exceptions in business rules, and volatility in the behaviour of normative agents. ... compares and contrast between rules and norms within an IS domain.
A SEMIOTIC APPROACH TO ORGANISATIONAL MODELLING USING NORM ANALYSIS Authors
: Simon Tan, Kecheng Liu, Zhiwu Xie
Affiliation: University of Reading Department of Computer Science PO Box 225, Whiteknights Reading, RG6 6AY, United Kingdom E-Mail
: {b.k.s.tan, k.liu, z.xie}@reading.ac.uk
URL
: http://www.cs.rdg.ac.uk/ais
Telephone: +44.118.9316024 Fax
: +44.118.9751822
Keywords : Norm Analysis, Agents, Process Modelling, Information Systems, Affordance. Appropriate stream: Developing the IS Discipline. Appropriate topic : Social and Cultural Issues of Information Systems.
1
Abstract This paper explores a theoretical perspective to information systems engineering, correlating requirements modelling techniques with Norm Analysis, to elicit organisational semantics and to formalise complex business rules. At present noticeable disparity exists between systems models, and the actual, ‘real’ organisation information systems. The modern multifaceted organisational disposition is complicated further with exceptions in business rules, and volatility in the behaviour of normative agents. Traditional systems approaches do not fully address these complications, and disparity in systems models; Norm Analysis, however, offers a viable alternative, which until now has not been extensively explored. These inherent systems design issues may be alleviated with the introduction of norms, to address the complexity of organisational information systems. We propose utilising Norm Analysis, a branch of Organisational Semiotics, for the elicitation of IS requirements to encapsulate rigorous business behaviour, and formalise intricate business rules. We believe this approach will contribute to the overall usability and coherence of organisational models. The second part of this paper considers the adequacy of representing norms in deontic logic, rules operands and temporal based systems, with an illustration of a constabulary crime-reporting case study. Keywords: Norm Analysis, Agents, Process Modelling, Information Systems, Affordance
1. Introduction Current organisational information systems have grown significantly in both size and complexity. Modern systems designs are often technically aesthetic, but functionally awkward and do not address the real needs of users. Empirical studies have since established a direct link between the poor understanding of information systems organisation and software failure. Unified Modelling Language, the de facto object oriented modelling language has gain widespread recognition as the leading systemmodelling standard. UML however, is not equipped to model the complexity of organisational systems behaviour. As such, we propose an extension to addresses this issue, incorporating normative elements. Norm Analysis (NA), which originated from Semiotics, may provide a viable alternative to deal with this aspect. Systems modellers are aware that gaps, in the form of “semantic gap”, exist between organisation and software (Liu 2000) with many undefined parameters. System analysts are constantly required to resolve these arduous tasks of tidying up loose ambiguous parameters. This quandary will inevitably fall on the system analysts’ discreet judgment, hence knowingly or unknowingly, or even worst in some situation overwrite crucial business decisions. There is an impending need to unify existing requirement engineering with Norm Analysis. Traditional system requirements and organisational semantics (Liu et al. 1999) have until now, evolved independently. This poses a serious problem on consistency issues in systems design. Norm Analysis will attempt to address and unify requirements engineering with semantics to ensure that systems model designs are not developed in isolation from the social aspect of organisational norm. Information systems from the viewpoint of semioticians are regarded as a holistic entity, as opposed to UML dependency based concept of compartmentalising IT business
2
process, with independent sets of loosely coupled software processes. Norm Analysis, we believe has the rigorous ability to model these complex organisational systems requirements. We propose incorporating a suite of semiotics tools: semantic analysis, Norm Analysis, and normative agents (Liu & Salter 2002). These components will undertake three crucial tasks of addressing the above systems issues. Firstly, it identifies the ontological dependency (Wand 1999) between systems processes, having established the ontology chart; it will then ascertain the semantic relations. Hence, formalising these relationships to model the behaviour of organisational systems design. It is not the intent of this paper to propose additional methodology to the existing field of information system practice; many renowned methodologies over time have demonstrated to be reliable and thorough. Over the last 30 years, there have been numerous attempts and endeavours to model organisational semantics (Peckham & Mryanski 1988, Wand et al. 1999), which were founded on entity-relationships. There is an impending need to re-focus and tap the rich organisational semantic layers. In recent