Dr. ir. G.J. Houben, Prof. dr. P.M.E. De Bra
Department of Computing Science,
Eindhoven University of Technology.
PO Box 513, 5600 MB Eindhoven.
tel: (040) 2472733
email: {houben,debra}@win.tue.nl
Dr. ir. W.R.Th. ten Kate, Prof. dr. E.H.L. Aarts
New Media Systems & Applications group,
Philips Research, Eindhoven.
email: {tenkate,aarts}@natlab.research.philips.com
Multimedia and Human-Computer Interaction group,
Centrum voor Wiskunde en Informatica.
Department of Mathematics and Computing Science,
Eindhoven University of Technology.
Philips Natuurkundig Laboratorium.
(For addresses, see point 2: Applicants)
An increasing number of organizations need to provide access to (textual and) multimedia objects stored and maintained in multimedia databases. Such access is provided through hypermedia structures and interfaces. When multimedia database information is made available over Internet or an organization's Intranet, World Wide Web browsers are the preferred user-interface platform. The World Wide Web consortium is highly active in enhancing the multimedia capabilities of Web components.
The biggest challenge in creating hypermedia presentations for databases containing multimedia objects is to automate the generation of these presentations (at least partially). Furthermore, the presentations must be tailored (or adapted) for the situation and needs of an individual user. This research project aims at the development of a design methodology for hypermedia access to and presentations of multimedia database objects. The design and generation of such presentations should require only a minimum of human intervention. The research will capitalize on the growing resources of multimedia objects stored and maintained in multimedia databases and use these as a resource for creating hypermedia presentations. The primary research question is to what extent we are able to automate the traditional hypermedia authoring process in order to generate acceptable quality, cost-effective, user-tailored presentations. Automatic generation of hypermedia presentations involves making design decisions with respect to information content, spatial and temporal layout and hyperlink navigation, modalities and media used, etc. Automatic generation of hypermedia presentations is currently only used in text-centric environments as an interface to strongly structured databases. Our research will address presentation generation in media-centric environments to interface semi-structured multimedia databases. It builds upon previous and ongoing research in multimedia presentation (see e.g. Ten Kate et.al. 1998), in timed hypermedia (see e.g. Hardman et.al. 1994), in adaptive hypermedia (see e.g. De Bra et.al., 1999) and in hypermedia presentations of volatile database information (see e.g. Houben et.al., 1998).
The goals of the research will be realized through the integration of current research results along with the development of new research results. As a base we take existing approaches to multimedia databases and knowledge representation to enable the integration of domain knowledge with existing media items. The development of an "authoring on demand" model for adaptable hypermedia requires investigation into the most appropriate knowledge representations for selecting and presenting media items in a coherent hypermedia presentation. This development can be taken one step further by adapting the representations and presentations to the individual user automatically, by using adaptive hypermedia techniques (Brusilovsky 96).
The innovation in the work stems from three separate issues: firstly the generation of a document corresponding to a rich hypermedia format, including temporal synchronization of its constituent parts and the generation of links within multiple media (for example, the SMIL format); secondly, the re-use of existing media items in a heterogeneous environment with widely varying client-side devices (palm top, mobile phone, television with set-top box or multimedia PC); thirdly, the automatic adaptation to the user's knowledge, preferences and task, as well as the compute and network environment. We will base our work on the Standard Reference Model for Intelligent Multimedia Presentations (SRM-IMMPS, Bordegoni et al. 97), and will explore the process of generating a hypermedia presentation, including temporal, spatial and linking aspects, from media items selected from the database. This requires an extension of the SRM-IMMPS to include design rules for temporal and linking aspects in addition to those for spatial layout. The SRM-IMMPS approach includes the development of appropriate user-modelling techniques.
Implementation of this work will require a broad base of technology, including multimedia database technology, knowledge representation formalisms, user models and document representation and manipulation models such as RMM (Isakovitz 95) and techniques such as XML, XSL and SMIL (W3C).
The research will be performed by one post-doc and two OIO's (phd students). One OIO will be funded by Philips Research, and is therefore not listed in the table below.
| year 1 | year 2 | year 3 | year 4 | |
| personnel (fte) | ||||
| oio | 1 | 1 | 1 | 1 |
| post-doc | 1 | 1 | 1 | |
| other | ||||
| expenses (kFl.) | ||||
| personnel (estimate) | 115 | 125 | 135 | 65 |
| material | p.m. | p.m. | p.m. | p.m. |
| other (travel) | 4 | 4 | 4 | 2 |
The cost (in kfl) of personnel is an estimate based on the "standard" salaries for OIO's and post-docs. (NWO will adjust these amounts.) The cost does not include the "standard" overhead which NWO pays to the "host institutes".
