FINAL REPORT. APE TRACK A, STUDY 1
Carl-Gustav Jansson and Ambjörn Naeve
Administrative overview
Experiment 2 of the initial set of SWELL experiments was named ´Content archives, student
portfolios & 3D environments (APE). APE has been coordinated by Upsala Learning Lab.The
work described here concerns track A:Content and Context of Mathematics in Engineering
Education. That track comprises two studies:
Study 1: Modelling of conceptual development in mathematics on the
Information Technology Program at KTH.
Study 2: Portfolio based reflection on the curriculum of the Media Technology
Program at KTH with focus on mathematics.
The initial plan was to coordinate these two studies. This plan was not fulfilled and the studies
has been carried out separately. This report does only cover study 1.
Principal Investigators:
Ambjörn Naeve, KTH (CID)
Carl Gustaf Jansson, KTH (DSV)
Other staff
Klas Karlgren, KTH
Olle Sundblad, KTH
Matthias Palmér, KTH
Mikael Nilsson, KTH
Daniel Pettersson, KTH, MSc student
Johan Olsson, KTH, MSc student
Sofia Olsson, KTH (KTH Online)
Kristina Edström, KTH (KTH Online)
International collaborative partners
Brad Osgood, Mathematics Dept. and Stanford Learning Lab, Stanford University.
David Hestenes, Physics, Arizona State University.
Project aim and goals
Important research questions in study 1 are the following. What conceptual structures in first
year academic mathematics courses do the students regard as most important, and how do
these structures evolve over the first academic year? How do these structures correlate with
the students’ perception of the most important mathematical concepts in high-school
mathematics courses? On a more abstract level, study 1 focuses on the students abilities to
document and reflect over their learning process, including courses and the connections
between them. Will knowledge capture, organization, re-use, self-coaching and collaboration
will enhance the learning experience.
"One does not understand mathematics, one gets used to it", von Neumann claimed. Students
at the IT- University in Stockholm seem to have difficulties with both getting used to and with
understanding certain math concepts introduced in the math courses of their engineering study
programme. They also have trouble seeing the relevance of the math concepts to other
subjects in their study programme. Students could, e.g., have trouble understanding concepts
such as Taylor-development, multivariable functions and multivariable equations and also to
see when there are, and are not, solutions to math problems. As a first step toward alleviating
students’ learning problems a study investigating the learning and understanding of math
concepts was carried out among the students.
There is extensive research in related areas on design tasks and learning. E.g.,
“constructionism” – a theory of learning and a strategy for education - suggests a strong
connection between design and learning (Kafai & Resnick, 1996). Related to designing
conceptual models are the learning-by-design (LBD) approaches using design challenges as a
pedagogical method: ”Construction and trial of real devices would give students the
opportunity … to test their conceptions and discover the bugs and holes in their knowledge.”
(Kolodner, Crismond, Gray, Holbrook, & Puntambekar, 1998). Others have studies the
advantages of using specific diagrams to support conceptual learning (Oshima, Yuasa, &
Oshimo, 1998).
Introducing conceptual modeling techniques and modeling tasks would be beneficial in three
different ways, namely:
- that conceptual modeling would support the learning of mathematical concepts. Engaging
students in modeling tasks hopefully supports learners’ conceptual development by
making important concept more explicit and by turning learners’ attention to related
concepts that they may have neglected otherwise.
- that conceptual modeling would support and encourage reflection on their learning in
general, i.e., that conceptual modeling may be an efficient technique to encourage metacognition. Hopefully students would reflect more on the concepts and the terminology,
how they are related, and maybe also about which concepts they had not yet mastered as
well as why some of theses concepts were causing learning problems.
Understanding why they have problems is the first step towards overcoming learning
problems that conceptual modeling would support “transfer” of math concepts to other
(computer science) subjects. Hopefully students would be supported and encouraged to
reflect on how the math concepts could be relevant to other subjects.
Evaluation plan overview
Focus and
general
objectives
Activity
Supportive
technology
Data
collection
Data
analysis/method
Timing
Concept
modeling
exercises and
knowledge
management
Conzilla
4 * 150
concept maps
Content analysis
Fall 2000 Spring
2001
STUDY 1
Promote metacognition in
mathematics,
including own
understanding,
difficulty and
applicability.
Promote
transfer.
UML
Web-support
(study material
and interactive
exercises)
Observations
( field notes,
participant
observation)
Questionnaires
Correlation analysis
on individual
concept submaps,
students and
alternative
observations
Goal accomplishment
After the 24 months of the study, we consider our goals to be reached. In fact, due to the
tools’ development (Conzilla), the archiving efforts, the conncection to the CyberMath system
and the construction of the virtual mathematics exploratorium, we even consider us to have
reached beyond our initial goal expectations. These results form an important part of the base
for the planned continuation of our project. In fact, they have been crucial in creating a formal
continuation of the project as the submodule “Personalised Mathematical Courselets” within
the PADLR-project (Personalized Access to Distributed Learning Resources). Further
publications based on the analysis of the empirical data will also add to the result.
Activities during January-June 2000
An overview of activities in this period is given in the following table.