years, we have seen rapid growth in enterprise systems organisation, huge technological advances and radical evolution in business IT organisation. These changes have exerted great demands on systems requirements engineering tools. Traditional systems, as such are overwhelmed, and ill-equipped to address the social behavioural aspect of organisational roles (Joaquim et al. 1999). In order to illustrate the functions of Norm Analysis, within an IS organisation model. We have instantiated Norm Analysis with several leading system requirement techniques: Role activity diagram (RAD) (Ould, 1995), Activity diagram and Scenario diagram (Booch et al. 1999). This paper is structured as follows: Section 1 identifies the limitation of traditional systems requirement approach, and proposes an extension to elicit complex organisational information. In section 2, we briefly describe norm concepts with respect to deontic operators, rule-based and temporal-based, and its significance to systems modelling. Norm Analysis in section 3, explains the concepts of Norm. NA structure and rules constructs, with emphasis on the four stages of NA framework: Responsibility analysis, Information identification, Pre/Post triggers analysis and Norm specification. Section 4, compares and contrast between rules and norms within an IS domain. Section 5 illustrates a case study of a crime reporting system, and demonstrates how norms and semantics are elicited by applying NA. Drawing the conclusion and discuss future work on Section 6.
2. What is Norm? Norms in essence are a set of rules and regulations, an underlying protocol governing the human communications network. These are etiquette of behaviour observed among people of different communities, cultures, groups and even sub-cultures unique within an organisation. These norms are embedded within a social context transcending the boundaries of explicit, implicit, formal and informal states, collaborating to attain certain goals. Norms revolves around agents, which influences the agents to execute a series of concerted actions to achieve a particular goal. In this respect, it can specify to a limited
3
extent how an agent should or should not behave, under a prescribe sets of triggers. Norms are highly regarded in many instances; it in reality controls the social and organisational domains, which exerts a major influence on affordance. Affordance and Norms are not mutually exclusive, yet inextricably intertwined within the social context of organisational behaviour. The term “affordance” was first coined from the perceptual psychologist (Gibson 1977), referring it to the actionable properties between the world and an actor. To Gibson, affordances are relationships of entities and observable pattern of behaviour occurring within a domain. Patterns exits, as such it could be captured, and taken to anticipate users and systems’ behaviour. It should be stressed that affordance exists naturally: it may or may not be apparent, visible, known, or even desirable, but it plays a significant role, and its effects will greatly influence behavioural norms and the order of patterns. Affordances are time dependent, thus it abides by the rule and concept of start and finish of an event. As such, defiance of affordance and norm are not to be taken without due consideration within a domain. This deviation from acceptable norms has its consequences; punishments and penalty may be imposed on agents compromising these protocols. Advocating Norm Analysis as a systems method offers a normative perspective to systems modelling and design, utilising rich semantics to depict the ontological dependency, and to facilitate the elicitation of system requirements. Norm Analysis also facilitates the deduction of organisation signs and norms (Stamper et al. 2000). Norm Analysis is well equipped to accommodate different systems methods, techniques and approaches. Norm’s application in information systems is not just a recent fad; it has seen wide spread adoption in various industries more notably in the field of Artificial Intelligences and legal systems. Norms are used in organisational systems modelling e.g. by Sergot (2001) and Ivan (2000), with varying degrees of systems complexities. This paper will present a feasible normative approach, which introduces roles that complements existing system requirements techniques. Secondly, norm with its pliable behaviour and rules, when applied appropriately would widen the spectrum of information requirements. Lastly, and more importantly this system approach will attempt to resolve the handling of business exceptions, utilising intelligent agents for the prognosis of systems behaviour, facilitating a proactive systems design responses as oppose to reactive model design. These characteristics will be manifested in the modelling and systems development phases, making it more communicative and reflective of a real business systems model. It is evident, that conforming to syntax, methodology and abiding good programming practices alone will not produce a system that users require, without thoroughly understanding the semantic abstract layer of an organisation. Norm Analysis will attempt to steer the developer towards a normorientated approach to systems model design.
