Distributed Association Rule Mining Algorithm Based On Related Interest And It’s Application In Diagnosis Of Acute Inflammation

With the increasingly development of internet communication technology,the medical industry has gathered a tremendous amount of structured and unstructured data.These data which contains potential and valuable information can not be obtained by simple collection or statistical analysis.So we can use relevant data mining techniques to explore the internal links between medical data and find the knowledge which is helpful to medical diagnosis.In this paper,we study algorithm of association rule and make some related improvements to extract the association between disease and symptoms in acute inflammatory medical data,and provide auxiliary decision for medical diagnosis.The main researches are as follows:(1)In view of the demerits of association rules mining technology which is based on the degree of support and confidence,we propose a new association rules mining algorithm which named I&ItemMine_AC(ItemSet&Item association correlation)based on the related interest measure.Experiments proved that the mining algorithm improves the quality of generated association rules.And it’s relevant measure is very effective in pruning.(2)Distributed and parallel computing based on Spark cluster is realized for the limitation of computing power and memory space for stand-alone processing and storage of large data.The experiment shows that the distributed and parallel computing is superiorly time-sensitive and extensible.(3)Based on the association rules mining algorithm and Spark,a medical auxiliary diagnosis system for acute inflammation is constructed.The system is tested by the data set from UCI,which proves the effectiveness and feasibility of the system.