Scholarlink

<p>Case-based reasoning systems (CBRs) are widely acknowledged for their scalability, as they excel in handling large volumes of cases. Model bases can be considered as more encompassing generalizations of case bases. This article provides a novel viewpoint on how knowledge is represented and how problems are solved in CBRs. It also presents a new approach termed ‘Middle-Out Generalization.’ The novel method of generating patterns using case bases distinguishes itself from previous studies. It does not combine CBR with PBR but instead demonstrates a novel technique for generalizing cases, improving case-based reasoning efficiency, and reducing computing time. Tests on chess endgames are a proof of concept rather than the primary focus; experiments demonstrate the effectiveness of our approach, showing that the system can successfully improve its performance, with no look-ahead, while reducing execution time. Key findings from the experiments highlight that it nearly mirrors Stockfish’s suggested best move 99% of the time.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Extending the scope of case-based reasoning through multiple case-based generalizations

  • 期刊论文

  • 发表日期：

    2025年6月3日

查看全文
  • Zahira Ghalem,
  • Djamel Eddine Zegour,
  • Thouraya Bouabana-Tebibel,
  • Stuart H. Rubin

摘要

Case-based reasoning systems (CBRs) are widely acknowledged for their scalability, as they excel in handling large volumes of cases. Model bases can be considered as more encompassing generalizations of case bases. This article provides a novel viewpoint on how knowledge is represented and how problems are solved in CBRs. It also presents a new approach termed ‘Middle-Out Generalization.’ The novel method of generating patterns using case bases distinguishes itself from previous studies. It does not combine CBR with PBR but instead demonstrates a novel technique for generalizing cases, improving case-based reasoning efficiency, and reducing computing time. Tests on chess endgames are a proof of concept rather than the primary focus; experiments demonstrate the effectiveness of our approach, showing that the system can successfully improve its performance, with no look-ahead, while reducing execution time. Key findings from the experiments highlight that it nearly mirrors Stockfish’s suggested best move 99% of the time.