The researchers will need office space, computer and laboratory equipment and other materials. However, this is not paid separately by NWO but is part of the unspecified overhead.
NWO pays a "standard" amount for travel expenses. This amount is more or less what is needed to hold meetings to coordinate the work between the three main researchers, and for report and feedback meetings with the project leaders. The researchers will publish their result at scientific conferences and in journals, and will therefore need more travel money. These additional travel funds will be provided by the "host institutes".
signature:
date:
The main goal of the project is to increase the level of automated adaptation during the process of creating hypermedia presentations. In particular, the generation process should adapt to varying user and system characteristics. For user adaptation, we consider such things as the current state of knowledge of the user, the task the user is involved with and characteristics and preferences of the user, including accessibility issues. Adaptation at the system level includes accounting for the end-user system the presentation will be played on. This may be a Windows based PC, but may also be a palm top or a mobile phone, or a television enhanced with a set-top box. In order to adapt for a particular system, the document format and memory size are important considerations. In addition to the hardware the user is engaged with, there are also considerations for the network bandwidth between the server generating the source document, the servers supplying the media items and the hardware at the client side.
While this gives an overview of the adaptation that may take place, we also need to investigate where the adaptation should occur. This can be considered both from a conceptual point of view and from an architectural point of view. We deal first with the conceptual point of view. When a user is seeking specific information, adaptation can take place at the database query level. There are multiple queries which can result in different pieces of information being presented to the user in similar looking ways. (For example, all issues of a newspaper look the same, but contain different news every day.) Alternatively, the same information can be presented in different ways. An example is showing the same HTML page on different sized windows. This can go further, where the same information can be presented using different (multimedia) style sheets, but even be presented using different media types (depending on the adaptation characteristics described in the previous paragraph).
Given the characteristics of adaptation and means for adapting, the final question to be answered is where the adaptation should take place within the software architecture. In a client-server model, adaptation can take place at the server side, the client side, proxy side or some combination of these. Part of the investigation will be to determine which adaptation can best be carried out where.
The project calls for a broad range of expertise in order to integrate the differing disciplines required.
Our approach is to coordinate the work among the three partners, making use of the expertise available directly from the partners, and also from related projects being carried out, in particular the extensive multimedia database work at CWI (with which Paul de Bra, TUE, is also associated), and the NWO(SION) project on hypermedia presentation of volatile multimedia database output, being carried out at the TUE. We describe our approach in terms of the roles of the three positions in the project (where two are being requested from NWO(SION) and one is funded by Philips Research). In broad terms, the post-doc at CWI will select, define and design the "underlying" conceptual and architectural infrastructure on which the OIO's will base their work. The TUE OIO work will be mainly "knowledge-based" in order to capture the conceptual aspects of adaptation, and the Philips OIO will be mainly "document format" based in order to generate the concrete presentations displayed on the (different) end-user platform(s).
The interdisciplinary nature of the research requires a senior-level researcher to coordinate and lead the work described by the three collaborating partners and to assist the applicants in guiding the two OIOs. The post-doc at CWI will be closely involved with determining the research direction of the project, as well as coordinating the project on a day-to-day basis. (A candidate for the post-doc position is already available.)
The scientific content for the post-doc position will concentrate on determining which parts of the document generation process can most usefully be automated, and ensuring that the methods and tools being developed can be integrated into the current manual authoring environment (already in existence at CWI). In addition, this role will include the design of the overall infrastructure of the project, including keeping up to date with developments within W3C, in addition to broadcast standards, in terms of appropriate technologies and document formats. The post-doc will advise the OIOs as to which technologies and standards to adopt. The post-doc will also play a key role in the joint design decisions about the distribution of the "work" between clients (the machine on the user's side which may be a powerful multimedia PC but also a simple palmtop) and the servers. There is a tight link between the distribution of work between client and server and the selection of the language in which to represent the multimedia objects to be communicated to the client for presentation. While design decisions may be influenced by the chosen "first application" which is an electronic program guide (see below), the post-doc will ensure that the infrastructure remains sufficiently generic to enable the development of other applications such as mail order catalogues, museum kiosk information displays and websites, and showcase databases by collaborating real-estate agencies.
