Tags, singular and plural: smoosh them all into one? August 13, 2014 1:04 PM   Subscribe

Is there a way to get rid of the consistent problem of a plural tag and a singular tag "splitting" the posts that appear under a tag subject?

For instance, I just ran a tag search to see what previous posters had used for "infographic" and got two results at the top of the list:

infographic
infographics

Right now, they're split evently (67 to 66), so I can't use the majority-rules dictum to decide which tag to use.

Can we get rid of singular vs. plural on the tags? Either through some tech wizardry or trying to get a culture of using one or the other?

This may seem like a silly request and I hope it isn't, but I've noticed it since I began posting more frequently and I wonder if it's been discussed? (I couldn't find a previous discussion here.)

I prefer the tech solution myself, where somehow when you clicked on a singular form of a tag it would also include the plural-tagged posts.

Thank you for your consideration.
posted by joseph conrad is fully awesome to Feature Requests at 1:04 PM (30 comments total) 1 user marked this as a favorite

Here's a previous conversation we had about this: tag = tags.
posted by pb (staff) at 1:05 PM on August 13, 2014 [1 favorite]


What he said; it's a complicated thing in a lot of ways.
posted by cortex (staff) at 1:07 PM on August 13, 2014


...so I can't use the majority-rules dictum to decide which tag to use.

I'm not sure that's the best way to decide which tag to use. Is your post about a single infographic or does it include several? If your post is about many I'd go with the plural. If it features a single I'd go with infographic. It might seem like the tags are interchangeable—and they might be for some uses—but they can have very different meanings. I think automatically combining them into the same tag results page would be a mistake.
posted by pb (staff) at 1:08 PM on August 13, 2014


And one thing I'd suggest is that if you're worried about a decision between two forms of a tag, just use both. Erring on the side of a couple more tags is probably the better option for discoverability under the current system.
posted by cortex (staff) at 1:10 PM on August 13, 2014 [8 favorites]


Sure, I can use both. Thanks for the comments so far, I'm going through the previous thread that pb linked in the first comment. I guess I don't like using both because it seems like it could be a bit more... "elegant"? It just seems a bit cluttery to use both singular and plural. I'll step back now because I'd like to hear the community's thoughts, from regular users.
posted by joseph conrad is fully awesome at 1:22 PM on August 13, 2014


some tech wizardry

This would require semantic understanding of the post. You can't just smoosh things together with unthinking rules. Consider the tag princes. De-pluralizing it automatically (even using a pluralization dictionary) would cause posts about royalty to get lumped in with posts about the artist.

Right now we have one kind of error (false negatives by omission), but automatic (de-)pluralization would introduce another (false positives).

In my experience the "related tags" sidebar does a pretty good job of picking up commonly used plural alternatives, so people searching by tag will usually see if there's a (non-)plural version they should check out as well. A small hassle but not too inconvenient.

I suppose a pluralization dictionary could be used to only merge tag search results in cases in which the other form also appears in the related tags results, suggesting semantic similarity.

For what it's worth I try to use singular and plural forms when tagging my own posts.
posted by jedicus at 1:25 PM on August 13, 2014 [1 favorite]


This is why everybody should just stick to only posting about things like sheep, fish, moose, deer, aircraft, and you.
posted by gauche at 1:41 PM on August 13, 2014 [7 favorites]


LOLbutt
LOLbutts
posted by cjorgensen at 1:46 PM on August 13, 2014 [3 favorites]


This is why everybody should just stick to only posting about things like sheep, fish, moose, deer, aircraft, and you.

You say that, and yet...
posted by jedicus at 1:49 PM on August 13, 2014 [2 favorites]


This "regular" user has a deep dislike of anything automagically being done to people's tags, even though I possess a librarian badge and toolkit and was a committed member of the MeFi backtagging effort a few years ago.
posted by donnagirl at 2:03 PM on August 13, 2014


Do you have any idea how many librarians and linguists have died fighting this battle?
DO. YOU?
posted by Foci for Analysis at 2:45 PM on August 13, 2014 [7 favorites]


pony
ponies
posted by Kabanos at 3:11 PM on August 13, 2014 [2 favorites]


Yeah, doing this automatically would be problematic in certain edge cases. For example, humor isn't at all the same as humors. And then there's the extra problem of humor vs. humour.
posted by Kattullus at 4:20 PM on August 13, 2014


What if we switched Metafilter to use a language without plurals?
posted by curious nu at 5:00 PM on August 13, 2014 [4 favorites]


Automatic depluralization is totally bananas, but not at all banana.
posted by aubilenon at 5:17 PM on August 13, 2014 [6 favorites]


infographicae
posted by odinsdream at 6:15 PM on August 13, 2014 [3 favorites]


In my experience the "related tags" sidebar does a pretty good job of picking up commonly used plural alternatives

It does, but only if (and only because) a subset of previous posts have used both the singular and plural tags.

