Embedded vector generation method and device based on buried point hierarchy, equipment and medium

Abstract

The invention relates to the field of artificial intelligence, and provides an embedded vector generation method and device based on buried point hierarchy, equipment and a medium, which can perform fusion on a dictionary tree to obtain a hierarchical tree structure, not only retain hierarchical information of buried point data, but also ensure that each hierarchy has overall significance and place buried points with the same prefix under a same branch. The similarity of the same prefix buried points is well described, automatic construction of a hierarchical tree is realized in combination with network training, buried point data to be processed is acquired, query is performed in a target hierarchical tree according to the buried point data to be processed to obtain a target embedded vector, and then automatic generation of the embedded vector is realized based on the hierarchical tree. The invention also relates to blockchain technology. The buried point data and the target embeddedvector can be stored in the blockchain.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a method, device, equipment and medium for generating embedding vectors based on buried point levels. Background technique [0002] Buried point is the record of user behavior process and results in order to meet fast, efficient, and rich data applications. In many companies, in order to generate and query buried points efficiently, buried points are often named with fixed prefix rules. [0003] In the existing technical solutions, in order to learn buried point information, the usual method is to perform one-hot encoding on the buried point id to obtain the embedding encoding representation of the buried point, and input it into the model to learn the relevant information of the buried point. Among them, one-hot encoding refers to encoding N states by using N-bit state registers, each state has an independent register bit, and only one bit of these regis...

AI Technical Summary

Problems solved by technology

[0004] However, since the principle of one-hot encoding is to assume that different buried point ids are independent of each other, there is no correlation

Therefore, the prefix rule information of buried points will be lost through one-hot encoding, resulting in the embedding encoding being unable to describe the correlation characteristics between levels

Images