Are there some free chess game databases where it is possible and allowed to download games? Is it possible to use this information for my own purposes? For example, would be allowed to make a machine learning app that learned from the downloaded games?
asked Jul 17, 2017 at 12:38
Add a comment
|
5 Answers
-
Is it possible to use this information for my own purposes
Absolutely! Nobody has the copyright. Enjoy your machine learning adventure!
-
Database rights may apply though, so I guess it's best to read the terms and conditions or ask for permission. en.wikipedia.org/wiki/Database_right– LucDec 14, 2020 at 22:26
Caïssabase is one of the biggest and best.
From the website:
What is Caïssabase?
When I got back into Chess a year or two back I couldn't find a decent free regularly updated database of Chess games.
There are some good ones out there like Kingbase and Millionbase (not updated anymore).
...
What is in Caïssabase?
CURRENTLY 4.02 MILLION GAMESI started from the following databases/sources.
Millionbase
Kingbase
I then merged in all of the TWIC back issues up to the date on the archive as well as all the master games from pgnmentor.
...
What about curation? Glad you asked. Roughly I followed these rules.
- Removed all duplicates
- Removed all games with less than 5 moves
- Aimed for all games to have one (or both) players at least master strength
- Fixed names and event spellings where possible
- Inserted ELO's where known
- Inserted extended ECO opening codes
...
Legal/Copyright
Where I live facts are not copyrightable, Chess moves played in a game are facts.You'll notice there are no comments on any of the games as comments are copyrightable.
The latest download is from September 2019, so perhaps the intention of a monthly download is proving a bit too much. Meanwhile I've downloaded it for use with Scid vs PC and will be a grateful user.
Edit: There is a new download available as of 14th November 2020 with a new goal of updating once a quarter.
answered Feb 16, 2020 at 11:37
I would like to add a caveat against the blithe assumptions that chess games/score are not possible to copyright.
Individually, probably not. But as a collection, 'catalogue copyright' just might apply.
In some places, catalogue copyright is used for works that normally could not be protected by copyright, but which still merit some form of protection. Lists, catalogues and directories are typical examples: the protection time is usually less than 10 years.
If it does apply or not depends on the situation: is it national law (say, US Copyright law) that applies? is it the Berne Convention? The World Copyright Convention? Or even individual trade agreements between two countries?
It would almost certainly not be applied for fairly small, random downloads: it might however be applied for full download/dump of a chess database. An organization that plans to use it may add identifiable material that makes it possible to say where the information was taken from.
I would add the Lichess Elite Database to the list :-) It is the best of lichess games: only higher-rated players in "decent" time controls (aka no bullet).
Mere aggregation is not protected by copyright in the US. Copyright only protects creative production, not a collection of facts as stated above. Annotations and comments might be copyrighted but it wouldn't add any copyright collection to the moves in the game.