Activity
Outcome/Results
Realized/Timing
Integration of the project activities
within the framework of the IT program,
year 1.
A positive and stable context for carrying out
the study.
Spring
Seminars and conceptual modeling
exercises for faculty involved in ITprogram
Making the involved teachers more familiar
with conceptual modeling
Spring
Design of study-material for IT students
A paper on conceptual modeling of
mathematical knowledge
Summer
Preparation of tasks
Short papers specifying the students tasks
Summer
Development of a lecture series
A specification to be integrated with the
descriptions of the IT program curriculum
Summer
Design of evaluations for the study
Scheme of evaluations
Summer
Software development
Usability tests by the end of the summer
Summer
Start of two masters theses
Spring
Design of interactive web-support
Conceptual modeling activities and links to
suitable interactive exercises (e.g. UML
Tutorial).
Summer
Dissemination and outgoing activities
such as: meetings, workshops,
conferences, talks, presentations etc.
(see section 6 for further details)
Contacts and future collaboration opportunities
January-June
Widening the range of the conceptual modeling
exercises to include other subjects than
mathematics on the IT-program. in a longer
perspective.
Integration of project activities within the framework of the IT program, year 1. Carl
Gustaf Jansson has in dialog with the involved faculty and Ambjörn Naeve planned the study
within the context of the IT program. The students will get a task split up into two parts
(solutions to be submitted at the end of each term), a series of lectures integrated with the
regular courses, studymaterial on modelling in general and modelling of mathematical
concepts, intercative studymaterial and computerbased tools for modelling. Formally the
students will get 1 credit for this course out of the 5 credits for the course Introduction to IT.
Design of a lecture series. Ambjörn Naeve, Carl Gustaf Jansson and Klas Karlgren has
planned a lecture series to be integrated in the different courses of the IT program. The
lectures will mostly be given by Ambjörn Naeve, but in some cases by the regular teachers.
Preparation of tasks. Carl Gustaf Jansson, Klas Karlgren and Ambjörn Naeve have designed
and compiled a conceptual modeling task to be presented to IT students in the autumn and
which will be the basis for their work within this study during 2000, 2001. The task should be
solved in two steps (term 1 and term 2) The tasks are of three kinds, producing maps of high
level conceptual maps, detailed maps on particular concepts and maps on the structure of
proofs. The conceptual maps are assumed to be annotated in such a way that the students
selfreflection concerning understanding, perceived difficulty and applicability is reflected. A
preparatory exercise has also been produced, with the aim of training the students on
modeling of the high school mathematics that is part of the curriculum during the introductory
weeks.
Seminars and conceptual modeling exercises for faculty involved in IT-program.
Ambjörn Naeve has on several occasions, illustrated the basic techniques of conceptual
modeling in UML for the teachers involved in the IT-program at Kista, and shown examples
of how this technique will be used in the mathematics courses. This has formed a good base
for the current studies and created interest in widening the range of the conceptual modeling
exercises to include all the different subjects on the IT-program.
Design of study-material. Ambjörn Naeve has written a study-material that presents
conceptual modeling at a suitable level for this experiment. No material meeting the
pedagogical requirements did exist. Apart from being a basis for this study, the produced
material will have a much wider applicability.
Design of Interactive web-support. Ambjörn Naeve has together with Sofia Olsson and
Kristina Edström at KTH Online developed a web-support for the conceptual modeling
activities. This web support includes a UML tutorial (Interactive UML) produced by the Open
Training company, some web-based animations produced within the Conzilla tool and some
additional interactive exercises.
Software development involves improvements and extensions of the Conzilla tool. This
work has proceeded according to plan. Mikael Nilsson has documented the code structure.
Matthias Palmer has worked on aspect filtering. Richard Wessblad from DataDoktorn has
worked with the the help-service system. Ambjörn Naeve have started up two master thesis
projects for two KTH-students, Daniel Pettersson and Johan Olsson. These projects both
represent extensions of Conzilla that will make the program more useful as an overall tool for
knowledge management.
Design of an evaluation scheme for the study. Klas Karlgren has in discussion with Carl
Gustaf Jansson, Ambjörn Naeve and the SWELL assessment team developed an evaluation
scheme including content and correlation analysis based on results from student task,
observations and input from questionnaires. An iterative revision of the evaluation scheme
has been carried out as scheduled during the first half of 2000.
Activities during July-December 2000
During the fall of 2000 the 150 students of the IT program have been carrying out a
conceptual modelling exercise as part of the "Intro to IT" course. The exercise is concerned
with creating conceptual models based on the mathematics they have experienced in the ITprogram curriculum. More specifically, three different concept maps are to be constructed:
The first one should describe the overall relationships between the most important
mathematical concepts that the students encounter in the mathematics courses, as well as the
relationships between these concepts and their applications in other courses. The second
concept map should focus on the function concept and describe the relations between the
different types of functions that the students encounter in the mathematics courses, and the
third concept map should describe the logical relationships between the different theorems
that are presented in these courses. The conceptual maps are assumed to reflect students’
understanding, perceived difficulty and applicability.
