Integrated Approach for Modelling of Semantic and ... - Semantic Scholar

1 downloads 1690 Views 117KB Size Report
If an actor has a social power to activate an action that changes a situation from the .... Some methods call this process of change as strengthening or .... IOS Press, ... Proc. of the second Int. conference on the Design of Cooperative Systems, ...
Integrated Approach for Modelling of Semantic and Pragmatic Dependencies of Information Systems Remigijus Gustas

Department of Information Technology, University of Karlstad S-651 88 Karlstad, Sweden [email protected]

Abstract. Traditional semantic models are based on entity notations provided by several links. Links are established to capture semantic detail about relationships among concepts. The ability to describe a process in a clear and sufficiently rich way is acknowledged as crucial to conceptual modelling. Current workflow models used in business process re-engineering offer limited analytical capabilities. Entity-Relationship models and Data Flow Diagrams are closer to the technical system development stage and, therefore, they do not capture organisational aspects. Although object-oriented models are quite comprehensible for users, they are not provided by rules of reasoning and complete integration between static and dynamic diagrams. The ultimate objective of this paper is to introduce principles of integration for different classes of semantic and pragmatic representations.

1 Introduction Any information system activity needs to be defined in a context of organisational processes. Thus, two levels of information system models are necessary [5]. Organisational level model that defines an ideal system structure. Further it will be referred to as an enterprise model. The implementation level determines data processing needs for a specific application. Most conventional semantic data models are heavily centred around the implementation level. The way in which organisational activity is conceptualised will define what information system is appropriate. Activities in an organisational system can be expressed in terms of actions of communication and collaboration between actors of an information system. This kind of knowledge is crucial to reason about purposeful implications of organisational processes. Most semantic models that have been used in traditional information system modelling approaches neglect such essential aspects of communication. Enterprise engineering is a branch of requirements engineering which deals with an early phase of integrated information system development. At the same time it can be viewed as an extension and generalisation of system analysis activity. Enterprise

2

Remigijus Gustas

modelling takes place in the early and middle phases of information system development life cycle. The most difficult part of it is arriving at complete, integrated and consistent description of a new system that sometimes is known as conceptual, semantic or requirements model. Despite of apparent clarity of the semantic models used by various Object – Oriented methods and CASE tools, the research has shown that a large part of the maintenance costs can be attributed to improper enterprise modelling or to misconception of real requirements [15]. Various graphical diagrams [19] are used to define semantics of information systems. It is obvious that all notations have been designed to describe one or a few, but not all aspects of information systems. This means that information system models should comprise a combination of several notations, each for some particular aspect. This may lead to a difficult question: ‘how is it possible to use several notations in a complimentary way to develop clear models?’ [14]. More importantly, it is often possible to employ a notation to describe some other aspects than those it has been designed to use for. The solution could be found in the identification of a set of the basic semantic and pragmatic modelling primitives that are adequate to analyse static and dynamic aspects of processes. In this study we will present and analyse a set of abstractions that can be considered as a necessary basis to built an integrated enterprise model. The focus of this approach is on modelling primitives which are not only are taking into account semantic models of traditional approaches, but also put a communication aspect into the foreground of information system modelling. Integration of semantic, pragmatic and non-traditional communication dependencies is considered as a most important feature of the suggested framework. Such enterprise models can be useful for the purpose of understanding and reasoning that is critical to the success of conceptual engineering activity in many areas.

2 Pragmatic Dependencies Starting point in a business process re-engineering research, are initial requirement statements that express the wishes of stakeholders about new organisation of system. These initial requirements are usually presented as a natural language text that is often ambiguous, incomplete and inconsistent. Although processes of information system adaptation can be driven by these pragmatic statements, usually traditional semantic models are not taking into account dependencies between activities and goals. Moreover interdependencies between goal models and process models are usually defined in a very fuzzy way. Some requirement engineering methodologies have already identified the problem of making system requirements precise, unambiguous, complete and consistent. The process of bridging goals to information system specifications was entitled 'from fuzzy to formal' [7]. Predominance of fuzzy thinking in a goal modelling has led to serious lack of interaction between semantic and pragmatic descriptions of processes. More often, a process goal is merely postulated rather than expressed in terms of semantic diagrams.

