Novelis ranked 1st in the Microsoft CodeXGLUE international challenge

Novelis ranks 1st worldwide in the international CodeXGLUE challenge, organized by Microsoft, on Java code generation from natural language.


Last March, Novelis was already in the spotlight thanks to its participation in two international challenges: the Spider Challenge organized by Yale University and the WikiSQL Challenge organized by Cornell University. In both challenges, Novelis took second and seventh place, alongside the biggest leaders in AI and RPA.


The Novelis R&D Lab team won 1st place in the international CodeXGLUE challenge on Java code generation from natural language:


The CodeXGLUE challenge – General Language Understanding Evaluation benchmark for CODE – organized by Microsoft, brings together large companies such as IBM or Microsoft and international universities such as Case Western Reserve University, UCLA/Columbia University, or INESC-ID/Carnegie Mellon University.

With CodeXGLUE, Microsoft seeks to “support the development of models that can be applied to various code intelligence problems, with the goal of increasing the productivity of software developers”. Microsoft wants to encourage researchers to take part in current challenges to further advance code intelligence.


According to Evans Data Corporation, there will be 23.9 million professional developers in 2019 worldwide, and the number is expected to reach 28.7 million by 2024. “With the growing population of developers, code intelligence, which aims to leverage AI to help software developers improve the productivity of the development process, is growing increasingly important in both communities of software engineering and artificial intelligence.”


The Challenge includes 14 datasets for 10 diverse programming language tasks covering:

  • Code-Code (redundancy/clone detection, code error detection, gap code (or text) completion, code autocompletion, code correction and code-to-code translation),
  • Text-Code (natural language code search, Text-Code generation),
  • Code-Text (code summary),
  • Text-Text (documentation translation).

Novelis has participated in the Text-Code task, which consists in automatically generating Java source code from natural language.

Currently, the Text-Code task leaderboard has 9 participants. Once we had built a model that met our expectations, we submitted our test results for official evaluation by the Microsoft community based on 3 criteria:

  • The Exact Matching (EM),
  • The BLUE Score,
  • CodeBLEU.

The Microsoft community then updated the ranking on the leaderboard that you can find below.



Abdelkader Rhouati, PhD Researcher & Datascientist at Novelis explains:

We have been working for more than two years on the problem of generating code in programming language from a need described in natural language. Our work adopts several approaches, designed and implemented by the Novelis R&D Lab team and has led to several results in the task of generating business code in Python and Java. Until now, we did not have a benchmark or a challenge that would allow us to evaluate our results in an objective way. Microsoft’s CodeXGLUE challenge allows us to gain this credibility because we could officially evaluate our results. Moreover, the 1st place obtained in the code generation task proves that we are on the right track. Note that the results published in this challenge are not very high because on the one hand the code generation task is very complex and on the other hand the proposed models are not yet mature enough.”


Novelis has placed innovation and R&D at the heart of its development strategy

Since its creation, Novelis has chosen to invest massively (30% of its turnover) in research and development.

For El Hassane Ettifouri, CIO and Director of the Novelis R&D Lab, this is no small matter:

Today very few companies are willing to invest ¼ of their turnover in research. It is this risk-taking that differentiates Novelis from other companies. We want to have a foot in the future and participate in the construction of this future by investing in research on technologies. Innovation is an integral part of Novelis’ DNA.

Moreover, our research work is concrete and has a real impact on our customers – who reap all the benefits of our technologies for the automation of their processes – but also for our employees who evolve in an innovative working environment.”

Read More
hicham December 17, 2021 0 Comments

RPA : From Process Robotization to Intelligent Automation

Posted on October 1, 2019 by :

Each year, the A CAPELLA Consulting teams meet for 2 days of seminars to share moments of conviviality, workshops, and acculturation sessions. This year, the consultants had the opportunity to accumulate and deepen their mastery of RPA (Robotic Process Automation) during a workshop conducted with their partner Novelis, represented by Walid Dahhane, the company’s CTO.

In this article, we invite you to take a look at the concepts backed by the acronym RPA and its use cases.

What is RPA?

RPA (Robotic Process Automation) is a subject of innovation and interest but whose background has origins in the birth of IT. It is therefore important to understand that the very topic of automation is not new.

Often the difficulty we have in defining RPA is to show how it differs from the automation processes we already know. The first difficulty resides in the very word “robot” which is not implied as an electronic robot operating alone but rather as software for automating a well-defined process. The second is due to the psychological brake placed behind the word “robot” which induces for some that they are doomed to replace humans in the execution of tasks.