The modelling activity was initiated by a lecture on conceptual modelling in UML given by
Ambjörn Naeve on September 4. As a support structure for the modelling exercise, the
students have had continuous access to the modelling web site and the interactive UML
course described above. The exercise stretches over the entire first year, and involves
handing in two sets of these maps - one set by the end of the fall term, and a second set by the
end of the spring term. The students have been allowed to work in groups of up to 4 persons.
Some minor deviations from the original plan have occurred. The modelling activities have
been "added on" rather than integrated with the mathematics courses as originally planned.
The initial plan was to collect four different versions of the concept maps. However, because
of the work load involved (and the credits given) we have reduced the number of versions to
two. Moreover, during the modelling process, the students requested to be allowed to work in
groups of up to 4 persons. We considered this to be reasonable, especially since conceptual
modelling benefits strongly from communication with others.
Implementation of interactive content and appropriate tools - Mathematical Resource
Components:
1. Constructing the components: Using programs like Mathematica, Projective Drawing
Board and the Graphing Calculator, we have constructed a number of mathematical
resource components that illustrate mathematical concepts and relationships. Some of
these components have been transformed into interactive web-graphics and some have
also been translated into CyberMath (the shared3D interactive learning environment for
mathematics that has been created as a part of the APE-track-C project).
2 . Archiving the components: There are different ways to archive mathematical
components of different kinds - including the ones described above. A newly developed
test-archive which can be updated dynamically and where the components are viewable
under the common browsers is available at http://www.nada.kth.se/cgibin/osu/dirlister2?math.
3 . Interacting with the components: Exploring how to interact with the components,
focusing on the ones constructed by using the Graphing Calculator, a program that is
available today for the visual display of mathematical formulas. We have acquired 250
user licenses for the Graphing Calculator at KTH (to cover the teachers and students of the
IT- and Media Technology program). The Graphing Calculator offers truly novel ways to
interact with the components of a mathematics archive, where frozen animations can be
downloaded and easily manipulated by users. This constitutes a very exiting graphical
way of conducting mathematical discussions between the teachers and the students as well
as between the students themselves. We have started to introduce this technique to some
of the mathematics teachers at KTH, and we are planning to introduce it for the students
of the IT-program in the spring of this year.
Activities during January-June 2001
During spring of 2001 the 150 students of the IT program continued carrying out the
conceptual modeling exercise described above as part of the "Intro to IT" course. The exercise
is concerned with creating conceptual models based on the mathematics they have
experienced in the IT-program curriculum. More specifically, three different concept maps
have been constructed: The first one describes the overall relationships between the most
important mathematical concepts that the students encounter in the mathematics courses, as
well as the relationships between these concepts and their applications in other courses. The
second concept map focuses on the function concept and describes the relations between the
different types of functions that the students encounter in the mathematics courses, and the
third concept map describes the logical relationships between the different theorems that are
presented in these courses. As a support structure for the modelling exercise, the students
have had continuos access to the modelling web site and the interactive UML course
described above. The exercise has stretched over the entire first year, and has involved
handing in two sets of these maps - one set by the end of the fall term, and a second set by the
end of the spring term. The students was allowed to work in groups of up to 4 persons.
Student tasks from the fall term have been collected and systematised. A systematic way to
encode the structure of the large variety of types of concept maps has been developed, which
will be presented in an upcoming report. This encoding will make it possible to investigate a
variety of correlations between different types of conceptual representations. The maps from
the fall term exercise have all been encoded in this way.
Student tasks from the spring term was collected. The work to encode of these maps in our
systematic notation is started.
The implementation of interactive content and appropriate tools has continued along the
lines specified above. These archives and tools will be introduced to the students on the ITprogram in the fall as a part of the continuation of the APE-project within the PADLRproject.
Fig. 1: A UML diagram describing the function concept drawn by a student of the IT-programme (fall 2000).
Educational evaluation/assessment results
A series of meetings with Monica Langerth of the Learning Lab assessment team was
conducted. Through these meetings we did establish a strategy of applying the theory
anchored evaluation model developed by Monica Langerth and Helge Strömdahl. The
analytic evaluation of the students submitted conceptual graphs, will be completed with a
series of deep interviews with selected students at the IT-program.
Activities during July-December 2001
During the fall of 2001 the main effort in the APE-track A project has been concerned with
the following three main activities:
1 . Analysing the context-maps that were collected in the mathematics modelling
experiment. This work is described briefly above. A more detailed account is available
in the form of a scientific publication.
2. Construction of mathematical components (joint work with APE-track C). A fuller
account of this work is included as an Appendix.
3. Developing the Conzilla concept browser in a way that makes it more suitable for use
within the PADLR-project – both the module 3.1 (Edutella) and the module 5.3
(Personalized Mathematical Courselets).
Analysis of conceptual graphs submitted by first year MSc students at KTH
This study is the first step towards developing and offering students support in their learning.
