A FRAMEWORK FOR THE ANALYSIS AND COMPARISON OF HYPERMEDIA DESIGN METHODS

Transcription

1 A FRAMEWORK FOR THE ANALYSIS AND COMPARISON OF HYPERMEDIA DESIGN METHODS Susana Montero, Paloma Díaz and Ignacio Aedo Laboratorio DEI. Dpto. de Informática Universidad Carlos III de Madrid Avda. de la Universidad Leganés, Spain ABSTRACT Due to increasing size and complexity of hypermedia and web applications, there is growing concern by how to develop quality, reusable and maintainable hypermedia systems. In fact, a series of well-defined design methods have been proposed to be used during the design process. In this paper, we propose a framework of reference based on software and hypermedia engineering fields in order to perform a constructive analysis of the most outstanding methods to date such as HDM, RMM, OOHDM, Autoweb, WebML, UWE and OO-H method. With this survey, we attempt to help developers to select which method is the most suitable for their needs. Moreover, from this analysis some lacks in the hypermedia development process are revealed. KEY WORDS hypermedia, design methods, software engineering 1 Introduction Hypermedia systems, and specially web applications, have been extremely demanded in different areas like commerce, education, health, and libraries, both to provide a hypermedia interface for existing information systems and to create new hypermedia applications. But all of this has been made in a very short period of time what has led most developers to skip the conceptual design phase and directly go to the implementation stage, producing applications of poor quality, usability and maintainability. The majority of these implementations have been made with software tools such as NetObjects Fusion product or DreamWeaver which support an automated implementation by contents and an easy set up [1], but do not pay attention to intrinsic features of hypermedia systems such as sophisticated navigational structures, interactive behaviours and multimedia compositions. Moreover, other feature which should be taken into account is security, since hypermedia applications are accessed by different users with different purposes [2]. Design methods, and hypermedia methods in particular, face up to the design of large and dynamic systems through a number of stages and products which make possible to approach modeling in a systematic and integrated way. As a result, systems will have a better quality, usability, maintainability and reusability. Several methods have been proposed including HDM, RMM, OOHDM, Autoweb, WebML, UWE and OO-H method. So that it is complicated to decide which one is the most suitable for a specific development. In this paper, we propose a framework of requirements to cover the modeling space of design methods for hypermedia systems. On the basis of this framework, the methods previously mentioned have been analysed. The rest of the paper is organized in the following way: in section 2 we propose a set of requirements to analyse or compare hypermedia design methods derived from the literature and from the experience of our group in the design of this kind of systems [3, 2, 4]. Section 3 presents the analysed methods as well as those requirements that are covered by such methods. In section 4, some research areas are opened from the requirements that have not been taken into account. Finally, in section 5 some conclusions derived from this work are drawn. 2 A framework for hypermedia design methods To analyse or compare different design methods, it is necessary to establish a framework of reference. In our case, this framework is determined by a set of requirements gathered from both the software and hypermedia engineering fields. The experience gained in years of research in the software field can help to improve hypermedia development as it is suggested by Lowe and Webby [5]. In fact, most of hypermedia methods borrow the notation of best-known data models, like the E/R (Entity-Relationship) model or OMT (Object Modelling Technique), in order to represent their system structure. However, these approaches cannot be used directly to model hypermedia systems since their development is much more incremental and iterative than in the rest of software systems [5] and, moreover, as Nanard and Nanard point out, software engineering lacks elements and mecha-

2 nisms to model aesthetic and cognitive aspects which are a basic concern in hypermedia design [6]. 2.1 Requirements derived from software engineering We will borrow some features of the design process from the software engineering area, but will not consider issues such as project planning or project management. According to [7, 8], any design software method must: S1. Describe a formal process to guide the development of a software application. That is, a method should describe what phases of the development are, how these phases are integrated into the overall development process and what artifacts are created. Typical process models for the software life cycle, including waterfall, prototyping, incremental development and spiral model [9] are not suitable for hypermedia since these kinds of applications pay more attention to organisation of information to promote their access [5] rather than to information processing or management, as it is in the case of software applications. S2. Contain a model to describe the real world and transfer it to a physical system. A model is a form of abstraction which allows us to represent the essence of a design. Several models belonging to the software engineering field, like Entity-Relationship model or UML techniques (Unified Modeling Language), could be used for that, as in [10], but they do not provide mechanisms to represent cognitive (e.g. navigation aid structures, see H2) and aesthetic (e.g. contents alignment and synchronisation, see H3) aspects. S3. Provide the designer with artifacts to specify the system requirements. Such requirements are categorized into: functional, i.e., what functions the system is expected to support; non-functional, i.e., how the system will work in terms of accessibility, accuracy, efficiency, integrity and so on; and usability, i.e, how the interaction process will be