Chinese named entity recognition method based on sequence interaction and syntactic integration

By introducing syntax integration and sequence interaction information into the Chinese named entity recognition model and combining the dynamic gating mechanism for information fusion, the problem of insufficient accuracy of named entity recognition caused by ignoring syntactic information in the existing technology is solved, and higher accuracy of named entity recognition and model understanding ability are achieved.

CN119204018BActive Publication Date: 2025-03-25JIANGXI NORMAL UNIV
1 Cites 0 Cited by

Patent Information

Application Number
CN202411451862.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-10-17
Publication Date
2025-03-25
Estimated Expiration
2044-10-17

AI Technical Summary

Technical Problem

The prior art ignores syntactic information in Chinese text in Chinese naming entity recognition, making it difficult for the model to accurately identify named entities, especially when there are ambiguity in the input sequence.

Method used

Using a method based on sequence interaction and syntax integration, a named entity recognition model including sequence interaction module, syntax integration module, information fusion module and classifier is constructed. The model obtains sequence interaction information feature vectors through the cross attention mechanism, and obtains syntactic information through the natural language processing toolkit, combines the dynamic gating mechanism to perform information fusion, and finally input conditions randomly performs entity tag sequence output.

Benefits of technology

It effectively reveals the internal structure of the text, improves the accuracy of naming entity recognition, especially when dealing with ambiguity, and improves the model's understanding of input sequences.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention discloses a Chinese named entity recognition method based on sequence interaction and syntactic integration. A retrieval algorithm is used to retrieve relevant context sequences based on an input sequence, and a Chinese natural language processing toolkit is used to obtain three types of syntactic information of the input sequence, namely, syntactic components, part-of-speech tags, and dependency relationships; the input sequence, the context sequence, and the three types of syntactic information are respectively input into a named entity recognition model to obtain sequence interaction information that fuses the input sequence and the context sequence, and syntactic integration information that fuses the syntactic components, part-of-speech tags, and dependency relationships; finally, the two types of information are dynamically fused and input into a conditional random field to obtain a named entity label sequence of the input sequence. By fusing sequence interaction information and syntactic integration information, not only is the missing entity information of the sentence-level input sequence effectively supplemented, but also the text structure information about the sentence in the model is introduced, and the importance of the two types of information is flexibly adjusted, thereby achieving higher recognition performance.
Need to check novelty before this filing date? Find Prior Art