Integrated Approach for Modeling of Semantic and Pragmatic Dependencies

3

Such ignorance of a goal power was recognised in the area of cognitive modelling. Goals are usually understood as states of affairs or situations that should be reached or at least striven for. Situations are resulting from the actions [6]. Goals can be defined as desirable situations that are interpreted by an actor as final. Such pragmatic notions as objective, vision, goal, etc. express the wishes and desires of actors concerning system they design or manage. A goal hierarchy can be formed of interconnected goals on different levels of abstraction ranging from high-level business objectives to low-level operational goals. Usually, objectives at the bottom level are situations that can be defined in terms of various semantic dependencies. On neighbouring levels of decomposition, goals are related by the composition dependency. The opposite of a goal is a problem. A problem describes a situation which is not desirable. The notion of a problem is used to refer to a problematic situation of an actor. Semantic specification of the problem is regarded as a part of the actual specification. This means that the problem can not be identified without stating the goal. If the designer has no predefined goal then the problem does not make sense [9]. Usually problematic situations denote restrictions that actors try to avoid. A pragmatic link between an actor and desired situation is entitled to as a goal dependency ( g ). The problem link ( p ) are used to refer a problematic situation of an actor. The goal and problem dependencies can be used to refer desirable or not desirable states or situations. In the following chapters, two pragmatic dependencies of influence between goals will be formally defined. They are entitled to as negative influence dependency ( - ) and positive influence dependency ( + ). The negative influence dependency from A to B (A - B) indicates that the goal A hinders to the achievement of goal B. The positive influence dependencies ( + ) between two goals means that the achievement of the first goal, would contribute to the achievement of the second. The negative and positive influences between goals are imposed by the conflicting interests of actors.

3 Semantic Dependencies Most of the semantic modelling techniques are based on entity notations provided by several links. Links are established to capture semantic detail about static and dynamic relationships. Typically semantic constraints have to be general enough to specify dependencies of a system in the different perspectives such as the "why", "what", "who", "where", "when" and "how" [16]. Semantic constraints can be described by using intensional and extensional [13] dependencies of various kinds. Semantics of static intensional dependencies can be defined as cardinalities represented by minimum and maximum number of individuals of concepts. Extensional dependencies usually specify constrains between classes and instances. Static dependencies of concepts stem from various semantic data models. Graphical notations of several associations in Martin/Odell’s style [13] are represented in Fig. 2.1.

4

Remigijus Gustas a)

A

B

(0 ,1 ;? ,? )

b)

A

B

(1 ,1 ; ? ,? )

c)

A

B

(0 ,* ;? ,? )

d)

A

B

(1 ,* ; ? ,? )

Fig. 2.1. Graphical notation of cardinality constraints. Note: the meaning of * is ‘many’ (i.e. more than one) and the meaning of ?is ‘not defined’.

Notations that are commonly used at the initial phase of concept modelling have to provide a clear understanding of cardinality constraints in both directions. The most common static dependencies that may be specified between any two concepts A and B are as follows: • (1,1;0,1) - Injection dependency which will be denoted by A ⇒ B , • (1,1;1,1) - Bijection dependency (A B), • (1,1;0,*) - Total Functional dependency (A B), • (1,*;1,1) - Surjection dependency (A B), • (1,*;0,1) - Surjective partial functional dependency (A ⇒> B ), • (1,*;1,*) - Mutual multivalued dependency (A B), • (1,*;0,*) - Total Multivalued dependency (A B), • (0,1;0,1) - Partial injectional dependency, (A |⇒ B ) , B), • (0,1;0,*) - Functional (partial) dependency (A • (0,*;0,*) - Multivalued (partial) dependency (A B). Many concepts have common constraints. The similarities can be shared between concepts by extracting and attaching them to a more general concept. In such a way, similar constraints can be inherited by several concepts. One way to represent generalisation hierarchies is by using of inheritance dependency. It will be denoted by a ). By means of inheritance similarities of concepts are shown. solid line arrow ( Aggregation is a conceptual operation which is useful for the formation of a concept interpreted as a whole from other concepts that may be viewed as component ). In the area parts. Aggregation can be specified by a composition dependency ( of artificial intelligence, sometimes, composition is referred to as a ‘part of’ relation [17]. Semantics of the composition link can be completed by cardinality constraints. Most behavioural diagrams put into foreground a dynamic link, which is very similar to the state transition in a finite state machine. Such transition link, constitute a modelling basis of various object-oriented diagrams that are used for specification of object behaviour. In our approach, a state transition is defined in terms of two states. If two states are connected by the transition dependency, then it means that by the action an object can be transferred from the actual state to the next state. Actual static constraints define a set of conditions for an object in the current state. The expected state

Integrated Approach for Modeling of Semantic and Pragmatic Dependencies

5

defines a set of conditions for an object in the desired state [11]. The graphical illustration of the transition dependency is presented in Fig. 2.2. ACTUAL STATE

ACTION

NEW STATE

Fig. 2.2. Transition dependency between actual and desired situation. Actions are represented by ellipse.

