Keywords: Allow public voting and downvoting. Allow the wider community to decide what gets tagged.

@bderoes 

heavily involved with keywords  ??

add a @ keyword_expert   (No space)
https://community-imdb.sprinklr.com/users/607a75d3934ad304571aa14b

Last Activity: 2 days ago

.

@Peter_pbn

You can only request a keyword, and then I assume a moderator then approves or rejects it. Or you can request a keyword deletion, and then a moderator approves or rejects it. The current system puts a lot of responsibility on moderators to know that the user suggesting the keyword is speaking accurately, when they probably haven't seen the show or movie in question.

Contrast with this, where the process is entirely user-led.

You can see who voted for things and how many votes they have. Maybe IMDB could keep usernames entirely anonymous in a similar system.

@bradley_kent

"Just don't think it would work on IMDb, and would probably make things even more confused and convoluted.  Besides, voting already seems to exist (in a way) just by the acceptance or rejection of a  keyword."

That's not voting though. That's suggesting a keyword and then hoping a site moderator accepts it. Once they've accepted it, it's there until someone else tries to get it removed. But I suspect this doesn't happen very often, leaving many (especially more popular series/films) a confused mess of inaccurate tagging.

"On one point, " A minimum threshold of +3 before a tag is visible on the main page," I strongy suspect that most keywords, Including the valuable "reference-to-" keywords, only have one title.  To eliminate a keyword until it has three titles would severely limit searching such keywords.  Oftentimes, those keywords with just one or two titles are more valuable."

If IMDB incorporated this system, I'm quite confident engagement in tagging would majorly increase. Rateyourmusic has a smaller pool of users and yet, substantially more tagging (even if it's not a perfect comparison).

"I doubt that turning keywords into a popularity contest would help.  Yes, keywords can be "general," but they mist also be "specific.""

If you look at The Expanse, the keywords are highly inaccurate in some cases - I suspect because the staff who approved them simply haven't seen the show, and took the explanation at face value.

@bradley_kent 

 I strongy suspect that most keywords, Including the valuable "reference-to-" keywords, only have one title.  To eliminate a keyword until it has three titles would severely limit searching.  Oftentimes, those keywords with just one or two titles are more valuable.

Can you clarify which types of keywords with the "reference-to" prefix you find "valuable," and why you find them valuable?

In my opinion, a couple of users on IMDb have been severely overusing (practically abusing) the "reference-to" prefix, applying it in many situations where it shouldn't be applied. As you have pointed out, this has resulted in thousands of "reference-to" keywords that have only been applied to one title.

I remember you previously saying that, when adding keywords, you try to use pre-established keywords whenever possible. I do the same.

I refer to keywords that have only been applied to a single title as "orphaned" keywords. And there is really no reason for them. If a keyword can't be added to more than one title, then in most cases it shouldn't be added at all. 

There's nothing "valuable" about orphaned keywords. For one thing, it disrupts searching based on multiple keywords in combination with each other. 

For example, if a movie includes an important plot point where a character speaks about the death of their pet dog (but the death or even the dog is not actually shown), the appropriate keyword to apply would be "death-of-dog" or "death-of-pet-dog." If a user instead adds the keyword "reference-to-death-of-pet-dog," then the user is doing it wrong. In that example, the use of the "reference-to" prefix bifurcates keywords and prevents that title from being picked up in search results using the keywords-in-combination search functions. In other words, people searching the keyword "death-of-pet-dog" in combination with other keywords will not be able to pick up the title with the keyword containing the "reference-to" prefix. There is no good reason to bifurcate keywords like this.

I suppose someone could argue that maybe someone would want to search for movies that merely discuss the death of a pet dog without showing it. But if that's the case, why limit the approach to the somewhat vague "reference-to" prefix? In other words, why not bifurcate keywords even further by applying prefixes like "discussion-of-death-of-pet-dog" or "textual-reference-to-death-of-pet-dog" or "spoken-reference-to-death-of-pet-dog" or "allegorical-reference-to-death-of-pet-dog" or "musical-reference-to-death-of-pet-dog?" Where would the absurdity end?