Therefore, the goal was to investigate the learning and understanding of math concepts. 150
engineering students participated in the extensive study involving modeling tasks stretching
over the entire first year of the study programme. In the fall, the students initially performed a
diagnosis task investigating which concepts they viewed as the most central concepts in
mathematics. During the first semester they were asked to construct graphical conceptual
models describing and relating all mathematical concepts they were confronted in the study
programme. Their views on the math concepts could be expressed in different ways. A
Unified Modeling Language (UML) notation was preferred, but if students experienced it as
too restrictive other notations were allowed, e.g., “mind maps”. These models were handed in
around Christmas, in December. During the spring semester students continued modeling how
they viewed the math concepts as well as all new math concepts introduced in the following
courses. New models were once again collected in the end of the spring semester. A typical
model could look like the one below.
The students’ answers to the diagnosis tasks and the two modeling tasks differ the most. The
study showed that students picked up and noted a number of new concepts from the math
courses which they included in their models. If the spring models are compared to the fall
models a number of differences can also be observed: New concepts are added. Specifically
concepts related to courses the students had attended during the spring semester. The models
also become more homogeneous. Perhaps because the students discuss the models with each
other, and perhaps because the students become more familiar with UML.
The table below shows the most central math concepts according to the students, at three
different occasions during their study programme; when initially enrolling in the study
programme, after one study semester, and after an entire study year. The lists are based on the
number of diagnoses and models in which each concept occurs.
Concepts after one semester
Funktion
Derivata
Gränsvärde
Polynom
Komplext tal
Differentialekvation
Felrättande koder
Homogena (ekvationer)
Inhomogena ((diff-)ekv)
Bipartit (graf)
Concepts after one year
Vektor
Matriser
Gränsvärde
Determinant
Derivata
Partiell/partialderivata
Funktion
Jacobimatris
Skalärprodukt
Polynom
In the last model students typically add new concepts from newly attended math courses (e.g.,
kedjeregeln, flervariabel, integral, extremproblem…) to the earlier model. Often students also
developed and elaborated on a concept that had previously been included in a model.
Rather often, however, students construct an entirely new model which do not include any
parts of their previous models even though much work was put into these. In many cases
students seemed to presuppose the parts not mentioned. Perhaps the parts were left out to save
the effort of relating these to new concepts in their new models. Another plausible
interpretation is that the connection between courses is not clear enough and concern uses of
the certain concepts (e.g., function) which do not overlap thereby making it difficult for
students to relate the different uses of the concept in their conceptual models. These
connections between courses should perhaps be made clearer and more explicit for the
students. Further analysis of the extensive data resulting from this study is on-going with a
special focus on the relationships between the different concepts and how these developed.
A more extensive description of the study can be found in the article ´Conceptual Modeling
as a Metacognitive Tool for Learning Math Concepts´ , Karlgren, Naeve and Jansson, to be
submitted to e-Learn 2002 (the World Conference on E-Learning in Corporate, Government,
Healthcare, & Higher Education), October 15 - 19, 2002, Montreal, Canada).
Creation of mathematical content
In parallel with the work of APE-track C, a large number of mathematical structures have
been visualized using special tools such as Mathematica and the Graphing Calulator. These
components include material from the basic courses in Linear Algebra, Differential and
Integral Calculus of one and several variables, Differential equations, Fourier Analysis and
Differential Geometry. They are presently being integrated in the Conzilla-based mathematics
exploratorium described below.
During the duration of the APE-project, we have been using Conzilla to build a virtual
mathematics exploratorium, which makes it possible to navigate a mathematical knowledge
manifold and explore its content in various ways. This exploratorium is available on the net
by downloading the Conzilla program [http://conzilla.sourceforge.net]. During the fall of
2001, we have recorded an entire lecture series on the history of ideas in mathematics, which
will be available as streaming video content components in the exploratorium. In the near
future, this exploratorium will be offered to students as a way to support the traditional
mathematics education and (hopefully) serve as a way to stimulate interest and motivate
further studies within the field of mathematics.
Extensions of the Conzilla program
Some of the major issues deemed to be strategically important have been to refactor the code
of Conzilla so that it can be used both as a standalone application and together with ontology
construction tools such as e.g. Protegé [http://protege.stanford.edu]. On the basis of the
refactored code, we have developed an applet version of Conzilla - which means that contextmaps can be navigated in an ordinary web browser - as well as RDF backend
[www.w3.org/RDF], which will serve as a basis for the ongoing adaptation of Conzilla to the
information standards of the emerging next generation of the Internet, the so called Semantic
Web [www.SemanticWeb.org], [Berners-Lee et.al., 2001], [Nilsson, Palmér, Naeve, 2002].
The RDF backend will also enable the use of Conzilla as a presentational tool and a graphical
query interface on top of the Edutella infrastructure mentioned above [Nejdl et.al., 2002].
Closely related to the refactoring work has been the necessary modularization of Conzilla. In
this phase Conzilla has been divided and separated so that parts can be exchanged and
excluded if needed. For instance the frame based appearance has been changed into a dynamic
solution with ViewManagers that now includes the old frame-view, internal-frame-view,
tabbed-view and single pane-view, which greatly facilitates the simultaneous presentation of
different context-maps as well as the semantic connections between them. This has been used
for developing the basic Protegé-Conzilla connection. The menu system has been reworked in
order to support modular design allowing modules and plug-ins to add menus everywhere.
