Learning Technology publication of IEEE Computer Society Learning Technology Task Force (LTTF) http://lttf.ieee.org/learn_tech/

Contents

	From the editor ..
	International Conference on Advanced Learning Technologies (ICALT 2000), August 6-8, 2001, Madison, Wisconsin, USA
	Supporting and Increasing University Faculty Use of Computers as Teaching Tools: "Geek Week!" (Abbie Brown)
	Project-Based Learning Just Became Easy: An Introduction to WebQuests (Adam Garry)
	Employing Case Based Reasoning in Asynchronous distance education (Avgoustos. A. Tsinakos, Kostantinos. G. Margaritis)
	Magic or Realism? Transforming learning styles into design features in net-based education (Carl Eneroth)
	ActiveMath, a Web-Based Learning Environment (Paul Libbrecht, Erica Melis and Carsten Ullrich)
	Chat Services with a Voice (Hira Sathu, Ranjana Shukla and Zhong Tang)
	Telemachus: A system for submission and assessment of students programs (Maya Satratzemi, Vassilios Dagdilelis and Georgios Evangelidis)
	A Pattern Language for Architectures of Intelligent Tutors (Vladan Devedzic)
	Webbus, Tailormade Transport on the World Wide Web: An Educational Webgame (Joop van Schie and Mark Oostendorp)
	Web-based Adaptive Testing (T. M. Sri Krishna)
	Context-bound evaluations of courseware in higher education (Peter Hosie)
	Students Building Communities—Funding Schools (Avis Marie Haynes)
	Visualizing Thinking with Digital Imagery (Brian K Smith)
	sitesALIVE! to Launch New Sailing Adventure for Classroom Study (Cindy Collins)
	Technology Based Training in Ancient Egypt (Jan Seabrook and Nick Rushby)
	Digital Environments: Monitoring Changes to Teaching (Robert Fox)
	Educational Software for Computer Engineering: A Case Study of an Interactive BDD Tool (Rolf Drechsler)
	Chaucer in Cyberspace: Linking the Medieval and the Modern Worlds through Technology (Thomas J. Napierkowski)
	Conference announcements

From the editor ..

Welcome to the July 2001 issue of Learning Technology.

The IEEE International Conference on Advanced Learning Technologies, Madison, USA (August 6-8, 2001) is turning out to be a very high quality conference. The website of the event is http://lttf.ieee.org/icalt2001/. The call for participation is available in this newsletter below.

You are also welcome to complete the FREE MEMBERSHIP FORM for Learning Technology Task Force. Please complete the form at: http://lttf.ieee.org/join.htm.

Besides, if you are involved in research and/or implementation of any aspect of advanced learning technologies, I invite you to contribute your own work in progress, project reports, case studies, and events announcements in this newsletter. For more details, please refer author guidelines at http://lttf.ieee.org/learn_tech/authors.html.

IEEE International Conference on Advanced Learning Technologies (ICALT 2000)
(Call For Participation)

6-8 August 2001
Madison, Wi, USA
http://lttf.ieee.org/icalt2001/

Proceedings published by:
IEEE Computer Society Press

Keynote/invited speakers:
1. Tim O'Shea, Master of Birkbeck, University of London, United Kingdom
2. Elliot Soloway, University of Michigan, USA

The registration form, accommodation and other details are available on the website.

Further inquiries:
John Klus (klus@engr.wisc.edu)
Kinshuk (kinshuk@massey.ac.nz)

Back to contents

Supporting and Increasing University Faculty Use of Computers as Teaching Tools: "Geek Week!"

Although the observation is drawn from a single case study, it seems there may be a correlation between the distribution of propeller beanies and an increase in the computer-based technology skills of university faculty members. This is based on a recent, week-long series of workshops held at Washington State University (WSU) on the subject of using innovative technologies for teaching purposes (the beanies were included in each participant’s registration package, and were worn with great pride during the week).

Washington State University’s teach.edu program (a U.S. Department of Education PT3 initiative) provides ongoing support to faculty members interested in restructuring courses and lesson plans in order to make better use of innovative technologies for the purpose of educating pre-service, K-12 teachers. This year, as a part of the program, faculty members in WSU’s College of Liberal Arts, College of Sciences, and College of Education were offered the opportunity to spend one summer month planning and developing the integration of innovative technologies into their own teaching practice. Twenty-three faculty members were selected based on proposals submitted to a review board; the selected proposals outlined their plan, referenced the National Educational Technology Standards for Teachers (NETS-T) and explained the plan’s impact on pre-service teacher education.

In order to support the efforts of those faculty members involved in restructuring their courses, a week-long series of workshops was offered in May of this year; referred to by the participants (and ultimately the planners) as “Geek Week.” The goal was to give faculty members an chance to work with new technologies and new teaching methods in a collegial and supportive atmosphere. Along with the faculty members who received summer funding, each college was allowed to invite seven graduate students to participate in the event.

Figure 1. Registration for “Geek Week”

Workshop activities were been planned to respond to the needs of the funded faculty members. A qualitative coding protocol was applied to the project proposals; critical issues and specific skill-development requests were identified, and a set of activities was developed based on the identified needs.

“Geek Week” Workshop Development:

• Step One: All of the accepted summer proposals were reviewed, paying particular attention to what each person was looking to learn more about in order to complete his/her summer project.
• Step Two: Using Post-It notes, we created an affinity digraph of the various topics revealed in step one.
• Step Three: 5 Superordinate headings were created to help organize the various topics (“The Basics,”  “Designing Instruction,” “Getting Geeky,” “Deep Geek” and “Evaluation and Assessment”).
• Step Four: A set of thirteen workshop topics were created that (it was hoped) would most directly address the needs of the group.
• Step Five: The teach.edu project coordinators and director met to review the proposed workshop topics.
• Step Six: Workshop facilitators were identified, as were locations for each workshop. Facilitators were asked to supply a few sentences describing the goals and activities for their session.

Many of the workshops are offered multiple times and by multiple facilitators in order to allow each participant to create a schedule that is most appropriate for his/her project, and to offer multiple perspectives on some of the more popular issues. The thirteen workshops offered were:

The Basics

1. The Internet, WWW and e-mail
2. WWW: HTML and web-page editors

Designing Instruction (Human Factors)

3. Principles of Instructional Design
4. Principles of Instructional Media Production
5. An Introduction to Online Learning

Getting “Geeky” (Introduction To Software And Basic Concepts)

6. PowerPoint Workshop
7. HTML and Web-editing Workshop
8. Macromedia Director Workshop
9. Digital Imaging Workshop
10. Digital Sound and Video Workshop

“Deep Geek” (Media Production)

11. Video and Audio in a digital environment    
12. Authoring Tools for CD-ROMS and Web-Based “Applets”

Evaluation and Assessment

13. Evaluating Instruction and Instructional Media

Workshop facilitators included members of WSU’s College of Education and the university’s Center for Teaching, Learning and Technology (CTLT). Most facilitators were expert or “super-user” level technologists. Since the participants came to the workshops with a wide range of abilities and experiences, some were complete novices, others had spent years using computing tools for teaching purposes, there was some concern that the presentations might be inadvertently daunting to some participants if the facilitators addressed the concerns of the more experienced (like most human endeavors, “showing off” seems to be a natural inclination during technology-oriented activities). For this reason, the following was included in the opening pages of the program provided to participants:

Unfortunately, even the most well-meaning geeks are sometimes difficult to understand (even among other geeks!). On behalf of all the workshop facilitators, here are two recommendations for making the most of the workshops and the labs:

Ask questions. We’re working under the assumption that, during this week, everyone (including the facilitators) will need clarification about some of the most basic principles of instruction and technology.  We’re working under the assumption that all questions are good questions.

Insist that everyone speak plain English. Don’t let anyone use an acronym or an abbreviation that you don’t understand. Geeky acronyms and abbreviations are intended to make life easier; URL is much easier to say than “Universal Resource Locator,” but they can sometimes be used (inadvertently, of course) to confuse people. At some point this week, just about everyone will need something, “translated from the geek.”

One more thing to consider: This week’s most important phrase is “User-Centered.” And you are the user! In every lab experience, facilitators and hypernauts will assume that anything that goes wrong is the result of shoddy design on the part of the hardware and software manufacturers. It’s not your fault if the manufacturers made things difficult to use! We will ask that you bear with us when problems arise; we’ll do our best to figure out and explain how these tools are supposed to work and what we can do to make them work to your best advantage.

The combination of a qualitative-analysis approach to the creation of the workshops and the advice offered in the program seems to have contributed to the success of the event. Although formal evaluations have not yet been analyzed, the anecdotal information gathered indicates that most people felt satisfied with the workshops and confident in their abilities to further their own personal development. A common statement made many times toward the end of the week was, “Now I know what questions to ask.”

Although analysis and a user-centered approach played a large part in the development of a successful program, it is the author’s contention that the propeller beanies distributed at the beginning of the week played their own very special part in the process.

