In summary

"Our customers’ satisfaction through innovation is a priority and an essential component of the group strategy". We develop many optimisation tools and thus simplify customers’ access to technologies.

Testimonies

François Chahuneau

Director of Technology

Through thorough and efficient monitoring, we use the best technologies in the market in order to constantly optimise digitisation procedures and therefore fight the ravages of time which damage media.

Testimonies

National Geographic Institute

Public Institution

The IGN (National Geographic Institute) is required by the State to produce geographic data for national territory and make it available to the public (in the form of maps, electronic files or via its website Géoportail).

In order to create a geographic database linking the cadastral data of the DGI (General Directorate for Taxes) with the Institute's geographical data, the IGN has entrusted Numen with the complex task of producing digital images. This involves georeferencing, transformation, the processing of connections and the assembly of scanned cadastral maps.                                                                    

The vectorial input of administrative boundaries required the processing of 200,000 cadastral maps, representing 10 million video encoded reference points.

Pacte Project

How to substantially improve the quality of the digitised text ?

The context of future investments acutely raises this question by asking to all holders of heritage documents, in a return on investment perspective, to imagine a model of economic use applied to digitised content.
The enhancement of this content is directly related to their high quality and their faithfulness to the originals.

R&D project objective:

The PACTE (text capture improvement project) project is designed to improve the performance of text capture processes (OCR, handwriting recognition, manual entry, writing) by using language processing, and more specifically statistical and hybrid linguistic methods.
It concerns the application field of written heritage digitisation projects, in a multilingual context.

A partnership of complementary players:

The PACTE consortium includes three companies (Numen, A2IA, Isako) and two laboratories (Alpage, LIUM).
It thus includes all business issues, lexical resources, scientific expertise and technology required to deal with the subject in depth.

SmartGed

The SmartGED project launch came shortly after the release of version 1.1 of the CMIS sector standard.
It emerged out of an industrial need that could not be satisfied by existing products on the market, yet crowded, of EDM and content management systems.

Indeed, the management of documents from publishing and digitisation chains, whose content is fixed but whose metadata that are associated with it are scalable, requires customer access to create or modify them, sometimes at several stages of the production workflow.
The associated EDM must also have high performance mass loading in order to support the throughput of industrial tools used to power it.

It is therefore the first EDM natively compatible with CMIS 1.1.

The implementation of SmartGED is based on the PostgreSQL object-relational DBMS whose advanced features and recent NoSQL extensions make it a sufficient support for an effective implementation, using an innovative design based on the use of objects of key-value type (Hstore).

Compliance with a standard recognised by the major players in the Electronic Document Management industry, associated with wise technological choices now allows us to assert that SmartGED is a fully developed product able to meet the needs of an industrial documentary production system:

  • high-performance mass loading capacities
  • service layer offering the opportunity to create industrial batches,
  • flexibility and scalability of the content-related metadata.

The operating principles of a production EDM require to specialise the CMIS standard to integrate additional features relating to this segment specifics.
The integration of specific features essential to the world of production to meet our needs:

  • Reporting-oriented data structures,
  • Combination of task lists with Folders, workflow “Basket”,
  • Connection to an external workflow engine.

Key Topics

Numen has been treating sensitive data, complex and multilingual textual content for more than 40 years now. Today, Numen is positioned as a leading player in the field of BPO and dematerialisation services

Numen is proud to rely on a solid team developing and industrialising state of the art processes of research in the following fields of application:

  • document images processing
  • OCR and Automatic Document Reading-Automatic Document Recognition technologies
  • statistical learning technologies, especially applied to the contextual spelling correction *
  • text-mining and automatic indexing of documents
  • data-mining and analysis of business data flow
  • dynamic composition of documents
  • automatic routing of documents by intelligent analysis of content (pattern matching)
  • technologies related to the digitisation security (identity management, electronic seal and signature, secure print)
  • computer distributed processing (cloud, big data, workflow engines)

7 % of turnover in R&D

Seeking to optimise the quality and costs of our customers’ services while willing to increase our production chains efficiency, Numen is annually investing 7 % of its turnover in research programme.

A team

R&D teams are part of the Technology Department, whose mission is to organise the technological skills capitalisation through the diversity of projects for customers and the multiplicity of areas of business application.
With own funds as well as partial financing of the State, different R&D projects are internally conducted. Others are conducted in collaboration with partner companies and university laboratories.

Numen is proud of counting on a solid team whose progress increases as it faces technological challenges.

Through its engineering and R&D department, Numen automates its customers’ processes, solves productivity issues through its innovations in capture but also in data processing (text-mining, statistical and automatic indexing), and develops complex information systems.
Numen regularly incorporates young engineers, graduates of “grandes écoles” and higher education.

Through its engineering and R&D department, Numen automates its customers’ processes, solves the productivity issues through its innovations in capture but also in data processing (text-mining, statistical and automatic indexing), and also develops complex information systems.
Numen regularly incorporates young engineers, graduates of “grandes écoles” and higher education.

Signature extraction solution (s) and authentication

Numen Europe has developed through its R & D activities an automated solution for the detection and authentication of handwritten signatures on any type of documents. The relevant documents are contracts, orders, money transfer orders, payment orders, etc. The goal is to verify the authenticity of signatures in order to participate in secure transactions.

The state of the art concerning the detection signatures on documents is quite poor, we have developed a solution dedicated to these same materials of poor quality.
After localization, the next step is to authenticate the signature against a reference set of signatures, in order to check if the author of the signature is the authorized signatory.

The signatures on these documents are obviously variable in size, the thickness of the pen used, the number of connected components that make them up. Further documentation may have multiple signatures.
The quality of the original document are very important factors in detection.
Indeed, a chopped signature, noisy, or contiguous to text or lines is more difficult to detect in its entirety (Figure 1)
With the algorithms developed by Numen Europe, we get the output of the software detected by a thumbnail signature.

With conclusive rates : Figure 2

FundOperations, transactional data extraction solution for Investment Fund Industry

FundOperations, transactional data extraction solution for Investment Fund Industry.

After numerous contacts with the industry from the Luxembourg Fund, we discussed a recurring subject for documentary treatment regarding transaction orders received by fax with relatively similar encountered problems, namely:

  • Quality issues performed extractions;
  • No automation, completely manual rekeying of important risk causing errors;
  • Operational and financial risks due to high error rates generated during manual fax transactions to conversion;
  • Too high cost for treating these incoming orders by fax;
  • An STP automation via SWIFT messaging that has reached its limits

Most of Transfer Agents have also reported having tried in recent years to implement through their internal IT system, a solution to this problem by using OCR technology but without any convincing success.
It is in this complex context that Numen Europe through its R & D activities have contributed.

In addition, the aspects linked to the reduction of operational risks through increased assisted automation , the main risks relates to errors caused by the human factor.
Thus, each of these types of errors has been the object of a process definition to reduce significantly or eliminate these risks which provides our customers sometimes substantial reductions in direct costs (financial damage).

X