A better configuration system for modules as well as cleaner interfaces between them have
been developed. Hence the former modular system has been turned into a full fledged pluginsystem greatly easing the pain later when designing specific functionality.
The Graphical User Interface, has been greatly improved reflecting the changes above. As an
example, it is now possible to use curved lines (cubic B-splines) in connecting the concepts,
which greatly improves the possibility to design intelligible context-maps.
Conzilla has been made into a tab-plug-in under Protegé. This has included a wrapper to their
knowledge base API (based on their OKBC standard). Some simple layout code has been
written for further adaptation to GraphViz. Several issues remain to be resolved: roundtrip,
project management in Protegé should be matched to some similar solution in Conzilla,
namespaces in Protegé need further investigation, etc.
Presentations and Publications
Presentations
Visit to Stanford and Arizona State University ( March 9-29, 2000)
Participation in SIGGRAPH 2000 (July 2000)
Presentation for the KK-foundation and the Swedish School Board (Skolverket) at Levintelligence
(Feb 1, 2000)
Presentation for Teachers at KTH at Dept of Didactics at KTH (April 4, 2000).
Presentation for Mathematics teachers from Kista (secondary school level) at CID (April 12, 2000).
Presentation at the inauguration of Uppsala Learning lab (May 25, 2000).
Presentation for 150 science and technology teachers at KTH (May, 2000).
Presentation for Human Machine Interaction graduate students at KTH (May 26, 2000).
Presentations at two planning workshops for the IT program (Kämpasten March 7-8 and Hässelby
May 9 – 10, 2000)
Presentation at the SweLL Evaluation meeting (May 9 and 10, 2000)
The Conzilla program was presented in Washington DC on October 28 at the CILT-2000
learning conference (www.cilt.org/cilt2000) arranged by the Center for Innovative Learning
Technologies. A report from this conference - in power point format - can be found on
http://www.learninglab.kth.se/library/presentations.
KTH Learning Lab, seminar 10/1 - 2001
Uppsala Learning Lab, seminar 21/3 - 2001
International Conference on open learning and Distance Education (ICDE) Düsseldorf, 3/4 - 2001.
Högskolan i Gövik, Norway, tele-presence mediated presentation, 3/5 – 2001.
Luleå Technical University, seminar on mathematical didactics, 7/5 – 2001.
Mitthögskolan Östersund, e-Learning conference, 16/5-2001.
WGLN workshop on performance learning, KTH, 18/6 - 2001
SIGGRAPH-2001, Los Angeles, USA, Aug12-17, 2001.
8:th Conference on Advanced Computer Systems (ACS-2001), Mielno, Poland, Oct17-19, 2001.
2:nd european web-based e-learning conference (WBLE), Lund, 24-26/10 - 2001.
Invited talks by Ambjörn Naeve at Umeå universitet, Sept 21, 2001, Ingenjörsskolan KTH-Haninge
(KTH), Oct 23, 2001, Högskolan i Kalmar, Oct 30, 2001 and at Riksutställningar, Dec 7, 2001.
Swedish Educational Television (UR) visits CID , Dec 13, 2001
Swedish National Board of Education (Skolverket) visits CID, Dec 14, 2001
National Centre for Flexible Learning (CFL) visits CID, Dec 14, 2001
The CyberMath system was presented to 200 high school students during the Kowalewski
mathematics days held at KTH on Nov, 9-10.
Publications
Karlgren, K., Naeve, A. and Jansson, C-G , Conceptual Modeling as a Metacognitive Tool for
Learning Math Concepts , to be submitted to e-Learn 2002 (the World Conference on ELearning in Corporate, Government, Healthcare, & Higher Education), October 15 - 19,
2002, Montreal, Canada.
Naeve, A. (2001a) Begreppsmodellering och matematik, CID-109, TRITA-NA-D0103,
Department of Numerical Analysis and Computer Science, KTH, Stockholm, 2001,
http://kmr.nada.kth.se/papers/MathematicsEducation/cid_109.pdf.
Naeve, A. (2001d) The Concept Browser - a New Form of Knowledge Management Tool,
Proceedings of the 2nd European Web-Based Learning Environment Conference (WBLE
2001), Lund, Sweden, Oct. 24-26, 2001,
http://kmr.nada.kth.se/papers/ConceptualBrowsing/ConceptBrowser.pdf.
Nilsson, M. & Palmér, M. (1999) Conzilla - Towards a Concept Browser, CID-53, TRITANA-D9911, KTH, 1999, http://kmr.nada.kth.se/papers/ConceptualBrowsing/cid_53.pdf.
Nilsson, M. (2000), The Conzilla Design - The Definitive Reference, CID/NADA/KTH, 2000,
http://kmr.nada.kth.se/papers/ConceptualBrowsing/conzilla-design.pdf.
Palmér, M., Naeve, A., Nilsson, M. (2001) E-learning in the Semantic Age, Proc. of the 2nd
European Web-based Learning Environments Conference (WBLE 2001), Lund, Sweden, Oct.
24-26, 2001, http://kmr.nada.kth.se/papers/SemanticWeb/e-Learning-in-The-SA.pdf.
