Instance retrieval method based on fusion of association function and D-HS index

Abstract

The invention discloses an instance retrieval method based on fusion of an association function and a D-HS index. The method comprises the following steps: proposing a concept and a corresponding calculation formula of a fuzzy correlation function in combination with a fuzzy number, and proposing a single-dimensional correlation function combination similarity measurement method in combination with an analytic hierarchy process. Compared with a full-dimensional correlation function calculation method, the method not only ensures index precision, but also improves index speed when facing high-dimensional retrieval. A Sigmoid function is used for carrying out nonlinear transformation on data, so that the problem of low index accuracy caused by non-uniform data distribution in a traditional D-HS index method is effectively solved. The improved method is fused, so that the preciseness and accuracy of the traditional D-HS index are improved. The complexity that an association function needsto perform similarity calculation on all instance library instances is avoided. The retrieval effectiveness problem is solved to a certain extent.

Description

technical field [0001] The invention relates to an instance retrieval method. Background technique [0002] How to measure the distance between instances is the core issue of instance retrieval. Correlation function, mixed attribute distance, nearest neighbor method and its improvement research have been proposed one after another, and have achieved remarkable results in improving index accuracy. However, instance retrieval is an organic combination of the basic contradictions of index accuracy and index speed. The increasingly complex calculation methods bring a huge amount of calculation while improving the accuracy, resulting in a decrease in index speed. In addition, the indexing speed is not only restricted by the index precision, but also affected by the number of instances in the instance library. During the construction of the instance library, in order to make the instance knowledge as rich as possible, the designer often updates the instance library continuously. T...

AI Technical Summary

Problems solved by technology

However, instance retrieval is an organic combination of the basic contradictions of index accuracy and index speed. The increasingly complex calculation method brings a huge amount of calculation while improving the accuracy, which leads to the decline of index speed.

In addition, the indexing speed is not only restricted by the index precision, but also affected by the number of instances in the instance library. During the construction of the instance library, in order to make the instance knowledge as rich as possible, the designer often updates the instance library continuously, so the instance library There are some redundant calculations in the retrieval that affect the indexing speed and do not benefit the indexing accuracy, that is, the retrieval utility problem

Images