Table information extraction method, device, storage medium and electronic equipment

Abstract

The invention discloses a form information extraction method, device, storage medium and electronic equipment, which preprocesses the merged data in the target form to obtain a two-dimensional matrix; and inputs the row and / or column data of the two-dimensional matrix to the table header Detect the model, obtain the header detection result, and determine the table style of the target table; generate a cell text sequence and a header text matrix according to the processing mode corresponding to the table style; obtain each text in the target field text and the header text sequence The matching result of the target field text is obtained, and the cell object corresponding to the target field text is obtained; based on the value sequence of the cell object corresponding to each target field text, the index information of the row or column corresponding to the matching result is established to realize the information extraction of the target cell , to get the extraction result. The present invention is more suitable for complex form information extraction through the header detection model and the text matching processing of the header text sequence, and improves the efficiency and accuracy of information extraction of complex forms.

Description

technical field [0001] The present invention relates to the technical field of information processing, and in particular, to a method, device, storage medium and electronic device for extracting table information. Background technique [0002] With the growth of data volume, data extraction technology is an effective means to mine target information. Among various forms of information, tables are an important form of data representation in documents, and are usually used to organize the basic information and statistical data of the described objects. Wait. At present, most of the table information extraction methods are manual extraction methods, which require a lot of manpower and time costs. [0003] Although machine learning methods are also used to extract table information, for tables with more complex information, the accuracy of the extraction results is not high. For example, tables in the financial field have complex headers, diverse announcement types and annotat...

AI Technical Summary

Problems solved by technology

At present, the information extraction method of tables is mostly manual extraction, which requires a lot of manpower and time costs.

[0003] Although machine learning methods are also used to extract table information, the accuracy of the extraction results is not high for tables with more complex information.

For example, tables in the financial field have complex headers, diverse announcement types, and annotation types. For each different extraction requirement, data needs to be labeled, and the amount of labeling is large, which reduces the final table information extraction results. accuracy rate

Images