Finding similar chess motifs in the endgames is a problem that chess coaches are often confronted with when designing personalised training for their students. Finding such motifs is often time-consuming, as reviewing and analysing entire games is a time-consuming process.
The approach we have introduced uses static, dynamic and heuristic properties of a chess position to textually describe the positions. Following the generated descriptions, we then used information retrieval methods to efficiently search for and identify similar positions. The methods were then further integrated into the server architecture to improve and facilitate the user experience.
We validated the performance and efficiency of the system through an experiment in the form of a survey completed by chess players.
|
