A Novel Attribute Reduction Algorithm for Incomplete Information Systems Based on a Binary Similarity Matrix

With databases growing at an unrelenting rate, it may be difficult and complex to extract statistics by accessing all of the data in many practical problems. Attribute reduction, as an effective method to remove redundant attributes from massive data, has demonstrated its remarkable capability in simplifying information systems. In this paper, we concentrate on reducing attributes in incomplete information systems. We introduce a novel definition of a binary similarity matrix and present a method to calculate the significance of attributes in correspondence. Secondly, We develop a heuristic attribute reduction algorithm using a binary similarity matrix and attribute significance as heuristic knowledge. In addition, we use a numerical example to showcase the practicality and accuracy of the algorithm. In conclusion, we demonstrate through comparative analysis that our algorithm outperforms some existing attribute reduction methods.