3. Norm Analysis Norms evolve in accordance to circumstances and change, akin to a living-breathing organism adapting to its environment, a habitat operating within a complex boundary of interdependent entities whose relations and properties are largely determined by their
4
functions in the organisation. It is a source of commitments and contracts that agents establish mutually; however, agents being autonomous may choose to violate (Liu et al, 1999). It is a composite of multiple actors performing different activities, capacity and objectives with a concerted goal. Agents in this context refer to both software and human entities, autonomous in characteristic (Chong & Liu 2000) facilitating the communication, and possible violation of norms. Human agents however take precedence over software agents, and as such are responsible for the delegation of duties. Agents either human or software interact actively within and across business domains. The understandings of norms are crucial to enables modellers to appreciate systems processes, and agent’s behaviour to reinforce the overall systems design. Thus, represents an organisation true complexity, and diverse mix of autonomous processes and behaviour. Norm Analysis Norms can be broadly categorised as cognitive, evaluative, perceptual and behaviour (Stamper 1992). These are further elaborated in (Liu 2002). Behavioural norms are highly applicable to organisation Norm Analysis; comprising of 4 stages namely, 1) Responsibility analysis, 2) Information identification, 3) Trigger analysis and 4) Norm specification. Norm Analysis (Liu et al. 2001) formulates an approach to elicited complex organisation arguments. It identifies and determines the pre-condition and post-condition of agents’ behaviour to ascertain responsible agents. A norm has the following construct (Stamper et al. 2000): whenever if then is to do This construct enables norm elements to capture, which are required for execution in lowlevel programming language, e.g. LEGOL (Liu 2000 & Stamper 1980). Deontic logic has different representations of a fundamental nucleus. It has a canonical criteria expressed antecedent. Users have to be cautious of this proliferation of expression based on normative context of each scenario.
4. Rules Versus Norms A rule determines a principle or instruction stating the way in which things should be done. Sets of rules e.g. IF, ELSE, THEN constructs. However, business rules are more tightly specified in this respect. They represent procedures, constraints and policies on the way an organisation should conduct its business (Liu & Ong 1999). Software agents serve to automate and enforce rules and isolate those instances in which the independent judgement of a human is required. These rules are embedded directly in the program
5
code. It is the procedural logic of how system workflow intertwined with business logic. Thus, when business rules changes it requires the arduous tasks of rewriting the program logic code. A norm in contrast represents a hypothetical construct using agents to associate actions to conditions. (Liu 200) Norms include deontic operator, rule-based and temporal- based operands. The antecedents of each norm will have a prescribe set of successive instructions on the next norm, structured on the pre/post notion of norm triggers. Agents are empowered to exercise and assert responsibility and authority over organisational norm. However, exceptions exist and rules differ. Refer to an example below: A police superintendent in-charge of an inspector, and the inspector over the junior officers. At a glance, it may appear similar to a typical hierarchy structure model; however, it is quite different with the inclusion of components as deontic operator, timebase and rules-base operands. E.g. Agents role: The superintendent has to abide by legislative and policies (rules) set by the governmental bodies and police force (agents). The Superintended should not instruct his subordinate to reopen a case without valid reason, contravening the police policies (violation of norm); if there are no supporting evidences pertaining to the case. (Rules overwrite > Agents roles) The superintendent can specify the rights of his subordinates by using deontic operators, such as obliged, prohibited and permitted. The subordinates are the responsible agents for the actions prescribed by the superintendent. The subordinates under the command of the superintendent may choose to follow or violate the norm. However, the latter may induce a punishment unless there is an acceptable ground. Temporal Elements of Norm Conditions of properties can and will change over time. These temporal-based norms would greatly influence the Norm Analysis constructs, in determining how to resolve the procedural algorithm. • • •
Absolute (calendar date) Relative (e.g. bill due-date) Event triggers (trigger via a process)
Absolute and relative-time triggers a state of changes on norms with the passage of time. However, associated agents may overwrite triggered event, causing a change of states and event.
