Diagram-Based CBA using DATsys and CourseMaster
Colin Higgins, Pavlos Symeonidis, Athanasios Tsintsifas
School of Computer Science and IT,
The University of Nottingham,
Jubilee Campus, Wollaton Road,
Nottingham, NG8 1BB, UK.
ltr@cs.nott.ac.uk
Abstract
Supporting the assessment of an ever-increasing
number of students is an error-prone and resource
intensive process. Computer Based Assessment (CBA)
software aids educators by automating aspects of the
assessment of student work. Using CBA benefits
pedagogically and practically both students and
educators.
The Learning Technology Group at the University of
Nottingham has been actively researching, developing
and using software to automatically assess programming
coursework for 14 years. Two of the systems developed,
Ceilidh and its successor CourseMaster, are being used
by an increasing number of academic institutions.
Recent research has resulted in a system for
supporting the full lifecycle of free-response CBA that has
diagram-based solutions. The system, DATsys, is an
authoring environment for developing diagram-based
CBA. It has been designed to support the authoring of
coursework for most types of diagram notations.
Exercises have been developed and tested for circuit
diagrams, flowcharts and class diagrams.
Future
research plans for authoring exercises in many more
diagram notations.
1: Introduction – Definitions
Learning technology is a widely used term that refers to
the application of technology to enhance the learning
process. Computer Assisted Assessment (CAA) is a field
of learning technology that studies the use of computers
to deliver, analyse and mark student coursework. A
specialisation of CAA is Computer Based Assessment
(CBA). CBA differs from CAA in that with CBA the
coursework’s solution is entered on-line.
Culwin distinguishes between two types of automated
assessment, fixed response and free response [5] also
referred to as objective and non-objective respectively [3].
The distinction between the two is based on the nature of
response to the CBA exercise. Fixed response systems
Proceedings of the International Conference on Computers in Education (ICCE’02)
0-7695-1509-6/02 $17.00 © 2002 IEEE
require the learner to choose a solution from a list of
available options. Such CBA systems use either: multiplechoice questions, simple text exercises, numeric value
exercises, or graphical hotspot exercises.
Further
categorisations are given by Charman and Elmes [4].
In contrast, free response systems can assess unanticipated
solutions.
Such systems have been employed for
assessing the implementation or design of computer
programs [7] and essays [16]. Typically, in free response
assessment a qualitative strategy describes the criteria
attributing to the evaluation of the exercise.
The automatic assessment of diagrams belongs to the free
response type of assessment since the solution has
potentially infinite solutions with varying degrees of
correctness. We define diagram-based CBA as “a type of
CBA in which the solution to a required problem or the
answer to a required question is drawn by the student
using a suitable graphical environment and is assessed
automatically according to appropriate marking
guidelines”.
2: The problem of developing Diagram-Based
CBA exercises
In what aspects does diagram-based CBA differ from
others? CBA systems, conceptually, consist of a database,
software
components
for
authoring,
marking,
administering, presenting, and submitting coursework, and
analogous user interfaces (UI) (figure 1). Three basic
types of data exist within the database in CBA systems:
• Authored material, such as coursework and marking
related data
• Class related material, such as student submissions,
marks, and user information
• Properties for administration and marking
The software components are the modules of software
that support the requests from the user-interfaces to the
database. Typically, these modules contain logic that
describes the execution and properties of the processes
involved in assessment. The user-interfaces provide the
means to use the system and support various options.
Figure 1: A high level view of the parts of typical
CBA systems
The variation between types of assessment does not
really affect the database, the software components for
setting, submitting and administering coursework, nor
much of the UIs. These can be implemented generically.
However, in three main areas, diagram-based
coursework exhibits differences. These are:
• The UI part that allows the entry and editing of the
solution to the exercise by the student
• The marking mechanism that directs the marking of
the student solution and returns appropriate feedback
• The process of authoring coursework and the
refinement of the generic CBA model to support the
full lifecycle of the CBA coursework for summative
purposes in a controlled environment
Each difference initiates a respective problem area.
The first problem area is the building of a graphical editor
customised to support the exercise and its domain.
