IMDb Data – Now easily available to contributors

  • 15
  • Announcement
  • Updated 6 hours ago
  • (Edited)
Today (20 Dec 2018) we are pleased to announce the IMDb datasets are easier to access and now directly from imdb.com. Using the new interface, contributors can bulk-access subsets of IMDb title and name data for personal and non-commercial use. Each dataset file is in a gzipped, tab-separated-values (TSV) format.

To access the datasets and for more information you can go here: https://contribute.imdb.com/czone

Stewart

Photo of Stewart

Stewart, Employee

  • 15 Posts
  • 16 Reply Likes

Posted 7 months ago

  • 15
Photo of Jeorj Euler

Jeorj Euler

  • 6966 Posts
  • 8911 Reply Likes
Thanks!
Photo of Jeorj Euler

Jeorj Euler

  • 6966 Posts
  • 8911 Reply Likes
However, I do hope that the extended datasets can be made available to a slightly broader span of frequent contributors.
Photo of Phil G

Phil G

  • 35 Posts
  • 81 Reply Likes
Chris? Stewart? Anyone?

I'd like answers to the questions I asked over 2 weeks ago, please.

What are the requirements for accessing the mysterious 'extended datasets'? Exactly how many contributions are there in 'a large number'? Exactly how many days/months/years ago counts as 'recently'?

But I suspect that whatever your answer, I'll be asking you to reconsider your policy on this. For my own case, I have contributed over 45,000 items to IMDb over the last 5 or 6 years (according to Col's end-of-year reports). I realise that's nowhere near as much as some, but I still consider it to be a very large number, and it was enough to get me into the top 250 contributors list three times in recent years. So I'm disappointed that you seem to be demanding more, before I can access data that wouldn't even exist without contributors like me. I'm sure I'm not the only person in this situation. For all the flaws in the old ftp-based system, at least the data was there and freely available. Why are you now so reluctant to share?

Also, why are you being so secretive about this data? I still have no idea what might be included in these 'extended datasets' and in fact I wouldn't even know that such a thing existed at all if I hadn't asked here. Neither the announcement in this thread nor the datasets page itself makes any mention whatsoever of what might be available or what mystical incantation is required to unlock it. Why is that?

(Edited)
Photo of Phil G

Phil G

  • 35 Posts
  • 81 Reply Likes
If restrictions are necessary, then so be it, but the criteria for access need to be designed carefully so that significant contributors are not left out. But without knowing the current criteria or IMDb's reasons, we're left guessing...
Photo of Vincent Fournols

Vincent Fournols

  • 2901 Posts
  • 4847 Reply Likes
To newcomers, I very often claim that IMDb is a collaborative site. But there is also the dark blind opaque side of IMDb: information that never make it out, concealed roadmaps, unanswered questions, which turn out to be extremely irritating (I leave the mysterious rating computation aside: this I can understand as it could be manipulated).
So I am afraid that the answer will never come.
(And thus I hope to tease and lead the bear to come out of his den !! :D)
Photo of ACT_1

ACT_1

  • 3284 Posts
  • 3102 Reply Likes


Jeorj Euler :
... There may have been some esoteric widespread incidents
of individuals or organizations who have not contributed much,
if anything at all, to IMDb yet built up their own websites
by siphoning off IMDb.
- - -


IMDb could add or already has added special names that are not real
and would not be found any other place but IMDb
IMDb could then do a Giggle Search for that name
and see who is publishing IMDb data as their own ? ?
.

Photo of Vincent Fournols

Vincent Fournols

  • 2901 Posts
  • 4847 Reply Likes
Funny, ACT: I was thinking exactly the same thing and was about to share it :)
Photo of ACT_1

ACT_1

  • 3284 Posts
  • 3102 Reply Likes

Hey! Vincent Fournols

my reply to your reply to: "IMDb could add or already has added special names that are not real"

IMDb Test Titles:
https://www.imdb.com/title/tt4684222/ Probably the Best Test Title Around
https://www.imdb.com/title/tt2418644/ Testing Movie
https://www.imdb.com/title/tt3239160/ Test
https://www.imdb.com/title/tt4242990/ Worldwide Title - A Test Movie (original title)
https://www.imdb.com/title/tt4844946/ A Test Series (2016-2019)
This title does not exist. It has been created for IMDb testing purposes only
https://www.imdb.com/search/keyword?keywords=imdb-testing
- - -

https://getsatisfaction.com/imdb/topics/feature-films-in-the-news-genre?topic-reply-list[settings][f...
1 day ago
by J.
I have no idea what, if anything, should be done with this curious entry:
Probably the Best Test Title Around.
- - -

1 day ago
by Ed Jones (XLIX)
That curious title is an IMDb Test title I believe.
Think I remember Col explaining this once.
It was made up names (nm) sequence numbers that had bogus actors on it.
Running across one of these titles or names is rare.
.

Photo of Stewart

Stewart, Employee

  • 15 Posts
  • 16 Reply Likes
Hello,

The Extended Datasets are available to those who have 1000+ approved contributions in the last 360 days, otherwise the Basic Datasets are available with just one approved contribution in the last 60 days.