As I think you will likely agree, consolidation of keywords is a very important goal. After all, you have advocated for consolidating keywords like "murder" and "murderer," even though these are two different plot concepts and it is possible to have one in a movie without the other.  The same could certainly be said of "reference-to-murder" and "murder" -- they are both plot keywords specifically involving a murder. If something is an important enough plot point to be added as a keyword, then it should be added without the "reference-to" prefix. And on the other hand, if a murder is briefly mentioned in a film and it's not important to the plot of the film, then it probably shouldn't be added as a keyword -- otherwise, it is arguably spam.

According to the IMDb guidelines themselves, the only use of the "reference-to" prefix that is definitely appropriate is to refer to something that is already (or could be) itself listed with its own entry in the database (e.g., titles, characters, people, networks, companies). Any other use of the "reference-to" prefix is probably overdoing it.

A minority of users apparently believe that "reference-to" should precede every keyword that is itself a proper noun. Other users (also a minority) apparently believe that "reference-to" should precede every plot concept that is spoken about but not shown on the screen. (This reminds me of the users who like to add the "character-says" prefix, which is also way overused on IMDb.)  Still other users (also a minority) apparently believe that "reference-to" should precede every plot concept that is spoken about and/or referenced in writing or in a picture in a film, but without showing it as an action in the film. In my opinion these various users are  misunderstanding how keywords work and are intended to be used.

The bottom line is that if a plot point is important enough (either within the context of the title or in the overall context of IMDb) to be included as a keyword, then it should be added, without applying the "reference-to" prefix. The only known exception (already expressly identified by IMDb staff) is when a keyword involves another title, character, person, or company with its own entry on IMDb.

@keyword_expert 

In your example from rateyourmusic.com, "Merseybeat" and "Pop Rock" are not keywords, but rather genres. 

Correct. RYM also has 'descriptors' which are closer to keywords which people also vote on similarly.

Both of these things are already available at IMDb. 

It doesn't function in the same way. There's no barrier to entry for keywords.

Only a very small percentage of keyword additions and deletions are actually reviewed by staff. This includes keywords that include certain adult terms like "sex," "rape," "rapist," etc. The rest of keywords will be automatically added or deleted in most cases, generally just a few minutes after they are submitted. Some exceptions can include keywords added by new accounts and by user accounts that have been flagged (either automatically or manually by staff) for further monitoring and/or review.

How does deletion work? If I were to add "Dystopia" to Black Sails, which is plainly and obviously inaccurate, would a staff member have to delete it?

I'm not sure what you mean by "main page." If you mean the title page for a movie or show, that page already displays only the top five keywords for a title based on user voting (so long as the keywords for that title have actually been voted on). So this is already the way the system works, except it simply displays the top-rated keywords, without any minimum threshold.

By "main page" I mean visible in tag filtering. If I add, say, "science fiction" to Game of Thrones it should only be actually be considered a tag if at least a couple of people agree with me. This reduces tag-bloat, but makes them much more accurate as Game of Thrones is quite obviously, not science fiction.

Only a very small percentage of keyword additions and deletions are actually reviewed by staff. This includes keywords that include certain adult terms like "sex," "rape," "rapist," etc. The rest of keywords will be automatically added or deleted in most cases, generally just a few minutes after they are submitted.

The trouble is, this means that people can inaccurately represent movies and series and dilute the relevancy of filtering via keywords.

Did you see my point about The Expanse? A lot of the tags there are a mess, I'll quote myself:

"One of the tags [on the Expanse] includes "interracial marriage" and like, well, yes - there's a character in S3 that is in an interracial marriage. It is a completely plot minor, borderline irrelevant part of their character though and I feel that specifically including that is very misleading as to its overall presence in the show, or even S3.

There's also a lot of other questionable tags on that series. "totalitarianism", "dystopia", "alternate history" (????), "shared universe" (with what?), "bounty hunter" (what bounty hunter???). If this was instead a straight upvote/downvote system all of them, I'm confident, would be downvoted."

@Skavau 

How does deletion work? If I were to add "Dystopia" to Black Sails, which is plainly and obviously inaccurate, would a staff member have to delete it?

Any other user can delete that keyword. To delete one or more keywords, just go to a title's keyword page, then click "edit," then choose "correct/delete."

By "main page" I mean visible in tag filtering.

You are still not being clear or specific enough. Since you mentioned "filtering," are you talking about keyword searches on IMDb? Or are you talking about the upvoting and downvoting of keywords on a title's keyword page?

