Research Topics

  • Requirements engineering

    The CRI has a long research history in this area in which it acquired a national and international recognition. For a long time, the tradition has been to define method techniques and tools that guide the discovery, specification, negotiation and validation of requirements that various stakeholders have regarding a software-intensive system. So far, the CRI has particularly explored the first theme, that of requirements elicitation. For the future, we have decided to deal with the following four topics: expressing requirements variability in a product family, matching requirements collections to the functionalities of COTS such as ERPs, strategic alignment, and the arbitration of large collections of requirements.

    • Mainly for economic reasons, the software engineering community has turned to developing software product lines and families. These are built in such a way that when implemented, they can be adapted by selecting and combining some of the variants they include. Variability is a key concept in software product lines that should be taken into account at the level of functional requirements as well as non-functional requirements. Modeling requirements variability and guiding the process of selecting the most appropriate alternatives to a given system is the first topic of the RE thematic pole. The challenge of existing approaches is to avoid the unbalance that systematically appears due to fact that we are directed either by organizational requirements or by the system's functionalities. The CRI has made a significant contribution to this subject by a balanced approach of alternatives selection. Our approach is characterized by an ambivalent modeling of variability that allows to specify both organizational requirements and the systems' functionalities at the same time. This approach is currently empirically validated in a project in which we are re-designing the product line of the STAGO company in the biomedical field.

     

    • Many information systems installed today are no longer developed on demand to satisfy specific requirements, but by adaptation of a pre-built system such as an ERP. In fact, development 'from scratch' has become an exception. Requirements engineering must be adapted to this new type of engineering. Our research hypothesis is that the problems arises in terms of looking for a mapping (‘matching’) between organization requirements and requirements that the system is already able to satisfy. In fact, problems arise in a similar manner in a variety of situations such as the evolution of a system, ERP configuration, the assembly of components, or the configuration of a product line. A distinctive feature of the matching approach developed at the CRI is that it assumes that the problem can be reduced either to an evaluation of similarities, or to an identification of gaps. For that purpose, we developed a generic typology of gaps and a collection of similarity metrics that can be instantiated for each specific matchings situation. The matching approach was validated in two industrial projects, namely ELAN (configuration of the PeopleSoft ERP) at SNCF (the French railway company), and by adapting an existing system at DIAC (a banking subsidiary of Renault).

     

    • The distinction between WHAT and WHY, which is at the centre of all the requirements engineering approaches, has brought the community to look into the alignment of information systems with enterprise business strategy. Most of the approaches address the alignment problem as a process of application of successive transformations that convert the expression of business strategy into information system models. Alignment is hence obtained implicitly through the application of methodological transformation rules. Our strategy is to study alignment as a concept, and therefore to model it. An alignment meta-model is being developed at CRI. It allows specifying couplings of various complexities levels, and at different levels of abstraction and detail. In addition, we think that it is not possible to deal with the alignment problem without measurement. We have studied a set of metrics and, to guarantee their generacity we defined it related to two ontology concepts of enterprise business and information system. The meta-model has is experimented in the context of a joint research project with the BNP-Paribas Bank to model and measure the strategic alignment of the information system of the bank.
    • Divergent positions introduced by the various stakeholders, and difficulties arising from constraints such as cost and time are traditionally treated in RE  using prioritization techniques such as AHP, cumulative voting, or the 100 dollars technique. As they deal with each requirement individually, these techniques do not allow to handle large quantities of requirements. However in real world situations, we are often confronted with a combinatory explosion problem that result from considering all possible requirements combinations. As an example drawn from industry, the number of requirements at Renault amounts to tens of thousands. It is not possible to compare them all, or even to consider them individually in a consistent way. The CRI proposes an innovative approach ofmulti-scaled arbitration that organizes requirements priorisation by groups of requirements. This approach allows to carry out only local evaluations with reduced numbers of alternatives.

