gromit82's profile
Champion

Champion

 • 

7.3K Messages

 • 

274.7K Points

Saturday, March 19th, 2022 5:14 AM

Solved

References to films as keywords

In the Keywords submission guide at https://help.imdb.com/article/contribution/titles/keywords/GXQ22G5Y72TH8MJ5#, there is a guideline which I believe is incorrect, or at least misguided:

Unacceptable keywords

... Repeating the title - For example adding shawshank-redemption as a keyword to the title Shawshank Redemption. If the movie Shawshank Redemption is mentioned or referred to during a title, you can use the keyword reference-to-shawshank-redemption

I agree with the first sentence generally; the title of a film should not normally be a keyword for the film. (There are exceptions when the title of the film is literally the common name for an aspect of the plot. For example, the plot of Alfred Hitchcock's Blackmail (1929) actually does involve blackmail, so blackmail is an appropriate keyword for the film.)

However, the second sentence, "If the movie Shawshank Redemption is mentioned or referred to during a title, you can use the keyword reference-to-shawshank-redemption", appears to be an error. If the movie The Shawshank Redemption (1994) is mentioned or referred to in another film, that should not be indicated by a keyword, but by the use of a Movie Connection. That's what we have the Movie Connections section for (see https://www.imdb.com/title/tt0111161/movieconnections#referenced_in for the 492 titles currently listed as referencing The Shawshank Redemption).

Furthermore, there is a significant advantage of using the Movie Connections section to indicate references to other films. In the Movie Connections section, all references are intended to be explained as to what kind of reference is being made. That's not possible in the Keywords section.

Finally, there isn't an active keyword reference-to-shawshank-redemption anyway. What IMDb does have is the keyword reference-to-the-shawshank-redemption (note the inclusion of "the" as in the film's title). However, it is only applied to 17 titles as of this writing, compared to the 492 references indicated in the Movie Connections section, and, as I'm trying to indicate here, it shouldn't be a keyword in any event.

Therefore, I am requesting that the sentence "If the movie Shawshank Redemption is mentioned or referred to during a title, you can use the keyword reference-to-shawshank-redemption" be removed from the Keywords guide.

2.7K Messages

 • 

47K Points

2 years ago

I have thought about this exact issue a lot. Here are my two cents, with some background before I respond to the specifics. Bear with me -- I promise I will eventually respond to your specific suggestions.

First, I believe that there has been significant overuse and even abuse of the "reference-to" prefix in keywords by a few contributors. Literally the only two mentions of the "reference-to-" prefix in the guidelines are for referring to titles ("reference-to-shawshank-redemption") and to names ("reference-to-marilyn-monroe"). The context of these two examples is that they are items that are otherwise listed elsewhere in the database (The Shawshank Redemption and Marilyn Monroe). My interpretation is that the guidelines are telling contributors to avoid potential confusion about whether a keyword is linking to those actual items in the database by instead using the prefix "reference-to-," to make it clear that this is a keyword instead (and not a link to the actual title or person).

I interpret all of this to mean that "reference-to-" need only be used for people, names, characters, titles, film-related companies, etc. that would otherwise be listed at IMDb. There is certainly no other mention of "reference-to-" anywhere else in the guidelines, so I think this is a reasonable interpretation.

Unfortunately, some contributors have took the "reference-to-" prefix and run with it, using it as a substitute for the "character-says-" prefix, and also for any minor place, thing, or concept that is spoken about, shown on the screen, or even alluded to. For example, if a character briefly mentions he once went on a vacation to China, the contributor(s) will add a keyword "reference-to-vacation-to-china." If a scene briefly shows a billboard that happens to advertise Marlboro cigarettes, the contributor(s) will add a keyword "reference-to-marlboro-cigarettes." If a character says they forgot to pay their utility bill this month, the contributor(s) will add a keyword "reference-to-forgetting-to-pay-a-utility-bill." And on and on.

I am not in the slightest bit exaggerating with those examples. Just to point out the absurdity of how "reference-to-" is being severely abused, here are a few actual examples:

reference-to-defecating-on-a-pillow 

reference-to-knitting-a-sweater 

reference to gay sex under the bleachers

reference-to-separating-one's-food 

reference to a bag of candy as a gift

reference-to-acid-being-thrown-on-a-boy's-face

I could give you thousands of absurd examples just like that, all of which are "orphaned" keywords (keywords that only apply to a single title).