designed. With regard to hypermedia applications, functional requirements include browsing capabilities, personalization, security constraints and interactive behaviours (see H2, H3, H4, H5 and H6 in subsection 2.2). Nonfunctional requirements include content and link validation, node size, cohesion and consistency of the information [11]. Finally, usability requirements are oriented towards improving the interaction between the user and the hypermedia application [12] and they include aspects like aesthetic, readability, consistency, self-evidence or predictability [13]. S4. Include validation rules for each design product. The method has to provide mechanisms to test the conceptual correctness of each design product and its completeness. For example, a hypermedia link has always a source and a target, usually referred to as anchors. A validation rule will check if for each source anchor embedded into a piece of content there is also defined a target anchor resolving to a node or content. S5. Maintain relations of integrity among the different design products. In order to provide an integrated framework, the method should support the traceability, both backwards and towards, of those design entities which are described in different design products, which can belong or not to the same development phase. For example, an integrity rule can check if for each node defined in a conceptual phase there is a set of presentation specifications defined during a detailed design phase (v.g., background colour, size and so on). S6. Support design reuse. Design is a creative process that cannot be learned from a book but from the designer s experience and existing systems. On the one hand, design patterns [14] describe problems that occur repeatedly and provide the core of the solution in order to use it many times and in different contexts and they can be combined with any design method. Hypermedia patterns are usually concerned with interface and navigation problems [15]. On the other hand, design components can reuse pieces of previously developed design as in [16]. S7. Count on software support tools to help in systems development process. In the development process of hypermedia applications a variety of users with different levels of knowledge and skills have to interact, including authors, designers, artists and engineers [17]. A method should integrate a toolset to be used by all these kinds of users. Such a toolset should facilitate modeling and design of the system in a graphical way, automate the documentation generation, validate the design products and generate code for the specified models. 2.2 Requirements derived from hypermedia technology Hypermedia applications differ from the rest of software ones due to the use of a non-linear structure whose information may be a combination of different types of media such as text, graphics, audio and video, and even it can hold interactive behaviours (e.g., applets embedded into web pages). Moreover, the user can navigate freely through such a structure selecting links. For these reasons, it is necessary to stress the following requirements took from hypermedia engineering [18, 19]. H1. Allow to describe the problem domain in terms of hypermedia components. These components are basically nodes, contents, links and anchors. A node is

3 an information holder able to contain a number of information pieces termed contents. A link is a connection among two or more nodes or contents. A link is defined between two set of anchors [4]: the source and the target, referring each anchor to a node or piece of content. Treating these elements as independent design entities allows to separate information from composition, navigation and presentation design and, in addition, it makes possible to provide multiple views of the same data. H2. Provide tasks to model the navigation structure. One of the most important tasks in hypermedia modeling is the design of the browsing which comprise nodes and links and, in fact, all hypermedia methods support the navigation design. Hypermedia methods should also provide elements to model navigation aids such as visual maps, active indexes, guided tours, marks, footprints and backtracking mechanisms [20]. Moreover, navigation modeling has to deal not only with declarative but also with event-based specifications (dynamic links). H3. Organize and harmonize multimedia contents. Another basic feature of hypermedia concerns presentation issues, since nodes include multimedia contents that need to be organized and harmonized in different dimensions such as time and space. In fact, the way and rhythm used to deliver the contents to the user determines to a great extent the hyperdocument usefulness. The method should allow us to describe the conceptual properties of the contents and their relationships as well as to decide where to place the contents on the screen (layout planning) and over the time, as for example in SMIL, where tags to synchronize contents are provided [21]. H4. Model the different types of users. Since most hypermedia applications are accessed by different kinds of users [11], the method should include mechanisms to deal with the conceptual design of users, that will be used to design adaptative applications or with security purposes (see next requirement). H5. Provide conceptual tools to formalize security policies. Nowadays hypermedia systems are used in multi-user environment where different types of users with different privileges access the hyperdocument [2]. Thus, the method should allow designers to model security issues such as which contents should be delivered to which users, who can modify the structure of the hyperdocument, who can personalize items or which constraints have to applied when creating or modifying a link. H6. Allow to describe the interactive behaviour of hypermedia systems. Hypermedia applications include complex functions such as dynamic generation of nodes, adaptive presentation and navigation, access to external applications (e.g., databases or existing information systems) or interactive contents (e.g., exercises in a hypermedia educational environment). Therefore, these dynamic features have to be taken into account