E.g. A suspect is apprehended for a crime committed in another county and the investigation reveal that this case is linked. This will automatically re-opened (pending active) via the trigger from an external source.
6
The Method of Norm Analysis Norm Analysis formulates a systematic approach to elicited norms, responsible agents and defines pre/post conditions of event triggers. The analysis is carried out in 4 stages to provide each process with a tabular normative unit of information. The NA table is designed to capture a series of norm “instance”. This modular design enables a clear and well-defined norm structure, representative of each characteristics pertaining to a specific norm. stages
tasks
outcome
Responsibilities Analysis
Identify responsible agents for the Start/finish of “an instance”
2
Information Identification
Select type of key information required by “an instance”
3
Triggers Analysis
Activities/actions effecting the Start/finish of “an instance”
1
4
Norm Specification
Preconditions
The conditions for acting and invoking norm
postconditions
The resultant after the successful execution of norm
Norms specified in the standard construct.
Table 1: Norm Analysis: Stages, Norms and Deliverables (Adapted from Liu 2000 and Liu & Salter 2002) Responsibility Analysis (Stage 1) Responsibility analysis identifies state association of entities and agents that are responsible for the start and finish of “an instance”. E.g. Affordance > Investigation. The period of an incident starts immediately from the instant a case code is issued but not prior to it classified as a Crime. The investigation period will only commence, once the police officer is assigned to the case, and terminated when the case have been resolved. These are examples of well-defined processes/actions with explicit period and agents. It is more evocative to link processes, which are time-bounded and responsibly linked, as compared to establishing links based exclusively on process that trigger a subsequent or concurrent process. Information Identification (Stage 2) “An instance” needs key source information for the preceding action. Without these vital bits of information, it is technically difficult however, not impossible for agents to derive prudent decisions. The investigating officer would need to know information as, details of the victims, location/time of crime committed and description of incident to proceed with the investigation. Else, the entire investigation process would not exist. These affordance and agents define in the NA tables, are elicited during the semantic analysis phase after the ontological chart are drafted.
7
Trigger Analysis (Stage 3) Triggers act as a mechanism to activate associated processes based on the pre-post conditions of existing social norm define in Stage 1; it could either take the form of Temporal, Substantive or Semiological to influence the trigger procedure. A relative temporal norm could take the form of a binding legislature. Specifying that a case should not be open for a period longer than 1 year without any development or new evidence. It must therefore be categorised under pending, as a case cannot remain open indefinitely. Substantive norm occurs e.g. when the witness recalls any key clues pertaining to the crime, where the status has been flagged as Pending, this action will automatically reopening the case. Semiological trigger happens for e.g. when a suspect is caught for another crime and is traced to implicate or influence the outcome of the existing case. Norm Specification (Stage 4) Norm specification is the final stage, where the earlier 3 stages of information are collated, decimated, formalised and structured to facilitate prudent behaviour decisions based on conditional norm presented. The complexity here is to deal with formalise norm when conditions are often violated. Refer to section 3. Conditions of higher order precedence will overwrite that of a lower. E.g. personal safety commands a higher precedence to reporting a crime. Norms would have to be formalised in this respect, it would then be incorporated in the above example to handle business exceptions and the “violation” of Norms.
