--- language: - en size_categories: - 1B ### Dataset Summary The Lichess.org database has been converted to UCI format, making it easier to analyze and interpret chess games. The `moves` configuration captures the sequence of moves using UCI notation, and the `headers` configuration provides comprehensive metadata for each game, enabling detailed statistical and strategic analysis. The data is subset based on the year and month (yyyymm format) the games took place. ### Supported Tasks and Leaderboards This dataset supports tasks related to chess game analysis, including move prediction, game outcome prediction, performance analysis, and opening strategy evaluation. There are no formal leaderboards associated with this dataset. ## Dataset Structure ### Data Instances An example from the `moves` configuration: ```json { "site": "abcd1234", "transcript": "e2e4 e7e5 g1f3 b8c6" } ``` An example from the `headers` configuration: ```json { "event": "Rated Blitz game", "site": "abcd1234", "white": "player1", "black": "player2", "result": "1-0", "utcdate": "2023-05-21", "utctime": "13:45:00", "whiteelo": 1500, "blackelo": 1400, "whiteratingdiff": 10, "blackratingdiff": -10, "eco": "C50", "opening": "Italian Game", "timecontrol": "300+0", "termination": "Normal" } ``` ### Data Fields #### Moves Configuration: - `site`: string, unique identifier for the game. Replays can be viewed by navigating to `https://lichess.org/` - `transcript`: string, sequence of moves in UCI format. #### Headers Configuration: - `event`: string, type of event. - `site`: string, unique identifier for the game. - `white`: string, white player. - `black`: string, black player. - `result`: string, game result. - `utcdate`: date32, date of the game. - `utctime`: time64[us], time of the game. - `whiteelo`: int64, ELO rating of the white player. - `blackelo`: int64, ELO rating of the black player. - `whiteratingdiff`: float64, rating change for the white player. - `blackratingdiff`: float64, rating change for the black player. - `eco`: string, ECO code of the opening. - `opening`: string, name of the opening. - `timecontrol`: string, time control format. - `termination`: string, reason for game termination. ### Data Splits The dataset is divided into monthly splits based on the year and month (yyyymm format). Each split contains both `moves` and `headers` configurations. ## Dataset Creation ### Curation Rationale The dataset was curated to facilitate research and analysis that use chess games, providing both move sequences and comprehensive metadata. ### Source Data #### Initial Data Collection and Normalization The data was collected from the Lichess.org open database, converted from PGN format to UCI format, and organized into `moves` and `headers` configurations. #### Who are the source data producers? The source data comes from games played on Lichess.org, an online platform where users from around the world play chess. This database does not distinguish between human-and bot-played games. However, it's reasonable to assume games played before April 2018 were overwhelmingly played by human players since Lichess.org released its [bot api](https://github.com/lichess-org/api) on April 1st, 2018. ### Annotations #### Annotation process The annotations include metadata such as player ELO ratings, game outcomes, and dates, which were extracted from the original PGN files and normalized into a structured format. #### Who are the annotators? The annotations were generated by the Lichess.org platform and curated by the dataset creators. ### Personal and Sensitive Information The dataset does not contain any personal or sensitive information. ## Considerations for Using the Data ### Social Impact of Dataset The dataset can be used to improve chess engines, develop training tools for players, and conduct research in game theory and artificial intelligence. ### Discussion of Biases The dataset may have biases related to the demographics of Lichess.org users, such as skill level distribution and regional representation. ### Other Known Limitations The dataset is limited to games played on Lichess.org and may not represent the broader chess-playing population. ## Additional Information ### Dataset Curators The dataset was curated by the Lichess.org team and contributors. ### Licensing Information The dataset is available under the Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license. ### Citation Information If you use this dataset, please cite it as follows: ``` @misc{lichess_uci, author = {Davis, Austin L.}, title = {Lichess.org Database in UCI format}, year = {2023}, howpublished = {\url{https://database.lichess.org/}}, } ``` ### Contributions Thanks to [@austinleedavis](https://github.com/austinleedavis) for adding this dataset.