A Virtual Knowledge Market-Place
M. Engelhardt, A. Kárpáti, T.
Rack, TC. Schmidt
Fachhochschule für Technik und Wirtschaft Berlin, Germany
© EURODL 2001
Abstract
Networked multimedia presently revolutionises teaching
as once did the invention of blackboard and chalk. Distribution and roll out of complex
multimedia objects however suffers from significant efforts any author has to fulfil in
production.
This paper presents a prototypic implementation of a virtual marketplace for multimedia
teaching objects, which by a simple user policy allows for trading knowledge material
between teachers. Knowledge entities thereby are considered as adaptable, compound objects
consisting of media data as well as annotations, background information and meta
descriptors. Our Knowledge Marketplace is based on the more general Multimedia Information
Repository MIR. The MIR system consists of an intelligent media database with
Web-authoring tool and is based on a hypermedia data model of reusable object components.
Further educational applications of our architecture are presented, as well.
Key words
Tele-learning, Hypermedia Modelling, Multimedia
Application,Knowledge Management
Introduction
A Knowledge Market-Place
The Media Information Repository
Application Examples
Conclusions and Outlook
References
Introduction
Web-based components for teleteaching and distant learning
are more and more considered a serious application framework within today's educational
systems. Many attempts are made to establish on the one hand administrative or virtual
community environments for organising classroom equivalents in the net. On the other hand
document- or content-management-like systems are employed to deliver hyperlinked, linear
pieces of educational material to the students. Frequently concern focuses on
presentational schemes and standardisation of content distribution.
In contrast to those common efforts it is our believe that hypermedia techniques may
mainly enrich teaching, if they are grounded on a multitude of individually styled highly
interactive educational applications. Specific didactic tasks presented in a joy-oriented
fashion oppose any standardisation process in knowledge presentation. Consequently our
work presented here concentrates on a high-level framework, concepts, technologies and
tools, for easy and rapid development of specific applications. This framework under
discussion we call the Multimedia Information Repository, MIR [1]. All applications
introduced further on are based on MIR technologies.
The support of different types of multimedia data in a teleteaching environment appears of
course to be of significant relevance. A frequent shortcoming, however, results from the
reduction of sight on pure media content. Media Objects form entities of knowledge as soon
as we restrain from the simple vision of purely binary content data such as texts, images,
sounds, videos etc. As information entities they are subject to structural classification
regarding thematic aspects, topic, key words and related fields., didactic annotations
concerning complexity, order, relevance to related objectives, and intrinsic meta
information regarding format, author, access rights etc, as well. Possible further
dimensions of structural annotations, e.g. rhetorical and narrative structuring, are
discussed in [2]. An appropriate multimedia information system should offer media content
and the corresponding descriptive knowledge, referentialy related but independently
stored.
Learning modules request for a coherent design of its contributing portions of
information. The information entities, being complex compositions themselves, need to be
set into semantic relations in order to form a teaching module or application. The
meaningful shaping of such structural overlays depends on the application and belongs to
the author. Therefore a desirable information system not only should exhibit capabilities
of embedding its contents into flexible structuring but also needs to strongly support the
process of authoring in accordance to its abilities, unique points of source editing being
the most prominent feature under request. The Multimedia Information Repository System MIR
summarised in this paper tries to cope with all the above mentioned aspects. In the
following section 2 of the present paper we describe a knowledge market-place as a
lightweight application of our Multimedia Information Repository, reflect on the ideas of
the media object model and including some references on related work. Section 3 presents
the underlying MIR system in short. Further applications of the multimedia environment are
discussed in section 4. Finally, section 5 is dedicated to conclusions and outlook of the
ongoing work.
A Knowledge Market-Place
Exchanging Media Objects
In recent times it has been widely understood that the preparation of qualitative advanced multimedia material ranges far beyond the scope of individual lecturers. The access to many different types of media such as text, images, animations interplaying with time-based material i.e. audio, video or online data processing, however, is fundamental to multimedia enriched teaching. It though requests for a specialised treatment which for the non technical oriented author is hard to fulfil. Facing the demand for good multimedia supplements in teaching on the one hand and recognising the difficulties in the production of such material a "market-place"-type idea to archive, exchange and reuse the valuable material meets a quite natural demand.