     

    The Requirements Engineering (RE) research group is headed by Dr. Camille Salinesi and consists of 7 members. The publications of the research group is an example of the excellence of our laboratory both on the national and international levels. This research group regularly publishes in renowned requirements engineering conferences such as RE (3 articles published since 2004), REFSQ (3 articles), the journal REJ (2 articles), and CAISE (5 articles). CRI has also recently participated to the publication of two books. In addition to international publications, the research group also publishes regularly in national journals (4 articles) and national conferences and international workshop with program committees (more than 30 articles). Our presence is also assured through the participation to several program committees such as those of RE, CAiSE, and REFSQ. We were also co-organizers of these international conferences during the past three years. CRI had the honour of organizing the IEEE RE conference in 2005 at Sorbonne.

  • Service engineering

    The research undertaken at the CRI aims at bridging the gap between the technical view and the business view on services so as to move from web services to services that support organisations. This problem is according to Papazoglou [1], one of the major challenges in service engineering.

    • The ISOA (intention service oriented architecture) approach developed at CRI is based on a semantic and executable service meta-model defined in the intentional perspective. The originality of this meta-model, called MiS, has been recognized by the Service Oriented Computing (SOC) community [1], [2], [3]. MiS focuses on the intention (“why”) of the delivered services rather than on the functionality they provide (“what”) . This usage-oriented vision of a service allows decreasing the distance between the customer and the supplier perspectives. It also allows introducing the variability in the specification phase of an intentional service. It provides a customizable enactment by permitting the dynamic choice of the variant most adapted to the service usage conditions. This proposition differs from semantic annotations of web services as defined by the “semantic web services” community, by defining the concept of intentional service added to an intentional guide of services enactment. Intentional services can be described, searched out, enacted, composed and distributed over the net in the same way as web services.
    • The executability of the MIS semantic meta-model is an strong axe of our researches because it allows to provide tools for dynamic orchestration of goal oriented services (in other words by the users requirements). This work merges the ones on semantic web services with the works of the SOC community and the requirement engineering applied to service composition. The enactment software platform uses semantic descriptors of services for dynamically composed software services that make the intentional services runnable.
    • The transition from the semantic level to the operational level is an important research question for this domain, as it will permit a convergence between the "semantic web services" and the SOC community. The transformation of intentional services into executable services is necessary in our approach in order to sustain the executability of technical services with tools (more particularly their composition, adaptation and search) in an intentional way. The achievement of this transformation is based on the MDE approach (Model Driven Engineering). We propose a model independent from the desired platform (similar to a PIM), some specific code generators for the desired platform, and some adapters that allow to call a technical service from an intentional orchestration tool. 
    • The last sub-theme in the SE thematic pole is related to the qualitative requirements generated by customers. Our position has two salient features: (a) web-services shared with the public through out framework are of a functional nature and expressed in generic terms in order to increase their reuse potential. To book a sit in a plane is an example of this kind of service; (b) the requirements of a customer at time ‘t’ show specific characteristics related to the quality that the customer associate with the fulfillment of the service. To make a suitable reservation is a manner to express this kind of requirement. In order to take into account the extreme dependence of a functional composition of services with its specific context of usage, we think that it is necessary to dynamically adapt the intentional services composition to the specific quality requirements expressed by the customer who uses it. Our approach for solving this problem is based (i) on a definition of functional composition of services as a set of hard goals; (ii) on the expression of quality requirements as soft goals, (iii) on an implementation of a context-aware platform combines web services (hard goals) and  aspects (soft goals).

    Research works undertaken in the IS pole are based on various collaborations in France: with the LSR-IMAG team (Grenoble) to deal with the transformation of intentional services into runnable services, and with the LORIA team (Nancy) for developing an approach of distributive and cooperative Information System development; with the LIFL-NOCE team (Lille) within the SAME2 project, which purpose is to translate intentional services of EAIH systems into interaction services. The CRI have also other collaboration with the RIADI laboratory (ENSI) based in Tunis. The CRI investigation in the competitive center “system@tic” allows us to increase our research network with well-known research teams on the area of software development, embedded systems engineering, real time, and mobile phone.

