9:31am.
>48,50
Yes, if you changed the algorithm to ignore the false/obscure link between
Great Expectations and Western Australia, it would also fail to show up in "Australia, convicts, fiction", where it is definitely relevant (if a little peripherally). And false positives are less of an evil in a search system than false negatives, so we should probably learn to live with it. It's too late now to develop a system that separates tags for the physical copy from tags for the work.
It seems the tagmash Russian, teen brings up the usual Potter, LOTR, Narnia books. I looked at one of the Narnia books and it has one Russian tag (maybe that persons book is in Russian). So maybe there should be some threshold below which a tag doesn't count.
It's too late now to develop a system that separates tags for the physical copy from tags for the work.
And it undermines the whole point. Various schemes have been hatched to "facet" tags—these are about the edition, these are about my opinions, these are about the content, etc. They undermine the point of tagging—extreme ease of use.
I'm beginning to see that the really useful thing about Tagmash Overlap could be that it educates us in how other people use tags, and thereby helps us use them more effectively for searching.
39: I'd like a tagmash watch.
The "bathtubs, Isle of Man, dark" tagmash is obviously a hint to detective authors that a book in which a corpse in a bathtub is washed ashore on the Isle of Man (in the dark) has yet to be written.
49>
Probably because they wanted to find out if there were any other books like the HP books in those particular ways.
52>
It's too late now to develop a system that separates tags for the physical copy from tags for the work.
For me, and for some others, Collections accomplishes that to some extent; things like read-status and owned-status lend themselves, IMO, more naturally to Collections than to tags. Of course, now that I've added my wishlist, I have them redundantly labeled (tags and collections), in case the collection data goes missing and because the collection information still doesn't export.
It would be boring if this were limited to just the most common tags.
Except it is, at least as far as I'm concerned. Until there's a way to eliminate YA, Young Adult, Children's, Juvenile, etc. so that I can actually see tagmashes about content, this feature is entirely useless to me.
While I agree with you that the
Sorted by our attempt at "interestingness." at the tagmash page isn't really working for me either, I think the ability to search for a tagmash is great.
(edited to correct html)
Message edited by its author, Sep 17, 2009, 12:37pm.
Playing with this feature caused me to fill up my hold quota at the library with books I wouldn't have thought of - so it gets a big thumbs up from me! :-)
Are we meant to be able to see the tagmash overlap of other members? If I visit someone else's tagmash overlap page, I initially get a "This page is only open to (their username)" message, but if I wait a few seconds, it still loads their tagmash overlap data.
Also, the page currently sorts by interestingness (score one for
hairy feet, rings.) Might we ever see other sort options, such as how many tags were used in the tagmash (I've personally found the ones
with the most tags to be the most interesting,) how many books we have in the tagmash, or what percentage of books we have out of the total number of books in the tagmash?
I
love really like this feature. Even with my rather small and narrowly focused library, it's a great way to find new books in ways I never would have thought to search.
Edited because, upon further consideration, I do not want to marry it. Also, lost brackets.Message edited by its author, Sep 17, 2009, 1:16pm.
A problem with it is that there are many repeats of tagmashes that bring up essentially the same books. So on my first page I have "fantasy, youth", "fantasy, kids", "children, fantasy", "children's, speculative fiction", "children's literature, fantasy", and on and on. Could the algorithm lower the priority of books that have already appeared in previous tagmashes so we can see a wider variety.
Wow. Why don't I get tagmashes like "hairy feet, rings" suggested? My #1 is "german, comics". Now being a cartoonist by trade I could find that mildly interesting if it actually dealt with German comics - a topic a know little enough about. But in fact mostly lists well known European comics (Tintin, Asterix) that some people may have tagged German because they happen to have a German edition. BORING!!!
The one that has me shaking my head is the fact that four people have tagged
The Other Boleyn Girl as "non-fiction".
Now, if there was just a way to see them in your catalog... :-)
53> That's not going to fix the problem; the fundamental issue is that a certain set of language and location tags have global usage, but also some fundamentally library-specific usage. But I want Thai, science fiction to come up with
The Fallen Country even if that means it comes up with
The War of the Worlds.