Coordination with other ongoing work at CWI on multimedia databases will also be carried out.
The TUE work will cover user modeling aspects to allow the generated documents to conform to the characteristics of the user. This will complement the Philips OIO, where the emphasis is on generating to different document languages for diverse end-user platforms.
The user modeling work will concentrate on those aspects of the user model which are most relevant to the selection of multimedia objects and to the navigation between objects. Work on user modeling for adaptive hypermedia (outside the database context) is being carried out by an OIO at TUE in another NWO(SION) project. That work solely considers user modeling as a representation of user knowledge in the context of a learning process. In the current project user modeling will not only be used to represent user knowledge but also preferences and situational aspects such as location, time, task, etc. Such user-related aspects may change almost randomly, whereas the evolution of user knowledge is more predictable and mostly considered as "augmentation" only.
In addition to a comprehensive user model, the OIO will also investigate design rules needed for creating a hypermedia presentation suitable for the user. These will be developed in close conjunction with the other work, since the design rules are the main connection between the knowledge repositories and the documents that are to be generated. However good the user model, if the design rules are inadequate then the generated presentation will not be acceptable.
An interface will be designed to allow a user to interact with the system as an end-user of the final system as well as for prototyping and testing purposes for changing the user characteristics or design rules manually. Parts of this design will be borrowed from work being carried out by an OIO at TUE in the NWO(SION) project on hypermedia presentation of volatile multimedia database output. (That research project however only considers static media items such as text and images.)
The work carried out at (and funded by) Philips will serve as a testbed for the research results of the project, using an Electronic Program Guide (EPG) as a vehicle. The Electronic Program Guide can include film clips, audio commentary and also textual information, such as transmission times. In principle, information on all programs that are broadcast anywhere in the world can be listed. Depending on system configuration (manual or automatic), access to the broadcast programs can also be supported, at least within the framework. Program information may be presented along various axes: per channel, per theme, per provider, etc., or in various ways: list, grid, etc., and on various devices: TV, PC, RC, phone, etc. The EPG thus appears to be an appropriate vehicle to experiment upon. The programs themselves and the information about them form a rich source of multimedia material from which sets of objects can be selected in order to test out the research findings. In addition, the EPG is a fairly straightforward hypermedia document, allowing a test environment to be constructed at low cost before attempting more complex scenarios. Finally, EPGs concern dynamic data, thus challenging system performance as well as necessitating the automation of the creation of presentations.
In addition to generating an EPG from the underlying materials, an important part of the work is to investigate suitable document languages and formats and transport protocols within the generation environment. In particular, for any single EPG generated, there are multiple options for the document format in which the guide could be sent to the end user.
The work of the OIO will include the development of hypermedia models, focusing on platform-centric adaptivity; integration of document-oriented (HTML, SMIL), object-oriented (MHEG-5) and video-oriented (MPEG) models; TV-oriented interfaces (EPG) and will address the relation and possible overlap between time-based data-structures/formats with transport protocols.
Implementation of the EPG will be performed with assistance of Philips staff and through graduate student projects. The OIO's contribution concerns the scientific part in further developing the conceptual and architectural framework, by investigating alternative approaches and comparing their merits. Those alternatives are based on literature and will be generated by all participants in the project.
The genericity of this work will be verified in a later stage, for instance by generating hypermedia presentations for mail-order catalogs, museum kiosk information systems or websites, or "showcase" databases of (collaborating) real-estate agencies.
The results will include concrete proposals for user models and design rules for use in generating specific types of presentations, and will include information on the trade-offs for different types of models and rules. These results will be visualized by applying the concepts and models to an EPG application prototype (and later to other applications as well).
Insight will be gained in the utility and suitability of different presentation formats for different end-user platforms, and to what extent these formats can be automatically generated or translated.
The deliverables will include scientific publications and demonstrations at relevant national and international conference proceedings and in journals, plus two doctoral dissertations (from TUE and from Philips).
The amount of information available to end-users is orders of magnitude more than can be processed. On the other hand, users need up-to-date and relevant information for the tasks they wish to carry out. Finding information, in particular multimedia information, is one step in providing relevant information, but the resulting information is currently presented at best as an ordered ranking of separate media items, and not as an integrated hypermedia presentation.