APPENDIX
Development of Mathematical
Content and the Conzilla t o o l
Ambjörn Naeve
CID, KTH
Background
Traditional educational architectures are teacher-centric and founded in curricular-based knowledge pushing. As
described in [Naeve, 1997] [Naeve, 1999b] and [Naeve 2001c], the Knowledge Management Research group at
CID [http://kmr.nada.kth.se] has developed a learner-centric educational architecture called a knowledge
manifold, which supports knowledge-pulling based on interest. A knowledge manifold can be described as a kind
of patchwork, consisting of a number of linked knowledge patches, each maintained by a knowledge gardener. A
knowledge patch is constructed from context maps, whose concepts are filled with content components, which
are designed with the overall aim to separate content from context by making use of multiple narration
techniques.
The knowledge manifold architecture is based on the following fundamental principles:
• Nobody can teach you anything. A good teacher can inspire you to learn.
• Your learning motivation is based on the experience of subject excitement and faith in your learning capacity
from live teachers.
• Your learning is enhanced by taking control of your own learning process.
• No 'problematic' questions can be answered in an automated way. In fact, it is precisely when your questions
break the pre-programmed structure that the deeper part of your learning process begins.
• Respect for ignorance, which is of fundamental importance in a non-elitist knowledge society, can only be
upheld when the ignorant person is uneducated1.
In order to support the knowledge manifold architecture, the KMR-group is developing a set of tools. This set
includes a concept-browser, which is a new kind of knowledge management tool designed in accordance with a
set of principles that are laid down and discussed in detail in [Naeve, 1999b] and [Naeve, 2001d]. These
principles include a strict separation of context and content, contextual descriptions in terms of a collection of
semantically visual context maps, which can be navigated by moving through contextual neighborhoods,
presentation of the content components through context-dependent aspect-filters, and contextualization of
content components that are themselves context maps.
During the last 3 years the KMR-group at CID has developed a first prototype of concept browser called
Conzilla [Nilsson & Palmér, 1999], [Nilsson, 2000]. It has the potential of being useful within the fields of elearning, e-commerce and e-administration, and the KMR-group is presently participating in national- and
international collaborations within all these fields. For more details on these projects, see http://kmr.nada.kth.se.
Within the newly started WGLN-supported PADLR-project, Conzilla will be used both as an interface to the
Edutella system constructed within PADLR-module 3.1 and as a tool for the creation of personalized
mathematical courselets (PADLR-module 5.3).
1
The term "educated" is being used here in the formal sense, meaning "graduated from an educational system".
Extensions of the Conzilla program during July-Dec 2001
Some of the major issues deemed to be strategically important have been to refactor the code of Conzilla so that
it can be used both as a standalone application and together with ontology construction tools such as e.g. Protegé
[http://protege.stanford.edu]. On the basis of the refactored code, we have developed an applet version of
Conzilla - which means that context-maps can be navigated in an ordinary web browser - as well as RDF
backend [www.w3.org/RDF], which will serve as a basis for the ongoing adaptation of Conzilla to the
information standards of the emerging next generation of the Internet, the so called Semantic Web
[www.SemanticWeb.org], [Berners-Lee et.al., 2001], [Nilsson, Palmér, Naeve, 2002]. The RDF backend will
also enable the use of Conzilla as a presentational tool and a graphical query interface on top of the Edutella
infrastructure mentioned above [Nejdl et.al., 2002].
Closely related to the refactoring work has been the necessary modularization of Conzilla. In this phase Conzilla
has been divided and separated so that parts can be exchanged and excluded if needed. For instance the frame
based apperance has been changed into a dynamic solution with ViewManagers that now includes the old frameview, internal-frame-view, tabbed-view and single pane-view, which greatly facilitates the simultaneous
presentation of different context-maps as well as the semantic connections between them. This has been used for
developing the basic Protegé-Conzilla connection. The menu system has been reworked in order to support
modular design allowing modules and plugins to add menus everywhere.
A better configuration system for modules as well as cleaner interfaces between them have been developed.
Hence the former modular system has been turned into a full fledged plugin-system greatly easing the pain later
when designing specific functionality.
The Graphical User Interface, has been greatly improved reflecting the changes above. As an example, it is now
possible to use curved lines (cubic B-splines) in connecting the concepts, which greatly improves the possibility
to design intelligible context-maps.
Conzilla has been made into a tab-plugin under Protegé. This has included a wrapper to their knowledge base
API (based on their OKBC standard). Some simple layout code has been written for further adaptation to
GraphViz. Several issues remain to be resolved: roundtrip, project managment in Protegé should be matched to
some similiar solution in Conzilla, namespaces in Protegé need further investigation, etc.
Creation of mathematical content
In parallel with the work of APE-track C, a large number of mathematical structures have been visualized using
special tools such as Mathematica and the Graphing Calulator. These components include material from the
basic courses in Linear Algebra, Differential and Integral Calculus of one and several variables, Differential
equations, Fourier Analysis and Differential Geometry. They are presently being integrated in the Conzilla-based
mathematics exploratorium described below.