If I add, say, "science fiction" to Game of Thrones it should only be actually be considered a tag if at least a couple of people agree with me. This reduces tag-bloat, but makes them much more accurate as Game of Thrones is quite obviously, not science fiction.

That's a bad example, because IMDb blocks genre keywords, so "science-fiction" literally can't be added as a keyword to any title. (Edit: I just checked, and apparently "science-fiction" is not blocked like some other genre keywords. It absolutely should be, though. I have posted before about the need to block more genre-specific keywords, but so far this has not happened.)

While keyword spam is definitely a problem, the way to deal with it is to downvote and/or correct or delete improper keywords. Any user, including you, can do this.

"One of the tags [on the Expanse] includes "interracial marriage" and like, well, yes - there's a character in S3 that is in an interracial marriage. It is a completely plot minor, borderline irrelevant part of their character though and I feel that specifically including that is very misleading as to its overall presence in the show, or even S3.

If a character in a title is in fact in an interracial marriage, then the keyword "interracial-marriage" would be perfectly appropriate to add to the title. If, on the other hand, no characters are involved in an interracial marriage, but a character quickly refers a single time to interracial marriages generally, then it may not be appropriate to add "interracial-marriage" as a keyword, especially if it is not an important plot point. It is all context-dependent and fact-dependent. 

Again, if a character in a title is actually involved in an interracial marriage, and that is part of the plot (even if minor), it is completely okay to add "interracial-marriage" as a keyword to that title. 

For me, the #1 benefit of keywords is keyword searches. I would want that title in your example to picked up in searches, for example, so I could identify titles that include "interracial-marriage" in combination with other keywords. For example, here is a search for "interracial-marriage" in combination with "racism." 

There are even researchers and statisticians who use IMDb keywords to explore different things, like for example, the historical prevalence of the keyword "interracial-marriage" within titles over time (comparing different decades to each other). It's important to have as many titles with that plot point included, as long as it's actually part of the plot and not a mere passing reference to something that is not truly part of the plot.

Back to your example, to complicate things a little further, I see that the title you refer to, "The Expanse," is a TV show. In this particular case, if the interracial marriage is not discussed in a majority of episodes, then it should probably be deleted from the title page for the overall series, and added to the specific episodes that discuss or involve the interracial marriage. That outcome would already be required by existing IMDb guidelines:

Episode-specific keywords submitted to the TV series page - Keywords submitted to the TV series page should be relevant to and describe the entire series. Keywords that are specific to a particular episode should be added to that episode only.

It would seem that the user who added "interracial-marriage" to "The Expanse" was not familiar with this guideline. That seems to be the real problem demonstrated in your example.

Back to your example, if you feel that an existing keyword currently applied to The Expanse should absolutely not be there, you can easily delete it yourself. And if you feel that a guideline is not necessarily inappropriate but is just not a good keyword, you can downvote it on the keyword page. That is the existing system for community vetting of keywords. Beyond that, I am failing to understand what specific improvements you believe should be made to the system.

@keyword_expert 

Any other user can delete that keyword. To delete one or more keywords, just go to a title's keyword page, then click "edit," then choose "correct/delete."

I suspected that. And then a moderator makes the final decision?

You are still not being clear or specific enough. Since you mentioned "filtering," are you talking about keyword searches on IMDb? Or are you talking about the upvoting and downvoting of keywords on a title's keyword page?

Yeah, I mean keyword searches. What appears on a keyword search, under my draft idea, would essentially be keywords that have accumulated enough support from other site members. So a keyword wouldn't be searchable just on the word of a sole person.

That's a bad example, because IMDb blocks genre keywords, so "science-fiction" literally can't be added as a keyword to any title.

"Space opera" then. 

While keyword spam is definitely a problem, the way to deal with it is to downvote and/or correct or delete improper keywords. Any user, including you, can do this.

Here are Game of Thrones keywords.

At the bottom we have "grim darkness" with two downvotes. It's a silly keyword that's been downvoted, but it still appears in the grim darkness keyword search. 

With the caveat that I am not familiar with this title, I think I disagree with you on this point. If a character in a title is in fact in an interracial marriage, then the keyword "interracial-marriage" would be perfectly appropriate to add to the title.

