1、IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 8, May 2010 ISSN (Online): 1694-0784 ISSN (Print): 1694-0814 1 Agent-Based -Tools Integrated into a Co-Design Platform Alain-Jrme Fougres 11University of Technology of Belfort-Montbliard 90010 Belfort, France Abstract In thi
2、s paper we present successively the proposition and the design of: 1) -tools adapted to collaborative activity of design, and 2) a multi-agent platform adapted to innovative and distributed design of products or services. This platform called PLACID (innovating and distributed design platform) must
3、support applications of assistance to actors implies in a design process that we have called -tools. -tools are developed with an aim of bringing assistance to Co-design. The use of the paradigm agent as well relates to the modeling and the development of various layers of the platform, that those o
4、f the human-computer interfaces. With these objectives, constraints are added to facilitate the integration of new co-operative tools. Keywords: multi-agent system, co-operative agents, co-design, distributed design, micro-tools, development process. 1. Introduction The objectives of this paper are
5、to present: first, the design of a platform adapted to the innovative and distributed design, and secondly, assistance applications to actors of design process that the platform supports. These applications are called micro-tools (-tools) 23, 15. The concept of -tools consists of software applicatio
6、ns which are light, easy to use, integrated in a shared environment, and connected between them using a database. This Platform (PLACID: platform for innovating and distributed design, Plate-forme Logiciel dAide la Conception Innovante et Distribue in French) is developed with an aim of bringing an
7、assistance to the work of co-design, guided or not by complex processes (like workflow) for their capacity to manage flows of co- operative work (control and execution of co-operative processes). In addition to these objectives, strong constraints of flexibility and adaptability are added, to facili
8、tate the integration of new co-operative tools. We remember that, in a general way, co-operative activities integrated in virtual spaces of design require tools for: - Interpersonal or group communication (synchronous and/or asynchronous communication tools). - Organization and cohesion of the group
9、s and the activities (coordination tools). - Distribution and division of information, applications and resources (distribution and sharing tools). - Space-time definition of co-operation: space distance between the members of a team (in a real or virtual room), and temporal distance in the interact
10、ion (sequentially or parallelism of tasks). The co-operative design platform that we present in this paper is based on a software agent approach approach well adapted for distribution of components. The principal characteristics of agents (autonomy, adaptability, co- operation and communication) mak
11、e it possible, first, to effectively manage distributed, heterogeneous and autonomous components, and secondly, to facilitate exchanges of information and resource sharing between the components (interaction, communication and co- operation). These agents are of type: application (-tools or other to
12、ols of assistance to Co-design), coordinator/mediator, system and interface. Agents based system must manage organization and control of the community of agents. The effective use of the co- operative design platform (via an interface itself agent- based) is done in a context of strong and multiple
13、interactions, multi-users and multi-modalities. The -tools supported by this platform will not be necessarily integrated in a preset process of design. Their use can be specific, bringing a quite precise service in a phase of design. In all cases, each -tool will be connected to the multi-agent syst
14、em (MAS) by the intermediary of a host agent. This one will be used as interface of communication (inputs/outputs) between -tools and co- operative information system. This article is structured as follows: in section 2 we present the concepts implied in Co-design activities. The following section d
15、escribes capabilities of agent to communicate and interact. Sections 4 and 5 successively present objectives of PLACID platform and design of the first set of -tools intended to validate this platform. Finally, in section 6, we discuss the prospects of our work. IJCSI International Journal of Comput
16、er Science Issues, Vol. 7, Issue 3, No 8, May 2010 www.IJCSI.org 2 2. -tools for Collaborative Work 2.1 Context of Co-design The development of data-processing technologies, the democratization of the Internet, the use of the new resources on Internet gave rise to new working methods. We speak of co
17、urse, of the Computer Supported Cooperative Work (CSCW) 5, 11, 21. One of the major topics in the field of the CSCW is the development of groupwares. By definition a groupware is software which assists a user group for realization of a joint project. Group members collaborate remotely, either at the
