Work in Progress: Visualizing a Semantic Search in a Room based Knowledge Space
Nadine Ludwig Berlin Institute of Technology MuLF - Center for Multimedia in e-Learning and e-Research Straße des 17. Juni 136, 10623 Berlin, Germany
[email protected]
Abstract
get access to her/his room. Furthermore he can constitute several rights level for actions in and through the room.
A semantic network acts as an expedient foundation to search for objects in a cooperative room-based knowledge space. But to make the search results useable and give users
1.2. Problems of Cooperative Work and Distributed Learning Objects
an intuitive instrument the search interface has to stick to several principles of usability and the search results have to
Thus cooperative knowledge spaces enable users to work
be visualised in an appropriate way. The paper proposed
jointly on content objects and concurrently communicate
is dealing with the problems related to the visualization of
about their work but also give a certain latitude to create
semantic networks and provides a concept for this eld of
rooms and content all over the platform. This fact leads to a
research.
This concept includes the analysis of different
variety of content objects - content pages, pictures, videos,
navigation designs as well as a comparison of different al-
chat logs, forum threads - scattered randomly on the plat-
ternative presentations of search results.
form - whereas all these objects could be interesting for users. But a multitude of these objects can remain undiscovered although they also could benet from the cooperation
1. Motivation
and collaboration of the diversity of users. Of course, the
1.1. Cooperative Room-Based Knowledge Spaces
powerful text-based search to let their users nd what they
most popular and successful elearning platforms are using
In recent years we witness a great success of e-learning platforms like Moodle [9] or ILIAS [6]. Besides this success a huge community of developers worked on the constant evolution of these platforms.
The platforms in the
beginning mainly content-oriented are now complemented
need to nd. But what if you don't know what to search for? A semantic linkage of all objects in the cooperative knowledge space concatenated with a meaningful presentaion of this network enables users to nd and search for content objects they could be interested in.
2. Semantic network of objects
with tools for communication and embedded wiki-concepts. One of the recent development is the provision of content
A semantic network is build as a result of the proporties
and information in so called course rooms. The main idea
of its nodes and their relations to each other. In a room-
of the room concept is to emphasize the growing aspect
based knowledge space the objects can correlate in a mul-
of cooperative work.
Nevertheless the concept of rights
titude of ways. Therefore different types of objects have to
within this room concept is still aimed at a hierarchic system
be dened. This work distinguishes ve different types of
of lecturers and students.
objects:
In contrast cooperative knowl-
edge spaces (cf. [5, 2]) are based on the concept of selforganization [12]. Every user has the right to create rooms, generate content, upload documents and link them together. But the concept of right contains also the possibility, that the creater of a room is able to decide, if and how much users
•
Room
•
User
•
IPP - Internal Produced Page (content page)
platform. E.g.:
•
an user edits an IPP - the association isEditedBy with the user and the IPP as association members is created OR
•
an user enters a room - the association isEnteredBy with the user and the room as members is created
This concept of associations based on topic maps is a very subjective approach the associations are created according the behaviour and the actions of the users on the platform. But usually content (IPPs, EPPs, labs) are also related by the topic they are about - which is a much more objective way of a relation. Therefore the objects will be ded-
Figure 1. Semantic Network of Objects
icated to nodes in an ontology of sciences (like physics, biology, literature etc.). As sciences are mostly split up in categories and sub-categories where a sub-category from one branch can be related to another category or sub-category from a different branch this ontology represents a second semantic network (besides the topic map network described above).
These two networks are stored separately in a
database and the important challenge is to bring them together for a semantic search.
3. Visualizing a semantic search
3.1. Information visualization The storage of the data conveniently in regard to the form and structure is only one important part of the work. In order to provide the created data in a meaningful way
Figure 2. Associations between Objects
the developed network has to be ready for interrogation, search and representation.
•
EPP - External Produced Page (pdf, doc, jpg, avi etc.)
•
virtual laboratory (as Java applets)
Therefore initially some gen-
eral considerations on aspects of Information Visualization have to be done. Information Visualization is the use of computer-supported, interactive, visual representations od abstract data to amplify cognition. [1] Thus the approach
The different types of objects are assigned with differ-
of this work is to amplify users' cognition on how the mul-
ent sets of metadata. Initially the metadata sets are based
titude of objects in a room-based system is related to each
on the Learning Object Metadata element set (LOM). This
other which of these objects could be interesting for the
metadata does not sufce from various points of aspects es-
user. Therefore it is important to use the knowledge of in-
pecially when it comes to describing a virtual experiment
formation visualization to present the semantic network(s)
in a meaningful way. Therefore the LOM element set was
of objects in the most meaningful way [10].