Also in the case of (multimedia) data contained in more "traditional" databases the presentation of query results requires the generation of hypermedia navigation and presentation. For database queries in (highly) structured databases (with static data such as text and images) this problem is already being investigated in a SION project at the TUE. The current project extends this research to "real" multimedia content.
The results of this research will be immediately disseminable through the use of W3C standard languages for the distribution of generated presentations, and through the use of prototypes that will be made available through Internet and that will influence the standardization process and the further development of these languages. Also, specific attention will be paid to the problem of making multimedia information available through non-computer equipment which is used by the general public, including mobile phones and televisions.
The proposed research is related to research issues in multimedia databases, knowledge representation (domain descriptions for labeling and searching multimedia items, encapsulation of design rules, user model), hypermedia document modeling, hypermedia authoring, document processing (using W3C compatible standards), intelligent multimedia presentation systems, distributed object technology, hypermedia design methodology (Garzotto & Schwabe 93), (Isakovitz 95), adaptive hypermedia technology (Brusilovsky 96).
Because of this multi-disciplinary nature of the project it is necessary to bring together researchers from different sub-areas of computer science. Not only is the research experience of the three participating groups complementary (and also containing some useful overlap), a close collaboration among the groups is guaranteed through part-time positions: prof. Aarts from Philips Research works at the TUE for one day a week and prof. De Bra from TUE works at CWI for one day a week.
The proposed research fits the profile of the research school SIKS, but is also related to research topics that are studied in the research schools ASCI and IPA. The OIOs will participate in the doctoral program of SIKS and obtain their degree from the TUE. The applicants from TUE and CWI participate in SIKS. Some researchers from CWI also participate in ASCI and can draw required expertise from that research school as needed. In particular, there are contacts with Marcel Worring and prof. Arnold Smeulders of the University of Amsterdam. Prof. Emile Aarts from Philips and TUE operates on the intersection of SIKS and IPA. (He is a former member of SIKS and currently working in IPA.)
The proposed SION project strengthens the ongoing collaboration among TUE, CWI and Philips. This collaboration was mostly bilateral until now (TUE with CWI, TUE with Philips, CWI with Philips). The current projects explicitly aims at bringing the three partners together in a single project.
All three participants have contacts (and contracts) with numerous industrial and academic organizations within the Netherlands. They have (although not jointly between all three) been involved in research projects with others, e.g. Gigaport, ICES, PRISMA, MAGNUM and AMIS.
CWI is a member of W3C, and personnel from the Multimedia and Human-Computer Interaction (MMHCI) group are involved with the working groups: SYMM (developing SMIL), XML Linking (developing XLink and XPointer), CSS and FP (developing CSS) and HTML (chaired by a member of the MMHCI group). These links to standards organizations are important to ensure the feasibility of implementation of the design principles that will be developed.
The MMHCI group at CWI has been involved with a number of Fourth Framework projects, of which the largest was the Chameleon project. http://www.cwi.nl/Chameleon The focus of the project was to develop an authoring environment for hypermedia, resulting in the GRiNS authoring system for creating SMIL documents.
CWI is a member of the ERCIM consortium, allowing collaborations and working visits to be set up easily with partners in, for example, GMD and INRIA. There are also contacts with INA (Institut National de l'Audiovisuel) who store video archives in Paris.
The TUE applicants are part of several European networks, including AgentLink (on agent technology) and Renoir (on requirements engineering). They have (co-)organized several international workshops on adaptive hypermedia and currently host the world-wide "Adaptive Hypertext and Hypermedia Homepage", which links all researchers in this field. In the field of hypermedia access to multimedia database output there are good contacts with researchers at Rutgers University (New Jersey) who developed the RMM methodology.
As the core research center for Philips, the research group at Philips Natlab has numerous links to academic institutes world-wide.
Since this is a four year research project involving 3 main researchers (and 6 applicants) it is not possible to accurately predict exactly when which activities will take place. Also, depending on the outcome of activities in Year 1, the actual activities in subsequent years may be very different. Therefore, the given time schedule should be considered as highly speculative. However, a more important part of the schedule is that it shows the division of tasks between the three main researchers.
Year 1
TUE OIO: In this initial period the OIO will investigate the state of the
art in adaptive hypermedia and in the presentation of volatile hypermedia
database output, by working closely together with two other NWO(SION) OIOs
at TUE who are researching these two areas.
