Semantic search implementation method and system, computer equipment and storage medium

Abstract

The invention discloses a semantic search implementation method. Comprising the following steps: inputting a to-be-recognized text into a first NER model to obtain a text fragment mark sequence; inputting the fragments into a second NER model to obtain a fragment entity mark sequence; performing format conversion on the text fragment mark sequence and the fragment entity mark sequence by using a preset classification symbol and preset node data according to a preset rule to obtain a text relation leaf node sequence and a fragment relation leaf node sequence; respectively inputting a relation tree generation model to obtain a text father node coordinate sequence and a fragment father node coordinate sequence; respectively generating a corresponding text relation tree structure and a fragment relation tree structure; using the fragment relation tree structure to replace leaf nodes corresponding to the context fragments containing the entities in the text relation tree structure, and generating a semantic search relation tree structure. According to the method, the generalization of semantic search can be enhanced, the robustness of coping with complex query logic can be enhanced, and the recognition precision of long and difficult sentences can be improved.

Description

technical field [0001] The present invention relates to the technical field of semantic search, in particular to a method, system, computer device and storage medium for implementing semantic search. Background technique [0002] Semantic search is a natural language processing technology widely used in search engines. After semantic understanding of natural language input by users, it is parsed into underlying databases, including relational databases, non-relational databases, and graph databases. to extract the data and information that the user wants to search for. [0003] Entity extraction is an important part of semantic search. At present, the existing entity extraction technology is mainly realized through template matching and rule system, but the generalization of this approach is not high, and the existing implementation methods of named entity recognition There is also a precision bottleneck, making it difficult to handle long sentences with multiple entity typ...

AI Technical Summary

Problems solved by technology

[0003]Entity extraction is an important part of semantic search. At present, the existing entity extraction technology is mainly realized by template matching and rule system, but the general The generalization is not high, and the existing implementation methods of named entity recognition also have accuracy bottlenecks, and it is difficult to deal with long sentences of various entity types and sentences that are difficult to understand

Moreover, the existing semantic search implementation methods are also difficult to deal with complex query logic, such as multiple combinations and nested combinations of AND or NOT relations, and the search accuracy is not high

Images