Co-Evolution of I* Models and 3APL Agents

Co-Evolution of I* Models and 3APL Agents

University of Wollongong Research Online Faculty of Engineering and Information Faculty of Informatics - Papers (Archive) Sciences 1-10-2006 Co-evolution of i* Models and 3APL Agents Aneesh Krishna University of Wollongong, [email protected] Y. Guan University of Wollongong, [email protected] Aditya K. Ghose University of Wollongong, [email protected] Follow this and additional works at: https://ro.uow.edu.au/infopapers Part of the Physical Sciences and Mathematics Commons Recommended Citation Krishna, Aneesh; Guan, Y.; and Ghose, Aditya K.: Co-evolution of i* Models and 3APL Agents 2006. https://ro.uow.edu.au/infopapers/491 Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library: [email protected] Co-evolution of i* Models and 3APL Agents Abstract Agent-Oriented Conceptual Modelling (AOCM), as exemplified by the i* notation [9], represents an interesting approach to modelling early phase requirements that is particularly effective in capturing organisational contexts, stakeholder intentions and rationale. Our objective in this paper is to define means for executing i* models by translating these into set of interacting agents implemented in the 3APL language. We also propose a hybrid modelling, or co-evolution, approach in which i* models and 3APL agent programs are concurrently maintained and updated, while maintaining some modicum of loose coupling via consistency constraints. This paper explores how these two otherwise disparate approaches might be used in a synergistic fashion for requirement engineering. Disciplines Physical Sciences and Mathematics Publication Details This article was originally published as: Krishna, A, Guan, Y & Ghose, A, Co-evolution of i* Models and 3APL Agents, Sixth International Conference on Quality Software (QSIC 2006), Beijing, China, October 2006, 117-124. Copyright IEEE 2006. This conference paper is available at Research Online: https://ro.uow.edu.au/infopapers/491 Co-evolution of i* Models and 3APL Agents Aneesh Krishna, Ying Guan, Aditya K. Ghose Decision Systems Laboratory, School of IT and Computer Science University of Wollongong, NSW 2522, Australia {aneesh, yg32, aditya}@uow.edu.au Abstract ing for FormalTROPOS [2] style conditions on dependen- cies. Third, we wish to define methodologies to support Agent-Oriented Conceptual Modelling (AOCM), as ex- the co-evolution of models in the two frameworks, such emplified by the i* notation [9], represents an interest- that distinct groups of stakeholders can concurrently model ing approach to modelling early phase requirements that is and specify behavior, while maintaining some modicum particularly effective in capturing organisational contexts, of loosely-coupled consistency between the models. Fi- stakeholder intentions and rationale. Our objective in this nally, we are interested in compositional, extensible and paper is to define means for executing i* models by translat- easily maintainable modelling frameworks. We claim that ing these into set of interacting agents implemented in the the combination of high-level modelling in i* coupled with 3APL language. We also propose a hybrid modelling, or high-level specifications of functionality using 3APL agent co-evolution, approach in which i* models and 3APL agent programs offers such a framework. programs are concurrently maintained and updated, while Understanding the organisational environment as well as maintaining some modicum of loose coupling via consis- the reasoning and rationale underlying requirements, design tency constraints. This paper explores how these two oth- and process formulation decisions are crucial to model and erwise disparate approaches might be used in a synergistic build effective computing systems [10]. The i* modelling fashion for requirement engineering. framework is a semi-formal notation built on agent-oriented conceptual modelling. The central concept in i* is the inten- tional actor agent [9]. The actor or agent construct is used 1. Introduction to identify the intentional characteristics represented as de- pendencies involving goals to be achieved, tasks to be per- formed, resources to be furnished or softgoals (optimisation Agent-Oriented Conceptual Modelling in notations such objectives or preferences) to be satisfied. The i* framework as i* [9, 10] have become very popular in the recent past. also supports the modelling of rationale by representing key Such notations are commonly used to model organisational internal intentional characteristics of actors/agents. The i* context and offer high-level social/anthropomorphic ab- framework consists of two modelling components: Strate- stractions (such as goals, tasks softgoals and dependencies) gic Dependency (SD) Models and Strategic Rationale (SR) as modelling constructs. It has been argued that such nota- Models (refer to Figure 2). The SD model consists of a set tions help answer questions such as what goals exist, how