Scholarlink

<p>High Utility Pattern Mining is a key area in data science for identifying significant patterns within databases, with applications in text mining, bioinformatics, e-learning, medical analysis and recommendation systems. Traditional high utility pattern mining algorithms often generate patterns that are uninteresting due to weak correlations, misleading due to skewed item support distributions, or inefficient in runtime and memory usage. To address these challenges, this paper proposes EICoHUPM, an efficient algorithm for mining interesting correlated high utility patterns. It introduces the novel TND-UTtree data structure and the Maximum Support Limit concept, used with the Kulczynski measure to identify truly interesting patterns, along with two new pruning properties to reduce the search space. Extensive experiments on six benchmark datasets demonstrate that EICoHUPM outperforms state-of-the-art algorithms in runtime, memory usage, and quality of extracted patterns.</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

Extracting interesting CoHUPs from dataset with skewed and non-skewed item’s support distribution

  • 期刊论文

  • 发表日期：

    2026年5月11日

查看全文
  • Rashad Saeed,
  • Azhar Rauf,
  • Fahmi H. Quradaa,
  • Sara Mumtaz,
  • Saqib Alam,
  • Mohammed H. S. Al-Hakimi

摘要

High Utility Pattern Mining is a key area in data science for identifying significant patterns within databases, with applications in text mining, bioinformatics, e-learning, medical analysis and recommendation systems. Traditional high utility pattern mining algorithms often generate patterns that are uninteresting due to weak correlations, misleading due to skewed item support distributions, or inefficient in runtime and memory usage. To address these challenges, this paper proposes EICoHUPM, an efficient algorithm for mining interesting correlated high utility patterns. It introduces the novel TND-UTtree data structure and the Maximum Support Limit concept, used with the Kulczynski measure to identify truly interesting patterns, along with two new pruning properties to reduce the search space. Extensive experiments on six benchmark datasets demonstrate that EICoHUPM outperforms state-of-the-art algorithms in runtime, memory usage, and quality of extracted patterns.