Acknowledgement

The author greatfully acknowledges the support of the U.S. Department of Education’s PT3 (Preparing Tomorrow’s Teachers to Use Technology) program; the work of event organizers Ian Quitadamo, Greg Hooks, Skip Paznokas; and Washington State University’s Center for Teaching, Learning and Technology.

References

• International Society for Technology in Education (ISTE) (2000). National Educational Technology Standards for Teachers. Eugene, OR: ISTE.

Abbie Brown, Ph.D. Department of Teaching & Learning Washington State University Tel: +1 509 335 1631 abbie_brown@wsu.edu

Back to contents

Project-Based Learning Just Became Easy: An Introduction to WebQuests

Much of the teaching profession is beginning to incorporate project-based learning (a comprehensive approach designed to engage student investigation of authentic problems) into the curriculum, and many agree that it would be nice to have an aid to make this task easier. Sometimes we struggle to create an environment that allows students to marry technology with higher order thinking skills. To make this happen, a number of teachers are turning to the WebQuest--an inquiry-oriented activity, which some or all of the information that learners interact with comes from resources on the Internet (Dodge 1995).

WebQuests allow students to complete authentic projects and use technology to find and present information and, at the same time, alleviate some of the barriers teachers may find in their attempt to work in the confines of project-based learning. In planning for such learning, we should always start with the end in mind. A WebQuest can provide the educator with the project for a unit of study. After reading through a specific WebQuest, the teacher can then begin to select the standards to be mastered. An alternative way to start would be to select the standards, and then find a WebQuest that would help master those standards. Subsequently, the teacher can decide which enabling activities need to be taught. Every WebQuest has an Introduction, a Task, Resources, a Process, Evaluation (with a Rubric), and a Conclusion.

The WebQuest Way of Doing Things

The introduction provides the student with a chance to recall some prior knowledge and to become excited about the project. In the task portion, they are given the assignments that they will complete for this particular WebQuest. The process walks them through a systematic approach for completion. Resources offer the students and the teacher Web pages, books, magazine articles, and so on that will help with the project. By creating an evaluation component, the WebQuest allows the students to know how their learning will be measured before they begin their project, and it also helps the teacher as the evaluation is there for them to use or modify. Lastly, it includes a conclusion to summarize what the students have done. Some WebQuests offer Enabling Lesson Plans, while others contain extension activities or perhaps a teacher page.

Why WebQuests?

Students enjoy WebQuests because they are given the opportunity to use the Internet to find and apply information. The students also learn to use presentation software, which allows the students to impart their information in a creative way while educating others.

Teachers benefit from WebQuests in a number of ways. For example, rubrics for each project and Web resource addresses are provided, authentic learning occurs, ideas for projects are supplied, and learning is fun for students while they integrate technology. WebQuests can and should be modified by the teacher to fit the needs of the classroom.

The WebQuest Experience

As you can imagine, having the students complete a WebQuest during a project can be quite a wonderful experience. Here are some tips to help you start and succeed with your first few:

1. Always perform the initial WebQuest as a group so that the students may become familiar with the process and ask any questions as you proceed. A good example for this is Create a Habitat (http://www.biopoint.com/WebQuests/dist204/Welcome.html).
2. Before you create a WebQuest on your topic, search the Internet to make sure it doesn’t already exist.
3. For your first WebQuest don’t try to create your own--ease your way into the process.
4. Preview your WebQuest before you use it.
5. Change the WebQuest so that it will meet the needs of your class. Sometimes you will have to alter the procedure to suit the one-computer classroom or break up the steps.
6. Remember to let the students explore and have fun. There will be times when the WebQuest does not proceed the way you intended, but learning is still taking place--just go with it.

Once you have used a WebQuest as part of the learning process, you will share the experience with your colleagues. The higher levels of learning that can occur during any given WebQuest are amazing.

If you do not feel that you are technologically savvy enough to take advantage of a WebQuest, remember that there are plenty of resources on the Internet that can help you with software and hardware questions. For Microsoft questions, visit http://www.microsoft.com/education/tutorial/classroom/default.asp, and Macintosh questions can be answered at http://ali.apple.com/.

Time to Create My Own WebQuest--Where Do I Start?

You have scoured the Internet and you cannot find the WebQuest that you need. It is time to create your own. There are many tools to get you started--for example, a useful design flow chart can be found at http://edweb.sdsu.edu/webquest/Process/WebQuestDesignProcess.html. If you like to plan away from the computer, you can access http://storywind.net/field/minisab/technology/ and print out the design sheet. Also, you can visit http://edweb.sdsu.edu/webquest/materials.htm for all the materials and resources you will need to create your own WebQuest.

Once you know what you would like to do and have created a plan, it is time to make your WebQuest. It is not necessary to know html (hypertext markup language) to complete a WebQuest. Just knowing the basics of a Web page will be fine. Below is a list of ways in which you can make and store your WebQuest.

1. You can create the WebQuest in Microsoft Word or AppleWorks and save as a Web page.
2. Knowledge In Knowledge Out (http://www.kiko.com) has a WebQuest creation tool that you can use online (Internet Explorer only). It takes you through the process and all you do is click in the boxes and type.
3. Homestead.com (http://www.homestead.com) is also a drag-and-drop Web page creator that allows you to make your page online. The best part about this resource is that you don’t have to worry about uploading.
4. Geocities (http://www.geocities.com) is a wonderful place to store your site. From the main page, you can click on Upload and FTP (File Transfer Protocol), and then you can upload your files.
5. Another great place to store your Web page is http://nbci.com This site is very unobtrusive and makes uploading easy. When you go to the site, you can click on My Web Site and it will take you through the process.
6. Of course, you can always use a Web editor like Microsoft FrontPage, Claris Homepage, Adobe PageMill, Netscape Composer, or Macromedia Dreamweaver. After you create it here you can use any of the above sites to upload your page.

Preparing a WebQuest to be published on the Internet can be frustrating, but there are many sites that make the process very easy. If you learn the basics of Web creation and uploading, then you will be well on your way.

Once you have mastered creating WebQuests, it will be time for you to transfer that knowledge to others. A great place to start is with your students. When students create their own WebQuests, they will be using many higher order skills. They will need to know a great deal about their subject area if they are creating a WebQuest that others must complete.

WebQuests are the curriculum drivers of the future. They provide teachers and students with the necessary tools to thrive in a project-based learning environment. Join the wave of teachers who are turning to WebQuests to facilitate such learning and create an environment in which students are using high order thinking skills to solve real world problems.

WebQuest Tips

1. A good place to start is the WebQuest about WebQuests: http://edweb.sdsu.edu/webquest/webquestwebquest.html
2. Always name your home page index.html.
3. Try to blend some pictures in with your text.
4. Look at the WebQuest rubric before and as you create your site: http://edweb.sdsu.edu/webquest/webquestrubric.html
5. Join the WebQuest listserv. This is a great community of people who can help you with questions. You can join by sending a subscribe message to webquest@egroups.com.
6. Visit your WebQuest periodically after it has been posted to the Internet and check for bad links within your resources.
7. If possible, create a teacher page that gives the teacher some tips and other resources that will help with the completion of the WebQuest.

Web Resources

The Main WebQuest Page
You will find all the material needed to complete and use WebQuests at this site: http://edweb.sdsu.edu/webquest/webquest.html

EdHelper
Here is a site that has a tremendous number of WebQuests sorted by subject area: http://www.edhelper.com

The Big WebQuest Collection
This site has tons of resources that have been created by educators and pre-service teachers. I particularly like the Indiana State Extensions and New Mexico State University (WebQuests in both English and Spanish): http://edweb.sdsu.edu/webquest/webquest_collections.htm

My WebQuest collection
WebQuest pages that I have compiled for teachers like you: http://www.geocities.com/agarry.geo/webquest.html

Grades K-2
http://www.esc20.net/etprojects/formats/webquests/summer99/northside/bug/default.html

Grades 3-5
Harry Potter WebQuests that ask students to take on some untraditional roles: http://www.plainfield.k12.in.us/hschool/webq/webq113/
Invention WebQuest--This one will take you away from the traditional inventors: http://www.esc20.net/etprojects/formats/webquests/summer99/northside/Inventions/default.html

Grades 6-8
Roller Coaster Madness--The students will be creating the fastest roller coaster ever, using the history and science of roller coasters to do this. Also, they will have a chance to simulate their creations online: http://www.esc2.net/TIELevel2/projects/roller/

Grades 9-12
Personal Reflections on Vietnam--This WebQuest will have your class debating the war for many weeks: http://www.richmond.edu/~ed344/webquests/vietnam/intro.html

Adam Garry School Consultant, Co-nect agarry@co-nect.net

Back to contents

Employing Case Based Reasoning in Asynchronous distance education

Abstract

The current paper describes the employment of Case Based Reasoning in the implementation of domain independent educational environment called See Yourself Improve (SYIM).

