Scholarlink

<p>Discovering spatial co-location patterns that are both prevalent and valuable is crucial for utility-aware spatial data analysis. High Utility Co-location Pattern Mining (HUCPM) incorporates utility into pattern discovery, but most existing methods neglect the possibility of negative utilities, leading to incorrect or incomplete results in practical scenarios where such values often occur (e.g., maintenance costs of free facilities). To overcome this limitation, we propose PSO-MUCP (Pruning Strategies Optimized Mixed-Utility Co-location Pattern Mining Algorithm), a novel algorithm that supports mixed utility pattern mining. PSO-MUCP introduces four targeted pruning strategies to effectively reduce the number of unpromising candidate patterns, and adopts a compact <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(k\)</EquationSource> </InlineEquation>-utility list structure to improve mining efficiency and scalability. These innovations not only ensure completeness but also significantly mitigate the inefficiency caused by the lack of anti-monotonicity in utility-based pattern mining. Extensive experiments on real-world and synthetic datasets demonstrate that PSO-MUCP accurately mines all high utility patterns, reduces candidate patterns by an average of 97.38%, and significantly outperforms state-of-the-art HUCPM algorithms in execution 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

Efficient pruning strategies for mining high utility co-location patterns with negative utility features

  • 期刊论文

  • 发表日期：

    2026年4月30日

查看全文
  • Xuguang Bao,
  • Shuaikang Yuan,
  • Yongming Huang,
  • Liang Chang,
  • Tianlong Gu

摘要

Discovering spatial co-location patterns that are both prevalent and valuable is crucial for utility-aware spatial data analysis. High Utility Co-location Pattern Mining (HUCPM) incorporates utility into pattern discovery, but most existing methods neglect the possibility of negative utilities, leading to incorrect or incomplete results in practical scenarios where such values often occur (e.g., maintenance costs of free facilities). To overcome this limitation, we propose PSO-MUCP (Pruning Strategies Optimized Mixed-Utility Co-location Pattern Mining Algorithm), a novel algorithm that supports mixed utility pattern mining. PSO-MUCP introduces four targeted pruning strategies to effectively reduce the number of unpromising candidate patterns, and adopts a compact \(k\) -utility list structure to improve mining efficiency and scalability. These innovations not only ensure completeness but also significantly mitigate the inefficiency caused by the lack of anti-monotonicity in utility-based pattern mining. Extensive experiments on real-world and synthetic datasets demonstrate that PSO-MUCP accurately mines all high utility patterns, reduces candidate patterns by an average of 97.38%, and significantly outperforms state-of-the-art HUCPM algorithms in execution time.