A Model-Driven Approach to Content Repurposing - Semantic Scholar

1 downloads 171 Views 73KB Size Report
Web standards such as the HyperText Markup Language (HTML), the eXtensible Markup Language (XML), and. Cascading Style Sheets (CSS) can help content ...
A Model-Driven Approach to Content Repurposing Željko Obrenovic, Dušan Starcevic University of Belgrade Bran Selic IBM Rational Software

Abstract In this paper we present a model-driven approach to content repurposing. Our work is inspired by the OMG’s Model Driven Architecture (MDA). The MDA is an approach to system specification and interoperability based on the use of hierarchically organized formal models. The MDA approach uses a platform-independent base model (PIM), and one or more platform-specific models (PSM), each describing how the base model is implemented on a different platform. In this way, the PIM is unaffected by the specifics of different implementation technologies, so it is not necessary to repeat the process of modeling an application or content each time a new technology or presentation format comes along. The essence of our solution is the Multimedia Metamodel (MM), which defines various platform-independent multimedia concepts. Using the Multimedia Metamodel we have explored novel design approaches for content repurposing solutions. The metamodel can be used as a conceptual basis for content repurposing and creation of platform-independent multimedia content allowing authoring for device and network independence. Using the Multimedia Metamodel, it is also possible to add metadata to existing content, which can then simplify content analysis to aid repurposing. Keywords: content repurposing, multimedia, model-driven architecture

1. Introduction Internet-enabled cell phones, PDAs, desktop, laptop, and wearable PCs have quite different requirements and presentation capabilities. Much online content was created independently of these modern devices and is often not suited to them. If we want to access this content efficiently through a broad variety of pervasive -access devices, we need to rethink how it is specified and created. Content repurposing can help us create a broad content base for various device profiles by reusing existing content and adjusting it to fit new situations [1]. However, leveraging content repurposing to its full potential is a challenging task. The number of device profiles available for accessing online content is measured in the hundreds. A profusion of different formats and standards for online content makes the situation even worse since there is no consensus on how to unify them [2]. The problems listed here exist in other domains too. Therefore, it is useful to analyze pertinent methods and experiences from these domains. For example, in the software development community, there is a general need for exchanging data among tools from different vendors. To allow efficient collaboration, the model-driven approach accounts for great diversity among data structures to be exchanged [3]. Can we borrow something from such solutions? In this paper we present a model -driven framework for content repurposing. The essence of the proposed solution is the Multimedia Metamodel, which introduces concepts and mechanisms used in the multimedia domain. Using this Multimedia Metamodel we have explored novel design approaches for content repurposing solutions. In next section we give a short overview of existing solutions of interest to our work. Then we present the core of a model-driven approach to content repurposing. After that, we describe in more detail the Multimedia Metamodel. Following that, we explain how a model-driven approach can be applied to c ertain content repurposing problems.

2. Existing solutions Many existing multimedia authoring tools provide limited content repurposing by allowing transformations of their native format to and from other formats. However, using these export/import content bridges is unsuitable for the highly dynamic online world. Developing content bridges is a difficult and costly process, as bridges must have detailed knowledge of proprietary formats and interfaces. If we want to allow broader content repurposing among various platforms, the number of bridges can be up to N2-N, where N is the number of target platforms [4]. Furthermore, the processing logic of a particular bridge is not necessarily reusable in the construction of other bridges, which greatly increases the development cost. Web standards such as the HyperText Markup Language (HTML), the eXtensible Markup Language (XML), and Cascading Style Sheets (CSS) can help content providers to face challenges of multipurpose Web publishing. The World Wide Web Consortium defined these document markup and style-sheet languages in the interests of Web device independence, content reuse, and network-friendly encoding [5]. Structured documents with style sheets allow the same document to be presented on a variety of Web devices. HTML, XML, and CSS are declarative data formats, which are easily converted to other formats. They are more likely to be device-independent and tend to live longer than programs. However, those standards cover only a subset of multimedia content space and usage scenarios. Researchers in the multimedia field have made considerable attempts toward separating knowledge content from presentation. Thus, Ram et al. used the Procedural Markup Language (PML) for this purpose [6]. PML allows specification of knowledge structures, the underlying physical media, and the relationships between them using cognitive media roles. The PML specification is translated into various presentations depending on the context, goals, and user expertise. However, these solutions do not directly address the problem of existing content and content repurposing. Posnak, Lavender, and Vin proposed a framework that simplifies the development of multimedia software components by facilitating the reuse of code, design patterns, and domain expertise [7]. Their solution enables the components to dynamically adapt the quality of the presentation to the available resources in heterogeneous environments. This framework is primarily aimed at decoding and processing of digital audio and video data. Fraternali proposed a methodology and tools for conceptual development of online Web applications. He defined an abstract notation for the specification of structure, navigation, and presentation semantics [8 ]. However, this approach is more suitable for conventional Web access and for data-oriented Web applications, and is not directly applicable to online access for pervasive devices.