States are resulting from the actions [6]. Specification of actual and desired states is crucial to the understanding of action semantics. Any state transition dependency indicates a possibility to change a state, and visa versa, a possibility to accomplish the action can be specified by a state transition dependency, i.e. (NEW STATE). (ACTUAL STATE) Communication dependencies between two actors involved in a particular action, describe the "who" perspective. Such dependency link between two actors (agent and recipient) indicates that one actor depends on the other for some flow. The agent can be any actor who is able to send a flow, for example, individual, group, role, position, organisation, machine, information system, etc. Graphical notation of the flow deFLOW RECIPIENT) is pendency between an agent and recipient (AGENT represented in Fig.2.3. FLO W AGENT

R E C I P IE N T

Fig. 2.3. Flow dependency

The flow dependency represents a transfer of the ownership right for a particular object of FLOW. Before sending the flow, it is owned by an AGENT and later, dependent on whether a flow would be accepted or not, the ownership is transferred to a RECIPIENT. The flows can be decision, information or material. Recipients by depending on agents are able to achieve their goals.

4 Interaction between Static and Behavioural Dependencies Any flow dependency between two actors may imply a communicative action as well. It is then considered to be an action and a communication flow. It should be noted, that many approaches in the area of business process re-engineering do not view actions in two different perspectives [8]. Cohesion of action and flow results into a more complex abstraction. Therefore, the flow dependency link between two actors specifies that a recipient depends on an agent not just only by a specific flow, but also by action. ) Actors are specific sub-systems of the overall system. The semantic link ( ACTION) indicates that the action can be initiated from actor to action (ACTOR

6

Remigijus Gustas

by any individual, which belongs to the class ACTOR. The presented dependency may define co-ordination, decision, control, etc. The dependency link from action to actor, ACTOR, means that an actor will be affected by the executed aci.e. ACTION tion. Often this dependency link is combined with the flow that is desired by a dependent actor. Graphical notation of the communicative action dependency is represented in Fig.4.1. A CTIO N

AG E NT

FLO W RE CIPIEN T

Fig. 4.1. Communicative action dependency between two actors

Underlying concepts and dependencies play an important role in various business modelling approaches which are based on communication [21]. A typical action workflow loop can be defined in terms of two communicative action dependencies. Sequences of communicative actions in workflow models [1] may serve as a basis to define obligations, authorisations and contracts [20]. Example of a typical workflow loop that is defined in terms of two communicative actions into opposite directions is represented in Fig. 4.2. ORDER

ORDER

CUSTOMER

SUPPLIER ITEM

SUPPLY

Fig. 4.2. Action workflow loop between Customer and Supplier

This graphical example shows that a customer is authorised to send an order to a supplier by using the predefined ordering action. If this order would be accepted then a supplier is obliged to supply an item. The supplier is also responsible to follow a contract, which can be defined in terms of relationships between incoming (ORDER) and outcoming flows (ITEM). A contract in the presented example can be defined as folORDER SUPPLIER then lows: If CUSTOMER SUPPLIER ITEM CUSTOMER. An agent carries out a specific action in order to achieve a predefined state. Existence of object x in some state would also impose the fulfilment of a set of static dependencies between states. Two kinds of changes such as disconnection and connection occur concerning the associations of an object during a transition of the object from one state to another. A disconnection removes an existing association from existence and connection adds a new association. The disconnected and connected associations are represented by entirely different relationships of two states. The definition of such noteworthy difference in a current and new state of action is very important to understand the nature of action. Only those processes, that conclude with a state change expressed by the disconnection or connection event, can be interpreted as ac-

Integrated Approach for Modeling of Semantic and Pragmatic Dependencies

7

tions. For instance, a graphical example of the noteworthy difference among three different states, that are important to understand actions of customer and supplier, is illustrated by Fig. 4.3. Product Not O rdered Product Cust om er Product Needs to be O rdered

O rdered Product

O rder

Item

O rder

Sup plier

Product Needs not to be O rdered

Not Supplied Item

Supply

Supplied Item

Item

Fig. 4.3. Constraints to Order and to Supply. Note: inheritance ( ) in this paper is used in a non traditional way. Strict definition of the inheritance dependency is presented in chapter 7.