extended by a more technical category lab. (see [8] for more information on this issue)
3.2. Concept of Selection
The semantic network composed of instances of the different types of objects (gure 1) is conducted automatically while users are moving throughout the platform and oper-
The activity of selecting one object from many that are
ate with the objects. In this concept eleven different associ-
available is a common one. [10] Initially the user has to
ations with different relationship members are designated
decide which object to use as initial object for a semantic
- see gure 2. These associations are based on the concept
search. For example, the user found an IPP very interest-
of the Topic Maps 2.0 Standard [7]. Every association is
ing and wants information, there is a decision to make what
afliated with a method or process that can be done on the
object would be the best initial object: the IPP itself, the
2
Thus the important research issues of this work in progress are the conjunction of the two semantic networks described above for a meaningful semnatic search, solving the presentation problems considering the knowledge from information visualization and usability aspects, and the decision on how much selection the user will have to do and how much will be left to the system.
4. Innovation & Outlook The innovative aspect of this Work in Progress lies in the combination of cooperative learning and work in roombased systems with aspects of the Semantic Web.
Figure 3. Semantic Search for Objects
E-
Learning platforms increasingly focus on supporting cooperative work of their students but on the other hand most cooperative room-based platforms lack of meaningful con-
author, the room the IPP is stored in. Is this a decision the
cepts for the content-oriented aspects for this type of work.
user has to make on his/her own or is it necessary to support
This research approach is part of a dissertaion about modu-
the user to provide the best results?
larization and Semantics in Cooperative Knowledge Spaces
Once this selection is made the search form has to pro-
is intended to be nished at the end of 2009.
vide different options the searching user can choose from. Is it reasonable to give the user the whole range of search
References
options or is it a better way to constrain these options and leave it to the system to give the best results? In the latter case probably a much more powerful search algorithm is
[1] S. K. Card, J. D. Mackinlay, and B. Shneiderman.
Read-
ings in information visualization: Using vision to think. The
required.
Morgan Kaufmann series in interactive technologies. Morgan Kaufmann, San Francisco Calif., 1999.
3.3. Presentation
[2] CURE. http://www.pi6.fernuni-hagen.de/cure/. [3] R. D¨ aßler.
The human cognition is dependent on external presenta-
Informationsvisualisierung.: Stand, kritik und
perspektiven.
tion forms and due to the visual ability of the human brain
Technical report,
Fachhochschule Pots-
dam, Fachbereich Archiv/Bibliothek/Dokumentation, Pro-
a graphical presentantion is recommended [4]. A graph - in
jektgruppe InfoViz, 1999. [4] M. Eibl. Visualisierung im document retrieval. Forschungs-
terms of mathematics - consists of points (or nodes) to illus-
berichte
trate items or objects and connecting lines to display the re-
7,
Informationszentrum
Sozialwissenschaften,
2003.
lations between the nodes [11]. In this case the nodes are the
[5] Fronter. http://de.fronter.info/.
objects that are connected by associations (from the Topic
[6] ILIAS. http://www.ilias.de/.
Map concept) as well as by the scientic relation (from the
[7] ISO 13250: Topic Maps. Topic maps xml syntax, 2006.
ontology concept). Even simple semantic networks are dif-
[8] N. Ludwig. Modularization and semantics of learning ob-
cult to be illustrated due to their complexity and the variety
jects in a cooperative knowledge space. In Institute of Elec-
of interconnections. Possible presentation forms [11] are:
trical and Electronics Engineers, editor, IEEE International Symposium on Technology and Society, 2008: ISTAS 2008 ;
•
tree diagram
•
tree diagram in H3
•
interconnection diagram
Fredericton, NB, Canada, 26 - 28 June 2008 ; proceedings, Piscataway NJ, 2008. IEEE Operations Center. [9] Moodle. http://www.moodle.org. [10] R. Spence. Information visualization: Design for interaction. Pearson Prentice Hall, Harlow, 2007. [11] D. Trunk.
Semantische netze in informationssystemen:
Verbesserung
A tree diagram soon reaches its limits in case of many
der
suche
durch
interaktion
und
visual-
isierung. Kolner Arbeitspapiere zur Bibliotheks- und Infor¨
objects and interconnections. Therefore the tree diagram is
mationswissenschaft, 2005.
not very applicable for such complex semantic network. In
[12] M. Wessner.
contrast interconnection diagrams and tree diagrams in H3
Kontextuelle Kooperation in virtuellen Ler-
numgebungen, volume 8 of Schriften zu Kooperations- und
are able to illustrate a complex semantic network but once
Mediensystemen. Eul, Lohmar, 2005.
a certain number of objects has been reached a confusing chaos of nodes and lines evolves [3].
3