3. A model-driven approach Our work is inspired by the OMG’s Model Driven Architecture (MDA). MDA is an approach to system specification and interoperability based on the use of hierarchically organized formal models [3, 9]. It uses a platform-independent base model (PIM), and one or more platform-specific models (PSM), each describing how the base model is implemented on a different platform [10]. In this way, the PIM is unaffected by the specifics of different implementation technologies, and it is not necessary to repeat the process of modeling an application or content each time a new technology or presentation format comes along. The essence of proposed solution is the Multimedia Metamodel, which introduces concepts and mechanisms used in the multimedia domain that are of interest for content repurposing. Consequently, we have focused on two main topics: •

The design of the unified Multimedia Metamodel, in which we have synthesized the knowledge and common concepts from different multimedia domains into a single uniform view;



Applying the Multimedia Metamodel to content repurposing problems.

Architecture-wise, the Multimedia Metamodel is placed within the metamodel layer (Table 1). The role of the Multimedia Metamodel in our solution is similar to that which the Common Warehouse Metamodel (CWM) plays in the domain of integration of business data [11]. The need for a more generic multimedia framework exists, as there is a lot of similarities and overlaps between various multimedia content standards, such as MPEG-4, Virtual Reality Modeling Language (VRML) and Synchronized Multimedia Integration Language (SMIL) [12].

Table 1. Mapping the Multimedia Metamodel to the OMG's MDA levels. OMG MDA Level

Multimedia Metamodeling Architecture

Description

M3 Metametamodel

The Meta Object Facilities (MOF)

The MOF is an OMG standard that defines a common, abstract language for the specification of metamodels. MOF is a meta-metamodel - the model of the metamodel, sometimes also called an ontology.

M2 Metamodels

The Multimedia Metamodel (MM)

The Multimedia Metamodel provides a common and standardized language for sharing and reusing knowledge about phenomena from various domains relevant to the design of multimedia solutions. It is called a metamodel as it is the abstraction (model) of platform specific models.

Platform-Specific Schemas (XHTML, WML, SMIL Schemas…)

Platform specific models of multimedia content.

M1 Models M0 Objects, data

Content data (WML, XHTML, SMIL files…)

Instances of platform specific models.

The Multimedia Metamodel serves as a conceptual basis for various content repurposing solutions. Firstly, it provides means for constructing uniform high-level views on various existing platforms, that can be used for repurposing or integration of platform-specific content. Additionally, with the concepts from the metamodel it is possible to create new content in platform-independent terms, or to add general metadata to existing content, which can simplify content analysis to aid repurposing. The views on multimedia content from different levels of abstraction can be derived by model transformations. The question of model transformations lies at the center of the MDA approach, and, therefore, we use it as a basis for model-driven content repurposing. In MDA, platform-independent models are initially expressed in a platformindependent modeling language, and are later translated to platform-specific models by mapping the PIMs to some implementation platform using formal rules. Figure 1 illustrates the basic idea of model-driven content transformation. As pointed out in the previous section, bridge transformations work directly with platform-specific metamodels (schemas). In MDA, however, the greatest possibilities lie in more general approaches. The transformation of the content models can be specified by a set of rules defined in terms of the corresponding higherlevel metamodels. The transformation engine itself may be built on any suitable technology such as XSLT tools.