A transition from the state ‘Product Needs to be Ordered’ to the state ‘Ordered Product’ can be performed by the action of Ordering. Existence of an object x in the state ‘Ordered Product’ would imply that it may have one or many ‘Not Supplied Item’ associated to the x Product. The noteworthy changes performed by actions are important for actors, because they create a need to react. The reaction mechanism is represented in terms of communication flows. For instance, if ‘Product Needs to be Ordered’ then a Customer has to react appropriately. In this particular situation, a Customer is supposed to send an Order to a Supplier. If an item is ‘Not Supplied Item’ then the Supplier is supposed to Supply an Item to the Customer. The semantic difference between two states ‘Not Supplied Item’ and ‘Supplied Item’ is defined in terms of two semantic links (see ‘Ordered Product’ and ‘Product Needs not to be Ordered’).

5 Dependencies in an Extended Action Workflow Loop A typical action workflow loop include two communication flows sent into opposite directions. The customer is an actor who initiates the work flow loop to achieve his goal. The receiver of a flow is a performer. Flow dependencies in two opposite directions imply that certain relationships are established between two actors. In the reality it represents either a commitment or a contract [8] between customer and performer.

8

Remigijus Gustas

Any business process defines a set of responsibilities as well as set of requests that a customer can ask a performer. Usually, the customer is an actor who initiates the action in order to achieve his goal that is referred by a desired situation (DS2). The goal corresponds to a final situation in a communicative action workflow loop. By using a flow in the forward direction, a customer is asking a performer for some action. If the request corresponds to a contract, it will always create a situation that is an opportunity for a performer. Pragmatic dependencies are represented graphically in Fig.5.1. p

Cust omer

o Problematic Situation (PS1)

Action of Customer

Desired Situation (DS1)

g Flow 2

Flow 1

Problematic Situation (PS2)

p

Perfor mer

o Action of Supplier

Desired Situation (DS2)

Fig. 5.1. Graphical representation of pragmatic dependencies

A pragmatic link between an actor and his desired situation is entitled to as a goal dependency ( g ). The problem link ( p ) is used to refer a problematic situation of an actor. An opportunity link ( o ) is referring to an intermediate situation between a problematic and desired. If an actor has a social power to activate an action that changes a situation from the problematic to intermediate, then this intermediate situation for some other actor may help to create new desirable situations. According to the presented schema, a customer has a possibility to initiate the action by sending Flow 1 to a performer in order to avoid a problem denoted by a problematic situation (Customer p PS1). If a performer is satisfied by the flow (it will be accepted) then the problematic situation would be replaced by action of customer to a desired situation (DS1). In the next step, a performer by sending Flow 2 has a possibility to change his problematic situation (PS2) to the desired situation (DS2) that is regarded to as a goal of customer (Customer g DS2). Graphical example of the semantic and pragmatic dependencies in a typical action workflow loop is depicted in Fig. 5.2.

9

Integrated Approach for Modeling of Semantic and Pragmatic Dependencies

p

Cust omer

o Not Ordered Product

Item

Order Ordered Product

g

Order

Available Item

p

Sup plier

o Supplied Item Supply

Fig. 5.2. Example of dependencies in an action workflow loop

Satisfaction of actors is closely related to their goals and problems. In order to activate an action, an agent has to know about the opportunities available to a recipient. If a recipient is viewing the intermediate situation as an opportunity to achieve his goal or to avoid problem, then the flow, that has been sent by an agent, would be acceptable by a recipient.

6 Dependencies of Positive and Negative Influence Any two goals can be contradictory, if one of them is interpreted as a problem to reach another goal. Contradictory goal can influence negatively or hinder the achievement of a desirable situation. This means that interpretation of goal and problem is relative and dependent on the actor objectives. The same situation can be interpreted as a goal for one actor or as a problem for another. The negative influence dependency has been introduced in F3 [7] to specify contradictions between goals. The negative influence dependency from A to B (A - B) indicates that goal A hinders to the achievement of goal B. Conditions for the existence of a negative influence dependency between two situations are as follows: if ACTOR p S1 and ACTOR g S4 then S1 - S4 . This axiom specifies that a problematic situation (S1) hinders to the achievement of a desired situation (S4). Moreover, in a context of an opportunity (S2), the following axiom is true: if ACTOR p S1 , ACTOR o S2 and ACTOR g S4 then S2 - S1, S2 + S4. According to this definition, an opportunity must influence negatively a problematic situation and influence positively a desired situation. The positive influence dependency between two goals means that the achievement of one goal would contribute to the achievement of the second. It should be noted that any problem that hinders to a problematic situation may be considered as an opportu-