I hope this helps.
Stewart
(Edited)
Photo of Vincent Fournols

Vincent Fournols

  • 2901 Posts
  • 4854 Reply Likes
Thanks a lot Stewart.
Photo of Phil G

Phil G

  • 35 Posts
  • 81 Reply Likes
Thanks very much Stewart, that's actually much better than I'd thought initially (I knew I'd taken a break for a while but didn't realise I'd been pretty much inactive for a whole year!)

Would it be possible to add a note saying this to the datasets page for people who don't yet have access, so that they understand that what they're seeing isn't everything available.
Photo of Brian Risselada

Brian Risselada

  • 30 Posts
  • 34 Reply Likes
Thank you for the response!
Photo of J

J

  • 9 Posts
  • 10 Reply Likes
Ok, it's nice to finally see that more data is available (to some people).
I see no way to qualify as I have a wide range of interests and I'm not from the industry so getting 1000 updates in 360 days will be never possible (I only made maybe 5 updates in the last 15 years and pointed to serveral issues with broken/incomplete exports of the LIST files via the normal support pages for ~19 years).

The Java Movie Database (JMDB) application...
What should make me qualify to access the data is the fact that I'm the author of the Java Movie Database (JMDB) which is available since 19 years (first versions only to a limited set of beta testers). --> http://www.jmdb.de/
Contact details can be found on the above website.

There is no other free application I know of processing the (public) IMDb data for that long. The only thing coming close is the IMDbPY project.
While the application was originally created by two persons I'm the only one left (since 15 years).

The JMDB application allows to import/process the old LIST files (which is still the base for the search inside the application) plus the TSV files (not yet used inside the application to search).

The reason that I didn't update the code to use the TSV content inside the application beside adding support to import the data is basically because the available content is so limited, that it is actually useless.
Sorry, but there is no other (nicer) way to sum it up when it comes to the TSV file format.
This is what the people already complained about in the original thread (https://getsatisfaction.com/imdb/topics/imdb-data-now-available-in-amazon-s3).

So I would really like to see how the "complete" stuff actually looks like and if it's actually for the following reasons:
  1. Extend the TSV data import that JMDB supports to the full data-set (not only limited to the "free" stuff available now - which is useless)
  2. Making the new TSV content searchable (if it's actually worth the effort - not for details limited to 10 entries (title.principals) per movie and TOP-4 movies/titles per name (name.basics) while the old LIST format offered the full list of people involved in a title)
  3. Maybe offering some functionality to the professional "contributors" to use the application to send updates to the IMDb website (API available?)

As this has been under the radar here other use of the IMDb data in the past...
I also want to share that the IMDb data has been and still is (at least the frozen LIST files) used together with JMDB:
  1. by students/postgraduates (postdocs) from universities for data analysis, etc. (also in master thesis) as can be seen e.g. here (there are more like TU Berlin, Germany): UiO University Oslo Norway - https://www.uio.no/studier/emner/matnat/ifi/INF3100/v16/undervisningsmateriale/filmdatabasen-og-post... and
  2. it is used in computer science teaching lessons at schools as can be seen here: https://www.swisseduc.ch/informatik/120-lektionen/principles/recollection/sql-imdb.html
The imported IMDb data has been used to create covers or other internal notes on titles that have been recorded from TV

Finally some technical issues with the format...
There are also some technical issues I have starting with the fact that inside each of the *.tsv.gz archives the name of the file is "data.tsv".
Normally the filename of the compressed file should be equal to the file found inside the archive minus the ".gz" compressor extension (Example: "title.crew.tsv.gz" --> "title.crew.tsv").
It's broken since the beginning. I you extract all files you have downloaded into the current/same directory you end up with only one file - that was extracted from the "last" archive processed.

When the data is imported with JMDB from the compressed gzip file and you try to add foreign key constraints to the filled tables containing the data it will not work for some cases as there are references in the e.g. title.crew.tsv (person reference) that has no matching entry in the name.basics.tsv file so the relational database tells you that your data basically is incomplete. There is more of this.

Kind regards,
Juergen Ulbts


Photo of K S

K S

  • 1 Post
  • 3 Reply Likes
Let me support Juergen's note.

I am an academic who has used his incredibly useful JMDB tools to manipulate and organize the IMDB LIST files as part of my research. His efforts have made using those IMDB flat files much easier for people like myself whose programming skills are sub-par, and as such I think advances the mission and visibility of IMDB.

I am pretty sure the target audience of the contributor files would include individuals like myself who are interested in having comprehensive files (in my case for statistical analysis). I imagine industry folks and those more interested in following certain titles would use either the regular webpage or (in the case of industry folks) using IMDB Pro.

So my hope is that these additional files are made available to volunteers like Juergen who provide the tools which many of us use to access the raw files. I think his software is a meaningful contribution to IMDB, and it would make sense both from the perspective of advancing IMDB and also for a reward for pro-social behavior behavior to expand the definition of "contributor" to include tool-making.

Best,
Koleman