MOF Meta-metamodel



MM Metamodel Domain #1



Transformations Rules

MM Metamodel Domain #2

PSM-PIM Mapping #1

PSM-PIM Mapping #2

Platform Specific Metamodels #1

Bridge

Platform Specific Metamodels #2

Figure 1. Models and transformations. For example, the first platform-specific model could be an XHTML schema, whereas the second model could be some of the speech synthesis platforms such as a Java Speech API. Building of (meta)model-driven transformation tools is much more efficient than bridge building as the number of modules is linearly dependant on the number of target platforms. For each new platform, it is necessary to develop only two new modules that map new format into the platform independent form and vice versa. As we already mention, the number of bridges can be up to N2 -N, where N is the number of target platforms [4]. While it is efficient to develop bridges for communication between two or among three formats, with five formats the number of bridges is 20, while with 10 formats you need to develop 90 bridges. With only 20 target formats you need 380 bridges. On the other hand, the number of model driven transformations for 5, 10, and 20 formats is 10, 20, and 40 respectively. Although conceptual design of multimedia content is not a new approach, a model-driven approach brings many advantages. Our approach is based on standard technologies such as the Unified Modeling Language (UML) and XML, which are familiar to many software practitioners and are well supported by tools. Therefore, it is not necessary to develop complex solutions from scratch, and it is possible to reuse existing model-driven solutions and experiences from other domains. In our work we rely on existing UML modeling tools, XML parsers and software frameworks, developing only code that extends, customizes, and connects those components according to common and standardized language defined in the Multimedia Metamodel. It is also important to mention that the concept of a metamodel is strongly related to the concept of an ontology used in knowledge management communities [13]. Thus, the Multimedia Metamodel can be viewe d as an ontology that provides a uniform view on important relations among multimedia concepts. Unification of different views on the field can give us a “fresh view” that could raise new ideas and approaches. The unified ontology of multimedia concepts can provide the context where we could perceive many relations that are not always obvious. In this way the Multimedia Metamodel could be used as a basis for collaborative development of knowledge about phenomena in multimedia [14]. Finally, it is important to note that by using metamodels and ontology concepts we inherit the possibility to use other knowledge management technologies such as data mining, which can help us in finding interesting patterns in data.

4. The Multimedia Metamodel The Multimedia Metamodel provides a common and standardized language for sharing and reusing knowledge about phenomena from various domains relevant to the design of multimedia solutions. Our aim was to identify a basic set

of concepts from each of the domains and the relations among them. Speaking practically, we wanted to achieve the following: •

A set of precisely defined terms and structured definitions of multimedia domain concepts instead of mere text-based information;



High expressiveness, enabling us to efficiently describe each aspect of multimedia;



Coherence and interoperability of resulting knowledge bases, using standard modeling and storage technologies;



Scalability of metamodels providing us with means for definition at different levels of abstraction.

We have used the MOF, a subset of UML, for description of the metamodel [3]. UML is an open standard and has standard mechanisms for defining extensions for specific applications contexts. We have used these mechanisms extensively in our approach. In addition, UML is widely adopted, and is familiar to many software practitioners, widely taught in undergraduate courses, and supported by many books and training courses [15]. We partitioned the Multimedia Metamodel into four packages (Figure 2): Physical Foundations, Computing Factors, Human Factors, and Content Repurposing Use Cases. Each package represents a different viewpoint on multimedia content and content repurposing.

Content Repurposi ng Use Cases

Content Transformation

Content Analysis

Computing Factors

Personalization

Human Factors

Logical Foundations

Multimodal Communication

Human Perception

Presentation Platforms

Human Cognition

Social Interaction

Physi cal Foundations

Physical Media

Figure 2. The Multimedia Metamodel structure.

