Recognition Method of Time Expressions in English Social Media Short Texts Based on Constraint Model

Abstract

The invention provides a constraint model-based recognition method for short text time expressions in English social media, belonging to the field of text time information extraction. The present invention aims at the problem that the features of the existing recognition methods are not specific to time expressions, and the number of them is small, and constructs a feature set from three aspects of word structure, grammatical structure, and combination features, so that the machine learning model can be used for time expressions. The accuracy is greatly improved; the present invention aims at the problem that conditional random fields share transfer features during the solution process, resulting in inaccurate identification of the boundary of the time expression. During the solution process, the extracted current position feature set is used to perform Amended, it is proposed to use the method of constrained random field to solve it, thus improving the accuracy of the identified time expression and making the boundary identification of the time expression more accurate.

Description

technical field [0001] The invention belongs to the field of text time information extraction, in particular to a method for recognizing short text time expressions in English social media based on a constraint model. Background technique [0002] With the rapid development of the Internet and smart mobile terminals, social media, a new product, is also developing rapidly at the same time. More and more people are used to publishing various text information on social media, and the time information in the text is very important. A part of time information in text is also often referred to as time expression. The time expression can express the relationship between the text content and time sent by people, so that the timeliness of the text content sent by people can be judged, and it plays a very important role in the application of event detection, automatic question answering, information extraction, etc. How to extract time expressions from short texts in social media ha...

AI Technical Summary

Problems solved by technology

But first of all, the traditional named entity method is aimed at many categories, including person names, organization names, place names, etc. In order to meet the needs of identifying various entities, its feature extraction has a certain degree of universality, and there is no feature for time expressions, resulting in It does not recognize temporal expressions well

Secondly, most of the traditional named entity recognition methods are aimed at formal texts, but the actual network texts are randomly generated by users, and the text length is generally short, and most of them are informal texts, which are noisy and have irregular structure patterns, which lead to traditional naming. Entity methods are less effective at identifying entities in web text

Images