Chinese comparative sentence recognizing method and device based on neural network

Abstract

The invention relates to a Chinese comparative sentence recognizing method and device based on a neural network. The method comprises the following steps that firstly, normalization processing is conducted on linguistic data and word segmentation results of the linguistic data, and then coarse granularity extraction is conducted through a method based on the combination of comparative feature word dictionaries, syntactic structure templates and dependence relationships; finally, a class sequential rule (CSR) extraction algorithm is designed, CRF is used for mining entity object information, namely, semantic role information, comparative feature word features and statistical word features are used for assistance, the BP neural network is constructed by using the four features, and a feature type with the optimal performance is found out for completing fine granularity extraction. Compared with the prior art, the Chinese comparative sentence recognizing method and device based on the neural network have the advantages that different levels of semantic information can be effectively used for conducting recognizing, therefore, a user can find out Chinese comparative sentences with high confidence coefficients from the linguistic data quickly and accurately and the recognizing rate of the Chinese comparative sentences can be increased.

Description

technical field [0001] The invention relates to a method for recognizing Chinese comparative sentences, in particular to a method for recognizing Chinese comparative sentences based on a neural network, and belongs to the technical field of natural language processing applications. Background technique [0002] Every day people are surrounded by all kinds of choices. In order to make better choices, we often choose to compare items of interest to us. At the same time, with the vigorous development of the mobile Internet, the Internet has gradually become more humane. Under the guidance of the open architecture concept of "user-centered and user participation", Internet users are gradually moving from passive browsing of Internet information to active creation of Internet information. Internet users are not only browsers of website content, but also producers of website content. Therefore, there are a large number of user-made comment information on the Internet, which exp...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the problem that syntactic and semantic information cannot be effectively used in the recognition of Chinese comparative sentences in the past, and to give full play to the characteristics of semantic information at different levels, and to propose a method for recognizing Chinese comparative sentences based on neural networks

Images