Scholarlink

<p>In this paper, a Bipolar Fuzzy Decision Information System (BFDIS) framework is developed to address complex decision data exhibiting bipolarity, orderliness, similarity, and noise interference. By introducing score functions and accuracy functions to reconstruct the dominance relation, and integrating this with a bipolar neighborhood relation, we propose for the first time a Bipolar Neighborhood Dominance Rough Set (NDRS) model that enables the unified representation of multidimensional features. Based on this model, a hierarchical attribute reduction framework is designed: the first level introduces an approximate reduction method aimed at preserving the model’s approximate classification capability; at the second level, the concept of Variable Precision Rough Sets (VPRS) is innovatively incorporated to formulate <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\alpha \)</EquationSource> </InlineEquation>-lower distribution reduction a novel approach targeting the consistency of individual lower approximation distributions. This method achieves a balanced trade-off between dimensionality reduction and fault tolerance through the adjustable parameter <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\alpha \)</EquationSource> </InlineEquation>. Experimental results demonstrate that the proposed approach effectively eliminates redundant attributes while preserving semantic and structural integrity, outperforming existing methods in both reduction efficiency and classification accuracy.</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

Dominance-based neighborhood rough set model in bipolar fuzzy information system and Its attribute reduction

  • 期刊论文

  • 发表日期：

    2026年4月6日

查看全文
  • Asiya Mijit,
  • Zengtai Gong

摘要

In this paper, a Bipolar Fuzzy Decision Information System (BFDIS) framework is developed to address complex decision data exhibiting bipolarity, orderliness, similarity, and noise interference. By introducing score functions and accuracy functions to reconstruct the dominance relation, and integrating this with a bipolar neighborhood relation, we propose for the first time a Bipolar Neighborhood Dominance Rough Set (NDRS) model that enables the unified representation of multidimensional features. Based on this model, a hierarchical attribute reduction framework is designed: the first level introduces an approximate reduction method aimed at preserving the model’s approximate classification capability; at the second level, the concept of Variable Precision Rough Sets (VPRS) is innovatively incorporated to formulate \(\alpha \) -lower distribution reduction a novel approach targeting the consistency of individual lower approximation distributions. This method achieves a balanced trade-off between dimensionality reduction and fault tolerance through the adjustable parameter \(\alpha \) . Experimental results demonstrate that the proposed approach effectively eliminates redundant attributes while preserving semantic and structural integrity, outperforming existing methods in both reduction efficiency and classification accuracy.