    The SE thematic pole of the CRI comprises 6 active members and is led by Carine Souveyet. Carine is one of the coordinators of the MADSI group (Advanced Methods on Information System Development) in conjunction with the GDR I3. She also participate as a member of program comities in various international conferences: CAiSE, INFORSID, RCIS, and some workshops like REFS (in conjunction with COMPSAC), RIGIM (in conjunction with ER) and MADSI (in conjunction with INFORSID) she is part of the founder group. The research activities of this center are regularly shown at major international conferences of the domain like the IEEE ICSOC conference (2 articles since 2004), the IEEE ISWS symposium (2 articles), the CAiSE conference (1 article), the SERP conference (2 articles) and the iiWAS conference (2 articles). Members of this pole participate actively in the SAME2 project and in two European project (IST ASCOMP and PPLM).

     

    [1] M.P. Papazoglou, P. Traverso, S. Dustdar, F. Leymann, Service-Oriented Computing Research Roadmap, Dagstuhl seminar on Service Oriented Computing, , March 2006.

    [2] B. Pernici, Service design and development, Dagstuhl seminar on Service Oriented Computing, March 2006.

    [3] L. Penserini, A. Perini, A. Susi, J. Mylopoulos. From Stakeholder needs to service requirements specifications. Technical Report, ITC-IRST, Automated Reasoning Systems, 2006.

  • Enterprise Systems Engineering

    nformation System Engineering has for a long time consisted in modelling the IS independently from its usage context. This situation is not adequate anymore to nowadays context in which all strategic, tactic and organisational aspects of an organization are linked with the IS. As a consequence, IS modeling should be at the core of organizational modeling. This is necessary to reason about the alignment of technical systems with the strategy and business processes, about the guarantee of compliance to norms and regulations, about risk and security management, about long term planification of business processes and other tasks that involve a matching between ICT and management.

     

    This axis is centred on enterprise modeling and on the reasonings needed to solve the aforementioned problems (that is why it is called ‘Enterprise Systems Engineering'). Researches in this thematic pole are organized around four sub-themes introduced below:

     

    • Our first research problem is to define a set of metamodels to represent enterprises. Our approach lies on a muli-perspective modeling that corresponds to the need to represent the multiple views of organizations. The CRI has developed the EKD-CMM method that offers a framework for modelling enterprise using a number of metamodels organized in three levels (objectives, process and IS) and that highlights relations between business and IS concepts. It has been used in several industrial projects in France and in foreign countries. Several publications [4] [5] attest those results. Multi-perspective modeling (intentional, organisational, decisional, functional, operational, informational, …) goes along with an organization of the various models at different abstraction levels, which eases their validation and the verification of the inter-models consistency. This proposition leads to (i) an entreprise architecture framework which consistency is ensured by interrelations clearly identified and justified between the different perspectives of representation; (ii) rules of systematic transformation between the different models at different levels of representation; (iii) verification rules to check inter-models consistency. The formalisation of these transformations is being developed according to the MDE paradigm. This approach ensures the Business/IS consistency ‘by construction’. The challenge is to define processes to maintain this consistency though time while facing changes.

     

    • Among the meta-models are process meta-models. Those available in the literature rely -almost systematically- on a functional perspective. The consequence is to freeze business processes as soon as they are modelled. However, business processes are submitted to numerous factors that influence their realizations. We propose to bring solutions to this problem by the means of two complementary mechanisms: variability and dynamicity

     

    • Recent works have highlighted the need for flexible, adapted and evolutive representations that we group through the term variability of processes. Our point of view is that existing process models are too detailed to represent the variability in a satisficing manner. Meta-models allowing more concision in the representation are needed. We chose a metamodel of the ‘actor/role’ type inspired from EKD-CMM and the RBAC (Role Based Access Control) standard. This metamodel relies on the concept of context and allow to express contextual business rules related to permissions, interdictions, obligations, recommandations and more largely to responsibilities.