Presentation Devices

Human Sensory Physiology

4.1 Physical Foundations package The Physical Foundations package introduces concepts for modeling of physical properties of the real world of interest for multimedia presentations. It consists of three packages: Physical Media, Human Sensory Physiology, and Presentation Devices. In the Physical Media package we have defined concepts of a stimulus and its sensible properties. Using this model we have described various stimuli, such as light and sound, as well as their characteristics, such as intensity (amplitude) or frequency (pitch). The Human Sensory Physiology package contains concepts that can be used to describe elements of the human sensory apparatus involved in processing of a stimulus. Using this knowledge, for example, it is possible to simplify a presentation of data excluding those signals that humans are not capable of sensing, exploiting effects such as frequency or time masking [16]. The Presentation Devices package defines the metamodel of multimedi a presentation devices (Figure 3a). This package is connected with the Physical Media package, as we define a presentation device as a stimulus source. The presentation device can be simple or complex, where the complex device integrates two or more other devices using merge or layout integration mechanisms. A merge mechanism creates new presentation device fusing two simpler ones. A layout presentation device arranges two or more devices on a common panel. With this package we have described various presentation devices used in our research, such as a raster screen display shown in Figure 3b. PresentationDevice

PixelPart



GreenPixel

BluePixel

Light

1..*

RedPixel

PrimitiveDevice

ComplexDevice



medium

IntegrationMechanism Pixel

BasicStimulus Layout

Merge

*



RefreshFrequency

Panel

(a)

RasterScreen



RasterMatrix

(b) Figure 3. Simplified presentation device metamodel (a) and an instance of it on the example of a raster screen display (b). 4.2 Computing Factors package The Computing Factors package introduces basic multimedia concepts from the computing technology world grouped in two packages: Logical Foundations, and Presentation Platforms.

The Logical Foundations package extends modeling elements with basic logical multimedia concepts [12]. It consists of four packages: Presentation Dimensions, Coding, Composition, and Quality-of-Service (QoS). The Presentation Dimensions package includes concepts of spatial and temporal dimensions of a presentation. These concepts include coordinate systems, and time bases. The Coding package defines concepts that define physical representation of content. For example, the same content can have textual or binary encoding. The Composition package defines mechanisms for seamless combining of multimedia objects of various types. This package defines temporal composition (synchronization), as well as spatial composition mechanisms. The QoS package defines means for quantifiable description of desired quality-of-service (QoS) of a presentation platform [15]. The Presentation Platform package consists of modules that define generic descriptions of standard presentation platforms. These concepts are derived by generalization of existing platforms. It includes five packages: 2D Graphics Platform, 3D Graphics Platform, Textual Platform, Text -to-Speech Platform, and Audio Platform. The 2D Graphics Platform and the 3D Graphics Platform packages define various standard elements for two -dimensional and threedimensional graphics such as shapes and transformations. The Textual Platform package defines common elements of various textual platforms such as HTML or Rich Text Format (RTF). The Text -to-Speech Platform package defines basic elements of speech synthesis platforms, while the Audio Platform package defines common concepts for work with non-speech audio. 4.3 Human Factors package The Human Factors package defines various concepts of interest for user interactions with online multimedia content. It consists of four packages: Human Perception, Human Cognitive Mechanisms, Social Interaction, and Multimodal Communication. Table 2 summarizes first three packages. In the Human Perception package we have identified perceptual concepts often used in contemporary user interfaces. For example, highlighting lets humans notice elements that stand out among the others. The Cognitive Mechanisms package includes several of the mechanisms that were recently used in HCI research such as spatial memory, attention, and curiosity. The Social Interaction package includes definitions of some common concepts found in online communities such as avatar [17]. Other categories in the table have similar explanations. Table 2. Human Perception, Human Cognition and Social Interaction metamodels. Metamodel

Human Perception

Sample categories of concepts (classes) Pattern recognition

Sample subcategories of concepts (subclasses) Visual pattern recognition

Grouping

Audio pattern recognition Visual grouping

Highlighting

Audio grouping Visual highlighting

3D cues

Audio highlighting Visual 3D cues

Examples (objects) Shape, letter, face Melody, phoneme Grouping by similarity, motion, texture, symmetry, proximity, parallelism, closure, good continuation Pauses between words and sentences, color of a voice Highlighting by color, polarity, brightness, orientation, size, motion, flicker, depth, shape

