An information acquisition method and system based on remote relationship extraction

Abstract

The present application provides an information acquisition system based on remote relationship extraction, the information acquisition system includes an information retrieval module, an intelligence acquisition module and an intelligence analysis module. The invention constructs a web-oriented industry domain information acquisition framework, and proposes methods such as knowledge fusion and domain rules to reduce the "noise" interference problem that occurs when the knowledge base "remote supervision" labels data; and utilizes the good performance of the factor graph model , which realizes the automatic extraction task of entity relationship in the process of intelligence perception. The above framework and methods can be effectively applied to intelligence perception, data analysis and acquisition, and are domain-transplantable.

Description

technical field [0001] The present invention relates to the technical field of data information acquisition, in particular to an information acquisition method and system based on remote relationship extraction. Background technique [0002] In the early stage, intelligence perception needs to rely on manual processing to obtain information, and then gain insight into perception intelligence and make decisions. With the development of information technology, a method based on webpage search has emerged for intelligence acquisition. This method mainly relies on web search engines and improved technology to search for relevant webpages. However, the amount of information collected on webpages is too large and complex, making intelligence processing and analysis difficult. Increased, the timeliness of intelligence perception is insufficient. For this reason, the intelligence acquisition method based on web text mining and information extraction has become a new research hotspo...

AI Technical Summary

Problems solved by technology

Semi-supervised learning methods are prone to "semantic drift" problems in the iterative process, while unsupervised learning methods are prone to associated problems such as instance aggregation errors in the mapping process, making the current three types of entity relationship extraction methods encounter certain difficulties

[0006] Website resources contain a wealth of intelligence information. Traditional intelligence acquisition methods directly use the results of web page retrieval as the research object, making intelligence perception work increasingly difficult in today's big data environment.

Images