     

    • Our research around dynamicity is recent. The concrete objective is to offer an obligation of result in tertiary organizations (for instance on delays guarantee) with a variable workload and invariable resources (in quantity). The research approach is to represent the dynamic weaving of work in the business process modeling and execution. Some has also to determine the paradigme of execution that would allow this dynamic management "around the actor". Finally, we have for objective to define an execution tool that knows dealing with dynamicity of ressources allocation.

     

     

    The ESE thematic pole is composed of six members. It is supervised by Selmin Nurcan who is the founder and co-animator of the "Modélisation et interopérabilité des entreprises et des systèmes d'information" ModESI working group of the "Ingénierie par et pour les modèles dans les systèmes d'information" thema of the GDR I3 and co-animator of the "Entreprise Communicante et Interopérabilité" ECI GT of GDR I3 and MACS of the CNRS. In this context, she organized and co-animated the second (2005) and fourth (2007) "Ingénierie et Gestion des Processus d'Entreprise" workshops and co-organized the "Ontologies, a smart way towards interoperability ?" Research school in 2006. These events have each gathered 45-50 researchers that participated to the emergence of a french community around themes about ESE. Selmin Nurcan is also one of the founding members (2006) of the Easy-DIM project at the initiative of the ECI GT and ERP/GDR MACS GT.

     

    At the international level, she is co-chair of the 8th and 9th workshops on Business Process Modeling, Development, and Support (BPMDS) associated with CAISE in 2007 and 2008 and inivited co-editor for a special number of the International Journal of Business Process Integration and Management. She was co-chair from 2004 to 2007 of the "Business Process Management" track of the international Conference IRMA who received the price of Best Track Award in 2004. She is program committee member of several conferences (CAISE, RCIS, IADIS, ME, ADVIS) and international workshops (BPMDS in CAISE, REFS in COMPSAC, RiGIM in ER) and INFORSID, and editorial committee member of international journals. She ensures the scientific valorisation of several international conferences (RE, CAISE, RCIS, RiGIM, REFS). A Phd thesis has been defended under her direction in 2001. Five thesis are being prepared, one will be defended in 2008. The activity in the thematic pole is significant in journals and in various conferences or workshops (29 since 2003). The following figures are representative of the research impact: CAISE (1), HICCS (1), ADVIS (1), international journals (4) and 2 book chapters.

     

    [4] J. Barrios, S. Nurcan. Model Driven Architectures for Enterprise Information Systems. the 16th Conference on Advanced Information Systems Engineering, (CAISE'04), Springer Verlag (pub) , June 7-9, 2004, Riga, Lettonie.

    [5] S. Nurcan, C. Rolland. A multi-method for defining the organizational change. Journal of Information and Software Technology, Elsevier. 45:2(2003), p. 61-82.

    [6] S. Nurcan. A survey on the flexibility requirements related to business processes and modeling artifacts. Proceedings of the 41st Annual Hawaii International Conference on System Sciences, Big Island, Hawaii, USA, 7-10 January 2008.

  • Reasoning Mechanisms

    This topic area deals with the development of theoretically grounded mechanisms to be reused in the researches undertaken in the three thematic poles of the CRI. These are (i) constraints based reasoning and, (ii) the intentional paradigm, which provide the basic elements that were necessary to develop methods and tools for RE, SE, and the ESE.