Human Cognition

Memory

Spatial memory

Highlighting by intensity, rate, speed Stereo vision, motion parallax, linear perspective (converting lines), relative size, shadow, familiar size, interposition, relative height, and horizon The inter-aural time (or phase) difference, the interaural intensity (or level) difference, Head Related Transfer Functions (HRTFs), a head movement, the echo, attenuate of high frequencies for distant objects Arranging documents in 3D model of a office

Attention Context Goals

Social Interaction

Person Avatar

Attention by pop-out Context + details Long-term goals Short-term user goals Person's descriptions Username 3D character

Attention by motion Exploring hierarchical structures Overall aim of user's interaction Current task in the application User profiles Line prefix in textual chat Network games

Audio 3D cues

Collaborative activity Session Space

Collaborative filtering

Recommendations for books or music albums

Temporary session Persistent session Physical space

Chat, video conferences, Email, discussion groups, news Home, job, classroom

The Multimodal Communication package defines the concept of a computing mode and multimodal communication. We defined a computing mode as a form of interaction that is designed to engage some human capabilities. The main concept in our ontology is the concept of a computing mode. We defined a computing mode as a form of interaction that was designed to engage some of human capabilities. While designing a user interface the de signer defines an interactive language that determines which messages and levels will be included in the interaction. We have classified messages that a mode can send in three main categories: sensual, perceptual, and cognitive . A computing mode can be simple or complex. A complex computing mode integrates other modes to create simultaneous use of various modalities, while a simple mode represents a primitive form of presentation [18]. 4.4 Content Repurposing Use Cases package The Content Repurposing Use Cases package provides concepts for description of service functions that can support operation and management of multimedia content. As this package defines mechanisms for transformation, analysis and personalization of multimedia content it can make the Multimedia Metamodel an active, well-integrated part of content repurposing solutions. Consequently, the package is organized in three packages: Content Transformation, Content Analysis, and Personalization. The Content Transformations package provides mechanisms for defining content conversions divided into two groups: mappings and transformations. Mappings are used between abstract layers, whereas transformations are used within them (see Figure 1). For example, it is possible to define mappings between platform-specific models, such as XHTML, and corresponding higher-level metamodels, such as the textual platform metamodel. Conversely, it is possible to define transformations between various presentation platform metamodels at the same level of abstraction, such as between textual and text-to-speech platform metamodels. The generic mechanisms defined in this package can serve as a foundation for more elaborate solutions. The Content Analysis package defines a framework for various examinations of content, such as data mining. The results of these analyses can be very valuable for the evaluation of content, especially if they are connected with perceptual and cognitive metamodels. It is often the case that contemporary user interfaces fail to create perceptually feasible artifacts. Analysis may be made upon various criteria such as perceptual, ergonomic, or QoS. The Personalization package defines abstract mechanisms for customization of multimedia content based on user profiles. Individuals process sensory stimuli in their own unique ways, and each individual has his or her own preferred sensory mode, which may change depending on context [19]. Therefore, to achieve effective personalization, this package correlates various concepts of human factors, media characteristics and presentation platform possibilities.

5. Applying the model-driven solution to content repurposing In this section we explain how a model-driven approach can be applied to two content repurposing problems: perceptual content repurposing, and designing of new multimedia content using high-level Multimedia Metamodel terms. 5.1 Perceptual content repurposing Perceptual content repurposing demonstrates the repurposing strategy that attempts to keep the perceptual preferences of the original content. To illustrate this, table 3 gives simple examples of possible mappings of some high-level perceptual concepts into primitives available on textual platforms such as XHTML or WML, and text-tospeech platforms such as JavaSpeech, Microsoft Speech API and VoiceXML.