10

Remigijus Gustas

nity for some other actor, i.e. if ACTOR p S3 and S2 - S3 then ACTOR o S2. Any opportunity that influences positively a desired situation may be considered as an opportunity as well, i.e. if ACTOR g S5 and S4 + S5 then ACTOR o S4. The negative influence dependency is useful to express contradictions between goals of various actors. If one of the goals hinders to the achievement of the other, then these goals are in conflict. The positive influence dependency between two goals is indicating that the achievement of one goal helps to achieve the second. It should be noted that the additional pragmatic dependencies are derived according to the following inference rules: if A +

B , B - C then A - C ,

if A - B , B - C then A + C , if A +

B , B + C then A + C ,

if A B then A + B . Here: is a goal composition dependency [12]. Semantic and pragmatic dependencies in an action workflow loop are very important to analyse viability of business processes. It should be noted that viability of a single communicative action dependency between two actors guaranties that the desired situations create new possibilities for the recipients of flows. The customer tries to initiate an action, because he wants to avoid a problematic situation or to achieve a new desired situation. By depending on a performer, a customer is able to achieve a situation that can not be reached without involvement of a specific performer. At the same time, if the performer fails to deliver the flow to a customer, then customer becomes vulnerable to the failure. The negative and positive dependencies between various situations are imposed by different intentions of actors. If influences in the action workflow loop are incompatible, i.e. A + B and A - B, then this situation is referred to as a contradiction. By using a set of influences between situations from the point of view of different actors, the contradictory goals can be identified. It is not so difficult to see in the example of the previous chapter that pragmatic dependencies are not contradictory. An overall set of semantic and pragmatic dependencies of a particular process constitute a formal basis for inconsistency analysis. Inconsistencies may be eliminated either through negotiation among actors, and by disregarding some of the goals or by disregarding some of the actors. In a case of existence of a cost effective development action, inconsistency among goals may serve as a driving force for business process re-engineering. Inconsistencies between actor goals in a context of the same process would mean that one of the actors is vulnerable to the failure in the achievement of his goal. Consistency of pragmatic dependencies in the action workflow loop guaranties that interests of two actors are not conflicting. If interests of actors are in conflict, then the action workflow loop may be not viable. Viability of action workflow loops can be studied in terms of semantically complete diagrams.

Integrated Approach for Modeling of Semantic and Pragmatic Dependencies

11

7 Semantic Incompleteness of Diagrams The way normally people analyze systems is by reasoning on a basis of the model for a particular part of the system. Many analysts in the area of information system development define their systems in terms of initial conceptual models, and later extend them by making a whole lot of assumptions. Very often these conceptual models are quite vague, because of several reasons. Sometimes, dependencies of conceptual diagrams may not be defined strictly and can be interpreted ambiguously. Even if the diagrams developed by experts are presented in a formal way, the system model may be still not clear enough. This can take place for the reason that the description of system is incomplete. Elimination of semantic incompleteness in a diagram, by the refinement of relations among concepts is important, if system analyst wish to reason automatically about the expected scenarios, contingent actions and opportunities available in a particular business process. An applicable set of dependencies allows us to avoid semantic holes [10]. In this study about conceptual dependencies, we concentrate on a particular subset of semantic links, which is entitled to as totally applicable dependencies. Applicability of dependencies can be achieved through appropriate transformations of concept diagrams. Some methods call this process of change as strengthening or restricting [2], [3]. In object-oriented approach, the transformation process is entitled to as sharpening meaning of concepts [13]. It improves the ability to understand and communicate conceptual models. Transformations of diagrams have been mostly studied in the context of the static dependencies. In this approach, we have introduced the common basis to deal with both static and behavioural parts of representations. It means that semantic transformations can take into account both aspects of the information system description. Semantic relationships of information system can be specified by using two kinds of abstractions: aggregation and generalisation. Abstraction of aggregation is based on the presented set of the totally applicable binary dependencies. These links are as follows: ⇒ is the injection dependency, ⇒> is the surjective partial functional deis the total functional dependency, is the composition dependency and is the communication dependency and is the transition dependpendency, ency. These dependencies will be referred to as basic. ). InThe generalisation abstraction is based on the inheritance dependency ( heritance links can be of various kinds. Although the inheritance constructs are wellunderstood, there is no complete agreement of an interplay between inheritance dependency and other types of basic constraints of semantic models. For example, some of researchers understand the inheritance in the same way as the inclusion dependency. To eliminate ambiguity, we will define the inheritance dependency in terms of the presented basic constraints. Let Ad be a set of static and dynamic semantic links that are specified for the dependent from A concepts. The inheritance dependency from concept A to B is defined B if and only if A ⊆ B and Bd ⊆ Ad. by A The inheritance is characterised by the following axioms: 1) if A B,B C then A C,

