ENHANCING UML CONCEPTUAL MODELING THROUGH THE USE OF VIRTUAL REALITY

Transcription

1 ENHANCING UML CONCEPTUAL MODELING THROUGH THE USE OF VIRTUAL REALITY Bin Zhang and Ye-sho Chen Louisiana State University {bzhang1, Abstract UML is one of the major conceptual modeling tools in Object-Oriented software engineering. However, recently more and more concerns arise from the modeling quality of UML. Among them, limitation of the two-dimensional presentation of its notations is reported to be significant. By bringing in the concepts of Steoropsis (3-D appearance), animation, interaction and immersion, Virtual Reality may improve the UML modeling quality. Specifically, in this paper we analyze the modeling quality of UML by using recent published literature in conceptual modeling. We then recommend the use of Virtual Reality to enhance the modeling quality of UML. Some examples of how Virtual Reality can be used to improve UML modeling quality are given. Finally, we propose a research framework for enhancing the quality of UML conceptual modeling through the use of Virtual Reality. 1. Introduction Unified Modeling Language (UML) has become one of the major modeling tools in the software engineering industry. It has profoundly changed the way software systems are designed and developed [19]. Because UML has the advantages of easier designing, larger reusability, and higher quality [20], it has been widely accepted and even been adopted by Object Management Group as Object-Oriented (OO) conceptual modeling standard since However, researchers and software designers has realized insufficiency of UML in expression ability eventually, since it is restricted in two-dimensional plane. These insufficiencies include lack of dynamic expression and interaction ability between groups of remote designers [5, 8, 11]. UML still has its limitations such as complexity, inconsistent semantics, and ambiguous construct [3]. Virtual Reality technology, on the other hand, is becoming increasingly sophisticated and cost effective [16], and could be applied to many areas such as architecture, education, medicine engineering and even Information Systems (IS) research, to simulate a real environment or represent complicated scenarios. The fast development in VR gives us an opportunity to extend the research field to improve the quality of UML. Through VR technology, we can add dynamic and interactive functions in UML. As far as we know, little work has been done in this area. As a prelude to examining the paper, in the next section, we review conceptual modeling using UML in the software requirement analysis phase. We then present an evaluation model for UML conceptual modeling quality. In addition, we introduce several important VR attributes which can be used in enhancing UML. In the third section, several scenarios of how VR could be used to enhance UML are illustrated. In the fourth section, we introduce the research methodology that will be used. Finally, in section five we propose a research framework of new conceptual modeling tool, Virtual Reality Unified Modeling Language (VRUML). 2. Literature review and theoretical foundation All Conceptual modeling is an analytical method to represent both static and dynamic phenomena of the user requirements. Its importance has been widely accepted since late 1960s. In the requirements analysis phase of information system development, Wand and Weber state [17], High-quality conceptual modeling work is important because it facilitates early detection and correction of system development errors. The Standish Group Project, an international consultancy, released a case study report [15] based on more then 30,000 organizations. It stated that for all the software projects developed, only 16% of them succeed. For the rest of 84%, 31% were totally failed, and 53% had cost and time overruns and missing features. According to another report by McConnell [10], the causes of software failure mainly concentrate on: objectives not fully specified (51%) and bad planning and poor estimating (48%). If we can understand customer s requirements more accurately in the requirements analysis and model the business context to facilitate the implementation of software, then we can increase the probability of success greatly. In order to increase the success rate of software, we need to increase the quality of conceptual modeling. Same issue could be applied in Information System development as well. So naturally, we need a standard to evaluate the quality of conceptual modeling. According to Lindland et al. [9], quality of final software products depend to a great extent on the quality of early conceptual 1

