Heterogeneous information network embedding method based on edge sampling

Abstract

The invention discloses a heterogeneous information network embedding method based on edge sampling, and the method comprises the following steps: S1, carrying out the preprocessing of the weight of each edge in an information data set, and obtaining a migration weight; S2, selecting the ith information element node from the information element node set as a migration starting point, and carrying out multiple migration with a rated length on the migration starting point based on edge sampling to obtain a migration array; S3, repeating the step S2 until i is equal to the total number of the nodes in the intelligence element node set V, namely '2jeemaa2' V '2jeemaa2', and constructing a plurality of walk arrays into a walk array set; S4, selecting one migration array in the migration array set, constructing a training sample based on the migration array, and training the single-hidden-layer neural network; and S5, inputting the intelligence element nodes in the intelligence element node set into the trained single-hidden-layer neural network to obtain an embedded vector. According to the method, the problems that the weighted graph cannot be processed and the migration is limited in the existing graph embedding algorithm are solved.

Description

technical field [0001] The invention relates to the technical field of data security, in particular to a heterogeneous information network embedding method based on edge sampling. Background technique [0002] In the prior art, there are few methods for hierarchical classification of intelligence data, and there are only methods for dividing according to user attributes or specific behaviors. This type of method depends on user attributes and behaviors, and the scalability is not strong. In the prior art, there is a method of intelligence analysis. In the push method based on content similarity, keywords and expert scores are used to establish a user interest model, and the recommendation is made after calculating the similarity. The essence is still the content of the basic matrix decomposition, and the disadvantage is that it cannot be solved. The problem of matrix sparsity; the method based on the retrieval system uses different international and domestic online joint ret...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the above-mentioned deficiencies in the prior art, a heterogeneous information network embedding method based on edge sampling provided by the present invention solves the problems that existing graph embedding algorithms cannot handle weighted graphs and limited walks

Images