SYIM aims to the provision of personalized distance education services during asynchronous distance education sessions, contributing also to the construction of student models.

Introduction

Asynchronous distance education appears as one of the most appealing instructional delivery methods as it combines flexibility to access teaching material with time to reflect, self-study techniques with peer-to-peer collaboration, and the use of low-cost technology. However, a number of pedagogical problems appear when confronted with asynchronous distance education (Ritchie and Hoffman, 1997; Wulf, 1996; Pernici and Casati, 1997; Relan and Gillani, 1997; Hunt 1999; Pritichard, 1998; Khan, 1997; Harasim, 1990;Hall, 1997b; Holden and Wedman, 1993).

Students may feel isolated due to the distance factor or due to the lack of live interactions with the tutor or with classmates. On the other hand, due to the discontinuous character of asynchronous distance education, these sessions may also suffer from lack of intense supervision and lack of specialized feedback to students, according to their individual learning needs and preferences. So monitoring students’ performance may become a problematic process, while conclusions about the teaching effectiveness of a particular instructor may also be difficult to draw.

SYIM is a domain independent educational environment which has been developed in order to remedy some of the educational problems appear in asynchronous distance education, such as lack of intense supervision, hazy monitoring of the students’ performance progress, inability recording individual learning needs etc (Tsinakos and Margaritis, 2001).

The core idea of SYIM is to help the tutors to monitor the individual learning needs and the misconceptions of the distance students and to keep a track of the feedback provided to each student. Additionally SYIM provides to the students the benefit of the intense supervision related to their individual learning needs and the effective support and guidance on how to overcome a misconception or remedy a performance gap in order to improve both their performance and their context comprehension.

The idea of employing  Case Based Reasoning

Case Based Reasoning (CBR) is a method used in the field of Artificial Intelligence, which explores a range of human cognitive behaviour, including learning, memory, planing and problem solving (Han 1993, p. 8).

A variety of definitions regarding CBR are available in the literature (Leake, 1996; Riesbeck & Schank, 1989; Aamodt & Plaza, 1994; Kolodner, 1991). A common node among these definitions is that CBR solves a new problem by recognize its similarity to a previously known problem and adapting the known solution to the new problem (Riesbeck and Schank, 1989; Kolonder 1990).

Research in CBR deals with a variety of problem solving techniques which make use of specific previous cases (use of experience) rather than general domain knowledge. Much of human reasoning is case-based rather than rule-based. An experience tutor can instantly recognise the student's errors without having to execute a number set of rules stored in a rule base. On the contrary the tutor based on the experience can recall cases form the past (previous instructional sessions) which are similar to the current tackled, and use the to solve or to remediate the problem.

"When the tutor encounters an unfamiliar situation he/she attempts to retrieve the most comparable cases in his/her experience, and modifies these cases to fit the new situation. If the modified case works, then the tutor memorize the case for future use by updating his/her memory." (Han 1993, p. 9)

Research has been done on the potentiality of CBR to improve the process of student modelling as some researchers studied domain knowledge representation schemes, which might assist students’ learning process by focusing on memory structures (Bumbaca, 1988; Riesbeck and Schank 1991).

The employment of Case Based Reasoning in the SYIM educational environment is believed that it will be beneficial both for the tutors and the students. Such employment aims to automate the process of replying to the student's misconceptions, and will contribute to the construction of tutoring paths in order to advice the student how to overcome a performance problem based on the SYIM's experience of previous cases.

The implementation schema

In the initial version of SYIM the communication schema between the tutor and the student follow the procedure that appears in Figure 1. According that procedure, each of the student's misconception was posted directly to the tutor.

Figure 1.Misconceptions’ communication schema

Similarly, each of the tutor’s reply was posted directly towards the student and by the same time, both messages were saved in a Misconception Data Base which was responsible to record the tutor-student interaction.

The process of answering students' queries can become a time consuming procedure for the tutor as the number of misconception posting increases during an educational session. Additionally the delay of receiving tutor's reply, on the student's side, is also increased by time.

To overcome this problem, in the new version of SYIM, CBR have been employed, resulting to the radical modification of the tutor/student communication schema (Figure 2).

Figure 2. Employment of CBR in SYIM

According the new schema, when a student posts a misconception-query, the posting does not reach the tutor immediately. Each student's posting is recorded in the Misconception Cases Data Base. The scope behind this procedure is to preserve the student's personalized information used for the construction of the student model by SYIM. Furthermore this procedure enables the tutor to be aware of each student's the misconceptions postings, regardless of the reply source (system or tutor). As a next step, a Diagnosis Process is triggered in order the SYIM system to be search for the location of a relevant case that might be the answer on student's original posting. The Diagnosis Process, in order to identify a similar case, applies controlled vocabulary search in addition with free text search among the contents of the "Educational Knowledge Base" where the misconceptions with further educational value are stored. Therefore two scenarios are possible:

Scenario a "No relevant case is fount": This scenario is valid when the Diagnosis Process fail to identify and retrieve a similar case. The reason for this failure may be that the student's posting is an original misconception that has never been asked in the past. In such a case, the misconception is posted directly to the tutor.

The tutor now is responsible not only to reply to the student's question but also to decide if the posted misconception is an important new case whichhas a further educational value (helpful for other students) and therefore should enrich the contents of the Educational Knowledge Base.

If so, tutor's reply along with the student's question, beyond reaching the student, is also stored in the Educational Knowledge Base. If the tutor decides that the misconception is educationally unimportant, then the tutor's reply goes directly to the student without being registered in the Educational Knowledge Base. In this way, the Educational Knowledge Base is enriched only with those cases that are critical and facilitate the instructional process, preserving in that way, its internal integrity and validity.

Scenario b "One or more relevant cases are fount": If this scenario is valid, then the retrieved set of cases is displayed to the student, in a ranked order according to their relevancy towards the posted query. Each case include the question posted in the past by a different student, which concerns a similar concept to the newly posted query, in addition with the tutor's reply on how the student could overcome the particular problem. Therefore the student can select among the listed cases, the most proper case which provide an answer to his query. Having done that, the process of "Reply to the Student" is automatically terminated without the interference of the tutor. Furthermore the student can vote for each case if it was really helpful or not.

If the student can not identify a proper case which answer the initial misconception (improper case), then the process of "Scenario a" is triggered.

It is worth to mention that a case can include as part of its contents a number student-tutor nested dialogue messages which formulate a chain of navigational tutoring steps (tutoring paths) on how a student can overcome a particular problem.

This feature is extremely useful and time saving for the tutor, considering that a number of students' misconception appears repeatedly during an instructional session. On the other hand, this feature is also beneficial for the students as they can easily find a pre stored answers to their question and therefore they can proceed in the instructional material without time delays due to the inability of the tutor to provide an immediate answer.

A further additional benefit of this feature is that a number of past related queries are displayed to the student and therefore the latter can identify, explore and resolve some other critical instructional concepts

In conclusion, the employment of Case Based Reasoning as described above, acts beneficial both for the tutor and the students. The Diagnosis Process is a time saving feature which assist tutors during their instruction by decreasing the number of the questions seeking for an answer, accelerates the instructional process and also contributes to the content comprehension on the students' side.

References

• Aamodt, A. and Plaza, E., (1994) Case-Based Reasoning: Foundational Issues, Methodological Variations, and System Approaches AI Communications, 7(1), pp. 39-59.
• Bumbaca, F., (1988). Intelligent Computer Assisted Instruction: A Theoretical Framework, Internetional Journal of Man-Machine Studies, 29, pp. 227-255.
• Hall, B. (1997b). Web-based training cookbook. New York: Wiley Computer Publishing.
• Han, J. K., (1993). Case Based Cognitive Modeling: A studend modeling methodology for an intelligent tutoring system. Phd Thesis, University of Nebraska-Lincon.
• Harasim, L. (1990). Online education: An environment for collaboration and interllecual amplification. In L. Jarasim (Ed.), Online education: Perspectives on a new environment. (pp. 39-64). New York:Praeger.
• Holden, M. and Wedman, J.F. (1993). Future issues of computer-mediated communication: The results of a Delphi study. Educational Technology Research and Development, 41 (4), 5-24.
• Hunt, D. L. (1999). Web-based training. In W. J. Rothwell& K.J. Sensenig (Eds.), The sourcebook for self-directed learning(pp. 121-145). Amherst, MA: HRD Press.
• Khan, B.H. (Ed.). (1997). Web-based instruction. Englewood Cliffs, New Jersey: Educational Technology Publications, Inc.
• Kolonder, J.L, (1990). An Introduction to Case-Based Reasoning, Technical Report, GIT-ics-9019, School of Information and Computer Science, Georgia Institute of Technology.
• Kolonder, J.L, (1991). Improoving Human Decision Making through Case-Based Decision Aiding, AI Magazine, 12(2), pp. 52-68.
• Leake, D., (1996) Case-Based Reasoning: Expriences, Lessons, & Future Directions Menlo Park, California: AAAI Press.
• Pernici, B. and Casati, F. (1997). The design of distance education applications based on the World Wide Web. In B. H. Khan (Ed.) Web-based instruction. (pp. 245-254). Englewood Cliffs, New Jersey: Educational Technology Publications.
• Pritichard, C. L. (1998). The ASTD handbook of instructional technology. New York: MacGraw-Hill, Inc.
• Relan, A. and Gillani,B. B. (1997). Web-based instruction and the traditional classroom: Similarities and differences. In B. H. Khan (Ed.) Web-based instruction. (pp. 41-46). Englewood Cliffs, New Jersey: Educational Technology Publications.
• Riesbeck, C.K., and Schank, R.C, (1991). From Training to Teaching: Techniques for Case-Based ITS, In Eds. Burns, H., Parlette, J.W., and Redfield, C.L, Intelligent Tutoring Systems: Evolution in Design, Lawrence Erlbaum Associates, Hillsdale, NJ, pp 55-78.
• Ritchie, D. C. and Hoffman, B. (1997). Incorporating instructional design principles with the World Wide Web. In B.H. Khan (Ed.) Web-based instruction. (pp. 135-138). Englewood Cliffs, New Jersey: Educational Technology Publications.
• Tsinakos, Α., and Margaritis, K., (2001).See Yourself IMprove (SYIM). An educational environment for the provision of personalized distance education services and the formulation of student models", accepted in European Journal of Open and Distance Learning (EURODL), ISSN 1027-5207.
• Wulf, K. (1996). Training via Internet: Where are we? Training & Development, 50 (5), 50-55.