The OIO will investigate how user-model-based adaptation
(from one SION project) can be combined
with the presentation generation methods for multimedia database output
(from the other SION project).
The result should be a framework for the generation of adaptive hypermedia
presentations for multimedia database output, in the case of static media
(meaning text, images, and possibly also video but without considering
synchronization aspects).
Philips OIO: In this initial period the OIO will investigate the
multimedia and hypermedia possibilities of different environments,
including PCs, Television (with set-top boxes) and palm top devices.
This includes considering how these devices are connected to servers through
(public) networks. The OIO will investigate standards for media objects
as well as presentation languages. Based on these findings the OIO will
design a framework that defines the translations and reductions that are
necessary to present (in a "reasonable" way) different media objects on the
different platforms. The reduction will include requirements for scaling,
compression, color reduction and frame reduction but may also include
changing media, such as the replacement of a video by a slide show (keeping
in mind the synchronization between the images and the sound track),
or possibly even the replacement by a still image and/or audio only.
CWI post-doc:
In the first year the post-doc will investigate which changes and extensions
to the CWI hypermedia authoring platform are needed in order to change the
authoring process from a completely manual process into a (partially)
automated one. The research will focus on automating adaptation as needed
for the different output devices, and on the generation of templated from
database structures.
Year 2
TUE OIO: The second year is devoted to the incorporation of "real"
multimedia aspects in framework for adaptive hypermedia presentation of
database output. This means that where the presentation framework of the
first year basically builds two-dimensional structures, the new framework
will include a third dimension which is time.
Also, alternative navigation methods will be investigated for
the more restricted user-interface devices. (The related NWO(SION) project
only deals with large PC-screens which have room for the presentation of
simple graphical indexes to objects, whereas in palm tops or mobile phones
alternative ways to provide indexes are needed.)
Philips OIO: The second year is devoted to assembling the pieces
necessary to create the electronic program guide prototype.
This involves gathering information from various sources (different channels
or providers) and storing it in a central database (which is kind-of a
data-warehouse).
The OIO will design presentation structures for the EPG for different
user-interface devices and create a prototype for generating presentations
automatically from templates.
CWI post-doc:
The post-doc will implement the framework for generating hypermedia
presentations for multimedia database output in the automated CWI hypermedia
authoring environment.
The post-doc will also investigate the communication needed between clients
and servers in order to determine the part of the adaptation that is needed
in order to accomodate the different user-interface devices and communication
channels.
Year 3
TUE OIO, Philips OIO and CWI post-doc:
In this year the OIOs and post-doc will work closely together in order
to bring pieces together and possibly modify parts of their designs so that
a working prototype can be created.
This will be done by collocating the researchers for a period of 3 to 6 months.
(The research school SIKS requires OIOs to spend such a period at another
SIKS location. By colocating the OIOs at CWI they can use this requirement
to their advantage.)
Year 4
During the final year the OIOs will finish up publications and write and
defend their doctoral thesis.
The Multimedia and Human-Computer Interaction group at CWI is led by Lynda Hardman and has a history of research in tools and methods supporting the manual creation of hypermedia presentations, and has expertise in document processing languages, systems and tools. Members of the group participating in various W3C working groups are contributing to the development of HTML, CSS, SMIL and XLink. The group seeks to extend its research with more automated aspects of generating hypermedia presentations. Work has already been carried out on implementing a demonstration framework for labeling items in a multimedia database, and making network-based connections with a remote thesaurus server. This research group has strong ties with the multimedia database research performed in the Database and Data Mining Research group at CWI, led by Arno Siebes, and of which Paul De Bra (from TUE) is a part-time member. Both the Multimedia and Human-Computer Interaction group and the Database and Data Mining Research group are part of CWI's Information Systems cluster, led by Martin Kersten.
At the TUE work has been carried out on adaptive hypermedia technology. (An NWO(SION) OIO is currently active in this field.) The focus has been on the design of hypermedia applications that adapt to the knowledge state of the user in order to facilitate an effective retrieval of information. The AHAM model (De Bra et.al. 1999) has been proposed as a Dexter-based reference model for adaptive hypermedia applications. Furthermore, work has been done on the semi-automatic design and implementation of hypermedia applications for ad-hoc queries on data contained in relational databases. (This is another NWO(SION) OIO-project, for which the position will be filled medio '99.) Cooperation on these issues is established with the Information Access and Presentation research group at the IPO Research Center for User-Interaction (at the TUE).