Figure 1: The Knowledge Market-Place
The Knowledge Market-place (s. fig 1) collects fundamental
information entities in the sense of our introducting remarks: Media content is
accompanied by a basic set of annotations, e.g. field, subfield, topic, title, keywords,
author, date, access-rights, mime-type, etc., where semantical descriptors need to be
provided by the authors. By use of a Web-frontend entries can be supplied, searched and
retrieved, previewed and extracted for external use. Within the system media objects are
ready for immediate access and display in a classroom facility ("Get and show the
periodic table when needed") and "by means of a powerful MIR object model"
may be flexibly reused in further applications. Indeed, grounded on easy accessibility and
a simple user policy ("Get as many objects as you donate") one major objective
of the market-place is to aquire a profound archive of media objects to be reused in
additional educational applications.
Object Model and related Works
The Knowledge Market-place application was built on top of
the MIR environment, thereby enheriting its strong, adaptable object model. Media Objects
here may be seen as the central constituents to comprise the data structure of our model.
As the basic design idea a Mob is a neutral container which may carry a configurable set
of descriptors (properties) as well as references on subordinate Mobs or data. Defining an
application at first requests for turning the Mob structure into a meaningful formation,
i.e. configuring an appropriate Mob class. Semantics can be brought onto Mobs in a twofold
fashion: A predefined set of (mandatory) attributes assignes an information structure on
the data which may serve directly as user interface input or as steering quantities for
dynamic content processing. The powerful approach however lies in interpreting the active
references native to Mobs. An application designer not only can choose from arbitrary
interelations such as trees, Petri-nets, circuits, ?, but can dedicate operative
instructions to those data links ranging from a simple automated Web-link generation over
spatial and temporal construction policies up to conditional interactions within arbitrary
scenes.
Object class definition in our market-place application remains quite simple: As we are
dealing only with first-step informational units, Mobs carry their basic set of annotative
properties accompanied by a single data reference. Even though information structure
suffices for this purpose future applications of arbitrary complexity may reuse all basic
object components by stuffing them into appropriate Mob hulls. In the contrary, a simple
reuse-by-reference of media content data remains possible, of course, since all content
data stays kept in separated data objects.
Enumerous activities rank around document structuring and authoring of more complex
information models than HTML-formatting. From the educational area we exemplarily mention
the group of Maurer [3],[4], who propose and implemented the Hypermedia Composite Model as
a semantic container for learning documents. Even richer research is going on in the area
of multimedia database systems. For an excellent overview we refer to [5]. As an
interesting, elaborate approach to modeling hypermedia structures we like to mention the
Nested Context Model (NCM) of Soares et. al. [6]; see [7] for a comparative discussion of
the NCM.
Lightweight Implementation using XML
The implementation strategy chosen for the Knowledge
Market-place application is caracterised by its extraodinary leightweightness, why we like
to draw attention to it. As mentioned in the introduction the objective of the MIR system
is to provide a high-level framework for an easy and rapid development of hypermedia
applications. To achieve this goal MIR offers all functionalities for data access,
retrieval and manipulation with respect to the underlying media object model. In this way
specific implementation efforts may be reduced to a minimum.
The minimalistic ansatz especially accounts for our market-place which entirely relies on
XML technology [8]. The application logic could be defined by addressing the XML API of
the MIR system. Briefly stated this API allows for a mapping of predefined XML command
tags on our JAVA middleware (see below) by means of XSP logic sheets. The MIR-XML API is
contructed as nested command stack,
<mir:[command1]> <mir:[command2]> ... <mir:[commandn]> <param*>[value]</param*> </mir:[commandn]> ... </mir:[command2]> </mir:[command1]> ,
with the ability to process chained logical operations. Resting on this fundament the implementation tasks merely reduce to the standard definition of XML structure- (DTD) and stylesheets.
The Media Information Repository
In the present section we intend to give a brief overview
of the MIR system fundamental to our work. For a more detailed description we refer to
[9].
The Media Object Database
The core of our multimedia environment is formed by a media object database system. It combines all operations related to data storage and at the same time keeps track of information structuring ensuring referential integrity. Although MIR fully implements the media object model it remains neutral with respect to applications built on top of the database layer. The intention in designing the media repository was to provide a robust, powerful basis, on which a multitude of educational systems may be established with rather limited effort.
Figure 2: Media Object Database
The MIR database divides into two functional groups: The
Media Object Lattice and the Data Store (s. fig 2). Objects in both repositories may be
addressed by symbolic names embedded in a virtual file system. Besides administrative
information concerning owner, group and access rights, entities from the object lattice
can carry arbitrary annotations by means of an property list defined through object class
membership. Technically only distinct by data type definitions properties may contain any
kind of meta information, e. g. content descriptors such as subjects and keywords,
didactic annotators concerning presentation order or information depth, and technical
markers being specific to the educational applications on top of the database, as well.