This comes to a core difference between you and I of the level of relevancy that keywords are supposed to represent. To me, an "interracial marriage" is only a relevant element of a plot if it actually drives the story forward, plays a strong role within the narrative - and I regard simply the phenomenon being, at some point, present within a narrative to be misleading. I doubt someone looking for a show where an interracial marriage is represented would find much from The Expanse as it's essentially a background detail of a single series character.

It would seem that the user who added "interracial-marriage" to "The Expanse" was not familiar with this guideline. That seems to be the real problem demonstrated in your example.

Well, the character involved in the interracial marriage was a single-season character, as opposed to episode - but I take your point there.

Back to your example, if you feel that anexisting keyword currently applied to The Expanse should absolutely not be there, you can easily delete it yourself.

I have submitted deletion requests for The Expanse keywords, and they're currently pending. What happens if they're deleted? Can they just be re-added by someone else?

Beyond that, I am failing to understand what specific improvements you believe should be made to the system.

That the entire process should be entirely user-driven, akin to RYM. It has some similarities with RYM but it doesn't work the same way. On RYM, users propose genres or descriptors and then other users can instantly vote it down if they disagree. There's essentially no staff involvement whatsoever with what is or is not proposed. What you may not have noticed is that on each release on RYM, the specific votes are logged. You can see the other genres that people have downvoted.

Again, keyword contributions on IMDb (whether we're talking additions, corrections, or deletions) are about 99% user-driven, with literally no staff involvement. It may take 20 minutes or 24 hours for your keyword contributions to be "accepted," but in the vast majority of cases they are not being reviewed by IMDb staff, but rather simply take some time to go through the automated system. If you find that a keyword contribution you have made is still pending after 24 hours, then it probably will in fact be reviewed by a human being for some reason (usually because it includes a specific term pre-flagged for review, like the word "rape"). 

And yes, if you delete a keyword from a title, someone else (or the user who originally added it in the first place) can always add it back in. If this keeps happening in a virtual tug of war, you can post about it on this forum and ask for staff to decide what should happen with the keyword(s). 

Finally, I think you and I do have a core disagreement about the functions of keywords. As you point out, one use of keywords can be to find titles for watching, if someone is interested in that keyword concept as a plot-driver within the film. I acknowledge this is a very valuable use of keywords, and I do it myself all the time. But with that said, if I am specifically looking for films where an interracial marriage is driving the plot, then I will probably also read the plot summary as a second step of previewing the title before I watch it, to make sure that this is a plot-driver.

My point is that there are other uses of keywords, such as keyword-combination searches, research, and compiling statistics on keywords. Believe it or not, numerous academic and research papers use IMDb keywords as part of their research. And sometimes it is interesting to know how many films or shows within a certain genre or within a certain decade include a specific plot point.

I do agree with you that not every little thing that happens or appears on the screen (or is referred to) in a film deserves a keyword. That would be ridiculous. Unfortunately, some users are doing exactly that with keywords, which is borderline keyword abuse. I have complained about that in a variety of contexts, including this recent post. 

I think you and I just disagree on where we would draw the line. Again, if a character in a title is in fact involved in an interracial marriage, then I believe adding that as a keyword is entirely appropriate and consistent with the keyword guidelines.

Finally, back to your example, because this was a TV show with multiple seasons and numerous episodes, and because the character in the interracial marriage was only in some of the episodes, it sounds like the real problem is that "interracial-marriage" should have been added to specific episodes, but not to the series title. Perhaps on that we can agree.

Again, keyword contributions on IMDb (whether we're talking additions, corrections, or deletions) are about 99% user-driven, with literally no staff involvement. It may take 20 minutes or 24 hours for your keyword contributions to be "accepted," but in the vast majority of cases they are not being reviewed by IMDb staff, but rather simply take some time to go through the automated system. If you find that a keyword contribution you have made is still pending after 24 hours, then it probably will in fact be reviewed by a human being for some reason (usually because it includes a specific term pre-flagged for review, like the word "rape"). 

Okay, then so there's no checks at all? (excluding "rape" and other preselected terms)?

And yes, if you delete a keyword from a title, someone else (or the user who originally added it in the first place) can always add it back in. If this keeps happening in a virtual tug of war, you can post about it on this forum and ask for staff to decide what should happen with the keyword(s). 

Well my response there would be that a RYM-like upvote-downvote system would essentially stop that, and a simple majority would either accept or reject it.