Avgoustos. A. Tsinakos, Kostantinos. G. Margaritis Department of  Informatics University of  Macedonia Thessaloniki Greece tsinakos@uom.gr kmarg@uom.gr

Back to contents

Magic or Realism? Transforming learning styles into design features in net-based education

Abstract

Advantages in comparing net-based education to traditional teacher lead learning are recognized. However, in the hype of “e-learning”, the real needs of users/learners are not given sufficient attention. The present article points to the importance of developing net-based education in accordance with principles for user-centered design. The purpose is to demonstrate the use of learner-centered design in accommodating users with different types of learning styles. This article gives an account of a theory of learning styles and it goes on to show how different learning styles may be transformed into design features

(This article is a short version of a paper submitted to the Nordic Interactive Conference in Copenhagen (30/10 – 3/11/01) with the same title. It was written by Carl Eneroth, Cecilia Katzeff and Rasmus Larsson within the project “explore:e-learning styles” in the Explore studio at Interactive Institute, Stockholm, Sweden)

Learner-centered design

To link theories to practical design, it is important to possess adequate methodological tools. Learner-centered design (LCD) provides a framework for such tools [8][7]. The central claim of LCD is that interactive media may embody learning supports, which can address the learner’s knowledge level, motivation, and diversity. These are important aspects when viewing users as learners and designing for usability.

Learning styles

The concept of Learning Styles originated in the attempt among educators and researchers to look for the source of individual differences between children [6]. One such source was Learning Style or learning orientation, which More defines as: “the characteristic or usual strategies of acquiring knowledge, skills and understanding by an individual” [6]. Empirical research indicates that there is a correlation between learning style and the effect of learning. Learners may have different, mostly unconscious styles for learning. 

Kolb’s Experiential Learning Model

Kolb found that the uniqueness of these learning styles or ‘most comfortable ways to learn’ is influenced by “the combination of how people perceive and how people process” [5]. By joining two dimensions of how we perceive (concrete experience and abstract conceptualisation) with how we process (active experimentation and reflective observation), Kolb established four learning styles: diverger, assimilator, converger and accommodator (Figure 1) [4].

Figure 1. Adapted from Kolb’s [4] Learning Style Inventory of how we perceive and how we process information with learning styles presented in italics

In Kolb’s model, each learning style is thus a combination of two preferred learning skills: diverger (interpersonal and information skills), assimilator (information and analytical skills), converger (analytical and behavioural skills) and accommodator (behavioural and interpersonal skills) [1]. The focus in the project ‘explore:e-learning styles’ was on two learning styles: diverger and converger, while the assimilator and accommodator was left for later prototype developments.

Design criteria

Explore’s intention when transforming learning styles into design features in Explore’s e-learning styles project, was to use the principles of learner-centered design to first identify design criteria for two learning styles and then use these criteria to build a prototype. These were the design criteria used:

Table 1. Design criteria for two learning styles based on Kolb’s model (Figure 1) divided into four themes

Design features – preliminary end results

This article is written at a time when Explore has presented its first functional programs adapted to the diverger and converger learning style, but before the observation, measurement and evaluation of end user behaviour has started. The empirical data gathered from these usability tests will be employed not only in the coming version of the diverger and converger programs, but also in adaptations made for individuals with the assimilator and accommodator learning style. The screen dumps in figure two to seven illustrates aspects design features in the educational program prototypes adapted to the diverger and converger learning style.

Dramaturgy

Figure two introduces an unpredictable virtual world to experience with interactive tools that the diverger discovers when clicking randomly on the computer screen. Finding a tool provides the user with more information. The task is to rescue people who are stranded on an island (Figure 2). The underlying idea is to present a meaningful task that answers to the diverger’s principal question: why?

Figure 2. First impression in the Diverger program: an unpredictable virtual world to experience

Figure three demonstrates a predictable world for the converger to explore with interactive tools displayed in the hierarchical menu on the top left side of the screen. The menu links to specific topics of direct importance when taking out a course on a sea chart (Figure 3). The underlying idea is to present relevant information that answers to the diverger’s main question: what?

Figure 3. First impression in the Converger program: a predictable virtual world to explore

Information architecture

Figure four illustrates the information architecture and user processes for the diverger in the form of non-linear presentation of information. This setting allows for non-relevant features to be discovered in the virtual world, as a mean to indirectly reach the educational goal of the program. The learning outcomes are thus unintentional from the outset (Figure 4).

Figure 4. Non-linear presentation of information for the Diverger with unintentional learning outcomes

Figure five illustrates the information architecture and user processes for the converger, who favourslinear presentation of information. Non-relevant features in the educational program are avoided to more directly guide the user towards the educational goal of the program. The learning outcomes are thus intentional from the outset (Figure 5).

Figure 5. Linear presentation of information for the Converger with intentional learning outcomes

The key in figure six explains the meaning of the boxes and arrows shown in figure six and seven (Figure 6):

Figure 6. Key explaining the boxes and arrows of figure 6 and 7

Collaborative learning

Figure seven shows a mariner to cater for the diverger’s social cravings. The engaging mariner introduces reasons for taking out a course on a sea chart, which is useful later on when solving the task to rescue the people stranded on an island (Figure 7).

Figure 7. The Mariner supporting the Diverger’s need for social relation building

Figure eight exhibits a sea chart to satisfy the converger’s need for factual information that leads to the right knowledge. The simple operations employed when taking out a course on a sea chart is shown in sequence with an instructional voice-over (Figure 8).

Figure 8. Sea chart supporting the Converger’s need for hands-on factual information

Conclusions

The article have focused upon two major themes:

1. Ideas motivating the work of designing individualised interactive media for learning, and
2. Description of the process of transforming characteristics of learning styles into design criteria. 

Although hypotheses have been made about our two prototypes being accommodated for users with two different learning styles, these hypotheses are yet to be tested. One of our prototypes has been designed to best suit divergers and the other prototype has been design to best suit convergers. To test whether we have succeeded in our design endeavours an experiment is planned involving learners predominated by either a convergent or a divergent learning style.  

Experiences from using learning-centered design in practice show that:

• Kolb’s model (Figure 1) provides target group segmentation that function well with the learner-centered design approach and is easy to communicate internally and externally, but Dunn & Dunn’s model (Section 3.6) offers in general more practical design information;
• Testing and evaluating the effect and function of the program content necessitates the user-interface of the application to be well designed, completely implemented and debugged;
• Depending on the design criteria selected for the application, different development methods for steering the production process is required;

The design criteria for Kolb’s learning styles (Figure 1) can be structured into at least four themes:

• Design concept that captures the overall idea;
• Dramaturgy that conveys the first impression and the plot;
• Information architecture that reflects the attitude towards the integration of text, speech and other features; and
• Collaborative learning that reveals the approach to learning with or without other persons.

The preliminary end results are that it is possible to design two different programs that transform two learning styles according to Kolb’s model (Figure 1) into specific design features.

• For the diverger the program is produced as an unpredictable virtual world to experience, where information is presented in a non-linear fashion, the learning outcomes are unintentional and there is room for social relation building.
• For the converger the program is created to reflect a predictable virtual world to explore, where information is presented in a linear fashion with intentional learning outcomes and hands-on factual information introduced early.