Table 3. One possible mapping of some perceptual effects to textual and text-to-speech platforms. Effect

Textual platforms

Text-to-speech platforms

Pattern recognition

Letter and word recognition

Phoneme, spelling and words

Grouping

Grouping by similarity (font family) Grouping of letters by proximity (words)

Gender, age, name Word breaking rules

Grouping by parallelism (rows of text) Grouping of words by proximity (paragraphs) Grouping by closure and surroundedness (frames and tables)

Sentence breaking rules Silence

Size (bold and font size) Orientation (italic)

Emphasizing of text Volume of a voice

Color (color of text) Flicker (blinking text)

Rate and speed of a voice Pitch of a voice

Interposition (z-index) Shadow

Available stereo formats

Highlighting

3D cues

Based on this mapping, we have been developing simple tools that parse XHTML and HTML files, create a new file with higher-level markups using terms such as group and highlighting, and which than transforms this high-level markup into text-to-speech presentation. Figure 4 illustrates this transformation on the example of XHTML, Microsoft Speech API XML, and Wireless Markup Language (WML). PIM XML fragment

XSL Transformations

This is sample text . />







? This is sample text . XML SAPI

?

This is sample text .



WML

Figure 4. Simplified version of a possible XML based implementation of perceptual content repurposing among HTML, Microsoft SAPI XML, and WML formats. The transformations are made using XSL. The figure shows XSL transformations which translates XHTML into platform independent form, and two transformations that translate platform independent format into XML SAPI and WML format respectively.

5.2 High-level modeling of user interfaces We now describe an example of designing new multimedia content using high-level Multimedia Metamodel terms. Most existing multimedia content is described at a relatively low level of abstraction, so it is not always possible to determine the original intent of the author. Higher-level models of multime dia interfaces can help us to keep track of the content author’s aims. As a general rule, it is easier to repurpose the content if the intention of the author is clear. As an example, we describe the environment for experimental evaluation of multimodal feedback in dynamic pursuit-tracking tasks. The dynamic pursuit-tracking paradigm of interaction has valuable applications in fields of surgery, night vision or low-visibility mission navigation, and flight navigation and orientation. We have examined the effects of multimedia presentation and multimodal feedback in this task, extending the visual feedback with various sonification paradigms [20]. Figure 5 shows the high-level model of our multimodal pursuit-tracking environment. We used the extension mechanisms (stereotypes) of the UML to describe such models

Headphones

Localizationin3DSpace



PerceptionOfIntensity



StereoSound

DistanceSound

MultimodalFeedback



ImproveTaskMentalPicture

Background Monitor

AnimatedTarget Target MotionDetection

Figure 5. A model of a multimodal pursuit-tracking environment described using UML stereotypes defined from the Multimedia Metamodel.

The multimodal feedback in the pursuit-tracking interface is a complex presentation mode that integrates a visual presentation, and two audio modes. A visual presentation integrates a static background presentation and an animated target to attract the user’s visual motion detection perceptual mechanism. The first audio mode is designed to produce three-dimensional stereo effects using stereo cues, while the second audio mode changes the intensity of sound to warn a user about the error that he or she introduces. High-level models may be used for automation of some phases of the design of new multimedia interfaces. For example, we have been developing tools for generation of Java -based multimedia interface frameworks. These tools take as an input the XML description of a high-level model exported from UML models, parse it, and produce Java code files with an abstract Java framework that represents a skeleton of a designed user interface. A designer can make use of generic mechanisms supported in the framework, and extend this framework with platform-specific implementations of modalities.

6. Conclusion Inspired by the OMG’s Model Driven Architecture (MDA) we have created an architecture for content repurposing. The essence of our solution - the Multimedia Metamodel – defines general multimedia concepts and mechanisms, and serves as a conceptual basis for solutions of various content repurposing problems. Although the conceptual design of multimedia interfaces is not a new approach, a model-driven approach brings many advantages. Since MDA is based on standard technologies, which are familiar to many software practitioners and are well supported by tools, it is not necessary to develop complex solutions from scratch, and it is possible to reuse existing model-driven solutions and experiences from other domains. The model-driven approach and proposed design solutions may be very useful for designers and researchers in the multimedia field, as well as for lecturers and students of multimedia courses. The unified metamodel of multimedia can provide them with a context of multimedia concepts where they can see many relations that are not always

obvious, while model-driven tools can help them create better multimedia interfaces. The metamodel could also be used as a basis for collaborative creation of broader knowledge about phenomena in the multimedia field. Our future work will include extensions of the proposed metamodel and inclusion of domains from related fields, such as user modeling and intelligent tutoring systems, which deal with high-level models of users. We are also working on the design of multimodal test environments, reusable multimedia components, and data mining tools for evaluation of various aspects of multimedia and multimodal communication, which can exploit the benefits of this model-driven approach.