18、 same moment (synchronous activity), or at different times (asynchronous activity). The fields of application are very numerous: products design, teaching, trade or games. Groupwares must make it possible to several users to collaborate in explicit shared spaces. The concepts to be considered are as
19、 follows 5, 6: - Time and space: to bring together several distant people geographically (office in proximity or distant) and/or not working at the same time (different rhythms, incompatibility of the timetables.). - Modes of co-operation: asynchronous co-operation (autonomous working method), co-op
20、eration in session (the objective being to decrease the times of interaction between the various actors of a project), co-operation in meeting (the roles of actors are defined and each one takes part in its turn), close co- operation (increase in co-production). - Flexibility in heterogeneous fields
21、: interactions, distribution of data, resource sharing, access control, representation of information, planning for tasks execution. Activities of co-operative and distributed design are exchanges, division and co-operation between participants. It is usual to present co-operative information system
22、s like being able to answer to the different needs of co-operation: - Facilitating the resource sharing. - Assisting the coordination. - Improving the communication of group. - Supporting the individual motivation. - Supporting the development of organization. Activities related to collaborative wor
23、k are mostly exchanges (language acts, transactions), sharing and cooperation among participants. Then, to memorize easily the functions of groupware applications, it is convenient to take into account five main functions, namely: 1) Communication between participants of the community; 2) Coordinati
24、on between them this function can be insured by a centralized way or not; 3) Co-memorization, which means communitys memory construction it may be a project traceability support, like the set of sheets of paper produced during the course of action; 4) Co-production of shared resources, like proxemic
25、 space of cooperation, shared objects; and 5) Control of processes, control data or files circulation it may be workflow, which is a convenient software application if photographed evaluators The following figure (see Figure 1) schematizes relations between these basic functionalities of groupware.
26、We have called this functional model: 5Co 15. This model supplements the model 3C (communication, co- operation, coordination) 6, defining spaces necessary to the artifacts of collaboration. Communication Co-production Coordination Control of process Co-memorizing Fig. 1 Basic functions for Co-desig
27、n The 5Co How to design software applications that should achieved the 5-Co model, and that both in a technocentric and in an anthropocentric point of view? CSCW approach can be complemented by a more “microscopic” one, focused on course of action, within many different operations are realized, at a
28、 very short term, and by cyclical and opportunistic way. To aid these tasks, a new software concept can be defined. It is called “micro-tool” (-tool). 2.2 Concept of -Tool The concept of -tool 23, 15 is opposed to the current tendency tools of design, which are often heavy, prescriptive, and not use
29、d. Ideally, these tools must be (see Figure 2): - Easy to learn (a few minutes) and easy to use. - Simple (even if they are developed on the basis of an elaborate theory). - Easily programmable, easily modifiable by designers or by users themselves, usable in an opportunist way. - Autonomous, but al
30、so reactive when they are defined for co-operative processes (CMT: Co-operative Micro-Tools). Those are distributed between actors who act according to their skills this corresponds to the needs for concurrent engineering. IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 8
31、, May 2010 www.IJCSI.org 3DB Collaborative platform (PLACID) Fig. 2 Concept of -tools. We have just defined the concept of -tool like which can bring a support to realization of an elementary and specific task entering in a well defined activity (table 1). Table 1: Adapted levels of co-operative inf
32、ormation systems according to the Activity Theory 24, 13, 15. Level Activity Theory Human centered System centered Macro Activity Motivation Co-operative system Micro Action/Task Goal Micro-tool Nano Operation Conditions Functionality The product of a task can be an intermediate object of design. De
33、velopment of -tools corresponds to an oriented step activity. We present below the various principles which lead to their data-processing structuring: - Use of -tool is individual or collective; tasks can be structured in action plan. Then, it is necessary to specify the conditions of use of a -tool