Future directions

Regarding future directions, net-based programs that support process modelling for Kolb’s assimilator and object modeling for Kolb’s accommodator are ways to further materialise the idea of exchangeable user-interfaces to be used by individuals engaged in computer supported collaborative learning. In a not so distant future, net-based educational programs may be equipped with controls that after being adjusted to an individual learning style dynamically alters the program’s user-interface accordingly (Figure 9). In practice, this may be constructed through a limited number of pre-set paths adapted to a specific learning style.

Figure 9. Future vision of a net-based educational program equipped with controls that after being adjusted to an individual learning style dynamically alters the program’s user-interface accordingly

Finally, integrating Jung’s theory on personality types into Kolb’s model opens up for a more precise correlation between individual learning styles and specific design features. This may be accomplished by inviting the test persons to take Myers-Briggs Type Indicator test [5] that is built on the personality types of Jung. Other complementary models to Kolb’s experiential model, such as Gardner’s idea on multiple intelligences [3] and Dunn & Dunn’s model [2] including external conditions, allows for a scientific analysis on separate aspects disregarded by Kolb.    

References

[1] Boyatzis, R.E. and Kolb, D.A. (2000). Learning Skills Profile (LSP), McBer & Company, [On-line]. Available: http://trgmcber.haygroup.com/Products/learning/lspus.htm [2001-05-09]

[2] Dunn, R. and Dunn, K. (1999). The Complete Guide to the Learning Styles Inservice System.USA: Allyn & Bacon

[3] Gardner, H. (1983). Frames of mind. New York: Basic Books

[4] Kolb, D. A. (1976). Learning Style Inventory: Technical Manual, Boston, MA: McBer and Company. In Hälsouniversitetet, Linköping. (1994). Learning Style Inventory: A Self-description of Preferred Learning Modes, abridged and excerpted from Kolb (1976).

[5] Kolb, D. (1984). Experiential learning: Experience as the source of learning and development. CITY, New Jersey: Prentice-Hall.

[6] More, A.J. (1987). Native Indian learning styles: A review for researchers and teachers. Journal of American Indian Education, 27 (1), 17-29, [On-line]. Available:  http://jaie.asu.edu/v27/V27S1nat.htm [2001-04-19]

[7] Nielsen, J. (1993). Usability Engineering. Academic Press.

[8] Quintana, C., Eng, J., Carra, A., Wu, H-S., Soloway, E. (1999): Symphony: A Case study in extending learner-centered design through process space analysis, Proceedings of CHI‘99, 473-480.

Carl Eneroth Tel: +46 8 783 24 85 carl.eneroth@interactiveinstitute.se

Back to contents

ActiveMath, a Web-Based Learning Environment

System Overview

In Saarbrücken we are developing the web-based, user-adaptive, interactive learning environment ActiveMath. ActiveMath does not rely on predefined courses but offers the dynamic construction of courses. Currently, its major features are user-adapted content, sequencing, and presentation, support of active and explorative learning by mathematical services, support of teachers by information about their students, and a semantic encoding of content that is the basis for reusability.

For an ActiveMath session, the user chooses her learning goals from the concepts in the knowledge base. Then the systems recursively retrieves all concepts the goals depend upon and applies pedagogical rules and information from her user model to select the most appropriate content and form to present the content. This yields an instructional graph whose linearization is converted into HTML pages and presented to the user (see Figure 1).

Figure 1. An ActiveMath session

Figure 2. Architecture of ActiveMath

Figure 2 depicts the client-server web-architecture of ActiveMath. Currently, ActiveMath integrates the following components: a session manager, the knowledge base, a presentation planner, a user model, a pedagogical module, and mathematical services such as the proof planner Omega [3] and the Computer Algebra System (CAS) Maple. Requests of the user and (in the other direction) HTML-pages are communicated via a web-server to the session manager. The session manager stores the generated courses and translates URL requests into actions (e.g., the request for a new course about the topic group) that are passed to the responsible component. The presentation planner generates the personalized learning documents by requesting and processing information from the knowledge base, the user model, and the pedagogical module. Information about the user's actions, such as the success of solved exercises, is passed from the session manager to the user model where it is used for updating.

ActiveMath was designed with the goal in mind that the different components can be easily exchanged. For example, we will soon replace our simple table-based user model by an Bayesian net. Furthermore, our architecture makes it relatively easy to integrate new mathematical services and to design and implement interactive exercises.

Knowledge Representation

The knowledge base contains mathematical knowledge represented in the XML-based OMDoc [2] format. This allows a fine-grained representation of mathematical content by items such as definition, proof, theorems, motivation, etc. The items may include natural language formulations as well as formal objects. These formal objects (e.g., symbols) relate to actual mathematical objects, i.e., semantics. For instance, independent of whether the presentation is ``plus'' or ``+'', both presentations relate to the unique mathematical operation. This semantics provides an ontology for the content of the course which is indispensable for a reuse of learning material and for a combination of material from different sources. In addition to the actual mathematical content, our knowledge representation contains meta-data for structures, dependencies, and pedagogical information which can be used for the dynamic generation of interactive documents.

Presentation Planning

The central component of ActiveMath is the presentation planner. It generates a personalized course in a three-stage process:

1. First, the content is retrieved from the knowledge base. Starting from the goal concepts chosen by the user, all concepts they depend upon and corresponding additional information (e.g., examples for a concept) are collected recursively. This grabbing process uses the dependency metadata information. The result of this retrieval is a collection of all concepts plus additional information about them that need to be known by the learner in order to understand the goal concepts.
2. Then pedagogical knowledge is applied. According to the information in the user model and the pedagogical module the collection of content items is processed and transformed into a personalized instructional graph of learning materials. ActiveMath employs pedagogical information represented in pedagogical rules (see Figure 3 for an example). Pedagogical rules are separately stored and can be employed to configure the system with pedagogical strategies and knowledge. This makes ActiveMath an excellent tool for experimenting with different configurations.
3. Finally, the instructional graph is linearized.

The result of the presentation planning is a linearized instructional graph whose nodes are OMDoc items. Filters transform this collection into HTML pages by XSL-transformations.

The rule in Figure 3 chooses exercises that involve the use of a mathematical service. It determines that if there is an exercise needed for the concept ?concept with difficulty ?difficulty and if there is an exercise for the concept with the appropriate difficulty level that requires the mathematical service ?service and the user knows how to use the service (indicated by (user-knows ?service)) then choose this exercise for presentation.

User Modeling

The user model consists of two subcomponents: the history that stores the data about user's actions, and the profile that stores the user's preferences and knowledge mastery data. When a learner registers, she can enter her estimation of her knowledge mastery values which will be constanly updated according to her actions. The user model is inspectable and modifyable. The knowledge mastery assessment is represented by values for a subset of the competence features from Bloom's taxonomy [1] namely knowledge, comprehension, and application.

Conclusion and Further Work

We briefly described the current version of ActiveMath. The next versions will increasingly offer suggestions for next learning steps, feedback for exercises (in particular interactive exercises), and communication with the user. A demonstration of ActiveMath is available at http://www.activemath.org/.

Bibliography

[1] B. Bloom (1956). Taxonomy of educational objectives: The classification of educational goals: Handbook I, New York: Longmans.

[2] M. Kohlhase (2001). OMDoc: Towards an internet standard of mathematical knowledge. In E. R. Lozano (Ed.) Proc. of Artificial Intelligence and Symbolic Computation, AISC'2000, LNAI, Berlin: Springer Verlag.

[3] E. Melis and J. Siekmann (1999). Knowledge-based proof planning. Artificial Intelligence, 115 (1), 65-105.

Paul Libbrecht paul@ags.uni-sb.de Erica Melis melis@ags.uni-sb.de Carsten Ullrich ullrich@ags.uni-sb.de Saarland University Postfach 151150 D-66041 Saarbrücken, Germany and DFKI GmbH Stuhlsatzenhausweg 3 D-66123 Saarbrücken, Germany

Back to contents

Chat Services with a Voice

Abstract

This paper investigates techniques to provide additional dimension to student-student and student- teacher interaction at an academic institute. The technologies adopted also take into account the constraints posed by the Internet infrastructure that is widely used for this interaction and the affordability of the solution for the user community.  The research adopts an approach that does not involve the user community to subscribe to higher bandwidth connections. The research involved a background study of current tools and software packages to integrate voice and image functionality into a chat service. The results presented in the paper indicate the possibility of using voice in addition to text chat. However, the results of the developed interface indicate that group discipline is needed to permit and facilitate a smooth voice and text chat conference. The concept involves the transmission of text over the Internet, which is suitably interfaced with text-to-speech software at the client end. This enables voice in addition to text at the receiving ends. In the future it is intended to investigate other applications of this technology.

Introduction

A previous study (Shukla, Sathu, Zhong, 2001) shows that adding voice and image helps by way of adding a human touch to the impersonal on-line text chat and thereby improves overall learning. Real audio over the Internet would be most desired. However, in view of the prevailing data communication environment this has significant shortcomings.  Firstly, it requires committed bandwidth and other quality of service features that are not required for ordinary text chat. Secondly, even where high-speed networks with quality of service are available these are expensive and beyond the scope of the student community. Finally, even where such access is made available the overall connection may not provide the desired features from end-to-end. Alternative means to achieve additional dimensions in the learning process need to be investigated at the same time avoiding any of the above mentioned constraints.