in a conceptual way and not be relegated till the implementation phases. With this purpose, event-base specifications are required, so that designer will be able to determine how the system will behave when some events occur [4]. H7. Support a bottom-up design. Some web applications are designed starting from a mock-up of their interface and creating the conceptual design in a further stage. Moreover, top-down and bottom-up cycles should be supported to deal with the abilities of different development teams as stated in [6]. H8. Make possible the evaluation of the system utility. Hypermedia applications are distinguished by the complexity of their interfaces and aesthetic and cognitive aspects of their contents. Therefore, it is needed a user-centred development where the evaluation plays a basic role to analyse whether the design solutions fulfil the user expectations and needs. Prototyping could help to improve the utility and usability of the application [12]. 3 Methods under consideration Recently, a huge quantity of development methods for hypermedia systems, and specially for web systems, are appearing. This reveals a web systems success as well as the necessity of controlling the modeling process. We have analysed a set of the most outstanding methods. For each of them a brief introduction and their most excellent features are presented. Moreover, those requirements that are completely accounted for will appear in brackets (Si or Hi). Hypermedia Design Model - HDM HDM [22] is based on techniques and models from database and hypertext field. Five dimensions are identified [13]: content, structure, presentation, dynamics and interaction (H6). To describe the system, they adapt the E/R model (S2) to introduce hypermedia concepts such as nodes, contents, anchors and links (H1, H2). Contents (slot) can be organized (collections), in addition to being synchronized and aligned by a set of dynamic relationships (H3). A toolset has been developed for web prototyping development and execution (S7). Moreover bottom-up design (H7) and XML are supported. HDM-Edit is the component that supports the HDM model, design patterns, reuse of design components (S6) and documentation generation. A set of evaluation criteria (H8) are applied to applications designed with HDM [13]. Relationship Management Methodology - RMM

4 RMM [23] is the unique hypermedia method that owns a complete software development cycle, from feasibility analysis to testing and evaluation of the system (S1). Moreover, it supports top-down and bottom-up modeling (H7) alike. This method is based on a model (S2), RMDM (Relationship Management Data Model) which is inspired in the E/R model and HDM. The method starts by representing the information structure via an E/R diagram. In the next step, attributes of entities are organized (H3) in m- slices which can be aggregated and nested. This allows different views of the information. Links and theirs anchors are also included (H1, H2). A Case Tool, RMCase has been developed to support the RMM stages (S7). To evaluate the application (H8), they suggest using the technique proposed by Garzotto [13]. Object-Oriented Hypermedia Design Method - OOHDM OOHDM [24] is a four-activity process that is performed in an iterative and incremental way (S1). The starting point is the Conceptual Design, based on OMT (S2) to model the application domain. The following is the Navigation Design (H2) that allows us to build different navigation views (Navigation Context) for the same conceptual model. In such navigation views anchors (H1) and behaviours (H6) can be defined. The third activity is the Abstract Interface Design that specifies the interface aspect. In this activity, contents (Abstract Data View) can be organized (H3). In the same way as in the preceding activity, different interfaces for the same navigation model can be built. The last two activities can be used for personalization purposes by means of the building of a specific view for each user profile. The system users can be modeled into the conceptual schema like another type of object (H4). An important point is design reuse by means of design patterns (S6). In the last activity, Implementation, a series of steps guide the mapping of the design into an implementation environment. In relation to CASE tools, the OOHDM-Web environment is under construction (S7). Autoweb Autoweb [25] proposes a top-down method based on three phases: the conceptual design, the database generation and the implementation of the Web application (S1). All these phases are supported by a tool suite so-called the AutoWeb System (S7) which automatically generates a Web application. The conceptual design is formalized by the HDMlite model (S2), an evolution of HDM [22] for Web. It allows us to describe the structure, the navigation and the presentation of the system (H1, H2). In contrast to the rest of methods, Autoweb allows us to define the contents (component-type) and structure of a node (entity-type) in the Structure Schema and to place them within the visualization area in the Presentation Schema (H3). To have different views of the system, various Presentation Schemas have to be generated. Web Modeling Language - WebML WebML [26] specifies data-intensive Web applications by four views: Structural, Hypertext (H2), Presentation and Personalization (S1). An iterative design process guides the development process from the requirements collection to the design customization. In relation to hypermedia elements, contents (units) of a node can be specified by the Composition Model. These contents can be organized in self-contained regions of the screen (page). In addition, it can be specified if regions will be displayed together or just one of them. Some navigation patterns are taken into account. In contrast to the others methods, WebML owns a collection of rules to build correct logical and physical hypertext (S4). WebML includes the notion of group (set of users) and user (individual) which are modeled as a special entity in the Structural Model allowing the personalization of the hyperdocument (H4). The interactive behaviour is supported by operation units which are linked to other units