Maximum frequent item set mining method based on adjacency list

Abstract

The invention relates to a maximum frequent item set mining method based on an adjacency list, and the method employs a storage mode combining the adjacency list and a Hash table to reduce the scanning frequency of a database and the spatial scale of traversal, can trim out an item set smaller than a support degree threshold as early as possible, and avoids the generation of all non-empty subsetsof a longer maximum frequent item set. According to the method, the established adjacency list is fully utilized, the original database only needs to be scanned once, and the method has the advantagesof low time complexity, memory consumption and the like.

Description

Technical field [0001] Maximum frequent itemsets mining is an important issue in many data mining applications. The present invention relates to a maximum frequent itemsets mining technology based on adjacency lists. In the process of studying the maximum frequent itemsets mining, considering the application of adjacency lists as data storage media The data is stored for later mining of the largest frequent itemsets. The technology can be applied in many fields, including health care, education, manufacturing, etc. It can also be used in daily customer relationship mining management, fraud intrusion analysis, online shopping analysis and other aspects, with a very large range of applications And market potential. Background technique [0002] Data mining is the process of extracting potential, unknown and not easy to be discovered intuitively from a large amount of data, and finally can be expressed as understandable knowledge. Association rule mining is one of the important co...

AI Technical Summary

Problems solved by technology

But it needs to create an FP-Tree containing all dataset items, which takes a lot of memory, and the memory consumption is proportional to the FP-Tree width and depth

The depth is generally the maximum number of items in a single transaction. If the number of frequent 1-itemsets in the database is large, and the memory cannot load the mapping information of all items in the FP-Tree, the algorithm will not work effectively

And scanning the database twice, generating and recursive FP-Tree multiple times also makes the space and time performance of the algorithm not high

Images