The Approach

The research was based on the premise that communication access to the Internet would be required as for ordinary text chat. Hence this does not involve the user community to subscribe to higher bandwidth connections. To permit reasonable quality of audio conferencing, the authors opted to integrate a text to speech engine (TTS) into the chat conference services.Several existing voice application software were studied and trialed. Use of available voice application software was not adopted since these did not provide the desired flexibility for customisation. The first part of this research involved establishment of a development environment for the chat service. This was followed by the detailed interface development to provide a secure and user-friendly environment. Text input by a member (student) at the client end is sent to a chat server to be multicasted to other members of the group. When group participants receive the text, it is converted to audio in addition to the normal text message. The conversion of the text to audio at the receiver end provides a synthesised voice to accompany the usual text message.

Overview of Available Web Based Voice Application Software

There are a large number of application software that empower a user to convert text to speech. A significant number of these are available for downloading free of charge (trial version). The authors downloaded some of these and related voice application software and studied their features. These included: SpeaksForItself¹, Digalo², ReadPlease³, Bell Labs Text-to-speech⁴, and Microsoft products like Speech SDK, Chant Kit and SR & TTS engine⁵.

Implementation

Keeping in mind the study and the premise as mentioned in the above approach, the implementation was carried out under the following heads:

Application Architecture

A web-based application was developed using client server architecture. The full nature of the developed interface is achieved through embedding ActiveX controls into HTML. Active Server Page (ASP) provides the basic platform for this dynamic interaction. 

Chat server

Chat server used for the project was set-up over the Windows NT 4.0 platform. Relevant Microsoft Internet Chat (MIC) Server was installed and configured from the Microsoft Exchange Services. An embedded chat client has been chosen to seamlessly integrate into the designed code for the web pages, which can tie into the MIC server.

Database

A database was created using the "MS access" software and was installed on the server. The Database can be easily upgraded and modified.

ILS

An ILS (Internet Locator Service) is also installed on the server for searching people across the Internet. The ILS directory service (email addresses) can be used to form a chat group.

User interface

The user-friendly interface for the client end comes up as depicted in the communication window below. The software for this is invoked by typing the specified URL. This brings up the first window (Fig.1) on the desktop. On entering the correct password the voice chat room with the images of the group members appears (Fig. 2). In the voice chat room, the name of the student initiating the session appears in the “username” text area. If the student/user would like to join in the voice chat room he/she can do so by clicking on "join the chat". There is a verbal announcement too to inform that he/she has joined the voice conference. From this time onwards student/user can hear and participate in audio conferencing.

Figure 1.

Figure 2.

To end the session the user/student has to click on "Leave Chat Room" button. An option is also available for private, one to one chat. It is possible to set up a voice option that suits a user. Accordingly one selects this through "set up your voice" button.

Conclusion

To achieve the additional dimensions of voice and image the authors have used Microsoft components since they are modular and could be modified to suit the project software design and development requirements. Further the authors intend to conduct field tests of the developed software. The field test will be conducted using students/staff. It is planned to use the feedback from the student to assess its success and modify it to suit a broader spectrum of students and the community at large.  The research group expects that the addition of voice and image components will improve the overall communication process between the students as compared to the plain text chat. The metrics for this would be the student and the staff feedback questionnaires.

Bibliography

• Shukla, R., Sathu, H. and Tang, Z. “Giving Voice to Discussion Boards” NACCQ 2001.
• SpeaksForItself. http://www.speaksforitself.com/speaksforitself
• Digalo. http://www.digalo.com/
• ReadPlease. http://www.znet.com/
• Bell Labs Text-to-speech. http://www.bell-labs.com/projects/tts/voices.html/
• Microsoft. http://www.microsoft.com/download

Hira Sathu hsathu@unitec.ac.nz Ranjana Shukla rshukla@unitec.ac.nz Zhong Tang Ztang@unitec.ac.nz UNITEC Institute of Technology Private Bag 92-025 Auckland, New Zealand

Back to contents

Telemachus: A system for submission and assessment of students programs

Abstract

The software tool called Telemachus has been developed to test and grade students’ programs. Students submit their programs via WWW and then the software compiles, tests and grades them and also generates statistical results. The aim of Telemachus is not only to grade the students' programs but more importantly to provide reliable performance data that could give a reasonable gauge of student knowledge and in this way contribute to teaching programming skills.

1. Introduction

In an introductory programming course students write programs in order to develop programming skills. The task of marking the solutions students produce in their programming assignments is laborious and error-prone. Thus, a number of researchers have been investigating the possibility of integrating technology into Computer Science examinations. Among others, Mason and Woit [7, 8] report their experiments from on-line programming examinations; Preston and Shackelford [9] describe a prototype for an on-line assessment software tool; Jackson and Usher [4] developed ASSYST a system for grading student programming exercises; Tinoco et all [11] develop QUIZIT a system for online evaluations in WWW-based courseware; Joy and Luck developed BOSS [5] a system for submission and assessment of students programming assignments.

Most of these systems interested us, but were inappropriate to be used in our courses for two main reasons: such a system must match exactly the requirements of the specific course on which it is intended for use and furthermore, the system must ensure compatibility with the University databases so that electronic marksheets can be integrated into the broader process of assessment administration. Thus we decided to develop a system called Telemachus handling not only submission, program testing and marking students' assignments but also providing reliable performance data that could give a reasonable gauge of student knowledge

Our system consists of two components: the first and the simplest one provides the means by which a student submits a program electronically for grading; the second component, which is used by the tutor, directs the assessment process. In the subsequent sections of the paper, we present the capabilities of our system and we show how technology can help the teaching process. We will show that this electronic marker does not only help in the process of testing and grading programs but can also give data for further didactic research. For example, using students’ performance data we can detect their misconceptions and further we can pinpoint any particular notion that students may not have grasped fully.

2. Motivation

In the Department of Applied Informatics every year about 130 students are required to attend the CS1 and CS2 courses. These compulsory courses are offered during the first and the second semester and are comprised of a two-hour lecture and a two-hour laboratory session per week. In the laboratory session students solve some programming exercises with the instruction of a tutor. They are given a number of programming exercises as homework, whose solutions they have to submit in the next laboratory session. Almost all the exercises are small or medium sized programming problems and the average number of the programming assignments is 35-40 per course.

Up until now (i.e. before Telemachus), the following rudimentary examination procedure occurred. Since in our department there was only one assistant (like in most Universities of our country, assistants are a rarity) it was impossible to check manually all the students’ programs. Thus, what happened was that at the end of the semester the tutor along with the assistant had to examine orally every student on a small number of programs (about 5 in all) as it was humanly impossible to check the entire listings (over 5000).

Obviously, this situation could not satisfy either our students or us and it was the weak point of both courses. We admit that accurate and meaningful assessment is vitally important for many reasons. First, it provides meaningful feedback to students and instructors; quality assessment informs students of their mistakes and successes and informs instructors of student knowledge. Second, it establishes confidence in the measurement of student performance; without accurate assessment, neither students nor instructors have a reasonable gauge of student knowledge. Third, it provides instructors and administrators with the ability to perform quality control; collecting reliable performance data enables examination of the instructional process for courses. Finally, accurate assessment makes new educational research opportunities possible; customized courses, better use of class time, and student performance trend analysis are a few examples of possibilities [9, 3].

Thus, we decided to develop Telemachus, a system that tests and grades students' programs and also provides reliable performance data that will help us to detect potential students’ misconceptions.

3. Description of the system

Telemachus consists of two main components: the one that students see and by which they submit programs electronically and access the results via WWW; and the second one that a tutor views and by which he/she can test and mark programs and obtain statistical results.

Students submit programs or access their results via WWW. Telemachus, in order to permit students’ access to these operations, asks for their normal login name and password and then it permits access only to those who are students of the Department and have to attend CS1 and CS2 courses.

The second component, that of the tutor view, is composed of 6 main modules (see figure 3): Exercises, Students, Options, Reports, Marker, New Semester.

3.1 The Module “Exercises”

We can see, in figure 1, the form that handles the exercises’ database. We have added until now 200 programming exercises into the database. We have categorized them into different topics (worksheets): basic statements, operations and types, selection structures, repetition structures, arrays, strings, records, files, pointers etc. Every year we choose a number of 35-40 exercises, among those included in the database, that are different from those of the previous year.

At the top of the form of figure 1, we see the buttons, which allow the tutor to add/ delete/ edit or find an exercise. In the section below on the left, we see the worksheet’s number that the exercise refers to and the total number of exercises in this particular topic. On the right, the tutor gives the data concerning an exercise, such as: the exercise code (worksheet No, exercise No, Question No); the total number of data sets (input, output data sets); if the exercise is included in the marking process. In the middle of the form, the tutor writes the exercise. At the bottom of the form, the tutor gives some extra settings concerning the data sets.