The MIR Authoring Environment
Figure 3: The MIR authoring tool
Easy access for authors the system grants through a Web
authoring tool (s. fig. 3). It is designed to guide through the different levels of object
complexity by means of several adapted views and editing components. The user interface
ranks around a class sensitive object browser, which reflects the file system view of our
database. Mir authoring was designed following a toolbox concept, the MIR Authoring
Framework, which allows for inplugging of application- or media-specific editors and
previewers.
Architecture and Implementation
The technical concept of the MIR environment is formed by an open multimedia architecture designed according to a 3-tiered principle as is shown in fig. 4. Client access is granted in manyfold ways: On the standard side the natural CORBA/IIOP-exchange of objects is offered to intelligent client apps complemented by standard Web protocol http for all public entities in the database. Equivalently for leightweight apps the above mentioned XML interface is offered. As an important feature of the platform introduced here may be seen its ability to deal with pluggable subservers (s. fig. 4). Subserving not only opens up the field for application dependent media streams, but also allows for incorporation of new, complex functionality such as online data processing without fattening a thin applet client.
Figure 4: Networked Architecture
The current implementation of the media object database runs in a relational database management engine, a Sybase adaptive server, with special tuning applied to it. This platform we chose as a robust, very fast and lean basis. All middleware components are written in JAVA, EJBs namely, and are primarily responsible for the session and transaction management and for a buffering cache layer which allows for latency hiding. Media specific operations such as compression/decompression, streaming or synchronisation tasks are performed by the middleware components, as well, since middleware services are scalable, support load balancing and in our case accommodate caching.
Application Examples
Media Objects in Time
As one major application the teaching and presentation system Media Objects in Time (MobIT) [10,11] centres about the idea of media objects synchronizable in time which may be linked to form fairly complex presentations (s. fig. 5). But at the same time any object remains self consistent and of independent use. Roughly speaking our basic concept consists of defining appropriate media object instances and a linking meaningful with respect to a time-line.
Figure 5: Time-based presentation in MobIT
In designing an educational system within our environment
structuring has to be given an applicational meaning. In the context of MobIT this is done
by the Compound Flow Model (CFM), which takes much care to define a simple structure of
straight forward logic, intuitively appealing to document authors. The CFM organises the
uniform hull entity Mob in a tree structure, where any branch reference expresses a
temporal and spatial inclusion relation. As bound to the basic design idea of MIR Mobs
include the subordinate object reference list and a screenplay script acting on the
references, thereby describing all parameters responsible for their behaviour in time and
space. The CFM concept assures self-consistency and thereby reusability of any subtree of
the data structure.
Virtual Design
The design studio of tomorrow will not contain
a computer anymore, but will consist of the computer network. Guided by this maxim a
completely different idea of computer based educational system has been developed in
collaboration with Bildo Akademie für Kunst und Medien Berlin. Interactive picture
networking has been adopted as a basic co-operative internet platform for designers of
digital images [12,13]. The project has been honoured in the meanwhile with the "New
Talents Award" at the direct marketing congress DIMA in Düsseldorf 1998 and the
special price Multimedia Transfer at Learntec 2000.
People from art and design communicate through their visual products. As it is rather
difficult to circumscribe representational and aesthetic contents in standard language
terms, a specific way of expression needs to be utilized: A Language of Pictures. Like any
stream of statements such a visual speech needs basic order principals, a timeline and
thematic assignments at minimal.
The Virtual Design project started from the idea of supplying a networked communication
platform which allows for creation of visual dialogs. Starting from a ?white canvas? each
participant is enabled to contribute data sets consisting of an image, a title and a
textual commentary to the system. The system itself requests such contributions to be a
reaction of a former entry. It thereby links entities and lines pictures in time chains,
optionally branching at nodes which invoked multiple reactions. As time evolves the
Virtual Design system will give
Figure 6: Virtual Design Visual Navigator
rise to a tree of pictures with each branch representing a
visual dialog between authors (s fig. 6).
Relying on MIR basic environment Virtual Design MOBs enclose images, thumbnails and
textual complements. The media object structure in the virtual design application is
defined by the virtual dialogs performed by using the system and is assigned automatically
as part of the work process. Note that no separate authoring is needed since VD combines
workspace and presentation.