2 models. Kung [7] elaborated that a conceptual model should support five features (understandability, expressiveness, processing independence, checkability, and changability) which we believe are the quality properties that conceptual model should possess. We summarize these properties in the table below: Table 1: Conceptual modeling quality properties Feature Description Expressiveness Processing independence Checkability Changability Descriptions of model must be easily and properly understood; The specification of everything in modeling must be in detail. Modeling must be independent of data processing considerations. The evolution of the model must be described. The reference to data and objects must be in detail. These aspects of modeling can be verified: 1. The model describes the application sufficiently and precisely. 2. The model has no discrepancy. 3. The implementation meets the information requirements The components of the model can be easily modified, added to, and removed. In sum, the objective of this conceptual model is to provide a formal representation of the observed reality. The quality properties aim to measure how a conceptual model represents some predefined specification characteristics. We will use this framework to evaluate the UML scenarios in the next section. According to Evermann and Wand [4], UML is an extension of facilitation of conceptual modeling. In the mid 1990's, Grady Booch, Jim Rumbaugh and Ivar Jacobsen invented UML, and in 1997 OMG (Object Management Group) adopted UML as standard of OO modeling. The core of UML is a set of diagrams that can be classified into 3 categories. The first category is a structural diagram that describes the structure of a system. The second category consists of behavior diagrams that describe the dynamic operations of the system. The third category comprises of a set of Model Management s that are used to organize and manage information about the system. According to a recent paper by Walsh and Pawlowski [16], VR is becoming increasingly sophisticated and cost effective, and VR research offers a unique and valuable perspective on IS research community. By using VR technology, we can bring in its properties of immersion, interactivity, presence, and stereopsis technology very easily. A user immersed in a VR environment can be isolated from the real work and is stimulated by the virtual world [21]. Immersion experienced in the virtual world can be of the same quality as our experiences in the real world [23]. On the other hand, the immersion allows the user to take advantage of their stereoscopic vision. Stereopsis, a normal one in VR field, can be a great benefit in disambiguating complex abstract representations [11]. If we display our data structure in 3D space, then it is easier for users to understand the data, and decrease the error rate in identifying path [18]. Interactivity helps users manipulate and control the research context. According to Dede et al. [24], the interactivity produces a profound sensation of being there, thus the learner quickly develops feelings of mastery, especially when first attempting to use an unfamiliar computer systems. Hubona et al. [6], in another research, found users understanding of a model improves when they can manipulate the structure. Some advantages of using VR for conceptual modeling are discussed below and summarized in Table 2: Table 2. Potential advantage of VR in conceptual modeling Conceptual Modeling Criteria VR functionality Effect Expressiveness Changability Checkability Processing independence Immersion Interaction Stereopsis Animation Users are isolated from the real world and are stimulated by the virtual world, thus eliminate ambiguous representation. Users can change the layout of the diagram and adjust the support with reference to architects use of VR systems. Present object in 3-D plane, increase the user s vision ability, thus can help user to figure out mistakes easily. Show the evolution of model, smooth the change of diagram (1) Immersion is defined as the extent to which the subject s senses are isolated from the real world and are stimulated by the virtual world [16]. It involves the representational richness of the mediated environment and the rage of senses employed. Users can eliminate ambiguous representation through immersion property of VR, which meets the and Expressiveness criteria of the conceptual modeling quality properties in Table 1. (2) Because of the interaction property of VR, we can improve the users understanding of data structure and every aspect of the model; in addition, users can adjust 2

3 the components in diagrams through the interaction with the model, which meets the understandability criteria of the conceptual modeling quality properties. (3) Stereopsis of VR is presenting objects in a 3-D plane, thus VR can describe a conceptual model more accurately and decrease the probability of modeling error. Stereopsis not only can decrease ambiguity of complex abstract representation, but also help the viewers to judge relative size of objects and distances between objects [11]. So adding stereopsis in UML can meet the Checkability and criteria of the conceptual modeling quality properties. (4) By using animation to smooth the transition of 2 diagrams, instead of showing 2 diagrams, user can see the evolution progress of the diagrams. take other perspectives than a front perspective or interactively rearrange the selected classes. By using VR, we can make notations more simple, clear, and easier to understand. We can express things that need to be modelled without much effort from the modeler. 3. Potential advantages of virtual reality for UML conceptual modeling In this section, we present two scenarios that show the weakness of current 2-D UML which can be enhanced by VR, the first scenario is a typical case in structural diagrams, and the second scenario is a typical case in behavioral diagrams. Figure 1. Class diagram for an academic system 3.1. Scenario 1: class diagram A class diagram can be very complicated. For example, classes in the class diagram are connected to each other intensively. However, usually we are only interested in part of the sophisticated class diagram, i.e., there must be some connections relatively more important to one group of people and less important to another. Suppose one group of software engineers only care about the classes relate to student, in a class registration system; another group, on the other hand, only care about the classes relate to instructor. An immediate problem is that a 2-D UML cannot provide such a requirement (see Figure 1) Analysis and potential enhancement of scenario 1 The classes in current UML are depicted as boxes. We can t emphasize the part of the diagram that we are really of concern. This scenario violates the conceptual modeling quality properties Expressiveness, Checkability, and. In contrast, perspective and emphasis in VRUML are different. In this case, user concentrates on the Student class by pulling the classes which are important for that aspect to the foreground (see Figure 2). Pulling classes to the foreground can hide the classes' details simultaneously. The attributes and operations of unimportant classes, in this case the People, Instructor, and Course classes, are left out. On the other hand, user can now interactively Figure 2. Scenario of using interaction to enhance class diagram in UML 3.3. Scenario 2: sequence diagram Inspecting a situation that one object wants to send multiple messages simultaneously to several different objects. Suppose a sequence diagram has simultaneous messages that are sent from one object to 2 other objects, as shown in Figure 3. Object Sender wants to send messages to Receiver 3 and Receiver 4 at the same time. In this case, it is not easy to understand a twodimensional representation. The trade-off introduced in the diagram to express this simultaneous sending is a small curve. For example, the curve under the Receiver 3 object indicates that there are two messages, Message 3 and Message 4, which go to the right and start at the same moment in the Sender object, one message for the Receiver 3, the other for the Receiver 4. However, the curve notation is not even included in the UML standards. 3