12

Remigijus Gustas

2) if A B , B ⇒ C then A ⇒ C , B,B C then A C, 3) if A 4) if A B,B C then A C, 5) if A B , B ⇒> C then A ⇒> C , 6) if A B, B C then A C, 7) if A B, B C then A C. Inheritance is defined in terms of two abstractions: intensional and extensional. If A inherits B, then the structure (intension) of concept B must be included as part of concept A intensional structure. Definition of inheritance in the extensional sense is deC then x ∈ C. fined as follows: if x ∈ B, B It should be noted, that the presented definition is more general than that is assumed in object-oriented approach. For instance, concepts A and B can be interpreted as categories of states and actors. Any communicative action is defined unambiguously if and only if it is expressed in terms of applicable constraints. Two states of the semantically unambiguous action must be connected to other concepts by the total static dependencies. As far as actor communication links are concerned, the flow dependency from an agent to recipient is considered as an applicable constraint as well. The same condition holds for a current state that is linked by a transition dependency to the new state. It means that any object, which belongs to the current state, is applicable for a specified transition link. The presented set of totally applicable dependencies is useful to assess the semantic ambiguity [10] of specifications and to reason about a particular part of the system. Very often information system specifications are quite vague, because some of the semantic dependencies are optional. If the diagrams can be defined in terms of basic dependencies, then certain formal inference rules may be applied to derive additional semantic links that can be used to check semantic consistency of information system requirements at the enterprise level.

8 Conclusions Interplay between semantic and pragmatic dependencies lie in the foreground of the suggested framework. In this approach, we have shown how to bring various semantic diagrams together and to combine insights from the point of view of different goals. An attempt of the generic framework is to bridge goals of various actors and a way various business processes are described. Goals and desires of actors constitute an important part of knowledge about business processes. The intentional relationships among actors are viewed with potentially common and conflicting interests. The presented pragmatic dependencies can explain freedom of a specific actor and the extent to which actors are exposed to a danger. The usefulness of great number of semantic dependencies in the area of information system design is an open problem. For instance, many dependencies introduced in database theory encounter problems with missing attribute values. These problems result from the fact that either the instance of the attribute is temporally unknown, but

Integrated Approach for Modeling of Semantic and Pragmatic Dependencies

13

applicable, or that its value can never be known, because the attribute is not applicable for a specific instance [4]. For the unambiguous definition of concepts, only applicable dependencies can be used. Despite of strictness of this requirement, it allows us to discover contingent actions in the semantic diagrams and to introduce rules of reasoning [10] at the enterprise level. The presented actor dependencies constitute a unified basis for modelling of dynamic relationships and can be regarded as an extension and integration of the state transition and interaction diagrams. Actor goal dependencies constitute a unified basis for modelling of pragmatic relations that are able to define actor intentions. Goals justify and explain the presence of the semantic dependencies, which are used to specify components of information system requirements. Thus, such integrated approach offers a novel perspective on semantic analysis of information systems. The main difference between this framework and Yu & Mylopoulos approach [18] is that any actor dependency may be considered at the same time to be the action and the goal dependency. The action dependency is defined in terms of state transition and flow dependencies. It has also been shown how the negative and positive influence dependencies, introduced in F3 [7], can be formally defined in terms of the semantic and pragmatic dependencies. There is a growing interest to integrate information system development methodologies from different areas such as requirements engineering, method engineering, workflow management, business process modelling, object – oriented approach, etc. Many system analysts recognise that it is not enough to describe semantics of information system by concentrating distinctly on one of the methods. When re-engineering information systems most of models tend to neglect communication aspects among several actors of organisation. Dependencies of communication and co-operation between actors and their actions describe a very important part of knowledge about business processes. Unfortunately, communication approaches often neglect some behavioural aspects of system modelling that are basic in information system engineering. The presented set of totally applicable dependencies can be viewed as an integrated semantic modelling technique to specify the deeper structures of relationships among concepts. Suggested generic framework focuses on the modelling of static and dynamic constraints, where several actors co-operate to achieve new desired states. The basic dependencies provide a uniform formal basis in the area of concurrent business process modelling, analysis and integration. The purpose of such introduced basis is that eventually information system diagrams can be used as a tool to assist reasoning and validate enterprise models before they are implemented.