7. References 1.

G. Singh, “Media Spaces”, IEEE Multimedia, IEEE CS Press, April-June 1999, pp. 18-19.

2.

R. Gonzales, “Disciplining Multimedia”, IEEE Multimedia, July-September 2000, pp. 72-78.

3.

OMG Model Driven Architecture resource page. (2003). Needham, MA: Object Management Group. Internet: http://www.omg.org/mda/index.htm.

4.

S. Brodsky, “XMI Opens Application Interchange”, IBM White paper, http://www-3.ibm.com/software/awdtools/ standards/xmiwhite0399.pdf, 1999.

5.

H.W. Lie and J. Saarela, “Multipurpose Web Publishing Using HTML, XML, and CSS”, Comm. of the ACM , ACM Press, Vol. 42, No. 10, October 1999;

6.

A. Ram et al., “PML: Adding Flexibility to Multimedia Presentations”, IEEE Multimedia, IEEE CS Press, April-June 1999, pp. 40-52.

7.

E.J. Posnak, R.G. Lavender, and H.M. Vin, “An Adaptive Framework for Developing Multimedia Software Components”, Comm. of the ACM, ACM Press, Vol. 40, No. 10, October 1997, pp. 43-47.

8.

Piero Fraternali and Paolo Paolini, “Model-Driven Development of Web Applications: The Autoweb System”, ACM Trans. Information Systems , ACM Press, Vol. 28, No. 4, October 2000, Pages 323–382.

9.

J. Poole, "Model- driven architecture: Vision, standards and emerging technologies", Workshop on Adaptive ObjectModels and Metamodeling, ECOOP 2001, April 2001. http://www.cwmforum.org.

10.

Bezivin, J., “From Object Composition to Model Transformation with the MDA”, Proceedings of TOOLS, IEEE CS Press, Santa Barbara, USA, August 2001.

11.

John Poole, Dan Chang, Douglas, David Mellor, Common Warehouse Metamodel, OMG Press, 2002, ISBN 0-47120052-2

12.

Stefano Battista, Franco Casalino, Claudio Lande, “MPEG-4: A Multimedia Standard for the Third Millennium, Part 2”, IEEE Multimedia, IEEE CS Press, January-March 2000, pp. 76-84.

13.

M. Gruninger and J. Lee, “Ontology Applications and Design”, Comm. of the ACM , ACM Press, Vol. 45, No. 2, February 2002, pp. 39-41.

14.

C.W. Holsapple, K. D. Joshi, "A collaborative approach to ontology design", Comm. of the ACM, ACM Press, Vol. 45, No. 2, February 2002, pp. 42-47.

15.

Bran Selic, “A Generic Framework for Modeling Resources with UML”, Computer, IEEE CS Press, June 2000, pp. 64-69.

16.

K.R. Boff, L. Kaufman, and J.P. Thomas, (Eds.). Handbook of Perception and Human Performance: Sensory Processes and Perception. Vols. 1 and 2. Wiley, New York, N.Y., 1986.

17.

L.M. Jessup and D. Robey, “The Relevance of Social Issues in Ubiquitous Computing Environments”, Comm. of the ACM, ACM Press, Vol. 45, No. 12, Dec. 2002, pp. 88-91.

18.

M. M. Blattner, E.P. Glinter, “Multimodal Integration”, IEEE Multimedia, Winter 1996, pp. 14-24.

19.

E.P.D. Pednault, “Representation is Everything”, Comm. of the ACM , ACM Press, Vol. 43, No. 8, August 2000;

20.

Z. Obrenovic, D. Starcevic, and E. Jovanov, “Experimental Evaluation of Multimodal Human Computer Interface for Tactical Audio Applications”, Proc. IEEE Int'l Conf. Multimedia and Expo (ICME 2002), Lausanne, Switzerland, 2002, Volume 2, pp. 29-32.