I think you and I just disagree on where we would draw the line. Again, if a character in a title is in fact involved in an interracial marriage, then I believe adding that as a keyword is entirely appropriate and consistent with the keyword guidelines.

Well you noted that a single episode of an interracial marriage would not be acceptable as a way of describing an entire series. The example in The Expanse is not what I call especially much more than that (a season out of 6, and only present as a background aspect of a single-season character).

Personally I'm less concerned with simple "this thing is present in the show/movie" descriptors like that, but more interested in the application of subgenres. For example, Dystopian Fiction is a recognised literary, television and movie subgenre. Where it begins and ends is obviously subject to some level of subjectivity, but it does have a relatively understood focus on fiction that presents a narrative of an oppressive, dark and often totalitarian system of government. But because on IMDB a keyword is a keyword purely if one person adds it, the actual list of "dystopia" titles is bloated. "Game of Thrones", "Oz" (a prison drama?), "Mayans MC" (a gangster drama), "The Transformers", "Samurai Jack" (?????). I'm quite confident that if the site transferred into an upvote-downvote system that a good chunk of the "dystopia" list would be downvoted.

And I know that you can go around and just delete tags, but it doesn't stop someone else returning them, undoing all your work. And I don't even think it should be entirely just *my opinion* until it's then someone else's opinion. On RYM, people who vote down or vote up a tag or a descriptor have their vote remembered.

@Skavau 

One thing I forgot to respond to earlier:

At the bottom we have "grim darkness" with two downvotes. It's a silly keyword that's been downvoted, but it still appears in the grim darkness keyword search. 

I believe "grim-darkness" is a bad keyword that should be deleted from the IMDb system. One of the problems with this keyword is that it's subjective. IMDb's guidelines discourage subjective keywords like this.

The solution is to simply delete "grim-darkness" from this title.

Oh sure, but the point was less the silliness of that keyword there but that it has been 'downvoted' (or tagged as "not relevant" by two people - yet it still is listed).

@Skavau I get that. I believe the only reason why that bad keyword is still on the title is that not enough users are aware that they have the power to delete keywords themselves, or they are too lazy to do so.

Maybe a hybrid approach of incorporating your suggestions would be for IMDb to set it up so that once a keyword receives 0 of 3 "not relevant" votes, that keyword would be automatically deleted from the title. I believe I would support that innovation. 

That would be quite good. It might help entice engagement too. I just checked Breaking Bad, one of the most well-known shows ever and maybe about 20 people have ever engaged with the keywords there. 

I suspect keyword_expert means the keyword receives 0 of 3 "relevant" votes, meaning 3 votes were for No, not relevant.

Edited to add: Of course, that means the person who adds the keyword needs to be blocked from voting it Relevant.

@bderoes Yes, that is what I meant: three people voting "not relevant" and zero people voting "relevant." Thanks for catching that.

Regarding your additional point, I believe people should be able to vote on their own keywords. I do this all the time after I add a lot of keywords. I carefully select the most relevant keywords and "upvote" them, so they rise to the top of the page, where they belong. Most users do not go to this trouble, but the option should be available for those willing to do so.

Well basically, if a keyword would receive 3 more "not relevant" votes to "relevant" votes it gets automatically removed and effectively blocked from returning. So whether it's 0/3 or 3/6 it goes.

@Skavau Something like that. If I were the one setting it up, for "mixed" votes (more than 3 votes, some "relevant" some "not relevant"), I would probably require at least two-thirds of the votes to be "not relevant" before it would be automatically deleted. But generally I agree.

Should I make a new post with this specific suggestion as opposed to the meta discussion here?

@Skavau Yes, that is a good idea. It is more likely IMDb staff would actually see it. 

Also, may I ask, can someone automatically delete highly "relevant" keywords per the current system? For instance, Breaking Bad has "cancer" marked as relevant by 10/10 people. Could I, theoretically just remove that?

@Skavau Yes, you can delete a keyword at the top of the keyword list for a particular title. But just because a keyword has a lot of "relevant" upvotes, that does not necessarily mean the keyword is in fact highly relevant.

Case in point: the "upvoting upskirt spammer."  All it takes is a few dozen sock accounts to game the system.

