Chinese enterprise name entity accurate identification secondary matching method

Abstract

The invention provides a Chinese enterprise name entity accurate identification secondary matching method, capable of identifying an accurate target entity from a plurality of entities mentioned by the public through a secondary matching method; extracting potential entities from public opinions through a BiLSTM-CRF model, and extracting feature words through feature words through TF-IDF; realizing first matching between the potential entity and the industrial and commercial enterprise catalog through a Jaro-Winkler distance algorithm; and as an industry feature word dictionary needs to be constructed for secondary recognition, recognizing the accurate entity is recognized according to the number of feature words matched with the industry feature word dictionary. Experiments prove that thehighest correlation rate and accuracy of the method are 0.93 and 0.95 respectively, and the accuracy of the method is improved by about 30% compared with the accuracy of only using a matching processonce.

Description

technical field [0001] The invention relates to the technical field of text recognition, in particular to a secondary matching method for accurate recognition of Chinese business name entities. Background technique [0002] Public opinion often contains key information pointing to specific entities of social cognition. Entity recognition is one of the core technologies of information retrieval and natural language processing. Traditional information retrieval and natural language processing technologies have been able to accurately identify Chinese name entities, but non-standard public expressions often involve multiple entities, confusing the real target. [0003] Aiming at information retrieval technology, Chinese patent application CN201910129259.9 discloses a method for automatic completion of enterprise names of contacts in an address book. Chinese patent application CN201511016095.7 discloses a method and system for data retrieval and retrieval result presentation, ...

AI Technical Summary

Problems solved by technology

But the problem is that the company name has a short name (name) and a full name (full name of the company), and the company name mentioned by the public uses a short name. When it comes to matching problems, it involves the short name and the full name. match comparison

[0008] In summary, the existing Chinese enterprise name recognition technology cannot achieve accurate recognition when matching with ambiguous expressions in public opinion.

Images