Cascaded conditional random field-based product name recognition method and device

Abstract

The invention relates to a cascaded conditional random field-based product name recognition method and device, and belongs to the technical field of internet data processing and analysis. According to the method, words are expressed by utilizing a word vector method; the semantic similarity of the words is measured through the similarity of vectors; the global context information is fused through a method for combining the word vectors and word clusters; and in allusion to the problem that the product names are complicated in structure and have nesting phenomenon, a cascaded conditional random field model is used for carry out product name recognition. Compared with the prior art, the method and device provided by the invention have the advantages that the problems that the context information is insufficient in the product name recognition and the product names have nesting phenomenon and are complicated in structure are effectively solved, the performance of recognizing the product manes which are complicated in structure is improved, and the product name recognition correctness and FI value are higher than that of the traditional method.

Description

technical field [0001] The invention belongs to the field of Internet data processing and analysis, and relates to a context-dependent product name recognition method and device based on a stacked conditional random field. Background technique [0002] In the Web2.0 era, with the rise of social networking platforms such as Weibo, every Internet user is no longer just a browser of information, but also a publisher of information. The Internet has changed from an information publishing platform to an interactive communication platform. . In the past ten years, my country's e-commerce industry has continued to grow and develop, and more and more companies conduct business on the Internet and promote and sell products through the Internet. As of December 2013, the proportion of enterprises that carry out online sales nationwide has reached 23.5%, and enterprises that conduct marketing and promotion through the Internet also account for 20.9%. More and more people are used to o...

AI Technical Summary

Problems solved by technology

At present, the research on proper noun recognition mainly focuses on the recognition of traditional named entities such as person names, place names, and organization names. With the development of the Internet and e-commerce, the recognition of product names is becoming more and more important. Recognizes relatively little

Images