Conclusions and Outlook
Having started from the observation that networked
mutlimedia telelearning is much more than a standardised collection of flat html-pages we
tried to present counter examples: A market-place type exchanger for multimedia knowledge
entities, a time-based web streaming system and a virtual design studio could scetch a
variety of possible educational applications. All our work is grounded on the open,
high-level multimedia architecture MIR which is ment to grant for rapid development in the
future.
Much work, however, has to be done in this ongoing project. The prototypic runtime
environment needs a reviewing and stabilization process. Much additional functionality is
needed in the environmental middleware and the authoring toolset. Finally the system opens
the field for a new interaction model, far beyond simple content linking as html provides.
Work on this is under implementation, now. Our efforts remain simple, but in the long run
wish to help answer the question stated in [14].
References
[1] Engelhardt, M.; Feustel, B.; Kárpáti, A.; Rack,
T.; Schmidt, TC.: The MIR
Homepage, Berlin 2001.;
[2] Rutledge, L.; Davis, J.; v.Ossenbruggen, J.; Hardman, L.: Inter-dimensional Hypermedia
Communicative Devices for Rhetorical Structure; Hashimoto, S.: Proc. Multimedia Modeling
2000; World Scientific, Singapore 2000, pp. 89-105.
[3] Helic, D., Maglajlic, S., Schaerbakov, N.: The HC-Data Model: A New Semantic
Hypermedia Data Model; Proc. WebNet'99; 1999; Charlottesville, VA, USA; AACE; 493 - 498
[4] Helic, D., Maurer, H., Scherbakov, N.: Authoring and Maintaining of Educational
Applications on the Web; Proc. ED-MEDIA'99; 1999; Seatle, USA; AACE; pp. 1792 - 1797.
[5] Aberer, K., Timm, H., Neuhold, EJ.: Multimedia Database Management Systems; Borko
Furht: Handbook of Internet and Multimedia Systems and Aplications;Boca Raton, Florida;
CRC Press LLC; 1999; 285 - 310.
[6] Soares, LFG., Casanova, MA., Rodriguez, NLR.: Nested Composite Nodes and Version
Control in an Open Hypermedia System; Intern. Journ. on Information Systems; 20; 1995; 6;
Elsevier; pp. 501-519;
[7] Antonacci, M.; Muchaluat-Saade, D.; Rodrigues, R.; Soares, LFG.: Improving the
Expressivness of XML-based Hypermedia Authoring Languages; Hashimoto, S.: Proc. Multimedia
Modeling 2000; World Scientific, Singapore 2000, pp. 71 - 88.
[8] Extensible Markup Language (XML) 1.0; W3C Recommendation, Feb. 1998; www.w3.org/TR/REC-xml.
[9] Feustel, B.; Kárpáti, A.; Rack, T.; Schmidt, TC.: An Environment for Processing
Compound Media Streams; Informatica; 25, 2; July 2001; pp. 201 - 209.
[10] Feustel, B.; Schmidt, TC.: Time Directed Internet Teleteaching; Proc. of ICL2000;
Kassel 2000.
[11] Feustel, B.; Schmidt, TC.; Marpe, D.; Palkow, M.; Cycon, HL.: Compound Media
Streaming in Time; Vaclav Scala: Proc. 9-th Int. Conf. Comp. Graph., Visual and Comp.
Vision WSCG'2001; Plzen; 2001.
[12] Kárpáti, A.; Löser, A.; Schmidt, TC.: Interactive Picture Network; IEE Electronics
& Communications; 99; 1999; 109; London; 1999; pp. 18/1-18/8.
[13] Born, T.; Hannecke, T.; Heine, A.; Kárpáti, A.; Kemnitz, T.; Löser, A.; Schmidt,
TC.; Vestner, J.: Virtual Design Update; 27-99; FHTW-Transfer; Berlin 1999.
[14] Maurer, H.: Can WWW be succesful?; IFIP - DS8; Jan. 1999.
Author(s):
Michael Engelhardt
FHTW Berlin, Hochschulrechenzentrum
Treskowallee 8, D-10318 Berlin, Germany
engelh@fhtw-berlin.de
Andreas Kárpáti
FHTW Berlin, Hochschulrechenzentrum
Treskowallee 8, D-10318 Berlin, Germany
karpati@fhtw-berlin.de
Torsten Rack
FHTW Berlin, Hochschulrechenzentrum
Treskowallee 8, D-10318 Berlin, Germany
rack@fhtw-berlin.de
Thomas Schmidt, Dr.
FHTW Berlin, Hochschulrechenzentrum
Treskowallee 8, D-10318 Berlin, Germany
schmidt@fhtw-berlin.de