>65
If it was only this... I keep finding Gregory's books in the non-fiction sections in a lot of places... Now - I like her style (might have something to do with the fact that I am a Tudor fan) but she is a novelist, not a genuine historian.
>68 Exactly - and takes huge liberties with history, so anyone relying on her works for historical information is riding for a fall.
It's possible that some "wrong" tags are just wrong to you. For example, if writing a term paper on non-fiction and fiction about Tudor England I might use "non-fiction" for every book I am going to use in the paper. That is, it has some special meaning to me.
True, and I might do that myself. However, I suspect I would remove them afterwards.
>28. "That's combined tags, so fiction is also Fiction and fictino."
You, sir, are clearly unaware that a "fictino" is a subatomic particle emitted by fiction. I demand a tag separation!!
It can pass through ten feet of lead without interaction, but is attracted to plotons.
The relationship is complex, fiction being generated by the collision of inspirons. These, sadly, are very rare and often strike in the wrong spot.
And if a fictino meets a non-fictino (its antiparticle) they'll annihilate each other in a burst of read-iation.
Wow, I just Ow-ed out loud.
It's possible that some "wrong" tags are just wrong to you.
People would quibble with my non-fiction tag on Beowulf, for example, but it's there because I have the NCE version which is forever being combined with the main work ... It's not always apparent why people tag and many don't do it with the site-wide connection implications foremost so it's impossible to say that something is tagged 'wrongly' even if you wouldn't tag it that way yourself. I'm doing a creative writing course this year and am starting to tag things that have useful bits of writing in them for whatever reason ... if one of my fiction books has something in it that will help my non-fiction writing, or my biographical writing, it will be tagged accordingly. That's not 'wrong'.
Right. That's the problem with using tags beyond a person. The cure is *numbers*—HUGE numbers. Tagmash, unfortunately, prefers "reaching" to being safe, but the statistics are all there for a better feature.
I work in a public library, and am finding that LT's tagmash feature is more useful than the somewhat scattershot subject headings that are assigned to books in my library's OPAC.
I'm wondering if there is a way to have LT reference an OPAC when doing a tagmash search, so that only things belonging to my library's system show up. I know that LT for Librarians is one way to do this, but I don't know if my system will be incorporating this anytime soon.
Is there a way to do this personally, not on an institutional level, but just by feeding LT an OPAC URL?
Sorry if this has been addressed before and I missed the answer.
Yeah, the only real answer is LibraryThing for Libraries (http://www.librarything.com/forlibraries). In the future, we may allow libraries to upload their holdings for free, and get various nice things out of it. I'd like to do this, but we have to figure out how to make some money for it—holding hundreds of millions of records for paying libraries is already quite a challenge.
Message edited by its author, Sep 26, 2009, 4:25pm.
>81
Thanks, Tim-- I do understand that the proposed task is pretty daunting, and it can't be done free. May I say, I have found LT so useful already for reader's advisory, my lifetime subscription is a great investment in your product.
What library are you with anyway? (if you don't mind saying)
San Mateo County, California. It is possible that we may incorporate LT at some point, but I am new to the system, and an extra-help librarian to boot, hence I have some hesitation about recommending OPAC improvements and such.
http://smcl.org/flash.htmlIs there any way to export the tagmash output in CVS, or some other format?
Message edited by its author, Sep 28, 2009, 6:11pm.
No, why?
Ok, I was just checking out this whole tagmashing thing, just another feature I seem not to have known about.
I was interested in the questions asked in #62 "Are we meant to be able to see the tagmash overlap of other members? If I visit someone else's tagmash overlap page, I initially get a "This page is only open to (their username)" message, but if I wait a few seconds, it still loads their tagmash overlap data."
I have read through the thread, but have not seen an answer. Which way is it suppose to work?
>87
Same message shows up sometimes shortly on the series page also if there are a lot of series to be loaded... so I think it is just some strange loading message. Might be wrong though.
>86 It would be interesting to import these results into a spreadsheet, and sort or analyze by various methods. I know that you have ordered them by some "interestingness" criteria, but perhaps we could discover some patterns in the data for ourselves. For instance, sorting by the raw number of entries in each tagmash could show us our proclivity for certain subjects.
(back to top)
