Association rule mining method based on concept lattice extension theory

Abstract

The invention discloses an association rule mining method based on the concept lattice extension theory. The method includes: extracting objects, attributes and attribute values in a data set according to concerned relations, building the formal context of a target data set, and using the basic concept lattice building theory to perform analysis so as to obtain concept nodes; judging elements on concept lattices, and calculating to obtain an irreducible element set on the concept lattice L (FC); reducing association rules according to set support degree and credibility to further increase algorithm calculation efficiency; calculating method complexity to indicate calculation quantity. The method has the advantages that the narrow-sense concept lattice theory is extended on the basis of traditional-form conceptual analysis for extracting association rules, and data set object differences are paid attention to while object similarity is explored; meanwhile, the corresponding attribute and rule reduction algorithm is used to guarantee the calculation efficiency of the method, and the method is fast, stable and efficient.

Description

technical field [0001] The invention belongs to the fields of computers, big data and cloud computing, and relates to a method for mining association rules between data set objects and attributes. Background technique [0002] In the field of data mining association rule discovery, the concept lattice constructed by binary relations is a very intuitive formal concept analysis method, which utilizes the connotation (attribute) and extension (object) of concepts and the generalization and relationship between knowledge concepts. The specialization relationship can fully reflect the implication rules between objects and attributes. However, the current concept lattice method puts too much emphasis on the differences between objects, but does not fully consider their similarities. In the field of business data analysis with strong sporadic nature, some subtle commonalities between objects are often ignored, resulting in rule Mining is too harsh and misses some subtle and key ru...

AI Technical Summary

Problems solved by technology

However, the current concept lattice method places too much emphasis on the differences between objects, but does not fully consider their similarities. In the field of business data analysis, which is extremely sporadic, it often ignores some subtle commonalities between objects, resulting in rule Mining is too harsh and misses some subtle and key rule discoveries; some emerging algorithms have added the similarity considerations between objects, but because the resulting association rules are too complex, the timeliness of the algorithm is too poor, and the resource usage is too high, so it cannot be used in Play a role in fields requiring high real-time performance

Images