A Configurable Information Extraction Method Based on Relational Table

Abstract

A configurable information extraction method based on a relational table. Firstly, the information extraction task is defined in a structured form, and then a combination of manual and machine learning is adopted to construct extraction rules to extract target pages; 1) Information extraction user interface: the user interface allows users to express information extraction requirements in tabular form, including the main body of information extraction; 2) Information extraction rule set generation, the extraction rule set of required information is divided into two parts: manually constructed Extraction rules and rules automatically generated by machine learning; 3) Data extraction: the extraction of information on the webpage and the persistence of the results: when extracting information from a webpage according to the information extraction table configured by the user, each attribute The content is extracted, and then the trained model is used to classify it.

Description

[0001] 1. Technical field [0002] The invention belongs to the Internet data extraction in the computer field, and in particular relates to a configurable information extraction framework technology based on a relational table. [0003] 2. Background technology [0004] With the rapid development of the Internet, people's lives are increasingly inseparable from the Internet, and the amount of information on the Internet is also increasing. [0005] The explosive growth of the Internet has made the Internet a huge information source, capable of providing massive amounts of valuable information. For users, how to effectively obtain and utilize such information has become particularly urgent and important. At present, most of the data on the Internet appear in the form of HTML. The information in HTML documents is mainly display-oriented, lacking a description of the data itself, and does not contain semantic information. Most of it is unstructured or semi-structured data. This...

AI Technical Summary

Problems solved by technology

[0013] 1. In terms of versatility, some can only extract information from the plain text content of web pages; some are only suitable for the extraction of semi-structured data; some rely on the inherent structure of web pages and can only be extracted from similar web pages, etc.

[0014] 2. Existing extraction techniques are mainly aimed at extracting the information that appears clearly on the webpage, without considering the information implicit in the webpage.

Images