Hokm and Shelem: event logs of human play in two Persian trick-taking card games

This dataset contains server-side event logs of human play from an online platform for two Persian partnership trick-taking card games: Hokm and Shelem (with jokers). Key Details Volume: 157,683 timestamped events across 433 games (222 Hokm, 211 Shelem), 2,340 hands, and 91 pseudonymous human players. State Reconstruction: Pre-action snapshots allow full recovery of initial deals and hidden hands for 98.7% of hands. Validation: Rules-based verification shows 0 mismatches across 38,462 checks. Important Caveat: Deals are non-uniformly random by design (Hokm deals are balanced; Shelem deals depend on the prior hand's play). Files & Contents (v1.1) raw/: Pseudonymized event logs in JSON Lines format (one per game). derived/: 8 relational CSV tables (games, hands, tricks, legal actions, bids, fixed-menu chat/emojis). code/: Python 3.9+ scripts (standard library only) for byte-for-byte reconstruction and statistical analysis. README.md: Data dictionary, game mechanics, and documentation. Privacy No PII, IPs, or free text (chat is strictly menu-based). Account IDs are salted SHA-256 pseudonyms (P001–P091), and dates are shifted to the year 2000 (preserving weekdays and UTC times). License: Data: CC BY 4.0 | Code: MIT