For Walid Dahhane “RPA is the acronym for Robotic Process Automation” which translates into French as “Robotisation des processes automatisés”. It is nothing but a software called robot, which performs data processing and manipulation of work, exchange between different computer systems, and executing internal and external communication.

The idea behind RPA is therefore to “robotize” business processes that would require the intervention of a human. The robot thus performs a succession of predefined tasks based on precise and identified information. It captures data in several systems then digitizes and performs actions usually performed by men.

What activities can be automated with RPA?

Once we better understand the concept of RPA itself, we wonder about the nature of the activities that we can automate.

The answer is : almost everything that happens on a computer and that a human can control via a mouse and keyboard (like for example : emulating clicks, entering data, copy / paste, navigate on several screens, connect to applications) and also perform purely computer processing (such as : calling web services, connecting to BDDs). We are in the context of RPA as long as technology can perform the action alone without the intervention of a human being and without a cognitive part in the process.

RPA therefore makes it possible to carry out transactions with any application in the same way as a human, and execute structured tasks.

If we take the example of Kiva robots from Amazon, we are in another field, that of electronic robotics, where the robots operate logistics tasks through learning. In the context of RPA, we always speak of process. The difference thus lies in the technology. In RPA, we are on a “computer” technology which does not present a cognitive function such as Artificial Intelligence in its execution process. For example, a chatbot can be part of an RPA process as a component, but will not be an executable process as such.

So we don’t find any intelligence behind RPA but we can have it in some of the components it will use (this is called cognitive RPA – answered later in this article is the different types of RPA). If we introduce intelligence into the RPA process, we kill the RPA and that reassures all the more to know that it will only reproduce what we ask it to do without taking initiatives and without autonomous decision making as we can have in Artificial Intelligence.

In most cases, RPA systems are not intended to live long without maintenance operations or adjustments because the processes they automate can constantly evolve. We thus naturally find a lot of RPA in transient phases of transformation of the IS.

RPA : Where are we now?

According to  Forrester Research,  “25% of tasks will be automated by 2020”.

In recent years, RPA has become an increasingly preferred option by companies to quickly digitize certain simple operations and rapidly respond to regulatory changes. RPA is indeed competitive compared to other and more costly solutions to implement such as IT integration projects or even resource outsourcing.


Companies that opt ​​for RPA in the digitalization of their processes thus find many advantages :

  • Quality and Precision : RPA effectively eliminates the risk of human error inherent in business processes and enables reliable and controlled results
  • Consistency and Quality of Service : RPA enables better control of processes and optimizes them
  • Financial Efficiency : RPA makes it possible to reduce the costs and time of execution of tasks, usually with an increased ROI
  • Audit, Security, and Compliance : RPA makes it possible to configure the processes of compliance with the necessary standards and rules (security, audit, GDPR, etc.) and provides traceability that may be required larer on during the audits
  • Productivity and Experience : by automating repetitive tasks, RPA allows employees to devote themselves to tasks that bring higher value to their work
  • Easy Integration : RPA technology can easily and quickly interconnect with existing systems while integrating advanced algorithms and machine learning tools into its components
  • Reliability and Availability : finally, RPA is reliable and available at all times, and the process does not suffer from any pause like an action carried out by a human being
What are some RPA used cases?

The choice of an RPA technology is particularly indicated for :

  • Generation of periodic reports, data entry, and data analysis
  • Performing repetitive tasks scheduled 50 to 60 times a day
  • Mass e-mail generation, archiving, and retrieval
  • Connection to ERP and executing other back-office transactions

To find out which process we can automate and if RPA technology is eligible, take a look at the following decision scheme :

Capture d’écran 2020-02-20 à 11.10.36

Other issues, however, must be taken into account, such as the number of robots to set up (we can quickly achieve a large number and render their maintenance a hassle), whether the existing system can manage them, and the impact on bandwidth and user interfaces.

The correct definition of the use case is essential to maximize ROI, the objective not being to replace teams but rather to improve their efficiency.

Example of used case presented by Novelis : A multinational food company with a presence on 4 continents wanted to reduce repetitive tasks that interfered with the main activities of the administrative and logistics department. The stakes were high since the error rate in the logistics department implied a total cost of USD 1.2 million in 2016. It has chosen RPA technology and today has 5 Robots that execute operations to both the financial department and the logistics department, without resorting to human intervention (unassisted processes) with more than 25,000 transactions carried out daily. The implementation of RPA thus resulted an annual saving of around USD 300k, but the most interesting achievement is the elimination of human errors quantified annually around USD 1.2 million. It also saved the equivalent of 75 FTEs in 6 months, allowing its employees to deal with other tasks with higher added value.