Again, in my view, "reference-to" should not be used except in the very narrow circumstances where a thing (person, character, title, company, etc.) is otherwise listed in the database and there is a need to avoid confusion about where the keyword is linking to. If something else in the plot is important enough to be covered in a keyword, then the "reference-to-" prefix should not be used. The reason is that when the "reference-to-" prefix is used for an important plot element, it will bifurcate the keywords in keyword searches. For example, if one contributor uses "cow" as a keyword and another contributor uses "reference-to-a-cow," the keywords have been bifurcated and there is no way (that I'm aware of) to combine them in a keyword combination search. This bifurcation problem is one of the biggest issues with using the "reference-to-" prefix.


And finally, if something is not an important plot point, then there is no need to add a keyword for it at all, let alone a keyword with the "reference-to" prefix. Otherwise it is spam to add a keyword for every single little thing that each character says, does, or wears in the film or show.

A related issue is that some people believe they must use the "reference-to-" prefix any time a keyword is a proper name (like a geographic place, a non-movie-related company, a consumer product, a news medium or social media website, etc.)  Instead of "idaho" as a keyword they believe they must use "reference-to-idaho," even for a film that takes place in Idaho. Instead of "kleenex," they believe they have to use the keyword "reference-to-kleenex." Instead of "wall-street-journal," they believe they must use the keyword "reference-to-the-wall-street-journal." Yet nothing in the guidelines say anything about the "reference-to-" prefix in these contexts, let alone requires contributors to use it.

Now, let's discuss the Shawshank Redemption example. On the one hand, this mention of "reference-to-shawshank-redemption" in the guidelines is good, because it provides the context I have discussed above about the proper use of the "reference-to-" prefix (as a means of avoiding confusion for items otherwise listed in the database).

On the other hand, it is a bad thing, because it is an express endorsement of the "reference-to-" prefix, which some contributors have misinterpreted as giving them free reign to use the "reference-to-" prefix as they see fit.

There is also the problem you have mentioned that it really should be "reference-to-the-shawshank-redemption," not "reference-to-shawshank-redemption."

 

As you point out, the preferable way to indicate a reference from one title to another is by adding it as a movie connection.  You have argued that only the movie connection approach should be used, and not the "reference-to-" keyword approach.  I think I would be okay with that change in policy. The "reference-to-" prefix has not been used enough for titles to have any meaningful impact on keyword combination searches if these keywords went away. And if contributors were instructed to use the movie connection approach instead, then it would definitely bolster that feature (because otherwise people can get lazy and only type connections as a keyword while ignoring the movie connections feature).

Regardless of what is done on the movie connections issue, I strongly believe the bigger issues of the "reference-to-" prefix really need to be clarified in the guidelines. Specifically, the guidelines should specify the limits for when this prefix should and should not be used.

If I had control over the policy here, I would probably eliminate the "reference-to" prefix entirely. It is at best confusing, it is has certainly been inconsistently applied, and its use only has the result of bifurcating keywords. If anything, I would only allow the "reference-to-" prefix for referring to people and characters, because in those situations, the keyword "marilyn-monroe" on its face doesn't make it clear whether Marilyn Monroe is in the film or is just referenced in the film. But that is the only narrow circumstance where I believe the "reference-to-" prefix is truly justified. Any other use of the prefix could be done away with immediately and IMDb would be better for it.

Thanks for allowing me to contribute my two cents (and then some).

(edited)

1.3K Messages

 • 

23.1K Points

@keyword_expert​ I disagree completely.  The "reference-to-" keywords have been very helpful in my research.  If someone or something is just talked about, or shown only in on-screen text or in a photograph/painting, without ACTUALLY being shown in moving image, the "reference-to-" keywords are extremely helpful.

If, for example, I was researching World War Two, I would want to discover not only those titles where World War Two scenes are ACTUALLY on screen, but where it was mentioned or discussed without ACTUAL scenes of World War Two.  This just one of millions of examples.

When it comes to keywords, I usually think of myself as a research librarian.  Anything that facilitates research is laudable.

2.7K Messages

 • 

47K Points

@bradley_kent​ As you admit, your approach results in "millions" of excess keywords. If everything "referenced" by a character in a title--even if not an important plot point within the title--were converted into a keyword, it would result in thousands of excess keywords per title. This doesn't "facilitate research"--it hinders research by bogging things down with marginally relevant keywords. It's the same thing with the "character-says-" prefix, which can have the same effect. If someone wants to research what is "said" or "referenced" in the dialogue of a title, they can search a transcript or screenplay. It makes little sense to use the keywords as a de facto transcript, with millions of excess keywords. This is spam, pure and simple.

In this sense, "reference-to-" prefixes are just as bad as "character-says-" prefixes. "Character-says-" prefixes also result in unwanted, unnecessary spam keywords for things that could and should instead be searched on the Internet within movie transcripts/screenplays, rather than bogging down the keyword sections of titles.