During the fall of 2001, Conzilla has also been connected to the CyberMath system developed with partial
support from WGLN in the project APE-track C. This connection is reported in the APE-track C scientific
progress report for July-Dec 2001.
The virtual mathematics exploratorium
During the duration of the APE-project, we have been using Conzilla to build a virtual mathematics
exploratorium, which makes it possible to navigate a mathematical knowledge manifold and explore its content
in various ways. This exploratorium is available on the net by downloading the Conzilla program
[http://conzilla.sourceforge.net]. During the fall of 2001, we have recorded an entire lecture series on the history
of ideas in mathematics, which will be available as streaming video content components in the exploratorium. In
the near future, this exploratorium will be offered to students as a way to support the traditional mathematics
education and (hopefully) serve as a way to stimulate interest and motivate further studies within the field of
mathematics.
What follows below is a series of snapshots of a “walk-through” of part of the exploratorium. Although the
pictures are chosen to try to convey an impression of what it is like to navigate a knowledge manifold with
Conzilla, they cannot convey the dynamic aspects of such an experience. Downloading the program and trying it
out for yourself is the only way to do that.
Figure 1. This is a context-map showing the different types of knowledge manifolds that are presently being
developed by the KMR-group at CID. The language used for the context-map is a dialect of UML
[http://www.uml.org] called ULM (Unified Language Modeling), which is designed to visually depict how we
talk about things. ULM is described in [Naeve, 1999b] and [Naeve, 2001d].
Figure 2. Right-clicking the concept e-Learning brings up the main navigation menu. Choosing Surf brings up
a sub-menu showing the contextual neighborhood of the e-Learning concept, which is all the different contextmaps where the concept e-Learning appears. In this case there is only one more context-map, namely the one
titled e-Learning projects. Choosing this entry from the Neighborhood menu brings up the corresponding
context-map, shown in Figure 3.
Figure 3. Since we have entered the e-Learning projects context-map through the e-Learning concept, this
concept is shown highlighted (in green) when we enter this context-map. The highlighting remains until we click
(anywhere) in the map. This is a feature of Conzilla that has been added during the fall of 2001. It is very useful
in supporting the “cognitive connection” between the different context-maps.
Figure 4. Pointing to the Exploratorium concept and hitting <space-bar> brings up information (meta-data)
about this concept. Conzilla has a meta-data editor called ImseVimse [http://imsevimse.sourceforge.net] that
supports full IMS [http://imsproject.org/metadata].
Figure 5. Hitting <escape> closes the meta-data window. Left-clicking the Exploratorium concept brings up an
iconized version of the detailed map connected to this concept without changing the context. This is another
feature that supports cognitive connection between the different contexts, since it allows a preview of another
context without committing to it. Double-clicking the Exploratorium concept changes the context and brings up
the context-map whose icon was displayed in the preview mode.
Figure 6. Pointing to the Mathematics concept and hitting <space-bar> brings up meta-data on mathematics.
The arrow at the bottom of the meta-data window indicates that there is more information below
Figure 7. Hitting option-space-bar expands the meta-data window and shows the full information text. Hitting
only space-bar (without the option) would have shown only the second part of the information text. In this way it
is possible to browse through a lists of entries, leaving only the relevant parts on the screen.
Figure 8. Right-clicking on Archives brings up the main navigation menu. Choosing View from this menu
brings up the content window on the right. This window shows a list of content components. Double-clicking
one of them will display its content in an ordinary web browser (such as Explorer or Netscape).
Figure 9. The content components also carry information (meta-data), which can be brought up by clicking their
names and hitting <space-bar>. Notice that several pieces of such information can be visible in parallel. The
figure shows meta-data about Ambjorns and Olles Math Archive, Cornell Math Library, EMIS and Math
and Liberal Art. Notice also that the concept Archives, whose content components we are inspecting, is shown
in the same color as the content window. This supports the cognitive connection between context and content.
Figure 10. Clicking Ambjorns and Olles Math Archive, and navigating to Differential and Integral calculus /
Several Variables / Gradients, displays an entry called FlyingCarpetSurfaceGradient.mov (not shown in the
figure). Clicking it will display a Quicktime movie in the web browser of your choice. However, dragging the
icon of the movie to the Graphing Calulator and dropping it there will invoke the mathematics behind the movie,
making it amenable to parametric changes. The figure shows the graphing calculator window after dropping the
movie in it.
References
Berners-Lee, T. & Hendler, J. & Lassila, O. (2001) The Semantic Web, Scientific American, May 2001,
www.scientificamerican.com/2001/0501issue/0501berners-lee.html.
Knudsen, C., Naeve, A. (2001), Presence Production in a Distributed Shared Virtual Environment for Exploring
Mathematics, Proceedings of the 8th International Conference on Advanced Computer Systems (ACS 2001),
Szcecin, Poland.
http://kmr.nada.kth.se/papers/TelePresence/CJKPresenceProd.pdf.
Naeve, A. (1997) The Garden of Knowledge as a Knowledge Manifold - a Conceptual Framework for Computer
Supported Subjective Education, CID-17, TRITA-NA-D9708, Department of Numerical Analysis and Computer
Science, KTH, Stockholm, 1997,
http://kmr.nada.kth.se/papers/KnowledgeManifolds/cid_17.pdf.