5. Case Study: Police Crime-Reporting This case study was based on a project of designing a crime management system for the police forces. An excerpt of this project, "crime-reporting" was highlighted to reflect the actual research conducted by our research team, which is still ongoing. This case study is provided, although for the sake of brevity the detailed workings and complexity of the model have not been fully reproduced here. The outputs from the model are sensitive to changes in key assumptions. We have not sought to be prescriptive about these assumptions, which will vary according to circumstances. However, this guide does identify the key sensitivities and the basis on which the assumptions underlying them might be made. The Crime Reporting Unit is the contact point between members of the public and the police departments. It is charged with the responsibility of collecting, maintaining, analysing, and reporting crime data for the nation wide crime-management. In the Crime Reporting case study, Role Activity Diagram (Ould, 1995) was used to capture the overview of the system outline with the multiple façade role characteristic of the Police Department (Fig. 1). Roles have been widely used for modelling the authority, responsibility, functions, and interactions, associated with agents within an organisations. RAD forms a regional road map of the systems design. However, these are ideal for an initial mapping, but insufficient to establish detail relationships between agents and the targets they manage. This structure is constructed on an abstract models of coordination, however social elements context that are crucial for organisational analysis are lacking, as behaviour or norms are not present in these dependency-based models.
8
CRIME REPORTING Victim / Reporter
Crime Assessor Start taking enquiry
Request Identification Submit identification Note reporter details Request incident details Describe incident
Appraise Incident Validity Crime Statement y
n
Valid Crime?
Verify no duplication
Refer to Relevant Dept
Fill crime details
Police
Write Statement Assign Case no.
Allocate Case
Assign to Investigating Officer
Investigation
y
n
Case solved?
Figure 1: RAD – Crime reporting process Role Activity diagram emphasise on roles, functions, activities and their interactions within events and logic that determines what activities are carried out and at what point in time. This derives a higher-level conceptual model to determine at best a sketchy superficial model of the system. It is not particularly concern in identifying and of patterning Norm behaviour at this stage. Agents in the form of token, resides on a particular state will represent the possible outcome of the next role instances. The functionality of token resemble a baton relay race, the next process can only occur once it receive the passed on token, satisfying the pre-conditions then only will it trigger the consecutive action and so forth. Thus, at every stage of modelling, the modeller has to make explicit decision, not allowing much margin to accommodate exception behaviour.
9
Public Domain
Crime Recorder
Crime Reporter
Other Departments
Crime Committed
Contact Police Crime Reporter appraise incident
1
Department (referral) [Norm 3] Reject
[Norm 1] Forward
[Norm 2] Accept Take written statement
2
1) Assign to Crime Recorder 2) Record Crime
Accept & record Crime
3A
3B
Figure 2: Activity Diagram – Crime reporting process detail. [Apprise - Forward Incident] Victim/ Witness
Crime Reporter
Police Force Info
Statutory/Other Department
Report Incident Describe Incident Furnish incident Does not constitute as a crime Apprise incident Waiting
Verifying
Verifying policies
Checking
Relevant Dept
Department Contact
Referral to dept
Figure 3: Scenario Diagram – Appraise Incident
10
[Verify New Crime] Victim/ Witness
Crime Reporter
Police Force Info
Report Incident Request particulars Furnish particulars Describe Incident Furnish incident Verify incident new (no duplication) Waiting
Verifying
Checking
Confirm new incident Take statement
Figure 4: Scenario Diagram – Verify Incident Extracting Norms (Activity diagram and Scenario diagram) Roles are further expanded to capture high-level normative elements. Activity diagram (Fig. 2) in itself is only sufficed to model at a superficial-level to identify the existence of high-level norms. Scenario diagram shows the existence of objects, their relationships in the logical view of the system, and how they execute a particular scenario or use-case. Let's look at a scenario (Fig. 3, 4) and how we can document it with the Activity diagram (Fig.2) crime reporting process. Norm An1alysis offers an extension to this perspective to capture norms, which is present but hidden.