In the example given: on Metafilter infographic 68; infographics 67; but in the related tags sidebar, only 6 (across all sites?) have used both.

In both cases too the plural/singular alternative is midway down the list of related tags. There's a bit of a discoverability iceberg here -- if you arrive at the "infographic" tag page it's not at all obvious that there's actually an equal number of probably-equally-relevant posts hiding behind the related "infographics" tag.

The tag search for infographic does surface this more -- and in this case it feels like the fuzzyness of the search is actually more useful.

So: could it be useful show tag search results in the sidebar, alongside the related tags? Something like "similar tags"?

Or if it's expensive to run a search on each request for a tag page, maybe at least add a link to a tag search, similar to how the "it looks like you're searching for a phrase" result page links to a Google search?

FWIW, I also find it cluttery and somehow inelegant to tag with both singular and plural forms; but if it helps (and it certainly seems that it does) then I'll do it.
posted by We had a deal, Kyle at 7:24 PM on August 13, 2014 [1 favorite]


> Is your post about a single infographic or does it include several? If your post is about many I'd go with the plural.

That's not a good example because when searching, rather than tagging, the user's interest is in finding infographics generally; the quantity in any post is less likely to be relevant.

I agree with the thrust of the argument, though; Automatically pairing singular and plural tags is going to be really hard to do well without dictionary rules. The easiest way to do that, short of being Google and providing functional search suggestions, would be to set up a second volunteer-based tag cleanup project based on suggested/prioritized tags. (This would also be a way to fix misspelled tags, remove spurious tags, etc.)

And by "easiest", I mean it would not be easy at all, so I'm not asking we do this.
posted by ardgedee at 3:57 AM on August 14, 2014


I recently asked a question about toys and ended up tagging it "toys toy gifts gift stickers sticker". I didn't think that was great, but I had no idea whether to use the singular or the plural version and thought the mefi consensus in general is that more tags is better than less, so I opted to use both.

I think it would be awesome if there was some way to do this better. Either a community consensus, or an automatic way that if you look for the tag "toy" you also see results for "toys" (it doesn't have to automatically add tags to post, it could just show the results for the plural in the search results if you look for the singular). I see the argument that people make that sometimes people may really intend the singular and not the plural, but then, I think that for searching/discoverability people will not exactly know what other peoples thought process was when they were adding tags and I actually think most people who click through the "infographic" tag will be equally interested in posts with the tag "infographics" and indeed may be surprised if they find out that there are so many tags about infographics that are not in the infographic result list.
posted by blub at 3:59 AM on August 14, 2014 [1 favorite]


CURATE ALL THE THINGS
posted by jeffamaphone at 10:28 AM on August 14, 2014 [1 favorite]


I'd like to hear the community's thoughts, from regular users.

The tagging system is broken. It works okay for some things, but it has an incurable problem of entropy. See above, where Cortex mentioned it being complicated and tellingly used the phrase "under the current system" [emphasis mine]? The mods know this. Tags need an eventual overhaul.

Last week somebody posted an AskMe about getting started in photography. It's an obvious topic that you'd think there would be a bunch of previous threads on, and in fact there are. But none of them are tagged in any way that is either (1) consistent or (2) apparent to someone searching. The photography tag itself has nearly two thousand items. Nobody is going to comb through all those, nor should anyone have to. This person created a new tag (beginningphotography) but I don't see any reason to think the next four people will think to look for that, so we'll keep seeing these threads posted as if they're new (as opposed to, "I've seen the 2009 and 2012 threads but am wondering if there's new information").

So with that said, my philosophy on tags is to just do my best, use common sense, and don't worry too much because the system is already broken in ways you or I probably can't make worse. As an individual concern, it's not a problem worth much attention.
posted by cribcage at 12:40 PM on August 14, 2014 [1 favorite]


The tagging system is broken.
I don't see it as broken, just not perfect, and all tagging I have interacted with works the same way. It is crowd sourced data and that will always have issues, but that does not make it useless. Whenever I am looking at tags (on MeFi or elsewhere) I always look at many versions of the same tag if I am doing general research. If I am looking for a specific answer, then I go to similar tags if I don't find my answer in the first tag.

The photography tag itself has nearly two thousand items. Nobody is going to comb through all those
Related tags exist to help you narrow down the questions to ones you are actually looking for.