Naeve, A. (1999a), IT-baserade matematikverktyg på KTH, CID-49, TRITA-NA-D0103, Department of
Numerical Analysis and Computer Science, KTH, Stockholm, 1999,
http://kmr.nada.kth.se/papers/MathematicsEducation/cid_49.pdf.
Naeve, A. (1999b) Conceptual Navigation and Multiple Scale Narration in a Knowledge Manifold, CID-52,
TRITA-NA-D9910, Department of Numerical Analysis and Computing Science, KTH, 1999,
http://kmr.nada.kth.se/papers/ConceptualBrowsing/cid_52.pdf.
Naeve, A. (1999c) Det Virtuella Matematikexploratoriet-en översikt, CID/NADA/KTH, 1999,
http://kmr.nada.kth.se/papers/MathematicsEducation/VirtualMathExploratorium/Exploratorium.pdf.
Naeve, A. (1999d) Matemagiprojektet - en översikt, CID/NADA/KTH, 1999,
http://kmr.nada.kth.se/papers/MathematicsEducation/Matemagi/Matemagi.pdf.
Naeve, A. (2001a) Begreppsmodellering och matematik, CID-109, TRITA-NA-D0103, Department of
Numerical Analysis and Computer Science, KTH, Stockholm, 2001,
http://kmr.nada.kth.se/papers/MathematicsEducation/cid_109.pdf.
Naeve, A. (2001b) The Work of Ambjörn Naeve within the Field of Mathematics Educational Reform, CID-110,
TRITA-NA-D0104, KTH, Stockholm, 2001,
http://kmr.nada.kth.se/papers/MathematicsEducation/mathematics_educational_reform.pdf.
Naeve, A. (2001c) The Knowledge Manifold - an Educational Architecture that supports Inquiry-Based
Customizable Forms of E-Learning, Proc. of the 2nd European Web-Based Learning Environment Conference
(WBLE 2001), Lund, Sweden, Oct. 24-26, 2001,
http://kmr.nada.kth.se/papers/KnowledgeManifolds/KnowledgeManifold.pdf.
Naeve, A. (2001d) The Concept Browser - a New Form of Knowledge Management Tool, Proceedings of the 2nd
European Web-Based Learning Environment Conference (WBLE 2001), Lund, Sweden, Oct. 24-26, 2001,
http://kmr.nada.kth.se/papers/ConceptualBrowsing/ConceptBrowser.pdf.
Naeve, A., Nilsson, M., Palmér, M. (2001), The Conceptual Web - our Research Vision, Proc. of the First
Semantic Web Working Symposium, Stanford University, July 30 – Aug 1, 2001,
http://kmr.nada.kth.se/papers/SemanticWeb/pospaper.pdf.
Nejdl, W., Wolf, B., Qu, C., Decker, S., Sintek, M., Naeve, A., Nilsson, M., Palmér, M, Risch, T. (2002)
Edutella – a P2P Networking Infrastructure Based on RDF, Accepted for presentation at the 11:th International
World Wide Web Conference (WWW-2002) in Honolulu, Hawaii, May 7-11, 2002.
Nilsson, M. & Palmér, M. (1999) Conzilla - Towards a Concept Browser, CID-53, TRITA-NA-D9911, KTH,
1999, http://kmr.nada.kth.se/papers/ConceptualBrowsing/cid_53.pdf.
Nilsson, M. (2000), The Conzilla Design - The Definitive Reference, CID/NADA/KTH, 2000,
http://kmr.nada.kth.se/papers/ConceptualBrowsing/conzilla-design.pdf.
Nilsson, M. (2001), The Semantic Web: How RDF will change learning technology standards, Feature article,
Centre for Educational Technology Interoperability Standards (CETIS), 2001,
http://kmr.nada.kth.se/papers/SemanticWeb/CETISSemWeb.pdf.
Nilsson, M. et al. (2001) IMS meta-data 1.2 RDF binding, Appendix to the IMS 1.2 metadata binding
specification, IMS, 2001, http://www.imsglobal.org/rdf/index.html.
Nilsson, M., Palmér, M., Naeve, A. (2002) Architectural Guidelines for Semantic Web Meta-data – Lessons from
E-Learning, Accepted for presentation at the 11:th International World Wide Web Conference (WWW-2002) in
Honolulu, Hawaii, May 7-11, 2002.
Palmér, M., Naeve, A., Nilsson, M. (2001) E-learning in the Semantic Age, Proc. of the 2nd European Web-based
Learning Environments Conference (WBLE 2001), Lund, Sweden, Oct. 24-26, 2001,
http://kmr.nada.kth.se/papers/SemanticWeb/e-Learning-in-The-SA.pdf.
Wilson, S. (2001) The Next Wave: CETIS interviews Mikael Nilsson about the Edutella project, Feature article,
Centre for Educational Technology Interoperability Standards (CETIS), Sept, 2001,
http://kmr.nada.kth.se/papers/SemanticWeb/TheNextWave.pdf.
View publication stats