@keyword_expert

Oh I agree, but *theoretically* more people voting positively in favour of a keyword should trend towards accuracy. Seems counter-intuitive that someone could come in and just destroy a highly 'relevant' keyword.

IMDB probably should add restrictions for new accounts voting to stop sock accounts. 

@Skavau An even better way to do it would be to limit voting by IP address. And not just keyword voting, but all types of voting, including voting on user reviews, quotes, trivia, etc.

@keyword_expert That too. Bizarre to me this isn't a thing already.

@Skavau 

All of my deletion requests for keywords got declined due to being "unable to verify" (not sure what this means) - how does one "verify" it beyond explaining that it isn't an accurate representation of the title? I provided comments explaining why I felt they were misleading, some longer, some smaller.

My best guess is this happened because you must have a relatively new account with not very many contributions. Am I right about that assumption? 

After you have made a few thousand (and possibly even a few hundred) keyword-related contributions, any further keyword contributions you make (including keyword deletions) will typically be automatically approved.

@bradley_kent The current system is not working. I can find you any popular TV and movie title that I've watched and outline all of the flawed, and inaccurate keywords present on the titles. Keyword watcher himself has gone through and discovered all of the duplicates and overly specific keywords.

Why would an upvote/downvote system work on RYM, but not on here?

Keywords should accurately and objectively report the content of a title, both general and specific.

In many cases this is simply not happening - because there's no barrier to entry and people simply do not understand the terms of reference for what they're using. It is already failing.

Keywords should be encompassing.  They should be an accurate reporting of the content of a title, regardless of the subjective predilections of a user or contributor.

I admit that keywords are not perfect, but you seem to be looking at just one example.  You need to look at "the big picture," and the millions of titles that are in the database.  

The flawed and inaccurate keywords should be corrected or deleted.  That's a simple, if time-consuming, process.

I have often thought that there should be some sort of "entry test" before anyone is allowed to contribute to IMDb.  That doesn't exist, so it is up to the staff to monitor what is going on.

@bradley_kent 

I admit that keywords are not perfect, but you seem to be looking at just one example.  You need to look at "the big picture," and the millions of titles that are in the database.  

I mean I can only really comment on things I've seen, and in many of the things I've seen - it's quite bad - and there doesn't really seem to be a way to remove poor keywords.

The flawed and inaccurate keywords should be corrected or deleted.  That's a simple process.

I tried to remove the flawed keywords of the Expanse, provided an explanation as to why, and got "unable to verify". 

How exactly am I supposed to verify, other than just explaining why they're inaccurate?

That's a problem with the staff.  They need to respond with an explanation.

@bradley_kent

I've submitted a general query on that issue.

Also, I've specifically tailored this submission into a different one here. 

@bradley_kent 

Voting on keywords is absurd.  This SHOULD NOT be a popularity contest.

Voting on keywords is already the default. Every single keyword page on IMDb allows voting on whether each keyword is or is not relevant to the title.

If I wanted to know if there was ever a title about the painter Alfred Mannings, or if the Damon Runyon character Nicely-Nicely Johnson, appeared in more than one film, keywords allow me to find the answers.

How exactly would keywords help you definitively answer those questions? Neither of those keyword searches picks up any results. And that fact doesn't answer either question, especially since keywords are highly fallible and far from complete. In other words, just because there are no keywords for either result, that doesn't prove that such titles do not exist.

https://www.imdb.com/find?s=kw&q=Alfred+Mannings&ref_=nv_sr_sm

https://www.imdb.com/find?s=kw&q=Nicely-Nicely+Johnson&ref_=nv_sr_sm

Keywords should accurately and objectively report the content of a title, both general and specific, regardless of the subjective predilections and judgments of a user or contributor.

There is such a thing, though, as keywords that are too specific. The more specific a keyword, the more likely it could be spam -- especially if it is an "orphaned" keyword -- a keyword that only applies to a single title.

You fell for my trap.  I submitted (and they were accepted) the keywords “alfred-mannings-character" for Summer in February (2013) and "nicely-nicely-johnson-character" for Guys and Dolls (1955) and The Big Street (1942).  They are included in the keyword list for each of these titles, BUT YOU CANNOT FIND THEM by entering a keyword search!  For some reason (a glitch? a time lapse?), these one-title and two-title keywords DO NOT come up when trying to search them.  Instead, you get: 

No results found for

Now, THIS is the kind of keyword problem that really needs attention.