At Philips Research work is being performed to enhance the television experience with interactivity, integrated with the support of Web access. Depending on the receiver's capabilities and the user's preferences different forms of presentation are required. The use of XML and XSL is being investigated to serve such needs. Expanding on this approach as well as investigating and comparing with other approaches leading to efficient techniques for automated generation of tailored presentations is of interest. This concerns both the fundamentals and the properties of the techniques involved, as well as the applicability for realizing next generation television systems.
Lloyd Rutledge, Lynda Hardman, Jacco van Ossenbruggen, and Dick C. A. Bulterman, "Mix'n'Match: Exchangeable Modules of Hypermedia Style", Proceedings of ACM Hypertext 99 Conference, February 1999.
Lynda Hardman, Jacco van Ossenbruggen, Lloyd Rutledge, K. Sjoerd Mullender and Dick C. A. Bulterman, "Do You Have the Time? Composition and Linking in Time-based Hypermedia", Proceedings of Hypertext 99 Conference, February 1999.
Paul De Bra, Geert-Jan Houben and Hongjing Wu, "AHAM: A Dexter-based Reference Model for Adaptive Hypermedia", Proceedings of the ACM Hypertext 99 Conference, February 1999.
Geert-Jan Houben and Paul De Bra, "Retrieval of Volatile Database Output Through Hypermedia Applications", Proceedings of the HICSS99 Conference, January 1999.
Geert-Jan Houben and Paul De Bra, "Generating hypermedia applications for volatile database output", Proc. of the AACE WebNet'98 Conference, pp. 220-225, November 1998.
Lloyd Rutledge, Lynda Hardman, Jacco van Ossenbruggen, and Dick C. A. Bulterman, "Implementing Adaptability in the Standard Reference Model for Intelligent Multimedia Presentation Systems", Proceedings of Multimedia Modeling 98, October 1998.
Lloyd Rutledge, Jacco van Ossenbruggen, Lynda Hardman, and Dick C. A. Bulterman, "Structural Distinctions Between Hypermedia Storage and Presentation", Proceedings of ACM Multimedia 98 , September 1998.
Ten Kate, W., Bulterman, D.C.A., Deunhouwer, P., Hardman, L. and Rutledge, L. "Presenting Multmedia on the Web and in TV Broadcast", Proceedings of the 3rd European Conference on Multimedia Applications, Services and Techniques (ECMAST 98), May 1998.
Licia Calvi, Paul De Bra (1997), "Proficiency-Adapted Information Browsing and Filtering in Hypermedia Educational Systems", Journal on User Modeling and User-Adapted Interaction, Vol 7, pp. 257-277, 1997, (Kluwer Academic Publishers).
Lynda Hardman, Marcel Worring, and Dick C.A. Bulterman, "Integrating the Amsterdam Hypermedia Model with the Standard Reference Model for Intelligent Multimedia Presentation Systems", Computer Standards & Interfaces, vol 18 (6-7) 497-508.
Lynda Hardman, Dick C. A. Bulterman and Guido van Rossum, "The Amsterdam Hypermedia Model: Adding Time and Context to the Dexter Model", Communications of the ACM vol 37 (2), pp 50-62, 1994.
Gwendal Auffret, Jean Carrive, Olivier Chevet, Thomas Dechilly, Remi Ronfard, Bruno Bachimont, "Audiovisual-based Hypermedia Authoring: using structured representations for efficient access to AV documents", Proceedings of ACM Hypertext 99, February 1999.
M. Bordegoni, G. Faconti, M.T. Maybury, T. Rist, S. Ruggieri, P. Trahanias and M. Wilson, "A Standard Reference Model for Intelligent Multimedia Presentation Systems" , Computer Standards & Interfaces, vol 18 (6-7) 477-496.
P. Brusilovsky (1996). Methods and Techniques of Adaptive Hypermedia. Journal on User Modeling and User-Adapted Interaction, Vol 6, 87-129, (Kluwer Academic Publishers).
F. Garzotto and D. Schwabe (1993). HDM-A Model Based Approach to Hypermedia Application Design. ACM Transactions on Information Systems, 11(1), 1-26.
Tomás Isakowitz, Edward A. Stohr, and P. Balasubramanian (1995). RMM: A Methodology for Structured Hypermedia Design. In: Communications of the ACM, 38(8) pp 34-44, August 1995.
W3C: See reports at http://www.w3.org/TR/