What are the different types of RPA?

We can distinguish 2 main types of RPA : Traditional RPA and Cognitive RPA.

Traditional RPA is itself divided into two subcategories :

  • Assisted RPA, which partially automates complex processes to obtain more operational efficiency but where we still need the intervention of a human being. This is the case, for example, of automated processes in a call center where RPA facilitates and speeds up the process by offering choices to the call center agent based on certain criteria. This type of RPA is often used for front-office processes
  • Unassisted RPA, on the other hand, fully automates business processes, thereby maximizing the expected ROI and without requiring human intervention. This is the example of a system that is fed through the  reception of  e-mails. This type of RPA is often used for simple back-office processes

Cognitive RPA will integrate intelligent components within the automated process in order to frame elements that involve human reasoning or decision-making, based on certain well-defined criteria. For example, RPA can be combined with OCR (Optical Character Recognition) * or Machine Learning technology. It must be understood that we are talking about adding cognitive / AI components but the skeleton that defines the business process cannot change and become intelligent. A robot cannot therefore decide on its own to add a new dimension to the process that has not been precisely defined before.

What automation tools are available in the market?

There are many RPA automation software out there, and the market is growing. In  Gartner’s 2019 Magic Quadrant  for Robotic Process Automation (RPA), we find at the top (see figure below) : 

  • Uipath, the leader, whose promise is to become accessible to all techniques without having knowledge. Their ambition for the future is that the software be present on all computers in the same way as Windows. But it is clear that even with packaged and very visual components, the solution is not yet appreciated by everyone. An open source version is offered strictly for personal use
  • NICE, conversely, requires at least 2 years of experience in object programming to be manipulated. It is a solution used for large automation projects
  • Blue prism
  • Automation Anywhere

Walid Dahhane’s presentation ended with a workshop where A CAPELLA Consulting teams were able to work on the development of use cases specific to the insurance and social protection sector.

In conclusion, Robotics Process Automation is particularly suitable for eliminating repetitive and tedious tasks, speeding up an operational process, and reducing costs. RPA will always be used to automate a rule-based business processes but not all of them are necessarily eligible for RPA, and right questions have to be asked in terms of issues and ROI before embarking on a project. Even if a POC can be achieved in a few weeks/months, the craze of CIOs for this type of solutions has led to the creation of several off-the-shelf robots that can now be purchased online to respond to a generic use case without replacing existing systems.


* The word OCR means optical character recognition or text recognition, a technology that converts different types of files such as scanned paper documents, PDF files or digital photos into editable and searchable files. (Source  Abby, one of the leading RPA solutions on the market for its OCR / Machine Learning alliance)

Read More
hicham February 17, 2020 0 Comments

RPA : Smart Automation for Operational Efficiency

Operational efficiency is a major issue that conditions the competitiveness of companies. Smart Automation is a relevant lever for optimizing business processes. RPA is the first step.

OCR and NLP are technological tools that allow us to push the field even further in terms of automation.

Technologies are evolving quickly, and so is the potential for automation. The expected benefits are many, and do not stop at a simple economy of means.

Securing high-stakes processes

Ex : Verification of the consistency of financial data by comparison of interface files.

Redirection of the workforce towards activities with higher added value

Ex : Time free for users after the removal of unpleasant tasks (manual invoicing)

Cost reduction

Ex : Alternative to Offshore Outsourcing of Certain Business Processes

Valuable creation

Ex : Automatic sending of emails / SMS for VIP customers of a hotel chain.

The automation of back-office processes is the easiest to understand. However, the potential of ‘Virtual Assistant Robots’ RAVs and Intelligent Automation considerably widens the scope of processes eligible for optimization and automation.


Intelligent Automation opens up unexplored perspectives. Yet, the positive impact of automation is not new : avoiding tedious manual tasks and improving the accuracy of the data being handled. On the other hand, we were faced with a glass ceiling : 80% of company data is unstructured. This is why the OCR and NLP are inevitable levers to overcome this constraint.


    Read More
    hicham February 3, 2020 0 Comments

    Novelis : R&D at the Service of an Innovative Digital Transformation.

    Publié le 18 septembre 2019 

    Meet  Mehdi NAFE , CEO & Co-Founder of Novelis, the startup that supports companies for a digital transformation at the cutting edge of new technologies. The secret ? R&D, R&I (Research and Innovation) and above all, team spirit!

