Breaking the Deadlock: Simultaneously Discovering Attribute Matching and Cluster Matching with Multi-Objective...

Area of Science:

• Data Science
• Bioinformatics
• Computational Science

Background:

• Integrating heterogeneous datasets from different research labs presents significant challenges.
• Existing methods struggle with attribute matching (finding correspondences between numeric features) and cluster matching (aligning patterns across datasets).

Purpose of the Study:

• To propose a unified data mining approach for solving attribute matching and cluster matching problems in heterogeneous datasets.
• To develop and evaluate a multi-objective optimization framework for effective data integration.

Main Methods:

• A multi-objective metaheuristics algorithm was developed to address attribute and cluster matching simultaneously.
• The proposed algorithm was compared against the genetic algorithm for performance evaluation.
• Experiments were conducted using both synthetic and realistic heterogeneous datasets.

Main Results:

• The multi-objective metaheuristics algorithm demonstrated effectiveness in solving the combined attribute and cluster matching problems.
• The approach successfully facilitated the integration of information from disparate scientific research results.
• Performance comparisons indicated the proposed method's viability against established algorithms like the genetic algorithm.

Conclusions:

• The presented data mining approach offers a robust solution for matching heterogeneous datasets.
• This work advances the field of data integration by providing a unified framework for complex matching tasks.
• The findings have implications for improving data analysis and knowledge discovery in multi-source scientific research.