Software engineering research has documented methods
for the construction of generic diagramming editors over
the past 10 years [15], [12], [2], [1]. Most of these
methods aim to empower programmers with libraries
powerful enough to ease the development of domain
specific graphical editors such as circuit, flowchart, and
drawing editors. Yet, these libraries are complex and
demand significant programming expertise. The effort
and resources involved in producing a single “per
domain” or “per exercise” editor prohibits their use in
CBA. Hence regarding the first problem area the
objective has been to design and implement a framework
for generic diagram editors in which the creation of a new
type of editor is a simple task that does not require
programming.
The second problem area concerns the marking
process. It extends to the description and execution of the
marking criteria and the creation of appropriate feedback.
Foxley and Zin have described a generic technique to
express marking schemes for assessing exercises in
programming languages using “Oracles” [9]. Oracles
represent the marking criteria that are used to mark an
exercise. They are described in a form that uses regular
expressions and they express the text matching behaviour
Proceedings of the International Conference on Computers in Education (ICCE’02)
0-7695-1509-6/02 $17.00 © 2002 IEEE
that can conveniently be used to assess an exercise
solution. The objectives in the second problem area have
been to investigate whether a similar technique can be
used to describe marking schemes for diagram-based
coursework.
The third problem is concerned with the provision of
support for the full lifecycle of diagram-based CBA
exercises. For many reasons, including a need to increase
scalability, performance, maintainability, extensibility and
usability, the original Ceilidh system had to be redesigned
and re-implemented [6], [10]. While restructuring Ceilidh
into its successor, CourseMaster, the objectives have been
to increase software quality and integrate the diagrambased facility with the generic marking mechanism.
3: Design and Implementation
Five types of users participate in the assessment
process: students, tutors, teachers, developers and
administrators. Tutors and administrators are not affected
by the type of CBA. In contrast, students, teachers and
developers have additional responsibilities. Students draw
the solution within an appropriate diagram editor, teachers
author new exercises using authoring tools, and
developers create elements for new diagram domains.
Figure 2: Conceptual plan for diagram-based
CBA
Figure 2 illustrates the overall scheme. The course
developer prepares diagram editor specifications and
marking tools for a specific domain. Within the teacher’s
environment these are customised further to meet the
requirements of the diagram-based CBA exercise. A
newly built exercise contains the student diagram editor
and marking customisations. The student uses the
diagram editor to complete the exercise that is marked via
a marking scheme. A marking scheme is a program that
specifies the invocation and configuration of appropriate
marking tools [11]. Finally, feedback is returned to the
student.
• The implementation built to support diagram-based
CBA consists of:
• Two platforms: DATsys and CourseMaster
• Two authoring environments: Daidalos and Ariadne
Both DATsys and CourseMaster are written
exclusively in Java and have been tested under various
platforms running Windows, Solaris and Linux. Daidalos
and Theseus are applications based on DATsys. Ariadne
is also based on DATsys, and additionally, requires parts
of CourseMaster. The CourseMaster clients and servers
are built on top of CourseMaster and are completely
independent of DATsys.
•
•
Oracles to describe the correct output values or ranges
of values per test-data
Feedback messages for all testing cases.
Marking Tools
Typography
Marking Tool
Checks the Typography of a program
Typographic Rules, Ranges,
Weights, N-Level Feedback
Program-Features
Marking Tool
Checks specific features of a program
Oracles for Features, Weights
2-Level Feedback
Compilation
Marking Tool
Compiles a program and checks the output
Oracles for CompilationOutput
N-Level Feedback
Dynamic Test
Marking Tool
Circuit Simulation
Marking Tool
Flowchart
Marking Tool
OO Design
Marking Tool
Figure 3: Relationship between DATsys parts
Figure 3 illustrates a detailed view of the data flow
between the implemented parts. Diagram notation
specifications are authored by the course developer using
Daidalos and are expressed as domain tool libraries and
application options. These are used in Ariadne by the
course teacher who develops the diagram-based CBA
exercise. Using Ariadne, the set of domain tool libraries
and application options can be refined to address the
intent of specific exercises.
In addition, Ariadne
produces exercise setting-up and marking files that are
used by CourseMaster’s marking server.
Theseus uses the tool library and the set of
application options to allow the student to draw the
exercise solution diagram.
Upon submission,
CourseMaster’s marking server uses the marking files
that have been created in Ariadne to mark and return the
marking results with feedback to the student. A
comprehensive guide to the marking process is given in
[17]and [11].
Figure 4 illustrates a range of programming and
diagramming marking tools in relationship to their
configuration.
