Multidimensional data space similarity matching method based on relation communication networks

Abstract

The invention relates to a multidimensional data space similarity matching method based on relation communication networks. The method includes the following steps that firstly, n relation communication networks in different data spaces are built, and a relation communication matrix RTN of the relation communication networks is built; secondly, a similarity relation importance weight matrix lambda[ij] of a data space S[i] and a data space S[j] is determined; thirdly, according to the relation communication matrix and the importance weight matrix, a constant matrix R is built; fourthly, K is assigned as 0, and an initial similarity communication matrix RTS0 is acquired; fifthly, the formula in the description is calculated; sixthly, whether ABS(SUM(RTS[(K+1)]-RTS[K]))>=F is achieved or not is judged, if yes, RTS is assigned to be equal to RTS[(k+1)], saving is performed, a similarity communication matrix RTS is acquired, the seventh step is executed, otherwise, K is assigned to be equal to K+1, and the fifth step is executed again; seventhly, elements in RTS are acquired, and matching between data objects in the multi-dimensional space is performed. Compared with the prior art, the similarity matching method can calculate the similarity value of the data objects in the multi-dimensional space can be calculated, and the advantages of high reliability, high expansibility, high feasibility and the like are achieved.

Description

technical field [0001] The invention relates to a similarity matching method, in particular to a multidimensional data space similarity matching method based on a relation propagation network. Background technique [0002] The calculation of similarity between data objects is a basic requirement in many scientific and application fields. For example, the basic method of face recognition is to judge whether two faces are the same face or not by calculating the similarity between two faces. A face is "similar" to another face. In biometrics, it is also necessary to calculate the similarity as a research basis. For example, to determine which plant a given plant leaf belongs to, that is, by calculating the given plant leaf and the plant information base. The similarity of various leaves in , if the similarity between a given plant leaf and the leaves of plant P in the plant information base is calculated by a certain method, it can be determined that the specified leaf belongs ...

AI Technical Summary

Problems solved by technology

The advantages of the SimRank algorithm are: the algorithm comprehensively considers the similarity relationship between different data objects, and calculates the similarity between the specified data objects through the similarity relationship between the data object and other data objects, but the SimRank algorithm does not calculate the similarity between different data objects. Classify the similarity relationship among them, so that the algorithm lacks a theoretical basis for the expansion to multi-dimensional data space

Images