This is an HTML version of an attachment to the Freedom of Information request 'EASE and Artificial Intelligence Study'.




Ref. Ares(2020)7898951 - 23/12/2020

 
PREAMBLE 
EASE (Electronic Access to European Commission Documents) is a project to ensure 
that the Commission is equipped with modern, electronic and integrated IT tools 
allowing the submission and handling of the requests for public access to documents. 
EASE supports the goal to bring the EU decision-making process closer to its citizens. 
The future system will : 
x  Improve the workflows linked to the submission, processing and preparation of 
replies to requests for access to European Commission documents  
x  Rationalise internal workflows and enhance consistency between replies. 
x  will cover the public interface for communicating with applicants, the internal 
Workflows within the European Commission, and the consultations of other EU 
Institutions, Member States and third parties, from the first request of the 
applicant to the final decision of the Commission.  
Regdoc is the Commission’s Register of Documents; it is a public application that allows 
citizens to look for Commission documents. Together with the document, it displays 
metadata of the document. In case the document is not publicly available, a web-based 
form can be filled-in by the citizen. The current application will be rewritten in the 
coming years but this project is currebtly in a very early stage so no specific 
documentation exists yet. 
The “Study on the use of artificial intelligence techniques for the electronic access to 
Commission documents” should examine the different possibilities to start using artificial 
intelligence (AI) and data analytics and visualisation techniques in information systems 
such as EASE and RegDoc. The introduction of these techniques is meant to enhance the 
EASE project in the first place, but SG would like to implement these tools also for the 
new RegDoc project as well as for the other registers. An example of an AI technique is 
Doris, a text-mining tool, used in the information system Better Regulation Portal. 
More information on the two projects can be found in the annexed documents: 
x  EASE - Project Charter 
x  Study on the Secretariat-General Registers - FinalReport 
1. SCOPE AND PURPOSE 
SG and DG DIGIT have agreed to collaborate in the context described in the preamble. 
The purpose of this Memorandum of Understanding is to set out the conditions for this 
collaboration. The MoU defines the high level scope of the work to be done by DG 
DIGIT and the required associated resources. 
2. SERVICES TO BE PROVIDED 
In order to support the introduction of AI techniques in the domain of electronic access to 
Commission documents, SG and DIGIT agree that the following activities need to be 
carried out: 
x  Identification and analysis of a set use cases where AI techniques can be applied, 
based on the functionalities foreseen for the two systems (EASE and RegDoc). 
DIGIT-00882-00 


 
x  Delivery of a prototype, implementing one of the use cases, which aims at 
validating the feasibility of the approach. 
For this purpose, DG DIGIT will deliver to SG: 
x  Four high-level use cases where AI techniques can be applied in the context of 
the EASE and RegDoc projects, based on the description of the foreseen 
functionalities described in the documents in annex. Each high-level use case will 
include at least:  
o  A description of how AI can support the specific functionality in terms of 
main benefits that the technology can bring to the users. 
o  A description of the specific AI technique that will be used to implement 
or support the functionality; 
o  A high-level estimate of the effort required to: develop (or apply) the 
relevant technique, train the related algorithms (if applicable) and deploy 
the solution in production. The effort should also take into consideration 
the resources required from SG, which will support the development 
process by providing business knowledge or validating the outcome of the 
AI technique. 
o  The list of use cases shall be provided in English, electronically, in 
Microsoft Office 2016 tools.   
x  A prototype implementing one of the four identified use cases. The specific use 
case will be selected through a joint workshop between SG and DIGIT. The 
purpose of this workshop would be to perform an analytical feasibility assessment 
on each of the four identified use cases and select the most suitable one. 
Specifically, the delivered prototype will contain: 
o  The implementation of the AI technique selected for the use case, 
including, if applicable (such as in case of machine learning techniques), 
an algorithm trained on the data sets provided by SG; this prototype 
should be a working tool; 
o  A user interace, detailed enough for the user to validate the outcome of the 
application of the AI technique; 
o  The technical infrastructure supporting the execution of the AI technology 
(e.g. cloud environment); 
o  The necessary documentation describing the technical aspects of the PoC.  
o  The proof of concept shall be delivered in source code that can be used 
within the scope of the EASE project.     The documentation on the PoC 
shall be delivered in English, electronically, in Microsoft Office 2016 
tools.  
 
SG will support the delivery of the above by providing to DIGT:  
x  The data sets (e.g. documents, spreadsheet, structured data stored in databases), to 
which the artificial intelligence techniques have to be applied; 
x  The needed business knowledge required by DIGIT to better understand the 
business context, the expected results and content of the provided data sets; 
x  Access to the users (in SG or elsewhere) who will need to test and validate the 
outcome of the prototype.  
DIGIT-00882-00 


 
DG DIGIT will ensure close collaboration with SG during the design and the 
implementation of the system. Input from SG is considered as essential for ensuring the 
right level of expertise in this business domain and in carrying out intermediary reviews 
of the artifacts developed, as well as performing the human quality control of the output 
of the prototype.  
 
 
DIGIT-00882-00