Process (Activities) Appraise Incident (will lead to 3 possible outcomes Refer to fig.2)
Norms [Norm 1] Forward: (Sub-Norm 1.1, 1.2, 1.3..) [Norm 2] Accept : (Sub-Norm 2.1, 2.2, 2.3..) [Norm 3] Reject : (Sub-Norm 3.1, 3.2, 3.3..)
[Norm 1 ] : Forward Process [Norm 1.1] Sub-Norm - 1.1: IF the (incident does not constitute to a crime); a case of domestic violence. THEN Crime Reporter Is Obliged To forward it to the family welfare departments. [Norm 1.2] Sub-Norm - 1.2: IF the (incident does not constitute to a crime); a case of customer/consumer disagreement. THEN Crime Reporter Is Obliged
11
To forward it to the consumer association organisation. Sub-Norm3:….. Sub-Norm4:….. [Norm 2 ] : Accept Process [Norm 2.1] Sub-Norm - 2.1: IF the (incident constitute to a crime); robbery committed. THEN Crime Reporter Is Permitted To take the statement and issue incident report serial number beginning with ROBxxxxx [Norm 2.2] Sub-Norm - 2.2: IF the (incident constitute to a crime); hit and run accident. THEN Crime Reporter Is Permitted To take the statement and issue incident report serial number beginning with HNRxxxxx Sub-Norm3:….. Sub-Norm4:….. Norm Analysis will attempt to formalise these unregulated normative behaviour (Liu et al. 2001) and agents. Many observable and obscure norms will be defined at this stage and the predicative behaviour analysed. Reusability of norm is achievable only when the modelled organisation is equivalent to the real system environment. NA information table clearly helps to state and structure responsible agents, key data, pre-post triggers and detailed behaviour norms. Appraise Incident: Process
1
Responsibility
Officer in contact with public
Info. Identification
Victim/witness info details, incident (date/time), etc..
Trigger
Public contact
(Conditions Pre/Post)
Report incident
Detailed Norm
WHEN the public made an incident report to the police IF the incident is valid and within the scope of police THEN the officer in contact OBLIGED to accept and filed an incident report ELSE selection process
Incident enquire
Table 2: Norm Analysis (Incident Process)
12
Determine Incident: Selection (Reject)
N3
Responsibility
Crime Reporter
Info. Identification
Missing person > 24hrs, family dispute, policies, etc..
Trigger
Incident Appraisal
(Conditions Pre/Post)
Incident appraisal
Detailed Norm
WHEN incident report is gathered IF the incident is invalid and outside the scope of police THEN the officer in contact OBLIGED to reject the incident
Incident rejected
Table 3: Norm Analysis (Determine Incident - Reject) N2
Crime Statement : Selection (Accepts) Responsibility
Crime Reporter
Info. Identification
Type of crime, category, urgency
Trigger
Determine Incident
(Conditions Pre/Post)
Valid Crime
Detailed Norm
WHEN a crime is confirmed IF it is urgent that it will be assign immediately THEN the officer in contact is to investigate
Statement report
Legend: selection
process
process #
N#
Table 4: Norm Analysis (Crime Statement - Accepts) Norm Analysis, present an unambiguous detailed description of the interrelated entities and its possible behaviour, which is structured on information collated from the high level RAD and Activity diagram. The results attained are abstract, yet comprehensive in its procedural logic, which considers entities behaviour, norms and agents. This allows the flexibility to model the very essences of complex organisation systems. Equipped with this information the modeller will have an invaluable glimpse of a significant insight in the communication network of the organisation.