If there are no posts tagged 'photography+beginner', then I would do a search. Also, when I preview a question about beginning photography, I get a list of some related posts:
Got baby, got DSLR. Now what? October 25, 2011
How to take these photos. July 16, 2011
Advice for a first-time DSLR owner? August 4, 2009
EASY secrets of taking great photos July 16, 2009
Pokémon Snap was about as far as I got with... October 27, 2008

It is interesting that the most recent one is 2011.
posted by soelo at 9:24 AM on August 15, 2014


decide which tag to use
Why not both?
posted by soelo at 9:26 AM on August 15, 2014


Yeah, searching by tags is largely an exercise in futility for me, as well. Suppose I specifically want to find posts about the infographics Randall Munroe, the creator of xkcd, has put together? I know there have been a few. Let's say I actually realize infographics are a good tag for that kind of stuff (not a given, because although it is a great term, I might not think of it):

67 posts tagged with infographic
68 posts tagged with infographics

Okay, well, how about xkcd?
53 posts tagged with xkcd

Ugh. Lots to sift through there. How to narrow that down more?

3 posts are tagged with RandallMunroe.

Now we are getting somewhere! Oh, wait. Only one Is about an infographic (the map of online communities). I know there are others! There was one I remember about wealth and how unequally it is distributed, with blocks in it?

2 posts tagged with wealthsdistribution...and neither one is on that Randall Munroe chart. Damn.

This is why I don't search under tags much. It's too frustrating, especially when I can go on Google and just type Xkcd infographics and find an entire blog devoted to nothing but that subject!

Just throwing this out there, but here's a thought (that might not help at all): would it maybe improve the search function if there was an added field for categorizing posts on Metafilter (blue), like there is on AskMe?

Currently, on the green, you can choose from these categories:

Clothing, beauty and fashion
Computers and internet
Education
Food and drink
Grab bag
Health and fitness
Home and garden
Human relations
Law and government
Media and arts
Pets and animals
Religion and philosophy
Science and nature
Shopping
Society and culture
Sports, hobbies and recreation
technology
Travel and transportation
Work and money
Writing and language

Seems like maybe being able to choose a category for your posts on the front end, when you make a post on the blue, would help with this from the other direction; someone searching for a specific tag could at least narrow the field down a little by searching tags only within that category.
posted by misha at 10:09 AM on August 15, 2014


An addendum: "Math, statistics and datawankery" might be a useful category to have, also.
posted by misha at 10:13 AM on August 15, 2014


Suppose I specifically want to find posts about the infographics Randall Munroe, the creator of xkcd, has put together?

That's probably a better use case for keyword search than tag-based search, though; you know the specific name of the creator, and his handle, and presumably his common domains, and can do a search for those specific things. Where tags have greater general utility is in search for a category or type of thing when total specificity is either unavailable or undesirable. Searching for photography and searching for the photography of Ansel Adams are, for example, two significantly different projects.
posted by cortex (staff) at 11:03 AM on August 15, 2014


http://www.metafilter.com/tags/xkcd+infographic

Why wouldn't you use that instead of hunting through the separate xkcd and infographic tags?

You can't compare metafilter's tags with google's search and claim the search is better because it found another page that is not in either metafilter or xkcd. At least compare it with metafilter's search that gives you four posts and 13 comments.
posted by soelo at 11:38 AM on August 15, 2014


My point was, and I obviously didn't explain this well because I have brain fog today, I DON'T USE TAG SEARCH. It is not efficient, and it is not worth the frustration to me. Personally, I use keyword search, like cortex suggests, or I do a Google search limited to Metafilter or whatever.

BUT, since this post was about someone being specifically frustrated with tag search, mainly, I was trying to come up with suggestions that might improve that specific function. I said right in my comment that my suggestion might not help at all. I really am not firing on all cylinders in any case.
posted by misha at 11:56 AM on August 15, 2014


And seolo, yep, you're right. Neither one of the searches you put up shows all the Metafilter posts about Randall Monroe's infographics, either. Just off the top of my head, the "map of internet communities" post is missing, for example. So I would agree that your examples better illustrate tag search limitations.
posted by misha at 12:06 PM on August 15, 2014


I DON'T USE TAG SEARCH
I understand that. You don't use it because a single tag doesn't find what you want. My point is that you need to use more than one tag sometimes.

Neither one of the "map of internet communities" posts 1 2 have the word "infographic" in them anywhere. A search using that term will not find them, either. Someone looking for them would have to change their search terms a bit to keep looking for it, so in that way I see tags and search as equally flawed, but both very useful.

To be fair to the posters, that was nearly four years ago (!) and infographic was not such a trendy word back then.
posted by soelo at 12:42 PM on August 15, 2014


« Older Killfile scripts?   |   What has anyone ever learned from Gawker? Newer »

You are not logged in, either login or create an account to post comments