3.2 The Module “Students”

The system handles students’ database with a form similar to that of figure 1. The tutor can add/ delete/ find a student and can edit some elements.

3.3 The Module “Options”

Using the form of figure 2, the tutor sets the options that refer to the electronic marker. This form is divided into 3 sections: the left top section contains information about the exercises that will be marked. In the top right section the tutor can choose the compiler that will be used for compiling the program and by mouse clicking the button “Compile Now” Telemachus starts the compilation process. The results of the compilation process are recorded. The bottom section contains information about students.

3.4 The Module “New Semester”

The form titled “New Semester” shown in figure 3, initializes students’ database every new semester or for each different course. Running this module, the tutor can add all students to the database by giving the total number of students who attend the particular course and the year of their enrollment in the Department (student’s IDs in the University’s database are formed in this way). This module also creates new students’ directories and deletes students’ directories of the previous year.

Figure 3.

3.5 The Module “Marker”

The button “Marker”, in figure 3, runs the module that checks the programs’ correctness. The executable code of a student’s program (which was previously generated by module “Options”) is run against the sets of test data. In the next step the module determines whether the program’s output is correct or not and marks the program. The checking approach is to match the student’s output produced by every set of the test data with the one produced previously by the system.

3.6 The Module “Reports”

Button “Reports” (figure 3) produces 3 different types of electronic reports. i) An extended report for every student, where he/she can see the following information: which programs were not successfully compiled; for every set of test data which program produced a correct or an incorrect output. ii) A report for all students with their grades. iii) A statistical report where the tutor can see for every exercise and every set of test data the rate of programs that were correct or had errors or were unsuccessfully compiled or did not gave an output due to an infinite loop.

“File” menu (figure 3, main form) handles the produced reports (open, print a report etc).

4. Test Data Adequacy & Students’ Conceptions

As we have already mentioned in the introduction, not only does Telemachus help the tutor in grading students’ performance but also in providing useful data concerning students conceptions.

As it is known, many errors in students’ programs [1, 2, 10] have an element of chance and are thus unpredictable. There are some errors, however, which are more systematic and more persistent and since they are due to students’ misconceptions can be predicted. Students produce programs that are correct for most of the cases but when these programs are tested for some data sets they beget incorrect results since students do not take into consideration all the cases. Telemachus validates students’ programs, running them against a number of predefined data sets rather than against random data sets so as to detect logical errors. We chose adequate data sets in such a way that a program with logical errors will produce an incorrect output or it will have an incorrect performance (infinite loop). Therefore, some data sets will cause students’ programs to give incorrect outputs whereas other data sets will cause correct outputs. Of course, the combination of incorrect and correct outputs does not guarantee the detection of a misconception; there are other types of errors that might be associated with the same combination of correct and incorrect outputs. Nevertheless, the combinations of the chosen data sets give valuable insight into students’ conceptions. Following, we give two examples in order to show the above.

4.1 Example 1

The first example is a series of programs which deal with the binary search [6]. We present 3 programs. The first is correct and the other two are incorrect. Table 1a gives the data sets and Table 1b summarizes the results.

Table 1a.Data sets

Table 1b. The programs and their performance

The second program gives, according to our observations, the most common error students make, while the third program shows the most usual modification that they make to the second program when students realize that it is incorrect.

4.2 Example 2

The second example shows a more trivial but equally frequent error in students’ programs. The proposed problem was to write a program which erases from a string any leading and trailing blank characters. Obviously the use of the appropriate repetition structure gives correct output and vice versa. Table 2 summarizes the results given when the applied code is the following:

Readln(st);
repeat delete(st,1,1); until copy(st,1,1)<>#32;
repeat delete(st,length(st),1); until copy(st,length(st),1)<>#32;

Symbols used: S =any string without any leading and trailing blank characters, B= a string of blank characters

Table 2.

5. Conclusions

Telemachus is very simple to use. Students use their normal email login names and passwords to log into the system, submit their programs electronically and access their report. Submitting programs electronically helps students save time, otherwise they would have to print their programs, which is a time consuming task. In addition in receiving a report on their submitted programs it gives them feedback on their mistakes and successes.

The help that Telemachus offers to the tutor is likewise invaluable. Besides the fact that the system saves the tutor from the laborious task of checking and marking students’ programs manually, it also gives information that a human might have completely missed: it spots errors that could be difficult to pinpoint from the visual examination of listings. Furthermore, students’ performance scores give the tutor the possibility to evaluate the success of a course. Finally, collecting reliable performance data over a long period of time the hypothesis concerning students’ errors will empirically be confirmed.

6. Acknowledgments

The "Operational Program for Education and Vocational Training" of the Second Community Support Framework, EC, financially supports this work. We acknowledge the significant help given by Theodore Folias and Maria Myari during the development process.

References

[1] Hoc J. M. Analysis of beginners' problem-solving strategies in programming, in Psychology of Computer Use, Green T.R.G., Payne S.J., van den Veer G.C. [eds], Academic Press, (1983), 143-158.

[2] Hoc J., Green T., Samurcay R., Gilmore D., Psychology of Programming, Academic Press , (1990).

[3] Hopkins K., Educational and Psychological measurement and Evaluation, Allyn & Bacon, Boston, (1998), 2-25.

[4] Jackson D., Usher M., Grading Student Programs using ASSYST, In Proceedings of SIGCE’97, ACM, 335- 339.

[5] Joy M., Luck M., Effective Electronic Marking for On-line Assessment, In Proceedings of ITiCSE’98, ACM, 134-138.

[6] Lesuisse R. Some Lessons Drawn from the History of the Binary Search Algorithm, The Computer Journal, Vol. 26, n° 2, (1983), 154-163.

[7] Mason D., Woit D., Integrating Technology into Computer Science Examinations, In Proceedings of SIGCE’98, ACM, 140-144.

[8] Mason D., Woit D., Providing Mark-up and Feedback to Students with Online Marking, In Proceedings of SIGCE’99, ACM, 3-6.

[9] Preston J., Shackelford R., Improving On-line Assessment: an Investigation of Existing Marking Methodologies, In Proceedings of ITiCSE’99 (Crakow Poland), ACM , New York, July 1999, 29-32.

[10] Soloway E., Spohrer J., Studying the Novice Programmer, Lawrence Erlbaum Associates, 1989.

[11] Tinoco L., Fox E., Barnette D, Online Evaluation in WWW-based Courseware, In Proceedings of SIGCE’97, ACM, 194-198

Maya Satratzemi Dept. of Applied Informatics University of Macedonia 156 Egnatia Str., P.O.Box 1591 54006 Thessaloniki Greece maya@uom.gr Vassilios Dagdilelis Department of Educational and Social Policy University of Macedonia dagdil@uom.gr Georgios Evangelidis Dept. of Applied Informatics University of Macedonia gevan@uom.gr

Back to contents

A Pattern Language for Architectures of Intelligent Tutors

Abstract. This paper introduces PLAIT, a specific pattern language for architectures of intelligent tutors. A pattern language is a structured collection of interrelated patterns in a specific domain. PLAIT is based on the idea of using patterns in the architectures of intelligent tutors, as well as on a number of patterns that have been discovered in the existing architectures of intelligent tutoring systems.

1. Introduction

In software engineering, patterns are attempts to describe successful solutions to common software problems [6]. Software patterns reflect common conceptual structures of these solutions, and can be applied over and over again when analyzing, designing, and producing applications in a particular context. Each pattern has a context in which it applies. When several related patterns are woven together, they form a pattern language. Pattern languages cover particular domains and disciplines, such as concurrency, distribution, organizational design, business and electronic commerce, human interface design and many more.

There are also patterns in intelligent tutoring systems (ITSs). Such patterns are, however, mostly implicitly present in ITSs. Patterns exist in architectures of ITSs, in the way learners learn from such systems, and in the way ITSs convey domain knowledge to the learners. This paper describes explicitly some patterns that exist in ITS architectures. The patterns described are all interrelated, and together represent the core of PLAIT, a Pattern Language for Architectures of Intelligent Tutors.

2. The PLAIT Pattern Language

A recently conducted research in the domain of architectures of ITSs has shown that designers of different ITS architectures have used a number of common solutions in their designs. As a result of conducting that research, seven patterns in existing architectures of ITSs have been discovered. They have got the names Inserted Layer, Top, Cascade, T-join, Cross, Multiplexer, and Store. They are basic constituents of the PLAIT pattern language, which is still evolving. Five of these seven patterns are briefly described below.

2.1. Inserted Layer

This is the basic pattern in PLAIT. Inserted Layer is used in ITS architectures whenever a new functionality is needed in the ITS, or the architecture must be adapted to a new requirement, or some translation (negotiation) between the existing layers has to be performed. In order to better decouple the existing layers, it is a good idea to insert a new layer to implement and encapsulate the new functionality, as in Figure 1a.

