From i* to istar 2.0: An Evolving Social Modelling Language

Transcription

1 From i* to istar 2.0: An Evoving Socia Modeing Language Lin Liu 1 Schoo of Software, Tsinghua University, Beijing, , China iniu@tsinghua.edu.cn Abstract. Conceptua Modeing, as a thought too, heps its adopters to describe an abstract observation to given rea word phenomena. i*, as a socia modeing anguage, was widey adopted by researchers in both requirements engineering and business information system anaysis. In order to further extend its adoption in future research and practice, istar 2.0 was conceived and pubished to reduce ambiguities and compexities. In this paper, I woud ike to share my observations as an i* modeer, about the major differences identified between i* and istar 2.0, about how to map an i* mode to an istar 2.0 mode, as we as how to further evove the modeing anguage to serve for the next generation modeing needs. Keywords: socia modeing, concepts, reations, evoution 1 Motivation My first contact with i* was in year 1999, a few months before my graduation and joining Eric s group, I downoaded two of his major i* modeing papers, in which he used i* to anayse business process of IKEA [1], and the mutua dependencies between members in software project team [2]. At that time, my word mode was aready agentoriented, due to my master s thesis, on Actor [3], the concurrent computing mode, and PhD thesis, on Agent-oriented requirements anaysis using Agent-Z [4] and process agebra to formaise use case scenarios. My first impression to the exampe modes was: it is a bit compex, and drawn artisticay, different from most other software engineering modes that I was famiiar with, such as, message sequence charts, state machines, and cass diagrams. Tonight, I finay get a chance to read the i* 2.0 Language Guide [5] by Fabiano, Xavier and Jennifer, word by word, which I shoud have done two and haf years earier, whie they sent numerous messages asking for comments and inputs. I fee synchronised with the evoving anguage finay, aas, it is better ate than never.

2 2 2 Reated work Bits and pieces in Retrospect of 17 Years persona modeing and extension experience with i* 2.1 A Successfu exercise - Modeing Trust in Smart Cards It didn t take me ong to faing in ove with using i* to mode rea word. As my first modeing exercise was a perfect match. Modeing trust in smart-cards systems can make use of many of the interesting modeing constructs in i*[6]. Incuding: Strategic dependency modeing captures the major roes in a genera smart-card system, and their dependencies. Roe, position and agent together with roe-paying and position occupancy inks, which are used to capture who is paying attacker, and which organisationa payer is paying the abstract roe of card owner, software owner, data producer, and how the trust and dependency distributed among different organisationa settings. The fexibe use of contribution inks, in particuar, the attacks are represented as a break contribution ignited from the attacker to the victim or vunerabe eement or ink. 2.2 A first attempt to extend i* capturing tempora orders between tasks with prior-to ink My second modeing exercise was due to the marriage of GRL with UCM [7]. GRL is a variation of i*, which emphasis more the goa-oriented perspectives of i*, I think I tried to combine the two anguage in a different way comparing to what the current URN standards suggested. I wanted to use i* mode as the container, or the pace-hoder of scenarios. I ony wanted the jogge ine of UCM, which traversing the tasks in i* payed by different actors. It captures the refining process of goas and generated the run-time execution scenarios invoving mutipe actors in the mode. In my mind, goas, actors, and scenarios are the three pieces to be fit together in the requirements engineering puzze. 2.3 Another successfu exercise modeing service reationships My next major move was using i* in services modeing. Again, i*, especiay, the socia, strategic dependency modeing construct, was a natura fit to mode service providers and service consumers, their needs and capabiities, their deegation of different types, their commitment and deivery of services, and quaity of service [8]. i* is expressive enough support my modeing objectives and provide meaningfu types of anaysis.

3 3 2.4 A second attempt to extend i* capturing context as annotations My second attempt to extend i* was due to the modeing needs of service adaptation. Where I wanted to say that when service text changes, service provider and consumer s choices wi change accordingy, so I associated context information on each of the service goa refinement ink [9]. I am doing this retrospect just to take a position of modeer, to see if the current istar 2.0 concepts and reations wi sti aow me to do simiar things in a cearer way, or it has decided to avoid such use or extension of the modeing anguage. 3 Redrawing i* modes with istar Discard organizationa positions together with occupancy and coverage ink As shown on the eft-hand side, in i*, position is used to capture organizationa positions, which can cover different abstract roes, and being occupied by different agents. As i* was designed as an organizationa modeing anguage, so position is considered as a first-cass modeing concept. In istar 2.0, organization is not stressed anymore, so we can define it as an abstract roe, participatedby agents, and it can speciaize either an attacker or a defender or both using is-a ink. In this case, a cover ink wi become an is-a, an occupies wi become a participates-in. It is not cear whether the INS ink is represented now, as it is a reation between actors of same nature. Athough I can make the mapping, I fee that it is a more natura and direct refection of the rea word meaning using pays and is-part-of than participates-in in this case. Fig. 1 actors and association inks in i* [6] 3.2 Changing means-ends and decomposition as AND/OR refinement In istar 2.0, the means-ends and decomposition inks in i* are unified to be caed AND/OR refinement inks, which oosens up the origina i* s strict enforcement on iterative eicitation of aternative ways to satisfy a given goa using means-ends ink. As a modeer, I had hated the constraint of imposed by this means-ends semantics, as I had to use some dummy tasks and goas when there are not meaningfu aternatives to choose from. However, I have to admit that we shoud keep it in mind to ask the

