Method and system for realizing full-text retrieval of electronic official document

Abstract

The invention discloses a method and a system for realizing full-text retrieval of an electronic official document. The method comprises the following steps: acquiring basic information of the electronic official document; determining whether the electronic official document is a picture type electronic official document or not according to whether a picture exists in the electronic official document or not; constructing an OCR (Optical Character Recognition) object for the electronic official document of a picture type to scan and recognize the official document; carrying out gray scale and amplification processing on the scanned picture; inputting the processed picture into a trained picture text recognition model to obtain a text recognition result of the picture; and realizing full-text retrieval of the electronic official document based on the text recognition result. By means of the OCR technology and in combination with the picture text recognition model, text recognition of the picture type electronic official document can be achieved, and the technical problem that the picture type electronic official document cannot be recognized and analyzed is effectively solved; and in addition, a non-picture type electronic official document recognition technology at present is added, so that full-text retrieval of different electronic official document types is realized, and the retrieval efficiency is comprehensively improved.

Description

technical field [0001] The invention relates to the technical field of full-text retrieval, in particular to a method and system for realizing full-text retrieval of electronic documents. Background technique [0002] Currently, the full-text retrieval of documents is often implemented through text analysis, and then the content is indexed to achieve full-text retrieval of documents. Most of the methods for document parsing in enterprise applications are based on Apache POI; the storage address of the document is obtained through the database connection, the content of the document is identified through the POI, and the document is tagged and indexed. [0003] At present, only the text type of the document can be identified through POI, and full-text retrieval cannot be realized when there is an image type. [0004] Electronic official documents refer to the electronic data of official documents in a standardized format formed by various regions and departments through the ...

AI Technical Summary

Problems solved by technology

[0005] In the actual work process, the content of electronic documents is often not limited to text-type PDFs. Some official documents scanned by printers or other scanning devices will be saved in PDF as pictures; Apache POI can only parse text-type PDFs. , when the picture type PDF appears in the electronic document, the parsing cannot be completed to realize the full-text search

[0006] In the existing scenario, more than 80% of the types of electronic documents have pictures. According to the existing methods, only the key information such as the title and time of the documents can be retrieved and the full-text search of 20% of the documents can be realized; It can be realized by manual input, which is bound to consume a lot of time, manpower, material and financial resources, and is inefficient

Images