Of these, three are new diagram-based marking tools:
• CircuitSimMarkingTool: marks logic design
based exercises
• FlowchartMarkingTool: marks flowchart based
exercises
• OOMarkingTool: marks object-oriented design
based exercises
The circuit-simulator marking tool needs as
configuration:
• Test data that will be fed to the inputs of the circuit
for the simulation,
Proceedings of the International Conference on Computers in Education (ICCE’02)
0-7695-1509-6/02 $17.00 © 2002 IEEE
Marking Tool Configuration
Runs a program, provides test-data
and checks the output
Simulates a circuit, provides test-data,
and checks the output
Test Data, Oracles for Program Output
N-Level Feedback
Test Data, Oracles for Circuit Output
N-Level Feedback
Translates the Flowchart to a program
and reuses the Dynamic Marking Tool
Dynamic Tool Configuration
Checks specific features
of the OOD Diagram
Oracles for Features, Weights,
N-Level Feedback
Figure 4: Marking Tools in CourseMaster
The flowchart-diagram marking tool, after translating
the diagram to a program, reuses the dynamic-test
marking tool. Therefore, it needs the configuration of a
dynamic-test tool.
The object-oriented design marking tool, investigates
the student diagram for particular features such as the use
of specific relationships between predefined classes and
objects, the naming of classes and methods, and various
other exercise dependant criteria.
4: Authoring Diagram-Based CBA using
DATsys and CourseMaster
The authoring of a diagram-based CBA exercise
involves the following stages:
• Using Daidalos to build a tool library for creating and
connecting diagram elements
• Using Ariadne to build a CBA exercise (by choosing
a subset of Daidalos’ tools for the student tool library,
selecting application features, developing the marking
scheme, and configuring the marking tools and the
CBA exercise)
The full lifecycle of CBA exercises involves the
following additional stages:
• Testing and deploying the exercise through
CourseMaster
• Running the exercise and marking student solutions
• Administering the exercise and evaluating the results
Diagram-based CBA exercises are authored with ease
once appropriate marking tools have been developed. The
task is lengthy, but straightforward and the outcome
benefits both students and educators alike. Figure 5
depicts the overall process. Daidalos is used to create tool
libraries. Ariadne is then used to author CBA exercises in
the domain for which the tools have been created.
Daidalos
1. create diagram
elements by defining :
- graphic view
- connectivity
- data-model
2. create tools and make a
tool library
3 create and make
application options
CourseMaster
Ariadne
4. author diagram-based
CBA by defining:
- theseus configuration
- CM exercise files
- Marking files
Marking System
Marking
Tool
Marking
Command
A diagram-based CBA exercise
Theseus configuration
Tool Library
Available Options
File Options
Undo-Redo
Clipboard Options
Alignment Options
Grouping Options
Zooming Options
... ... ...
External Tool Options
CourseMaster exercise configuration files:
- title
- question
- exercise properties
- setup properties
- client project properties
- solution
- skeleton solution
- scaling information
Marking tool configuration
Marking scheme (mark.java)
Figure 5: Steps for authoring diagram-based CBA
Application options for Theseus must be selected, and
liked the uncluttered feel and the speed with which
Ariadne is then used to develop the marking scheme and
Theseus responded during the drawing of their solution
configure the appropriate marking tools and the
and the marking of their work. The students also
properties of the specific CBA exercise.
mentioned that learning Theseus was easy and that the
The exercise is then ready to be deployed and tested
intuitiveness of the interaction with the diagram induced a
through CourseMaster and Theseus.
playful state of mind.
The second tested exercise type incorporated
flowcharts. This exercise requires students to draw a
5: Experience
flowchart for comparing three numbers. Although simple,
In the academic year 2000-2001, exercises in three
this example uses all the nodes of the flowchart diagram
different domains were automatically assessed using
notation. The students were taught the basics in flowchart
DATsys and CourseMaster at the University of
diagrams. Being novice programmers, they had to design
Nottingham. The exercises were part of a diagramming
a simple algorithm.
course that taught a range of diagram-based concepts.
The flowchart exercise proved to be very popular.
The authored exercises domains were:
The majority of the students entered their solution directly
• Logic design
into Theseus. Others got the order of input wrong and
produced mixed results, but with the help of
• Flowchart design
CourseMaster’s feedback, they rectified their mistake to
• Object-oriented design
complete the exercise.