References 1. Action Technologies. Action Workflow Analysis Users Guide. Action Technologies, 1993. 2. A T Borgida. Generalisation/Specialisation as a Basis for Software Specification. On Conceptual Modelling, M Brodie, J Mylopoulos, J W Schmidt (eds.), Springer-Verlag, New York, 1984, pp.87-112.

14

Remigijus Gustas

3. R Brachman, J G Schmolze. An Overview of the KLONE Knowledge Representation System. Cognitive science, 9(2), pp. 171-212, 1985. 4. E F Codd. The Relational Model for Database Management. Addison-Wesley Publ. Co., 1990. 5. G B Davis, M Olson. Management Information Systems. McGraw Hill, New York, 1985. 6. E D Falkenberg et al. A Framework of Information System Concepts. The Report of the IFIP WG8.1 Task Group FRISCO, 1996. 7. F3 Consortium. ’F3 Reference Manual (Esprit III Project 6612)’, SISU, Kista, Sweden, 1994. 8. G Goldkuhl. Information as Action and Communication. The Infological Equation, Goteborg University, Sweden, pp. 63-79, 1995. 9. R Gustas, J Bubenko jr., B Wangler. Goal Driven Enterprise Modelling: Bridging Pragmatic and Semantic Descriptions of Information Systems. Information Modelling and Knowledge Bases VII, Y Tanaka, H Kangassalo, H Jaakola, A Yamamoto (eds.), IOS Press, 1996 , pp. 73 -91. 10. R Gustas. Towards Understanding and Formal Definition of Conceptual Constraints. Proc. of the European-Japanese seminar on Information Modelling and Knowledge Bases VI, 1994, IOS Press, pp. 381-399. 11. R Gustas. A Basis for Integration within Enterprise Modelling. Second Int. Conference on Concurrent Engineering: Research and Applications, Washington, DC Area, August 23-25, 1995, pp. 107-120. 12. R Gustas. A Framework for Description of Pragmatic Dependencies and Inconsistency of Goals. Proc. of the second Int. conference on the Design of Cooperative Systems, June 1214, 1996, Juan-Les-Pins, France, pp. 625-643. 13. J Martin, J J Odell. Object-Oriented Methods: Foundation. Prentice-Hall, New Jersey, 1995. 14. W E Riddle. Fundamental Process Modelling Concepts. NSF Workshop on Workflow and Process Automation in Information Systems, May 8-10, 1996. 15. K Siau, Y Wand, I Benbasat. The Relative Importance of Structural Constraints and Surface Semantics in Information Modelling. Information Systems, Vol. 22, No 2/3, pp 155-170, 1997. 16. J F Sowa, J A Zachman. Extending and Formalizing the Framework for Information Systems Architecture. IBM Systems Journal, 31(3), pp. 590 - 616, 1992. 17. V C Storey. Understanding Semantic Relationships. VLDB Journal, F Marianski (ed.), Vol.2, pp.455-487, 1993. 18. E Yu, J Mylopoulos. from E-R to ’A-R’ - Modelling Strategic Actor Relationships for Business Process Reengineering. 13th Int. Conf. on the Entity - Relationship Approach, P Loucopoulos (ed.), Manchester, U.K., 1994. 19. E Yourdon. Modern Structured Analysis, Prentice-Hall, Englewood Cliffs, N.J., 1989. 20. H Weigand, E Verharen, F Dignum. Dynamic Business Models as a basis for Interoperable Transaction Design. Information Systems, Vol. 22, No 2/3, pp 139-154, 1997. 21. T Winograd, F Flores. Understanding Computers and Cognition: A New Foundation for Design. Ablex Norwood, NJ, 1986.