and can be predefined or built by the designer (H6). The whole design process is supported by a CASE environment called ToriiSoft (S7). Besides, WebML supports XML syntax to get a platform-independent design. The object-oriented-hypermedia (OO-H) method The OO-H method [27] is UML-compliant and allows us to model both statics and dynamics (H6) of the system (S2). These two views are extended with the Navigational Access Diagram (NAD) (H2) and the Abstract Presentation Diagram (APD) in order to capture web features. The designer has to construct as many NADs as user types when information personalization is required. The two strong points of this method are its CASE tool (S7) and a Design Patterns catalog (S6). Prototypes can be generated by means of its CASE tool. UML-based Web Engineering methodology - UWE The UWE [28] is an object-oriented, iterative and incremental approach whose core modeling activities are the requirements analysis (S3), conceptual, navigation and presentation design (S1). Task models and statecharts of Web scenarios are included to model the dynamics aspects of the application (H6). The UML (S2) and a set of stereotypes defined for the modeling of navigation (H2) and presentation aspects (H3) of Web applications are used for the artifacts representation. The design models are transformed into XML representations. In relation to hypermedia elements, nodes are represented by classes, links by stereotyped association and navigation aids by stereotyped classes (i.e. <<index>>) (H1). Furthermore, semiautomatic generation of web applications from design model is produced by ArgoUML tool (S7). 4 Outstanding requirements: some research open areas During the analysis of the aforementioned methods, some requirements of our framework are not addressed by them.

5 Requirements for design methods of hypermedia systems HDM RMM OOHDM Autoweb WebML OO-H UWE S1. modeling process P C C C C P C S2. based on a model C C P C P C C S3. functional, no-functional and usability requirements P P P P P P P S4. validation rules for each product N N N N P N N S5. integrity relationships among phases N N N N N N N S6. design reuse C N C N P C N S7. software support tools C P P C C C P H1. hypermedia components C C C C C C P H2. navigation structure C C C C C C C H3. organization and harmonization of multimedia contents C P P P P N P H4. user modeling N N C N C N N H5. formalization of security policies N N N N N N N H6. description of interactive behaviour C N C N C C C H7. bottom-up design C C N N N N N H8. evaluation of system utility C C N N N N N Table 1. Comparison between requirements and methods Next, we summarize the most important aspects that have been discussed in the previous sections in the table 1. The left part enumerates the software and hypermedia requirements and on the right side their rate of performance is shown for each method. Three kinds of notations have been used: C, if the method fulfills the whole of requirement; P, if the method takes into account the requirement partially; and N when the requirement is not taken account by the method. From this table, some weak points of the analysed methods are outlined below: Non-functional and usability characteristics. These aspects could improve the quality and efficiency of the system. Validation and integrity rules. These rules could help determine the correctness and the completeness of design and the integrity of the designed elements. Design reuse. Navigation patterns are usually used, but none relates to the interface and the structure of the system nor design components. They would allow us to reuse designers experience. Content modeling. Most of the methods do not model contents in a separate way and do not take into account their multimedia nature, therefore different data views, synchronizations and alignments among contents cannot be specified. User modeling. System users are not taken into account like another type of elements into modeling process so the personalization of the system is made by means of cut-paste technique. Security modeling. Security policies could be applied as access control mechanisms to different users. Bottom-up design. A design is not allowed from prototype of the application interface. Evaluation stage. This stage could help improve the utility and usability of the application and make possible a development more incremental and iterative. 5 Conclusions It is difficult to find a hypermedia design method that is fit for the development of any hypermedia system due to rapid change in this technology and the combination of different techniques and fields (database, object oriented, hypermedia, web,...) which have to be applied. However, the methods presented here are a proof of the increasing concern about covering modeling process of hypermedia systems in a more complete way, making an effort to establish the steps and techniques which have to be carried out in order to get a scalable, maintainable and usable application. With this survey, we want to reveal both the benefits and limitations of design methods in order to provide some descriptions that can be relevant for the designers, researchers and even users who are considering developing hypermedia applications using design methods. Furthermore, we have been developing a hypermedia method called Ariadne [18] which applies an iterative process based on the evaluation of design solutions with potential users to determine their utility and usability. The method provides mechanisms to define time- and spacebased constraints among contents, to model the users structure and security policies, and to guarantee the correctness and completeness of the design by means of validation and integrity rules. Finally, for a greater acceptance and feedback from designers and developers with the aim of refining the modeling process, we are working on a support software tool for automatic generation of applications. Acknowledgements This work is supported by Dirección General de Investigación del Ministerio de Ciencia y Tecnología (TIC ). References [1] P. Fraternali. Tools and approaches for developing data-intensive Web applications: a survey. ACM Computing Surveys, 31(3): , 1999.