of nodes and links. Each node represents an “actor”, and key actors depend on each other and what alternatives must each link between the two actors indicates that one actor be considered [6]. Our contribution in this paper is to define depends on the other for something in order that the former means for executing i* models. This exercise has been moti- may attain some goal. An SR model represents the inter- vated by the following observations. First, we seek to utilise nal intentional characteristics of each actor/agent via task the benefits of executable specifications. Second, we wish decomposition links and means-end links. The task decom- to view agent-oriented conceptual models and high-level position links provide details on the tasks and the (hierar- agent programs as jointly constituting a hybrid modelling chically decomposed) sub-tasks to be performed by each notation that leverages the complementary representational actor/agent while the means-end links relate goals to the re- capabilities of the two approaches. We are interested in sources or tasks required to achieve them. The SR model leveraging the (well-known) benefits of executable speci- also provides constructs to model alternate ways to accom- fications. This approach permits us to analyse early-phase plish goals by asking why, how and how else questions. We system models by performing rule-/consistency-checking at shall use the example of online shopping service from [7] higher-levels of abstraction. A version of this is check- Proceedings of the Sixth International Conference on Quality Software (QSIC'06) 0-7695-2718-3/06 $20.00 © 2006 throughout the rest of this paper to illustrate the i* frame- • TestGoal: if ϕ ∈ BF,thenϕ? ∈ GOAL. work and consequently how these models can be executed. Readers are encouraged to read [7] for the details of this • SkipGoal: skip ∈ GOAL. example. 3APL (An Abstract Agent Programming Language) [1, • SequenceGoal: if π1,...,πn ∈ GOAL,then 4, 5] is a programming language for implementing cogni- π1; ...; πn ∈ GOAL. tive agents. 3APL is based on a rich notion of agents, that is, agents have a mental state including beliefs and goals. • IfGoal: If ϕ THEN π1 ELSE π2 ∈ GOAL. Each agent has a number of basic capabilities. The basic capabilities of an agent are the basic actions an agent can • ∈ perform. An agent can have a number of practical reason- WhileGoal: WHILE ϕ DO π GOAL. ing rules for planning and revising its current goals. In this paper, we adopt 3APL platform [1] to support our work. These programming constructs of goals can be used in Our work is mainly based on 3APL definitions from [1, 4]. the body part of a practical reasoning rule and make 3APL Definition 1 A 3APL agent is defined as a tuple more flexible. n, B, G, P, A , where n is the name of the agent, B is a In a 3APL agent, P is a set of rules in the form: set of beliefs (Beliefbase), G is a set of goals (Goalbase), P πh<−ϕ| πb is a set of practical reasoning rules (Rulebase) and A is a set In this formula, πh and πb belong to a goal variable set of basic actions (Capabilities). [4], and ϕ is a belief. When the agent has goal πh and As described above, each agent is supposed to have be- believes ϕ then πh is replaced by πb. liefs about its mental state. Beliefs of 3APL are represented A set of beliefs, a set of goals and a set of rules of an using first order logic representation language. For exam- agent compose the beliefbase, goalbase and rulebase of this ple, a belief of a location of an agent can be written as agent. For a 3APL agent, Beliefbase is dynamic. It is up- ( ) agent x1,y1 . The programming constructs for beliefs are dated with executing basic actions from capabilities set. Ba- defined in [5] as below: sic Actions are mental actions that an agent can perform, Definition 2 (Programming constructs for beliefs)Given whose basic form is represented as: a set of domain variables and functions, the set of domain {ϕ1} Action(X) {ϕ2} terms is defined as usual. Let t1, ..., tn be terms referring where ϕ1 is precondition and ϕ2 is postconditions, both to domain elements and Pred be a set of domain predicates, of them are belief formula, empty is allowed here. Ac- then the set of programming constructs for belief formula, tion(X) is action formula. The execution of the mental ac- BF,isdefinedasfollows: tion will result in the update of beliefbase through replac- • p(t1,...,tn) ∈ BF ing preconditions by postconditions. Note that, Capabili- ties set is not compulsory to an agent, sometimes, an agent • if ϕ, ψ ∈ BF, then ¬ϕ, ψ ∧ ϕ ∈ BF. does not have a mental action. In addition, beliefs can be For example, an agent is at a certain position, written as generated from the communications between two agents agent(x1,y1), it has a task to lift a box at certain position, (sent and received). 3APL has a mechanism to support written as box(x0,y0).

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    10 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us