The exercises were authored using Daidalos and
Object-oriented design was the third domain for a
Ariadne and were deployed via CourseMaster servers.
diagram-based CBA exercise. Students were required to
The class consisted of 167 first year undergraduate
design a hotel management application according to a
students majoring in computer science.
well-defined specification. This exercise is harder to
The two logic design exercises were to draw a simple
solve than the circuit or flowchart exercises because of the
circuit for an elevator control board and to design a
expressiveness of the object-oriented diagram notation.
circuit for a switchboard that controls a nuclear facility.
The students had to perform critical thinking on which
A demonstration was important to familiarise the students
object-oriented elements to use and how to connect them.
with Theseus, although students were already familiar
The object-oriented design exercise has been popular due
with CourseMaster.
to its clear and well-focused question. The complex
These exercises proved to be successful. Once
nature of the solution drove a small number of students to
students understood the question, they were able to
draw the solution on paper first and then to enter it into
identify the necessary components and draw the solution.
Theseus. Upon closer inspection, the root of the problem
The vast majority of the students came up with the correct
seemed to be that these students preferred to design on
results. Some students experiencing difficulties were
paper in general. Overall, diagramming exercises met
helped by CourseMaster’s feedback and as a result were
with considerable success.
able to draw the correct circuit. Students particularly
Proceedings of the International Conference on Computers in Education (ICCE’02)
0-7695-1509-6/02 $17.00 © 2002 IEEE
6: Future Work
Many other types of exercises in various diagrambased domains can be created using the combination of
DATsys and CourseMaster. Figure 6 illustrates 16 types
of possible coursework for which tool libraries have been
authored. The time spent authoring for these 16 examples
was minimal. It took just three hours to make all 16 tool
libraries using Daidalos. An additional three hours were
needed to draw the diagrams using Theseus.
Creating a tool library within Daidalos is sufficient
only if the developer is satisfied with the similar-looking
types of Theseus editors. A course developer with some
programming experience could also extend DATsys with
a new package of classes representing a completely new
editor. In this case, the concepts modelled within
DATsys could be directly reused.
Devising a marking strategy and appropriate marking
tools is an essential step in authoring the CBA exercise.
In some cases, common diagrammatic marking tools can
be reused.
In others, completely new tools and
configurations are necessary.
Diagram-based marking tools can be developed to
mark a selection of diagram notations that have been
illustrated in figure 6. The feature tool can often be used
to address any shortcomings in other marking oracles.
• Data-flow diagrams could be marked in a similar
manner to flowcharts.
• Database scheme diagrams could be marked by using
a suitable tool that converts the diagram to a database
table, runs SQL queries, and tests the output data
using oracles.
• Network diagrams might be converted into formats
understood by various network simulator tools. Such
tools can perform a variety of tasks including load
balancing distribution examination, data throughput
analysis and performance scaling investigation.
• Pert diagrams can be marked with the combined use
of a pert simulation tool and a feature tool. The
simulation tool would calculate and evaluate time
dependencies and identify any discrepancies in the
students’ solutions.
• Medical diagrams (and any other picture-based
diagrams) can be assessed by developing a marking
tool of the graphical/hot spot category.
• Analogue circuit diagrams could be marked with the
use of an analogue simulator tool. A converter would
1. Arrows and Cells
2. Data-Flow Diagram
3. Database Schema
4. ERD
5. Structure Diagram
6. Process Diagram
7.Network diagram
8. Pert Diagram
9. Mind Map
10. Medical Diagram
11. Petri Net
12. State Transition Diagram
13. Graph – Tree
14. Chemical Diagram
15. Concept Map
16. Analog Circuit Diagram
Figure 5: Various views of Theseus for sixteen notations
Proceedings of the International Conference on Computers in Education (ICCE’02)
0-7695-1509-6/02 $17.00 © 2002 IEEE
•
•
need to convert the diagram to a net-list in a format
that is understood by an external simulation
tool such as Spice [14]. Spice could be used directly
or through a Java wrapper such as JSpice [13].
Concept maps can be marked by latent semantic
analysis tools similar to Lou’s work on essay-based
assessment [8]. A feature tool can then comment on
the presence or absence of certain types of concepts.