Figure 1. a) the Inserted Layer pattern b) the Top pattern c) the Cascade pattern d) the T-join pattern e) the Cross pattern

This pattern has been used in a dozen of existing ITS architectures. An example is introducing the believability layer in ITSs [1].

2.2. Top

The name Top reflects the idea of putting a new layer on top of other layers, Figure 1b. It is typically used when it is necessary to hide the complexity of the layered architecture from the outside world, as when the user or another system should have a specific or a well-defined "interface" for communication with a layered ITS. For example, A. Mitrovic has put the Constraint-Based Modeling (CBM) layer on top of the pedagogical module and the user interface in her SQL-Tutor, in order to overcome computational intractability of learner modeling [2].

2.3. Cascade

Here we have a cascade of new, distinct layers, hence the name Cascade, Figure 1c. This pattern is good to use when a new functionality is needed in a layered intelligent tutor, and it can be represented as a strict series of simpler modules. Also, sometimes a pipeline-style agent communication should be provided in an agent-based ITS, or a complex existing module should be broken into a sequence of simple functions (and there must be a strict order of function calls). In such cases, the solution is to represent each distinct functionality/agent/function as an individual module with strictly specified input and output, and make a cascade of such modules according to the observed strict processing sequence. An example is Knowledge Awareness (KA) agent of the Sharlok learning environment has its History observer, KA generator, KA filtering and KA monitoring connected in a cascade [3].

2.4. T-join

Figure 1d itself explains the name of this pattern. At a certain layer, a new functionality is needed but it doesn't only decouple existing layers - it has to communicate with another, "lateral" module. In fact, T-join can be viewed as an extension of Inserted Layer to accommodate communication with another module at the same level of abstraction in the hierarchy. For example, decomposing the learner's input in a Web-based ITS may require a new layer - a parser - between the user interface and the translator (see [5] for a concrete instance of this common problem).

2.5. Cross

This pattern has got its name from the fact that "vertical" and "horizontal" communication between some modules cross at a specific module (see Figure 1e). A new functionality is needed and it has to communicate with two distinct "lateral" modules. The Vincent animated pedagogical agent [4] is a good example of using Cross in an ITS architecture - its Action module is at the cross between the Mind module, the Body module, Micro Learning Environment, and the user interface (Vincent's appearance).

3. Conclusions

The knowledge of patterns can provide simple and elegant solutions to specific problems in ITS design. It can also help better organize the vocabulary of different learning and teaching styles and strategies, or the vocabulary of design elements and styles in ITS architectures. Patterns also have a great potential in generating and structuring explanations, hints, simulation, and other feedback that learners require from ITSs. They could represent the cores of solutions to analysis, design, architectural, instructional and other problems in ITSs that have been used more than once in different systems. Patterns can capture both static and dynamic structure of these solutions in a consistent and easily applied form. The PLAIT pattern language presented in this paper is the first attempt to define a specific pattern language in the domain of ITSs.

4. References

[1]      S. Abou-Jaoude and C. Frasson, Integrating a Believable Layer into Traditional ITS. In: S.P. Lajoie and M. Vivet (eds.), Artificial Intelligece in Education. ISBN: 90-5199-452-4. IOS Press, Amsterdam, 1999, pp. 315-324.

[2]      A. Mitrovic and S. Ohlsson, Evaluation of a Constraint-Based Tutor for a Database Language. International Journal of Artificial Intelligence in Education 10 (1999) 55-72.

[3]      H. Ogata and Y. Yano, Combining Knowledge Awareness and Information Filtering in an Open-ended Collaborative Learning Environment, International Journal of Artificial Intelligence in Education 11 (2000) to appear.

[4]      A. Paiva, I. Machado, and C. Martinho, Enriching Pedagogical Agents with Emotional Behavior: The Case of Vincent. Proceedings of the Workshop Animated and Personified Pedagogical Agents, Le Mans, France, July 1999, pp. 47-55.

[5]      S. Ritter, PAT Online: A Model-Tracing Tutor on the World-Wide Web. Proceedings of the Workshop Intelligent Educational Systems on the World Wide Web, Kobe Japan, August 1997, pp. 11-17.

[6]      D. Schmidt, M. Fayad, and R.E. Johnson, Software Patterns, Communications of The ACM 39 (October 1996) 37-39.

Vladan Devedzic Department of Information Systems FON - School of Business Administration University of Belgrade, POB 52, Jove Ilica 154 11000 Belgrade, Yugoslavia devedzic@galeb.etf.bg.ac.yu devedzic@fon.fon.bg.ac.yu

Back to contents

Webbus, Tailormade Transport on the World Wide Web: An Educational Webgame

Abstract

Webbus is an educational webgame based on Letterbox, a nature-game originated in Dartmoor, Cornwall. In this nature-game people are searching  for little boxes hidden in the hills, woods and mountains. Webbus is played on the World Wide Web. Its aim is, as in Letterbox, to wander but this time through the resources on the educational web. In the Webbus players are provided with  a virtual landscape in which ‘busroutes’ give directions for wandering. Important fun & learning  elements are: - solving a puzzle, which pieces are delivered once hidden ‘busstops’ on the  visited sites are found and - the opportunity to wander through the educational ‘webscape’. While playing the surfer learns to browse, search & find the webbased information. Kernel of the game is a database on a Windows NT- server that  holds information on the players, the busroutes, the hidden busstops and the status of the game for every participant. It also logs the traffic on the websites and the visitors’ appreciation of the websites that take part in the game.

1. Rationale and aim of the game

Most novice and even advanced users of the (educational ) World Wide Web find it difficult to find their way. Most teachers, very often fairly unexperienced users of the web, hesitate to apply content and applications in their daily classroom practice because of this lack of guidance and its potential risk of use. This reluctance also applies for parents of young children. In education there is also growing need for curriculum and/or subjectrelated structuring of potential educational content. Content providers on the other hand are in constant search for means to attract the attention of websurfers to give access to  their information, services and products.

Webbus offers entertaining and safe routes along a selection of rich webbased educational content. The quality of the selected sites is assured by the editiorial board of Kennisnet. By developing and maintaining the game Kennisnet draws attention to the content and services of educational suppliers and enriches the content of the educational portal Kennisnet. 

2. About Kennisnet

Kennisnet is an initiative of the Dutch Ministry of Education. The project Kennisnet includes:

• the development of a national educational intranet; Kennisnet aims to provide inter- & intranet access to over 10.000 schools and almost 3,5 million users by 2002. 
• a variety of intranet services and content-portals for children in schools & colleges, teachers, parents, managers and ICT-coordinators.

The Kennisnet organisation employs approximately 60 people occupied with the delivery of access and content. For the development of content and services Kennisnet contracts public and private content & service providers. The development of special services like the development of the Webbus game is usually outsourced to external partners such as Bit-IC.

3. Target group

The game is targeted at nine-year olds in education or at home. Together with Rowan Atkinson however we believe that there is 'a nine-year old in all of us'. The look and feel of the game and the interface are not specifically targetted. Webbus includes, and hopes to attract players of all ages and background.

4. The Game

Webbus is played on the World Wide Web. Its aim is, as in Letterbox, to wander but this time through the resources on the educational web. Underlying metaphor for Webbus is the public transport system. The look &feel of the game interface is closely related to the bustransport system in one of the major Dutch cities. In this way the game hopes to attract the passenger to particpate in an exciting ‘magical mystery tour’. In the Webbus players are provided with a virtual landscape in which ‘busroutes’ give directions for wandering. To participate the surfer creates him-/herself a ‘trafficcard’ and gets into the Web’bus’.

The busroute brings the surfer from one site to another. The route is thematic, e.g. museums, funsites, educational content, music, etc. Players are stimulated to proceed by offering them the possibility to  solve a puzzle, which pieces are delivered once hidden ‘busstops’ on the visited sites are found.  To complete the puzzle at least 7 pieces must be found. To give the player some idea where to look a hint on the hiding place of the busstop is given. Once a piece of the puzzle is found the player is asked to give a short evaluation of the website visited.

The main attraction however is the opportunity to wander safely through the educational ‘landscape’. While playing the surfer learns to browse, search & find the webbased information. The play ends when the surfer found all the hidden elements on the visited sites or decides to stop. In some cases there is an extra reward on finding all pieces of a puzzle within a set time or contest.  Gamers can skip busstops and change routes while playing. A separate window on the busroutes provides the players with data on progress and number of players (passengers!) in a busroute.

5. Particpants

In the Webbus game three kinds of participants are active:

• the Webbus-editor, a member of the editiorial office of Kennisnet, who designs the busroutes & invites the providers of websites to take part in the game. the player who gets in the webbus, wanders and plays the game
• the provider of the website who shows particiption, hides the busstops in the site and gives a hint.

6. Technology

Webbus uses javascripts and SQL. Kernel of the game is a database on a