BTC Document Classification

Fast and efficient classification of documents.

We use two technologies to ensure full data security.

Do you have questions? Contact us!

BTC Document Classification

Complete data security

Data security

Document classification is a guarantee of full data security in the organization. The tool recognizes, analyzes and classifies both text and image files with great efficiency. Indicates documents that should be secured due to their sensitive content.

Effectiveness

BTC Document Classification uses two methods based on artificial intelligence – machine learning and deep learning. The solution analyzes the file for both image and text content. The use of two methods makes it possible to quickly and efficiently classify data in an organization and identify documents containing confidential data.

Analysis of confidential documents

Document classification is a key solution used to mitigate data privacy breaches, as well as in identifying and securing confidential documents within a company. It allows the use of breakthrough technologies, it finds application in professional systems for IT management and security.

Comprehensiveness of the solution

Documents are analyzed for text, layout and special graphics such as stamps and signatures. By using machine learning models, categories are obtained, related to the type of document in question. Deep learning technology analyzes images for layout and helps recognize signatures that are applied to a document, among other things. Both plain text documents and scans of documents in graphic form are analyzed.

Secure the data in your organization.

IT Security Guarantee.

BTC Document Classification

Features

Use of innovative technologies

The solution uses two methods based on artificial intelligence – machine learning and deep learning – to ensure high performance.

A variety of file formats

The tool effectively recognizes both text files and image files in many formats, viz: .pdf, .docx, .xlsx, .jpg, .png, .jpeg. More than 100 extensions are recognized.

Language recognition

BTC Document Classification effectively recognizes text in Polish and English.

Operation 24h/7/365 days

Document classification is possible at any time, 24 hours a day, 365 days a year.

2 main categories, 50+ subcategories

The analyzed document receives one of two categories, Document or Other. While the algorithm recognizes that a given file contains document-typical content, a second classification stage is triggered. At this stage, you will get one of more than 50 subcategories to which the document belongs, such as applications, complaints, regulations, statements, etc.

5 special graphics

The deep learning methodology allows for the recognition of graphics closely associated with documents, such as signatures, stamps, signatures along with stamps, fingerprints, as well as QR codes.

Quality

The BTC Document Classification solution analyzes and classifies both text files and image files, thus ensuring high efficiency and quality performance.

Recognition of documents with special data

The use of artificial intelligence methods allows a computer network administrator to recognize documents with certain characteristics of interest, such as documents containing PESELs, TINs, but also with categories, invoices, applications, etc. Which allows you to block the sending of these types of documents outside organizations precisely because of certain characteristics and/or categories.

BTC Document Classification

How does it work?

STEP 1

Recognizing the type of document to go through the analysis.

STEP 2

Three types of files are analyzed:
– A file containing only text,
– A file containing only images,
– A file that combines both text and images together.

Depending on what data is in the document, the method of analyzing the file is automatically selected.

STEP 3

The analysis of the text with the indication of the category is done by machine learning models, which first return information on whether it is a document and then what type of document it is.

The analysis of the images is done with the help of deep learning, which returns category information based on the visual garment. In addition, text extraction from an image is triggered, with the text analyzed by machine learning models, which determines categories based on the text.

STEP 4

Determine the number of data protection-related features in a given document. The use of regular expressions, artificial intelligence, and communication with APIs (https://api.stat.gov.pl/) makes it possible to effectively detect features such as ID card numbers and passport numbers.

BTC Document Classification allows you to monitor sensitive documents to protect confidential data.

BTC Project EU