4 4 question each time we face a goa. In other words, I woud rather see the current istar 2.0 treatment as a simpification of the i* syntax rather than a change of semantics. 3.3 Adding Quaification ink between a quaity attributes with its associated eements A new kind of ink, quaification ink, represented as dotted ine, inks the quaity attributes with the corresponding goa, resource and task eements expicity. This is a major extension in istar 2.0, which can ony be represented using the naming convention of subject[object] in softgoas in i* or NFR framework [10]. This extension carifies the semantics of quaity attributes and suggests a proper way of its usage. 3.4 Adding DependerEmt, dependeeemt, rues and constraints on socia dependencies There is a rues and constraints section in the specification of dependency reationship in istar 2.0. It gives cearer guideines to modeers which encourage the proper use of the ink. This incude: (1) adding definition of dependeremt, dependeeemt, and confine depender and dependee as actors; (2) when a depender deegates a dependeremt to others, it cannot be refined or contributed by other eements within the actor boundary. Dependency reationships are not aowed to share the same name, which means there is exacty one depender and one dependee for each dependums. This extension is aso a very good move in making the semantics of dependency ink cear and easy to use. 3.5 Adding NeededBy reation between a task and its required resource Resources are considered a sub-component of a task in i*, which is not distinguished from a sub-goa, sub-task, or sub-softgoa. In istar 2.0, as resources are different from goas and task in nature, a different kind of ink NeededBy is suggested. Its impication is that resources are eave nodes, they wi not be further refined, they are ony checked for avaiabiity or not. 3.6 Removing some of the contribution inks in i* In istar 2.0, four types of contribution inks are defined: hep, hurt, make and break. Other contribution ink types, such as: some+, some-, unknown, AND, OR are discarded. The impication of this change is that, contribution inks are evauated individuay, we ony consider weak or sufficient, positive or negative contributions from a source eement to a quaity attribute. This aso improves the carity of the modes.

5 5 4 Discussions In summary, istar 2.0 carifies ambiguities in i* modeing framework, which makes the adoption by students and engineers earning the modeing syntax and semantics easy. For i* users, it wi not require much effort to understand the changes and adopt it in new cases and appying the changes in existing i* modes. Some minor points require further deiberation are as foows: Is a roe aowed to participates-in an agent? Whie there is no ogica expanation for it, in the istar 2.0 meta-mode, it is not prohibited? Is instantiation reation in actors aowed in istar 2.0? Sometimes the modeer may want to express scenarios at an instance eve? Is a quaity attribute evauated by itsef or together with the eement it quaifies? In other words, whether a quaity attribute is a standaone eement or is ony meaningfu together with an entity? A major probem yet to be addressed are using views of actors as a measure to contro scaabiity of mode. In today s organization, socia and strategic anaysis is often supported with operationa data as evidence for socia dependencies and infuentia factors. Thus, automated eicitation of socia modeing concepts and reations are considered an effective way to obtain socia reation modes as in i*. Acknowedgements This paper is partiay supported by the Nationa Natura Science Foundation Project (Grant No ) and Nationa Science and Technoogy Support Program (project no. 2015BAH14F02). Tong Li acknowedges the support of BJUT Startup Funding No References 1. E. Yu. Strategic Modeing for Enterprise Integration, Proceedings of the 14th Word Congress of Internationa Federation of Automatic Contro (IFAC 99), Juy 5-9, 1999, Beijing, China. pp Permagon, Esevier Science. 2. E. Yu and J. Myopouos. Modeing Organizationa Issues for Enterprise Integration, Proceedings of Internationa Conference on Enterprise Integration and Modeing Technoogy, October 28-30, 1997, Turin, Itay. 3. Agha G A. Actors: A mode of concurrent computation in distributed systems[r]. MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB, d'inverno M, Kinny D, Luck M, et a. A forma specification of dmars[c]//internationa Workshop on Agent Theories, Architectures, and Languages. Springer Berin Heideberg, 1997:

6 6 5. Dapiaz F, Franch X, Horkoff J. istar 2.0 anguage guide[j]. arxiv preprint arxiv: , Yu E, Liu L. Modeing trust for system design using the i* strategic actors framework[m]//trust in Cyber-societies. Springer Berin Heideberg, 2001: Liu L, Yu E. Designing information systems in socia context: a goa and scenario modeing approach[j]. Information systems, 2004, 29(2): Liu L, Liu Q, Chi C H, et a. Towards a service requirements modeing ontoogy based on agent knowedge and intentions[j]. Internationa Journa of Agent-Oriented Software Engineering, 2008, 2(3): Sun J, Liu F, Zhang H, et a. Understanding the Diversity of Services Based on Users Identities[C]//Internationa Conference on Advanced Information Systems Engineering. Springer Berin Heideberg, 2011: Chung L, do Prado Leite J C S. On non-functiona requirements in software engineering[m]//conceptua modeing: Foundations and appications. Springer Berin Heideberg, 2009: