Proceedings of the workshop "Intelligent Educational Systems on the World Wide Web",
8th World Conference of the AIED Society, Kobe, Japan, 18-22 August 1997

Intelligence in Educational Technologies for Distance Case

Alexei Dovgiallo
International Research and Training Center
of Information Technologies and Systems,
prosp. Glushkova, 40, Kiev - 22, 252650 Ukraine
E-mail: alex@tel.dlab.kiev.ua
Phone: +380 (44) 266 4315 Fax: +380 (44) 266 1570 URL:http://www.tel.dlab.kiev.ua


Abstract: Distance education provides a wide spectrum of technological decisions to support high quality teaching. It becomes a reality for both learners and adult learners with rapidly developing telecommunication technologies. The primary a tention to "intelligent" solutions in distant case is evoked by both availability of traditional forms of education and restricted resources and technical difficulties, such as the quality of communication channels and bandwidth. The aim of this paper is to present an activity of Computer-Based Didactic Lab and discuss the future plans of incorporating "intelligence" into Web-based education.

Feedback and adaptation on the Web

Adaptation to learner's abilities, preferences and current knowledge state in ITS is based on learner modeling techniques (Wenger, 1987), which are aimed at learner model construction and update. Learner model is used in conjunction with the knowledge about the domain of instructions and instructional strategies to modify the order of presentation of material, selection of hints and corrections, to provide better choice of exercises and tasks for paticular learner.

However, these techniques are difficult to implement within the WWW technology, since the system never knows the path by which a learner has arrived to current hypermedia document. In addition to this, the Web server should work with all of the various Web browsers available, so the interaction with the Web has to be supported by all of those standard browsers.

One of the possible solutions incorporated in our first Web-based course is a formal registration of learner, which helps providing constraints on material and adapting the interaction according to learner's performance. The underlying mechanisms are the Common-Gateway Interface (CGI) and interactive HTML forms. Such a form transmitted from the HTTP daemon to the Web browser should be sent back to the server when completed. Each form is associated with a particular CGI program, which is activate by the Web server when the form is received. The CGI has access to the data in the form and can manipulate them freely. In particular, the CGI can produce an output to the HTTP daemon text which represents another page of information to be transmitted to the Web browser for review by the learner.

For the purpose of Web-based instruction, the above-mentioned forms are the mechanism to return a particular response from the learner. The CGI provides a program mechanism to record the input and to relate it to previous inputs, as well as to select a particular set of choices sent back to the learner for subsequent hypermedia browsing.

The CGI forms combination allows for offering a variety of assessment methods such as true/false questions (radiobuttons) multiple-choice questions, short-answer and simple essay questions. The test is a set of above forms that is related to certain topic or subtopic and should be processed simultaneously. The tests are inserted into the course material on the Web in order to supply the main presentation chunks with evaluation facilities. Each learner's testing attempt has been evaluated by CGI program, which processes the learner's inputs. The program generates an HTML document containing current scoring results and on-line help that becomes available to a learner's browser. The CGI program also generates new version of the test for the next learner's attempt. A special database has been implemented to record all the learners inputs. So, it is possible to control the presentation of material based on the learning history.

CIT Course : from interactivity to intelligence

According to Copernicus project N1445 Kiev Copernicus team is developing the first in Ukraine distance on-line course on Communication and Information Technologies (CIT).

Course is aimed at:

The course has a modular structure, including Introductory module (covering communications in general outline), Technological module (introducing basic concepts of the Internet literacy) and Applied modules (dealing with using CIT in different areas: education, engineering, business). Educational material of the course includes information about all Internet services and technological aspects of telecommunication technologies use. After finishing the course a learner will gain experience in teacher-mediated discussions, one-to-one and one-to-many communications by e-mail, group collaborations and will be able to work with the Internet resources.

The course modules are divided into lessons which include tests and exercises as an obligatory activity to proceed to the next lesson. Currently a very simple learner model is used in CIT course which allows to attribute a learner to one of the types: novice or expert according to his/her test scores. For each type specific exercises are designed which corresponds learner's knowledge level and thus provide better motivation. In the future more elaborated course structure will be implemented enabling an intelligent adaptation based on a kind of indexing technology described in (Brusilovsky, 1996).

The first version of CIT course in Russian and Ukrainian was suggested for Ukrainian and CIS-countries users via E-mail. Subscription results showed a vivid interest to the Internet and telecommunications within various professional groups. This experiment which was supported by Eurasia Foundation allowed to update and tune the course material based on a feedback from participants, to include an additional material about local information sources and services according to requirements and to obtain a more detailed picture of potential Internet users, their background, interests and preferences.

Registration forms, tests results and questionnaires from about 300 participants form a unique database containing information on basic education, current professional tasks, learning objectives, difficulties and gaps in understanding the learning material. Preliminary analysis of this information showed that user stereotype approach (Rich, 1989) could be a nice solution to tailor course information for particular user. As the basic "dimensions" of users classification basic computer literacy, professional field and learning goals are suggested. Another substantial outcome for course adaptivity can be gained by using techniques of uncertain active and passive learner evaluation (Sinitsa, 1996). Suggested for knowledge level assessment they can be extended for other discrete learner parameters evaluation and used to tune and update learner model/stereotype during the course learning.

Future plans

Until recent time Internet was practically restricted to relatively narrow professional groups working in technical applications and natural sciences. These groups have a sufficient background in telematics, computing, and, naturally, in major international "technical" languages, such as English, practised both in the Internet applications and in the scientific communications.

Nowadays, Internet keeps extending to more and more countries and is becoming available to new groups of users with various non-technical professional and/or cultural needs. These new users may not have any specific skills in informatics or networking. At the same time, they may have substantial problems with the international languages (i.e., English) involved in the on-line Internet applications and software descriptions.

Above-mentioned CIT-course aimed to solve the first problem, i.e. "Internet literacy". As the next step we are planning to find, evaluate and suggest for potential users a set of linguistic tools, such as on-line dictionaries, glossaries, translators and multilingual browsers which could be used at their working place facilitating information extraction and communication with colleagues around the world. Some of these tools can be suggested for incorporation into everyday practice which requires additional intelligent combination of specific software. Learner model for this purpose can be extended to include additional parameters for tuning these tools to leanrner's needs.

In general, Didactic Lab plans are connected rather to searching prospective combinantion of existing powerful tools and media to enhance learning than to design architecture or shells for intelligent Web-based education. Hopefully, an attempts of leading companies to install a standard on user's profile data for Web will form a basis to share information about learner across various platforms and programs and thus promote learning modeling.

References

Dovgiallo, A., Gritsenko, V., Petrushin V. (1997) Intelligence In Educational Computing: The Impact For Ukrainian Informatization Programme. Media and Telematic Technologies for Education in East European Countries. P.A.M. Kommers, A. Dovgiallo, V.Petrushin, P. Brusilovsky (eds.). Univ. of Twente, 1997. p. 267-274.

Wenger, E. (1987) Artificial Intelligence and Tutoring Systems. Morgan Caufman Publ.

Brusilovsky, P., Schwarz, E., Weber, G. (1996) ELM-ART: An intelligent tutoring system on World Wide Web. Proc. Third International Conference on Intelligent Tutoring Systems, ITS-96 (Lecture Notes in Computer Science, v.1086) C. Frasson, G. Gauthier, & A. Lesgold (eds.) Springer-Verlag, Berlin, p. 261-269.

Rich, E. (1989) Stereotypes and user modeling. User models in dialog systems,, ed. Kobsa A. and Wahlster W., Springer-Verlag, Berlin, p.35-51.

Sinitsa, K. (1996) Numerical models for adaptive knowledge assessment. Artificial Intelligence and Education . Proc. Int. Workshop. - Kazan, 1996. - vol.2.P.69-72.