    Mehdi, tell us about yourself and your journey ...

    I had the “classic” path in engineering school (ENSEIRB-MATMECA, in Bordeaux). My interests are mainly in the fields of telecommunications and software engineering; in which I specialize. I worked with companies like CGI GFI , and SQLI in delivery management and digital transformation consultancy. I learned a lot, discovered various ecosystems, and realized that human qualities are as important as technical or operational qualities, they actually make the difference.

    Why entrepreneurship? What was your click?

    I have a lot of admiration and respect for those who take risks, those who are daring, that has always inspired me. So I’ve always had entrepreneurship in mind. I like to create, innovate, and bring added value. What pushed me to move on to the concrete was the technological acceleration that I noticed around me. Some companies that I have encountered during my career path did not anticipate the deep technological transformation that we are experiencing. I wanted to be part of the development of a structure capable of doing so, and of supporting our clients in this new adventure.

    Tell us more about Novelis.

    We combine two essential aspects, more and more inseparable : Research and Development, as well as the world of Services. The idea through  Novelis is to support companies in their digital transformation and allow them to work more efficiently. This is why we invest heavily in R&D and R&I, to participate in the development of technological tools, master them to create the most innovative capabilities available to our customers, and provide them with the most efficient solutions. We have our own R&D Laboratory and work a lot on AI-oriented subjects, more specifically on NLP * (Natural Language Processing).

    Other themes that also interest us : connected objects, virtual reality, Blockchain, etc. All the subjects that are at the heart of tomorrow’s challenges for businesses, and in which we wish to invest.

    * NLP is a multidisciplinary field which involves : linguistics, IT and AI. The objective is to create tools for processing natural language, for different applications. Not to be confused with computer linguistics, the aim of which is to understand languages ​​using computer tools.

    And why digital?

    It was obvious to me, I have been involved in it my whole career, and it is through this area that we can bring added value. In particular thanks to R&D, which is a necessity in order to be relevant, and effectively support our customers on very specific issues. We accumulate technological and scientific knowledge over time, and that’s what makes us more competent and different.

    What has been the biggest success for Novelis so far?

    As a small company, we have to prove ourselves and double our efforts to win markets and work with our clientel over time. Winning their trust, and convincing them with our skills, is our greatest pride. It remains a constant challenge and each success is a collective victory.

    Any particular difficulties?

    Managing temporality is a major difficulty in the life of the company. You have to be patient, sow seeds, and wait for the results to come in later. We must have a well-established plan, be able to repeat efforts, persevere, and have a spirit of innovation that allows us to find solutions to move forward. During which teamwork is a real strength, because it helps maintain this consistency.

    What is the secret of team cohesion?

    Team spirit is a particularly important point. Teamwork must be at the heart of the business. I have observed in my career too much damage from individualism. You have to think about your teammates, help each other and show solidarity in all circumstances. The best way to achieve individual goals is to achieve them through collective goals.

    The next steps for the startup?

    Novelis is 18 months old, we already have around fifty employees. It’s a fairly significant progression. Our objective is to strengthen ourselves, by doubling the number of employees at the end of 2020. We will do this by consolidating our presence with our major customers and by seeking new markets within our industry. We also want to provide new offers, and be able to support our customers on new themes and issues. We want to make Novelis a benchmark in the field of innovative technologies, and through our company help those who wish to implement their ideas and make it them happen for them by providing the necessary resources.

    What is the worst advice you've ever been given?

    “It’s too dangerous, it won’t work, you shouldn’t get into it.” It’s the worst when people push you to stay in your comfort zone. If I could go back in time and give myself some advice, it would be to not wait to get started and just go for it.

    How would you rate your experience in the X-TECH Booster program at the X Innovation Center?

    In short, excellence. We share the same values ​​conveyed by Ecole Polytechnique and their innovation centre : persistence, because it is essential to improve on a daily basis, and humility. X-TECH Booster is also a means of creating relations with other program members with X laboratories. We live this with a lot of enthusiasm and a lot of pride.

    One last word ?

    "If you think the adventure is dangerous, try routine ... It is deadly! "

    Read More
    hicham February 1, 2020 0 Comments

    Interview #5 : “At The Heart of Automation”, A Focus on Novelis Consulting

    For this 5th interview, I had the pleasure to have a chat with the founder of Novelis Consulting. Innovation being the heart of its strategy, the subject of Automation does not leave Novelis Consulting indifferent, but let Mehdi Nafe express it in its own words!

    Read More
    hicham February 1, 2020 0 Comments