4 4. Research method Figure 3. Scenario of sending simultaneous message in a sequence diagram (adapted from Gogolla et al. [5]) 3.4. Analysis and potential enhancement of scenario 2 This scenario violates the conceptual modeling quality properties, Expressiveness, Processing independence, and Checkability. First, the curve in the diagram is not a standard notation in UML. Second, such a curve cannot express the simultaneous concept, so this expression is not precise and correct, it could give ambiguous ideas to developers. A solution to this situation is to use a three-dimensional representation. As Figure 4 show, the objects are placed in a threedimensional space, and the lifelines between objects can be placed in different planes. Simultaneous messages can now be expressed easily, because in three-dimensional space, we can arrange objects that receive simultaneous message not to be closed together, and to show lifelines clearly enough. Furthermore, we can display the messages sent from one object to another in the animation, labeled with the message name. An animation can convey simultaneous ideas more accurately and be more easily visualized by the users. And more important, messages can be easily identified. This paper extends and combines current research both in conceptual modeling and virtual reality. Our research methodology is briefly outlined below: 1. A survey will be sent to 300 conceptual modeling practitioners, i.e. system analysts and software engineers. They will be asked about the most apparent weakness of UML, the prospects of usefulness of enhancing UML with VR. We select survey as our approach because survey research is most appropriate when the researcher has very clearly defined independent and dependent variables, and a specific model of expected relationship which are tested against observation of the phenomenon. [13]. 2. Our posited research hypotheses, the potential enhancement of VR to UML in several aspects, will be tested by a latent structural equation modeling (SEM) software, LISREL. SEM is a confirmatory factor analysis technique. Researchers who use SEM have a model and use SEM to determine whether that model is valid. Our model will be based on Witmer and Singer s presence questionnaire (PQ) and immersive tendencies questionnaire (ITQ) [21]. PQ can measure presence in Vitural Environments and ITQ can measure differences in the tendencies of individuals to experience presence. Both of them have been tested that they are internally consistent measures with high reliability. Considering sample size for reliable factors, Guadagnoli and Velicer [22] recommended that, first, components or factors with 10 or more low loadings (below.40) will be reliable when sample size is greater than 150; second, components with only a few low loadings should not be interpreted unless sample size is at least 300. Following their rule, we target our sample size at Conclusion and future work We have discussed the problems of UML in conceptual modeling by using a conceptual modeling quality framework in Table 1, and have given some examples for UML diagrams which were designed to demonstrate that VR technology could enhance the quality of modeling. Our examples include the concepts of stereopsis, immersion, interaction, and animation. However, our work presented here is only the beginning of an ambitious research agenda, i.e. a research framework for VRUML. We list below the feasible research in each diagram type of UML (Table 3). Figure 4. Scenario of using 3-D and animation to enhance Sequence (adapted from Gogolla et al. [5]) Our next step is to identify the relevant information systems domains/knowledge. Many things remain to be done, both at the conceptual level and the implementation level. Our research framework could be applied in the information systems research fields shown in Table 4. 4