34、 (life cycle of generated objects, share conditions.). - Interaction between actor and -tool relates mainly to data acquisition (objects of activity), their relations, their accesses and their management, with graphic tools for example. - Identification and description of -tool being the result of a
35、 collective and multi-field work; realization of models is recommended to facilitate exchange of ideas. - -tools are developed according to a procedure of software quality which we defined in our laboratory. -tools make it possible to divide software into modules adapted to achievement of simple tas
36、ks. Several tools in association can fulfill more complex functions. We defined a complete process of development of -tools for all actors cooperate (Figure 3). This process begins with the analysis of activity and leads on the corresponding software products and on the delivery of seven documents c
37、onstituting the memory of their designs. Three great phases structure this process which we called IDI (Identification, Design, and Integration): - Phase of identification of -tools, referring on the higher levels of engineering system and requiring a large collaboration of all actors, consists: 1)
38、in identifying among the tasks which make a specific activity, those which could be instrumented, 2) then, to specify them. - Phase of design aims, according to an incremental approach which facilitates permanent dialogue between all actors of development process, to define the architecture of -tool
39、 and these components, to develop them (UML/Java), to test them, and finally to validate the user interface. - Phase of integration into PLACID, agent platform that is connected to an ORB (Object Broker Request, CORBA in fact) for exchanges management and information sharing. Integration Space of co
40、-design Analyze Identification Design . Activity Theory . SADT A_UML Incremental development of -tools Identified -tools Prototype of -tool Co-Actors . PLACID . Corba . Web services Fig. 3 Development process of -tool. 3. Agents for Co-operative Activities Multi-agent systems have been proposed as a
41、 new approach for distributed artificial intelligence 27. The main characteristics of agents (autonomy, distribution, adaptability, flexibility, cooperation and communication 29 permit, on the one hand, to manage efficiently the heterogeneous, autonomous and distributed solutions, and on the other h
42、and, to facilitate exchanges of information and the sharing of resources between solutions (communication and cooperation). The idea of using the paradigm agent to design complex, interactive systems, IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 8, May 2010 www.IJCSI.o
43、rg 4 either distributed or cooperative is very attractive. Indeed, very early, researchers found in the set of characteristics of agents, the means to design efficiently some cooperative information systems 26, 25, 3. Other researchers, as Jennings 12, justified the adequacy of the approach agent fo
44、r distributed system modeling and design (adequacy hypothesis). MAS make it possible to distribute agents (processes) which are communicating entities, autonomous, reactive and having competences 22, 7. To design MAS according to these criteria, it is necessary that each agent owns the three followi
45、ng properties: independence, communication and intelligence (expertise, skills, know- how). We also must define the architecture of agents (cognitive functions, interactions), and structure the knowledge necessary for their various activities. These properties correspond to those definite as well fo
46、r an assistance platform to the collaborative design, as for the applications which it supports (all the more when they are co-operative -tools). 3.1 Agent Modeling Agents being heterogeneous entities with various modes of interactions and complex behaviours, it is necessary to define their type of
47、organization, and their capacity of evolution. Many definitions of paradigm agent have been proposed, one of the most consensual was made by Wooldridge. According to Wooldridge 28 an agent is an encapsulated computer system that is situated in some environment and that is capable of flexible, autono
48、mous action in that environment in order to meet its design objectives. Furthermore, a software agent is rational, finalized and co-operative 29. The autonomy of an agent is technically implemented by: 1) an independent process, 2) an individual memory (knowledge and data), and 3) ability to interac
49、t with other agents (perception or reception, emission or action). The systemic model of MAS is defined as follows: the agents of MAS evolve in an environment and interact with each other, respecting the roles assigned to them in an organization. Then MAS are described by the following quadruplet (1): MAS = (1) Where Agt is the set of agents, Int is the set of interactions defined for these MAS, Ro is the set of roles to be played by agents and Co is the organization of agents into communities, when they are def
