Research And Application Of Similarity Measurement Between Normal Cloud Models Based On Overlap Degree

With the rapid development of information technology, the era of big datacome quietly, to using such a huge amount of data, we need a fast and effectivemethod to give a scientific analysis and decision or recommendation. However,with the continuous development of the data acquisition and processingtechnology, uncertain knowledge representation and processing has become aproblem to be solved in the network computing. And as a two-way cognitivecomputing model cloud model by given samples randomly determined degree todescribe the concept of randomness, fuzziness, the cloud model in machinelearning, data mining and artificial intelligence have been widely applied insome fields.Go up century ninety time metaphase, academician Li Deyi put forward themathematical model--cloud model that is a combination of qualitative andquantitative. After twenty years of development, cloud model has extended tothe cloud theory. On the cloud theory, the researchers put forward the concept ofsimilarity measure of cloud model that is used to express the degree ofcorrelation between the similar concept of multiple values of the cloud indifferent languages. This is to meet the uncertainty of data classification, opened up a new road for optimization of uncertain data classification andrecommendation system. In this paper, using the characteristics of the similarityof cloud model and aiming at the instability and lack of high time complexity,the Overlap Based Expectation Curve of Cloud Model algorithm and theOverlap Based Maximum Boundary of Cloud Model algorithm were putforward. The OECM and OMCM algorithm were applied to the classification oftime series and the collaborative filtering recommendation system. Experimentalresults show that the measure can not only give full play to the uncertainty of theconcept of similar cloud data model, but also can make up for the lack of moreof the existing algorithm, improve the accuracy, stability and reduce the timecomplexity.In this paper, the main research work includes the following aspects:Firstly, the cloud model similarity measurement methods are divided intothree types, and analysis method based on the cloud, surrounded by a public areamethod and vector method and explain the advantages and disadvantages;Secondly, according to the cloud model rules of3En, this paper gives somedefinitions, including Below Cloud, Contain Cloud and the Degree Overlap ofCloud Modelï¼› In other words, from the horizontal, two position between thecloud and the logical relationship are defined, and this enrich basic algorithm ofcloud model;Thirdly, aiming at the shortage of high time complexity and instability ofthe results, make full use of the excellent characteristics of cloud model expectation curve and the boundary curve to propose OECM and OMCMalgorithms;Fourth, to explore the similarity characteristics of cloud model, thedefinition of bounded, symmetry, reflexivity, uniqueness;Fifth, the OECM and the OMCM algorithm is applied to time seriesclassification in this large data set, excellent characteristics respectively from theexperimental results, the correct rate of the stability and the time complexity ofthe algorithm in three aspects;Then, the OECM and OMCM algorithm is applied to the collaborativefiltering recommendation for MovieLens sites, the Mean Absolute Error (MAE)analysis the OECM and OMCM algorithm and the widespread applicationprospect. According to the nearest neighbor number equal to forty of the optimalvalue of the results, deep mining the relationship between users and projectsdata.Finally, the new algorithm was applied to Jester Joke data set that was agreater data and prediction score results and the actual score was normalizedmean absolute error calculation. Verify that the new proposed algorithm in thefeasibility in the field of collaborative filtering recommendation system.