Event key information extraction method combining domain synonym dictionary and pattern matching

Abstract

The invention discloses an event key information extraction method combining domain synonym dictionary and pattern matching. The method comprises the following steps: from real application scenarios and for a specific field, performing information extraction to clarify a target and design a filling groove; constructing an exclusive information dictionary in combination with the uniqueness of specific field information, and expanding a knowledge base through seed words by adopting a word vector method based on deep learning; conducting a set of rule system capable of performing shallow syntactic analysis through pattern matching on the basis of the knowledge base; inviting a linguist to perform feature analysis on related texts in the specific field, compiling rules, and finally completing key information extraction in the specific field. According to the method, rules and systems are separated by executing the rules and extracting the related information, linguistics and computer science are combined in the aspect of rule writing, and the accuracy of information extraction is jointly improved by integrating the force of language workers and the force of computer science technicians.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to an event key information extraction method based on the combination of domain thesaurus and pattern matching. Background technique [0002] With the widespread popularization of computers in various fields and the rapid development of the Internet, the total amount of information in society is increasing exponentially. The magnitude of the total amount of information has transitioned from MB (10^6) in the early 1990s to GB (10^9) to the current TB (10^12). After entering the 21st century, the total amount of information in the world is doubling every three years. According to statistics, 60% to 70% of these massive amounts of information exist in the form of electronic documents. In order to cope with the challenges brought by the information explosion, there is an urgent need for some automated technologies to help people quickly find the information they really need i...

AI Technical Summary

Problems solved by technology

The shortcomings of this method are mainly reflected in the complexity of the training process of the model, and the efficiency is not very high.

[0010] At present, most of the information extraction methods take the texts publicly released on the Internet as the main research objects, such as Wikipedia and Baidu Encyclopedia data. The extracted information has no clear target, which belongs to the general method research stage.

Images