5 Table 3. Possible research framework for VRUML UML s Class Object Component Deployment Use Sequence Collaboration Statechart diagram Case Activity Research Orientation How immersion can improve user s vision. How interaction can enable user to adjust the design of class. How stereopsis can eliminate ambiguity of class representation. Same as Class How animation can visualize the state and interaction of objects How immersion can improve user s vision. How interaction can enable user to adjust the design of component. How stereopsis can eliminate ambiguity of component representation. How immersion can improve user s vision. How interaction can enable user to adjust the implementation of the system. interaction between actors and the use cases. interaction between object instances over time. How stereopsis can show the simultaneous message between objects. How animation can show a sequence of object interactions. How stereopsis can show the organization of the objects during the interactions. transition of various states of object. flows of activities within processes. Table 4. Possible IS research fields of VRUML research framework Possible IS research fields Conceptual Modeling Virtual Teams Group Support System Human-Computer Interaction Collaborative Software Design System Development Methods Computer Mediated Technologies Requirement Engineering Knowledge Sharing/Knowledge Management Cognitive Science References [1] C. Batini, S. Ceri, and S. B. Navathe, Conceptual Database Design: An Entity Relationship Approach, Benjamin Cummings Publish, Redwood City, California, [2] S. S. Cherfi, J. Akoka, and I. Comyn-Wattlau, Conceptual Modeling Quality - From EER to UML Schemas Evaluation, Proceedings of the 21st International Conference on Conceptual Modeling, Tampere, Finland, [3] J. Erickson and K. Siau, Unified Modeling Language: Theoretical and Practical Complexity, Proceedings of Ninth Americas Conference on Information Systems, Tampa, FL, [4] J. Evermann and Y. Wand, Towards Ontologically Based Semantics for UML Constructs, Proceedings of the 20th International Conference on Conceptual Modeling, Yokohama, Japan, [5] M. Gogolla, O. Radfelder, and M. Richters, Towards Three-Dimensional Representation and Animation of UML s, Proceedings of Second International Conference on the Unified Modeling Language: UML 99, Fort Collins, CO, [6] G. S. Hubona, G. W. Shirah, and D. G. Fout, 3D Object Recognition with Motion, Proceedings of Computer Human Interaction 97, Atlanta, GA, [7] C. Kung, An Analysis of Tree Conceptual Models with Time Perspective, Information Systems Design Methodologies: A feature Analysis, T. Olle et al., North- Holland, Amsterdam, pp , [8] J. Leigh, A. E. Jognson, C. Vasilakis, and T. DeFanti, Multi-perspective Collaborative Design in Persistent Networked Virtual Environments, Proceedings of IEEE 5

6 Virtual Reality Annual International Symposium 96, Santa Clara, CA, [9] I. O. Lindland, G. Sindre, and A. Solvberg, Understanding Quality in Conceptual Modeling, IEEE Software, 11, 2, pp , [10] S. McConnell, Prospects for a True Profession of Software Engineering, Construx Software, URL: [11] J. Maletic, J. Leigh, and A. Marcus, Visualizing software in an immersive virtual reality environment, Proceedings of ICSE'01 Workshop on Software Visualization, Toronto, Ontario, Canada, [12] K. Park, A. Kapoor, and J. Leigh, Lessons Learned from Employing Multiple Perspectives In a Collaborative Virtual Environment for Visualizing Scientific Data, Proceedings of the Third International Conference on Collaborative Virtual Environments, San Francisco, CA, [13] A. Pinsonneault and K. L. Kraemer, Survey Research Methodology in Management Information systems: An Assessment, Journal of Management Information Systems, 10, 2, pp. 75, [20] R. A. Johnson, The Up and Downs of Object Oriented Systems, Communications of the ACM, 43, 10, pp. 68, [21] B.G. Witmer and M.J. Singer, Measuring presence in virtual environments: A presence questionnaire, Presence: Teleoperators and Virtual Environments, 7, pp , [22] E. Guadagnoli and W. F. Velicer Relation of sample size to the stability of component patterns, Psychological Bulletin, 103(2), pp , [23] W. Winn, A Conceptual Basis for Educational Applications of Virtual Reality, HITL Technical Report R-93-9, Human Interface Technology Laboratory, University of Washington. [24] C. Dede, M. C. Salzman and R. B. Loftin, ScinceSpace: virtual realities for learning complex and abstract scientific concepts, Proceedings of IEEE Virtual Reality Annual International Symposium, Santa Clara, CA, [14] R. Schuette and T. Rotthowe, The Guidelines of Modeling An Approach to Enhance the Quality in Information Models, Proceedings of the 17th International Conference on Conceptual Modeling, Singapore, [15] The Standish Group International, The CHAOS Report, 1994 URL: os_1994_1.php [16] K. R. Walsh, S. D. Pawlowski, Virtual Reality: A Technology In Need of IS Research, Communications of the AIS, 8, pp , [17] Y. Wand, R. Weber, Research Commentary: Information Systems and Conceptual Modeling A Research Agenda, Information Systems Research, 13, 4, pp , [18] C. Ware and G. Franck, Viewing a Graph in a Virtual Reality Display is Three Times as Good as a 2D, Proceedings of IEEE Symposium on Visual Languages, St. Louis, MO, [19] E. Yourdon, Rise and Resurrection of the American Programmer, Prentice Hall PTR, Indianapolis, IN,