@bradley_kent 

How long ago did you add these keywords? Was it within the past week? The reason I ask is that over the past week I have observed a noticeable time lag for keyword searches to pick up changes in keywords.

Title searches do pick up your keywords. 

https://www.imdb.com/search/title/?keywords=nicely-nicely-johnson-character

https://www.imdb.com/search/title/?keywords=alfred-mannings-character

I believe this implies that these are relatively new keywords. But if you added them weeks ago, then there is indeed a problem.

Yes, they were in the last week.  I wonder what the time lapse is between when a keyword is accepted and when that keyword becomes searchable?

@bradley_kent The time lag is typically about 36 hours for keyword searches, but it can be up to 60 hours. Over the past week or so, the lag has been a few days. Title searches for keywords always work faster than keyword searches for keywords. I previously provided additional details on this topic here.

@bradley_kent 

The keyword searches now pick up your newly added keywords.

https://www.imdb.com/find?s=kw&q=Alfred+Mannings&ref_=nv_sr_sm

https://www.imdb.com/find?s=kw&q=Nicely-Nicely+Johnson&ref_=nv_sr_sm

Lately the lag has been pretty extreme -- more than a week in some cases -- for IMDb keyword search results to be repopulated.

Because title searches are updated more quickly, title searches should be used during any of these keyword search lags.

So... Does this problem reside with (i) the submitter who offers inaccurate keywords, or (2) the staff's keyword list managers who accepted the inaccurate keywords, or (3) both of the above?  Probably (3).

Solution?  Possibly, instruction of submitters of keywords (a qualifying test before keywords offered are accepted?  better instruction and supervision of keyword list staff members?)  What else?

It seems that anyone should be allowed to submit keywords, so the problem may well reside with the staff.

@bradley_kent  wrote:

I wonder what the time lapse is between when a keyword is accepted and when that keyword becomes searchable?

After you asked this, I have continued to monitor keyword searches on IMDb to get a better sense of the answer to your question.

As best I can tell, the results and numbers that populate the keyword search result display pages are currently only being updated once per week, on Sundays. This is the sort of page I am referring to: https://www.imdb.com/find?s=kw&q=in-title&ref_=nv_sr_sm

I have no idea why the display pages are being updated less frequently than they used to be. This could be a temporary glitch, or it could be a conscious, permanent change. 

The actual keyword search results are being updated much more frequently -- more like once per day. These are the results I am referring to: https://www.imdb.com/search/keyword/?keywords=character-name-in-title&ref_=fn_kw_kw_2

As always, the most accurate, comprehensive, and up-to-date search results for keywords can always be obtained by running a title search (as opposed to a keyword search), and also by including Adult titles in the search query. This is the sort of search I am referring to: https://www.imdb.com/search/title/?keywords=character-name-in-title&adult=include

Those title search results are updated very quickly -- about every ten minutes or less.

I should mention one exception to this part of what I said a couple days ago:

As best I can tell, the results and numbers that populate the keyword search result display pages are currently only being updated once per week, on Sundays. This is the sort of page I am referring to: https://www.imdb.com/find?s=kw&q=in-title&ref_=nv_sr_sm

Here's the exception: when a keyword is entirely deleted from the database (i.e., all instances of the keyword are removed from all titles), that change will be reflected within 24 hours on search result display pages. In other words, once a keyword is deleted, that keyword will disappear from IMDb entirely within 24 hours.

But when a new keyword is added, or when specific instances of specific preexisting keywords are either added or deleted (without removing all instances), those changes will (apparently) only be reflected on the search result display pages once per week (apparently each Sunday).

@bradley_kent The keyword mapping that populates keyword search result display pages continues to be updated once per week, at around 11:00 AM GMT each Sunday (give or take an hour).

@bradley_kent   Since I wrote that (the past two weeks), the keyword mapping has been updated on Mondays instead of Sundays. The site is currently due for an update Monday morning GMT.

@bradley_kent  Strangely, the keyword mapping has not been updated in about 9 days now, and counting. This is an unusual delay. Previously (for the past four months or so) the delays have never been longer than 8 days.

@bradley_kent And the keyword mapping was finally updated, right at 11 a.m. GMT, like clockwork. For some reason the updating was skipped last Sunday.