    Constraint based reasoning 

    Constraints based reasoning by its ability to deal effectively with partial information, has proved to be a powerful mechanism for the treatment of various problems in each of the three poles. Indeed, it offers a good power of expression (by the declarative aspect) while providing excellent speed of execution even on a large scale (through optimized constraints solvers). Our research concerns domains of constraints and resolutions aspects, via the setting-up of effective solvers, as well as application aspects. Two uses of this mechanism by the RE and ESE thematic poles can be cited:

    • The University of Evora (Portugal) IS was completely developed using GNU-Prolog (the constraints based system developed at CRI). This work relates to the concerns raised in the thematic pole ‘enterprise system engineering’. It showed that the constraints are an elegant and efficient way to implement requirements such as the management of the schedule and the occupation of rooms.
    • We have recently shown with members of the requirements engineering thematic pole the contribution of constraints to configure product lines. This approach has proven innovative both in terms of modeling the product line (the constraints provide a good power of expression) that its exploitation (the solver ensures an effective satisfaction of the product line constraints during configuration). A prototype based on GNU-Prolog is underway.

    The intentional paradigm 

    Two elements are the base of development of an intentional Meta-model and its specification language, called MAP. On one hand, the systems we are facing today whether softwares or business systems are becoming increasingly complex in terms of their structure and behavior. On the other hand, it is known that the mechanisms of abstraction are one of the few ways to assist in mastering complexity.

    We believe that the intentional paradigm is a powerful tool of representation and reasoning to handle the problems raised by the engineering of complex systems by abstraction mechanisms appropriate to the fields of investigation of CRI. The intentional abstraction consists in escaping the details of the operational world (for example business processes and the detailed description of who does what, when and with what resources) by setting out problems in intentional terms, i.e., guided by the goals to be achieved without mentioning the detailed procedures of achievement.

    CRI has developed a language, called MAP, that is based on the intention-strategy coupling and a mechanism of abstraction. MAP is based on a meta-model allowing to represent problems with a flexible organization of the intentions to achieve (goals) and strategies for doing so (approaches to achieve the intentions). The abstraction mechanisms allows specifying the intentions-strategies couples at different levels of abstraction.

    The intentional paradigm underlies many researches of CRI in each of the three thematic poles. For examples:

    • MAP is central to our strategic alignment approach. Our position is that two elements can be aligned only if they have the same purpose. Therefore, the alignment is based on a pivot intentional model, expressed with MAP. The pivot model represents the common purpose of the information system and corporate strategy.
    • The services oriented systems engineering thematic pole used the intentional paradigm to develop the MiS meta-model which allows to specify services in intentional terms and hiding the web services that allow their implementation. The requirements studied in the requirements engineering pole are expressed in intentional terms.The intentional paradigm was essential to develop the required abstractions to solve the problem of conceptual mismatch met particularly in the development of our 'matching' approach.
    • The services oriented systems engineering thematic pole used the intentional paradigm to develop the MiS meta-model which allows to specify services in intentional terms and hiding the web services that allow their implementation.

    SME: Situational Method Engineering

    Method Engineering, and Modular Method Meta Model Method Engineering, is the discipline that aims at building and adapting IS engineering methods and tools, if possible on the fly, by reusing method components from a component data base, to adapt them to the specific context of projects or companies. Method Engineering was born from the observation that methods are not applied as prescribed and from the dissatisfaction of method users, i.e. application engineers.

    The CRI has been a precursor in method engineering by recommending a modular vision of methods which allow an improvement of methods diffusion, reuse, and adaptation. The various method engineering approaches developed at the CRI since 1995 have contributed to define the modular method meta model which is the CRI’s international hallmark.

    This meta model was applied to our own researches each time we needed to represent a method. For example with the following methods:

     

    • ACEM (Anne Etien PhD thesis) to ensure the co-evolution of IS requments and the needs of organizational changes.
    • NENO (Emmanuel Papadacci PhD thesis) for requirements arbitration in a projects portfolio  context (applied at Renault),
    • MIBE (Iyad Zoukar PhD thesis) to align an ERP with organization requirements (applied at SNCF to customize Peoplesoft),
    • EKD-CMM (Judith Barrios PhD thesis)  for organizational changes management, based on our multi-perspective model for enterprise modeling.