7: Conclusions
Evaluation results on diagram-based domains, such
as circuit design and software design, indicate that the
automation of the assessment of diagrams can be
effective and useful. CourseMaster has been available
within academia since 1999, providing support for the
marking of programming coursework. The diagram
authoring extension, entitled DATsys, was integrated into
CourseMaster in May 2000. Together they provide a
novel and realistic foundation towards authoring, running,
and administering diagram-based CBA. Exercises for
three domains have been authored, deployed, marked,
administered and evaluated with 167 students at the
University of Nottingham. Many more example exercises
can be created. Daidalos has libraries for many types of
diagrams. With the addition of appropriate marking tools
diagram-based CBA can be created for most types of
diagrams. Although developing new marking tools is a
development task that needs planning, reuse of existing
marking tools with minimal configuration modifications
is possible.
References
[1] Beck K., Gamma E., Advanced Design with Patterns in
Java, OOPSLA’97, Tutorial No. 30
[2] Brant J., HotDraw, MSc thesis, University of Illinois,
Urbana-Champaign, 1995
[3] Bull, J., Update on the National TLTP3 Project: The
implementation and evaluation of computer-assisted assessment,
Keynote, Proceedings of the 3rd Annual Computer Assessment
Conference, Loughborough, UK, 16-17 June, 1999
[4] Charman D., Elmes A., Computer Based Assessment: A
guide to good practice, Volume I, University of Plymouth, 1998
[5] Culwin F., Web hosted assessment--possibilities and policy,
Proceedings of the 6th annual Conference on the Teaching of
Proceedings of the International Conference on Computers in Education (ICCE’02)
0-7695-1509-6/02 $17.00 © 2002 IEEE
Computing/3rd Annual ITiCSE Conference on Changing the
Delivery of Computer Science Education, 1998, Pages 55–58
[6] Foxley E., Higgins C., Hegazy T., Symeonidis P., Tsintsifas
A., The CourseMaster CBA System: Improvements over
Ceilidh, 5th Annual Computer Assisted Assessment Conference,
Loughborough, UK, 2-4 July, 2001, Pages 189-201
[7] Foxley E., Higgins C., Tsintsifas A, Symeonidis P. Ceilidh,
a System for the Automatic Evaluation of Student Programming
Work, Proceedings of the 4th International Conference on
Computer Based Learning in Science (CBLIS99), University of
Twente, Netherlands, July 2-6, 1999, Section I6
[8] Foxley E., Lou B., A Simple Text Automatic Marking
System, Artificial Intelligence and Simulation of Behaviour 94
Conference for: Computational Linguistics for Speech and
Handwriting Recognition, Workshop in Leeds University, UK,
April 12th, 1994
[9] Foxley E., Mohd Zin A. M., The Oracle output recogniser,
LTR Report, Computer Science Department, The University of
Nottingham, 1993
[10] Higgins C., Hegazy T., Symeonidis P., Tsintsifas A., The
CourseMaster CBA System, submitted to the Journal of
Education and Information Technologies (official Journal of the
IFIP Technical Committee on Education) published by Chapman
and Hall
[11] Higgins C., Symeonidis P., Tsintsifas A., The Marking
Sybsystem for CoureMaster, to be presented at the 7th Annual
Conference on Integrating Technology Into Computer Science
Education, University of Aarthus, Denmark, June 24-26, 2002
[12] Johnson R., Documenting Frameworks using Patterns,
Proceedings of OOPSLA’92, ACM SIG-PLAN Notices, Volume
27, Number 10, Vancouver BC, Canada, October 1992, Pages
63-76
[13] Souder D., Herrington M., Garg P., DeRyke D., JSPICE: A
component-based distributed Java front-end for SPICE,
Proceedings of the 1998 Workshop on Java for HighPerformance Network Computing, 1998
[14] Vladimirescu A., The SPICE Book, John Wiley, New
York, 1994
[15] Vlissides J. Generalized Graphical Object Editing, Ph.D.
thesis, Stanford University, 1990
[16] Williams R., Automated essay grading: An evaluation of
four conceptual models, In Herrmann A. and Kulski M.
(Editors), Expanding Horizons in Teaching and Learning,
Proceedings of the 10th Annual Teaching Learning Forum, Perth,
Curtin University of Technology, February 7-9, 2001
[17] Symeonidis P., An in-depth review of CourseMaster
Marking SubSystem, Technical Report, LTR Group, University
of Nottingham, UK, 2001