6 [2] P. Díaz, I. Aedo, and F. Panetsos. Modelling security policies in hypermedia and web-based applications. In Web Engineering: Managing diversity and complexity of web application development, volume 2016 of LNCS, pages , [3] P. Díaz, Aedo I., and F. Panetsos. Labyrinth, an abstract model for hypermedia applications. description of its static components. Information Systems, 22(8): , [4] P. Díaz, I. Aedo, and F. Panetsos. Modelling the dynamic behaviour of hypermedia applications. IEEE Transactions on Software Engineering, 27(6): , June [5] D. Lowe and R. Webby. Utilisation of process modeling in improving the hypermedia development process. The new review of hypermedia and multimedia, 5: , [6] J. Nanard and M. Nanard. Hypertext design environments and the hypertext design process. Comm. of the ACM, 38(8):49 56, [7] D.E. Avison and G. Fitzgerald. Information systems development: techniques and tools. Blackwell Scientific Publications, Oxford, [8] I. Sommerville and P. Sawyer. Requirements engineering: a good practice guide. Wiley, Chichester, [9] R. H. Hayer and A. M. Dorfman, editors. Requirements Engineering. Chapter 1. IEEE Computer Society, [10] J. Conallen. Modeling Web application architectures with UML. Communications of the ACM, 42(10):63 70, [11] D. Lowe and W. Hall. Hypermedia and the Web: an engineering approach. Chichester: John Wiley and Sons, [12] J. Preece, Y. Rogers, H. Sharp, D. Benyond, S. Holland, and T. Carey. Human computer interaction. The open university. Adidison Wesley, [13] F. Garzotto, Mainetti L., and P. Paolini. Hypermedia design, analysis, and evaluation issues. Communications of the ACM (CACM), 38(8):74 86, [14] E. Gamma, R. Helm, R. Johnson, and J. Vlissides. Design Patterns, Elements of Reusable Object- Oriented Software. Addison-Wesley, [15] G. Rossi, D. Schwabe, and F. Lyardet. Integrating patterns into the hypermedia development process. New Review of Hypermedia and Multimedia, 5:59 80, [16] M. Gaedke and G. Gräf. Development and evolution of web-applications using the WebComposition process model. In International Workshop on Web Engineering at WWW9, volume 2016 of LNCS, pages 58 76, [17] Y. Deshpande and S. Hansen. Web engineering:creating a discipline among disciplines. IEEE Multimedia, 2(8):82 87, [18] P. Díaz, I. Aedo, and S. Montero. Ariadne, a development method for hypermedia. In proceedings of Dexa 2001, volume 2113 of LNCS, pages , [19] W. Retschitzegger and W. Schwinger. Towards modeling of dataweb applications - a requirements perspective. In Proc. of the Americas Conferenc on Information Systems (AMCIS) Long Beach California, volume 1, [20] J. Nielsen. Hypertext and Hypermedia: the Internet and Beyond. Academic Press, [21] SMIL: Synchronized Multimedia Integration Language. [22] F. Garzotto, P. Paolini, and D. Schwbe. HDM- a model-based approach to hypertext application design. ACM Transactions on Information Systems, 11(1):1 26, [23] Isakowitz, T. and Kamis, A. and Koufaris, M. The extended RMM methodology for web publishing. Technical Report Working Paper IS98-18, Center for Research on Information Systems, [24] D. Schwabe and G. Rossi. An object oriented approach to web-based application design. Theory and practice of object systems, 4(4): , [25] P. Fraternali and P. Paolini. Model-driven development of web applications: the autoweb system. ACM Transactions on Office Information Systems, 18(4): , [26] S. Ceri, P. Fraternali, and A. Bongio. Web modeling language (WebML): a modeling language for designing web sites. WWW9 / Computer Networks, 33(1-6): , [27] Gómez, J. Cachero, C. and Pastor, O. Conceptual modeling of device-independent web applications. IEEE MultiMedia, 8(2):26 39, [28] N. Koch and A. Kraus. The expressive power of UML-based Web Engineering. In Second International Workshop on Web-oriented Software Technology (IWWOST02), Malaga, Spain, June 2002.