6. Conclusion and Further work. At this juncture, it is evident that systems information engineering is not sufficiently equipped to model organisational behaviour, and may require a revisit and extension to the fundamental field of requirement engineering. The above case study illustrated the significances of ontological dependencies, affordances and normative agents on an organisational business domain. These techniques are at its infancy, but it has nevertheless gained widespread attention in many fields agreeing for a need to address the organisation semantic layers. However, having said that, it has the ability to capture
13
rigorous system behaviour and model the complexity of modern business organisation. The next phase of systems study is to define, regulate and model norms, as well as to develop the interoperability of agents. This approach would structure on behavioural norms for systems optimisation and reuse. It is with the aim of bring requirement engineering a step closer to realising the above objective with the advent of tools as, semantic analysis and Norm Analysis; more effort however, needs to be done in the following areas: 1) Formulising norms for applications in software simulation. 2) Validation of norms; and 3) Elicitation of high-level abstract organisational norms for system reuses. We are aware of the current norms systems limitations, attributed to the complex and unstructured autonomous behaviour of entities. Equipped with the semiotics tools, modellers at present will still have to rely largely on individual experiences and thorough understanding of the specific organisational domain. We aim to improve future work and formulise the current requirement system technique, unifying system requirement design with Norm Analysis, with further enhancement.
Acknowledgement This research is partly supported by EPSRC – SEDITA project GR/S04840/01.
7. Reference: Carlos, Ivan (2000) Logical Models of Argument. ACM Computing Survey Vol 32, No. 4. pp. 337-383. Chong, S. & Liu, K. (2000) A Semiotic Approach for Modelling and Designing AgentBased Information Systems Based on Roles and Norms. Proceedings of AOIS-2000 at CAiSE*00. iCue Publishing, Berlin, pp. 552-567 Liu, K. (2000), Semiotics in Information Systems Engineering, Cambridge University Press, Cambridge (UK). Liu, K, Alderson, Qureshi, Z (1999). Requirement Recovery from Legacy Systems by Analysing and Modelling Behaviour. International Conference on Software Maintenance Oxford, England (UK), pp. 3-12. Liu, K. Andy Salter (2002) Using Semantic Analysis And Norm Analysis To Model Organisations. Proceedings of ICEIS 4th International Conference On Enterprise Information Systems Ciudad Real - Spain 2002. pp. 847-850 Liu, K, Ong Tina, (1999) A Modelling Approach For Handling Business Rules And Exceptions. The Computer Journal, Vol 42, No.3. 221-231
14
Liu, Kecheng, Lily Sun, Alan Dix & Mohan Narasipuram (2001). Norm Based Agency for Designing Collaborative Information Systems, Information Systems Journal, 11, 229247. Filipe, Joaquim, Kecheng Liu & Bernadette Sharp (2000) Organisation Simulation Based on Normative Knowledge and Role Modelling, AAAI Spring Symposium workshop "Bringing Knowledge to Business Processes", Technical Report SS-00-03, Stanford, March 2000, pp.136-138. Joaquim, F., Liu, K. & Sharp, B. (1999), “A Semiotic Approach to Organisational Role Modelling for Intelligent Agents”, Kluwer Academic Publishers, The Netherlands. Pp.187-211 Martin Fowler, Ken Scott (1999), UML Distilled Second Edition A Brief Guide to the Standard Object Modelling Language, Addison-Wesley, Reading, Massachusetts. Ould Martyn A.Wiley (1995). Business Processes, Modelling and Analysis for Reengineering and Improvement. John Wiley & Sons, Chichester. Peckham, J. & Mryanski, F. (1988), Semantic Data Models, ACM Computing Surveys 20(3). Vol. 20, Issue 3. pp. 153-189 Sergot Marek (2001), A computational theory of normative position. ACM transaction on computational logic. Vol. 2, Issue 4. pp. 581-622 Stamper, R., Liu, K., Hafkamp, M. & Ades, Y. (2000). Understanding the Roles of Signs and Norms in Organizations – A Semiotic Approach to Information Design. Journal of Behaviour and Information Technology. Vol.19, No.1, pp. 15-27 Wand, Y., Storey, V.C. & Weber, R. (1999), An ontological analysis of the relationship construct in conceptual modelling, ACM Transactions on database systems. Vol. 24, Issue 4. pp. 494-528
15