There is literally no need for the "reference-to-" prefix in your World War Two example. If World War Two is an important plot point within a movie, whether or not a war scene is actually depicted in the movie, then it may very well deserve a "world-war-two" keyword. But if World War Two is not an important plot point within the movie, but rather is briefly mentioned in passing, then there is simply no point in making that into a keyword.  Keywords are for relevant plot points that carry the story forward, not for every little thing "referenced" or "said" in a movie. 

Apparently your approach is that if something is only "talked about" or "shown" in a still image, but not shown in a "moving image," then it deserves a "reference-to-" keyword, no matter how minor to the plot. But how are other users to know that your keywords are designed this way? 

After all, there are many different philosophies and approaches to how "reference-to-" should and should not be used. Some contributors apparently believe that all things with a proper name must be given a "reference-to-" prefix, even if they are depicted in moving images (for example, "reference-to-the-wall-street-journal").  

Other contributors apparently believe that anything said out loud by a character is worthy of a "reference-to-" prefix (as an even sloppier substitute for a "character-says-" prefix), even if that thing is also depicted in moving images. This can result in "world-war-two" and "reference-to-world-war-two" being added to the same title. In fact, there are 30 such titles in the database right now. After all, if a character both participates in World War Two and refers to World War Two, there is nothing in the guidelines that expressly says both types of keywords cannot be used for that title.

The point is that there are numerous interpretations of what "reference-to-" even means, let alone when it is acceptable to use it. If more contributors were to make active use of "reference-to-" prefixes with their own interpretations, this would quickly turn the keyword sections on IMDb into chaotic Towers of Babel. How are users to reconcile these subjective and conflicting interpretations of what "reference-to-" even means?

If someone were researching World War Two, then it probably is not going to help them very much to know about a movie where a character briefly "references" World War Two in passing, without World War Two being an actual plot point in that movie. That is why the vast majority of contributors do not use "reference-to-" in the sense that you are apparently using it. And if someone were researching titles where characters "say" or "reference" a certain thing or concept (without it being an important plot point), then the researcher should be able to find such titles by simply searching screenplays and transcripts.

The simplest way of summing all this up is that keywords are not intended to be used as de facto transcripts for a title. That is why we have actual transcripts and screenplays. Rather, with a few limited exceptions, keywords are meant for describing important plot elements, and if something is important to a plot, it doesn't matter the form (whether the thing is spoken about, depicted in a still image, or depicted in a moving image). The "character-says-" and "reference-to-" prefixes should have very limited usage, and they definitely should not be used to describe every little thing said, shown, or alluded to in a title.

(edited)

1.3K Messages

 • 

23.1K Points

Weren't "character-says-" keywords banned long ago?  I have submitted thousands of deletions of this kind of keyword, and they have all been accepted.

Part of the argument for this decision was that such information might more correctly exist in the "Quotes" section.

(edited)

2.7K Messages

 • 

47K Points

@bradley_kent

Weren't "character-says-" keywords banned long ago?  

That is not correct. You keep asserting this, but you haven't produced any evidence, and as I have previously reminded you, the exact opposite is true. Specifically, Michelle has made it very clear that "character-says-" keywords are acceptable and technically within the guidelines.

I have submitted thousands of deletions of this kind of keyword, and they have all been accepted.

I'm sure they were approved automatically, just like the vast majority of keyword deletions. Although "character-says-" and "reference-to-" keywords may be technically within the guidelines, most of them could be easily deleted for the reasons I have stated in this thread, and IMDb would be better off for it.

Overall, the "reference-to-" keywords are probably even worse than the "character-says-" keywords, given the lack of clarity about what the "reference-to-" keywords are even intended to mean, the idiosyncrasies of the "reference-to-" keywords as prepared by a very small number of contributors, and the egregious typographical errors from one contributor in particular who has been adding most of the idiosyncratic "reference-to-" keywords.

This contributor(s) is essentially spamming IMDb with bad keywords. The fact that there are so many typos in these keywords reflects that the contributor(s) are more interested in submitting a high quantity of keywords, even if the keywords are of low quality.

Again, it just makes little sense to use either "reference-to" keywords or "character-says" keywords in most instances.

If "character-says-ivory-soap-floats" is a bad keyword (and it is), then so is "reference-to-ivory-soap-not-sinking."

If "character-says-burn-the-witch" is a bad keyword (and it is), then so is "reference to burning a witch at the stake."

If "character-says-go-to-hell" is a bad keyword (and it is), then so is "reference-to-hell."

