DeepREXT

Intelligent digital engineering solution that automatically extracts, classifies and analyses requirements coming from raw documents and existing requirements databases

CONTEXT & STAKES

A key stage in the life cycle of industrial projects, particularly in the case of nuclear power plants, is to take into account the applicable normative and regulatory reference framework via project and system requirements. In other words, the efficient translation of existing generic documentation into a project-specific requirements framework.

This stage is complex, as it requires the involvement of several professions and experts, time-consuming, as the applicable documents are often dense and diverse, and above all crucial, as failure to take into account the expectations of stakeholders can result in very significant additional costs and delays.

the solution

DeepREXT enables the automation of requirements management via algorithms based on the following elements:

  • Creation of a centralised database for the documents to be taken into account
  • Automatic recognition - after calibration - of the characteristic formulation patterns of a requirement-carrying text element with the help of AI models
  • Automatic classification, rationalisation and comparison of requirements based on deduced elements / project (equipment, activity, etc.)
  • Search and export tools to interface with an existing digital ecosystem

Compared to its competitors, DeepREXT requires almost no human intervention to extract, classify and analyse requirements. DeepREXT's algorithms use existing data to derive thousands of rules.

Discover the solution in video

CHARACTERISTICS OF THE SOLUTION

DeepREXT is an innovative solution that allows you to capitalise on expertise by automating repetitive and low value-add tasks (extraction of requirements from a regulatory corpus, allocation to technical scopes, etc.).

DeepREXT transforms document formats into data using Deep Learning (DL) technologies that analyse the structure of document pages. Character recognition (OCR) and natural language processing (NLP) enable the recognition of the characteristic patterns of a requirement, their classification and rationalisation. As the tool is used, the algorithms adapts to the type of requirement being processed and improves over time. The detected elements are integrated into a dedicated digital environment, with search engine capabilities and the possibility of export and collaboration.

Discover the whole DeepREXT solution and our case study on the RTI² project

objectives

EXTRACT automatically a set of requirements structuring the project
INTERFACE efficiently a requirements management tool in an integrated ecosystem
ENABLE the transition from a time-consuming implementation scheme to an expertise-based verification scheme
OPTIMISE engineering requirements data efficiently through classification and rationalisation
CENTRALISE and organise access to a project requirements database with a digital platform
GUARANTEE the adequate consideration of an applicable context

PROJECT BENEFITS

Reduction of the time needed to process applicable documentation
Centralisation of technical data allowing quick access to information
Better control of projects in terms of deadlines, risks and costs
Better traceability of engineering choices
Optimisation of impact analysis

A QUESTION, A PROJECT?

Contact us