Using the last example, if hell is an important plot point in a film (for example, a character repeatedly says they don't want to go to hell), then the keyword "hell" should be added. Otherwise, the "character-says-" and "reference-to-" keywords should be avoided.

2.7K Messages

 • 

47K Points

2 years ago

Here's another reason why I support @gromit82's suggestions in this post: with "reference-to-" keywords, it is impossible to add context of how one title references the other. But with the movie connections option, the database allows (and even encourages) contributors to be specific about the nature of the connections. For this reason, the movie connections method is superior.

Champion

 • 

7.3K Messages

 • 

274.7K Points

2 years ago

Since we have not received a response from the staff about this after two months, I'm bumping this thread.

2.7K Messages

 • 

47K Points

2 years ago

I just wanted to point out that another guideline specifically regarding movie connections is potentially relevant here:

Generic references that do not link specific titles are not considered movie links. i.e. just because a character mentions Bond/Harry Potter/Frodo/etc. does not mean a reference to a specific Bond/Harry Potter/Lord of the Rings title or even all of them. Unless of course it is clear that a specific title is meant (i.e. poster/T-Shirt/specific cast reference). Otherwise, please consider submitting a plot keyword instead.

Setting aside the vague language and poor grammar in this guideline, this guideline does mention keywords. In my opinion the guideline doesn't really answer the question at hand, though.

My interpretation of this guideline is that if a specific title is not referenced, then you should use a keyword (e.g., "reference-to-marvel-cinematic-universe," "reference-to-harry-potter," and "reference-to-harry-potter-movie-franchise"). 

But if a reference is to a specific title, there are currently two options: (1) a plot keyword and (2) a movie connection. The guidelines do not specify whether one of these options is preferred over the other, nor whether both of them are preferred.

2.7K Messages

 • 

82.6K Points

@keyword_expert​ "But if a reference is to a specific title, there are currently two options: (1) a plot keyword and (2) a movie connection."

 

The sentence "unless of course it is clear that a specific title is meant" to me makes it clear that a reference to a specific title SHOULD be added as a movie connection. And the sentence "Otherwise, please consider submitting a plot keyword instead" makes it clear that only generic references (and not specific ones) should be submitted as keywords.

That being said, I do agree the wording could be better.

2.7K Messages

 • 

47K Points

I can see that interpretation, for sure. But there is room for different interpretations as well.

Champion

 • 

7.3K Messages

 • 

274.7K Points

2 years ago

I think this thread could use another bump.

Employee

 • 

17.2K Messages

 • 

310K Points

2 years ago

Hi @gromit82 & All -

My apologies for the delayed staff reply, we are currently working through a backlog of older, missed threads, aiming to address any outstanding issues.

Thanks also for all the thoughtful comments on this topic, I have filed a ticket for a policy review of these specified "reference-to" keywords.  As soon as I have an update I will post the policy confirmation here.

Cheers!

Champion

 • 

7.3K Messages

 • 

274.7K Points

1 year ago

I'm going to bump this thread again because the problem discussed in my original post has not been resolved yet.

Champion

 • 

7.3K Messages

 • 

274.7K Points

1 year ago

As it's now been a year and a half since I started this thread, I think it's time for another bump.

Employee

 • 

17.2K Messages

 • 

310K Points

Hi @gromit82​ -

I just reviewed the ticket and can confirm that it is still open and pending review by our Policy team.  I also commented on the ticket to inquire on the status and to help try and push it forward.  Once I have any update I will be sure to post an update here.

Champion

 • 

7.3K Messages

 • 

274.7K Points

6 months ago

This thread has been open for two years now, so I think it needs another bump.

Please keep in mind that what I'm asking for doesn't require any programming skill. It's barely even a change in policy. I'm just asking to change the following text at https://help.imdb.com/article/contribution/titles/keywords/GXQ22G5Y72TH8MJ5#:

  • Repeating the title: For example adding shawshank-redemption as a keyword to the title Shawshank Redemption. If the movie Shawshank Redemption is mentioned or referred to during a title, you can use the keyword reference-to-shawshank-redemption

to say instead something like this:

  • Repeating the title: For example, the-shawshank-redemption should not be added as a keyword to the title The Shawshank Redemption. If the movie The Shawshank Redemption is mentioned or referred to during a title, please indicate that with a Movie Connection instead of a keyword.

(edited)

Employee

 • 

1.6K Messages

 • 

16K Points

Hi @gromit82,

Thank you for your update to this thread.

I have passed your comment on to the relevant teams for further review, to try and expedite the appropriate actions to resolve this issue. I will add any updates to the thread once I have them.

Cheers!