An Approach to Intelligent Information Handling in Web-based by yaq43915


									Dicheva, D. & Aroyo, L. (2000). An approach to intelligent information handling in web-based learning
environments. Paper presented at ICAI 2000, San Antonio, Texas.

 An Approach to Intelligent Information Handling in
       Web-based Learning Environments

              Darina Dicheva                                                    Lora Aroyo
     Department of Computer Science                                 Faculty of Educational Science and
      Winston-Salem State University                                Technology, University of Twente
 601 Martin Luther King, Jr. Drive Winston-                         P.O. Box 217, 7500 AE Enschede,
        Salem, N.C. 27110, U.S.A.                                            The Netherlands

Abstract This paper discusses an integrated                   The traditional way of looking at intelligent
approach to information handling and knowledge                tutoring systems (ITS) research is undergoing an
management in web-based open-ended learning                   evolution due to the rapid development of new
environments. It supports both learners and                   information and communication technologies and
instructors in information structuring and task -
                                                              their penetration into educational settings. Recent
oriented processing and usage. AIMS, a web-based
intelligent tool for task-based information and               research indicates that the classical ITS-paradigm
performance support is implemented to exemplify the           no longer dominates the field of AI and
theoretical assumptions of this approach. AIMS                Education. ITS are viewed not so much as stand-
focuses on three important aspects of the information         alone but as embedded in larger educational
handling process - information structuring,                   environments, which support the student in the
information visualization, and user centered                  learning process. Among the most significant
approach. We employ concept maps (CM) to build a              current trends in ITS area is the shift of emphasis
subject domain ontology and use it as a basis for
                                                              towards open domains and conceptual
defining course structures. The domain CM is also
used for attractive non-linear visualization and
                                                              understanding [1]. The environments are
conceptual graphical navigation of the subject                expected to support the learner in performing
domain and the search results, thus allowing for more         open learning tasks with well-defined results.
efficient information searches. In order to provide                This paper discusses an integrated approach
appropriate adaptation to the individual information          to information handling and knowledge
needs and preferences of the learners AIMS models             management in web-based learning environments,
their behavior.                                               in which learners perform open learning tasks. It
                                                              supports both learners and instructors in
Keywords: intelligent information handling,                   information     structuring    and     task-oriented
concept maps, subject domain ontology, web-                   processing and usage. AIMS, a web-based
based training systems, intelligent agents                    intelligent tool for task-based information and
                                                              performance support is implemented to exemplify
1 Introduction                                                the theoretical assumptions of this approach.
                                                              AIMS focuses on three important aspects of the
                                                              information handling process - information
structuring, information visualiza-tion, and user
centered approach. In the paper we present             3 The AIMS System
AIMS objectives and information handling
approach,      emphasizing    the      agent-based     We have developed a knowledge-based
architecture of the system and the way it realizes     informational support system called AIMS
a knowledge-based approach for informational           (Agent-based InforMational Support) for use
support within web-based learning environments.        within the context of distance education. It is
                                                       aimed not only at enhancing learner's conceptual
2 Information Support in On-line                       knowledge in a specific subject area but also at
Course Environments                                    providing knowledge verification tools for the
                                                       teacher. The system can be used to aid learning
The availability of intelligent information handling   and teaching in different subject areas and to
support in web-based course environments is            provide user-oriented support in searching course-
essential for both students and instructors.           related information, concept teaching and learning,
Students need to be helped in searching,               and conceptual and task-oriented domain
accessing, and finding information rela ted to their   structuring. AIMS has an agent-based system
course work, as well as in comprehending and           architecture, which accommodates a team of
memorizing the retrieved information. In this          agents mediating the process of providing
relation it is important that students get a clear     information to the individual user.
and structured overview on the course content               AIMS is a knowledge-based system, which
and course materials. Instructors need to be           helps the users to come to solutions of their
supported in their efforts to build a course           problems by supporting information structuring
structure and find, organize, and structure on-line    and task-oriented search, but it doesn’t contain
course materials.                                      the solutions. It contains explicit knowledge
     We call an informational support system           representation of the domain (subject domain
(ISS) such a tool, which provides the users (in this   concepts and relations among them) and a user
case learners and instructors) with immediate on-      model. The user model represents system's belief
line access to a broad range of structured             about the current user's knowledge. The system
information and with domain-related help in the        uses it whe n supporting the user to find, collect,
working context. Such a system enables the user        and organize information necessary to perform a
to identify the information necessary for              course assignment.
performing a particular course task. ISS can be             AIMS comes as an example of how
used in educational context standalone (as an          educational systems try to meet the challenges of
extension of a traditional or on-line distance         multimedia, hypertext and internet technologies in
course) or integrated in a larger electronic           general and proposes a new model at the cross-
learning environment that allows the users to          road of the electronic performance support and
perform open learning tasks in a specific subject      course support systems. Conceptually, AIMS
domain. The goal is to provide an appropriate          combines advantages of some existing information
framework and architecture for informational and       management and educational approaches. It
conceptual support, which will improve student's       attempts to set up an adequate balance between
performance in a course thus making the learning       system and user locus of control, content learning
process more efficient. Such a system is a             and conceptual learning, graphical and textual
cognitive tool [2] helping the learners not only to    visualization, ‘dry’ course material presentation
complete their course work more efficiently and        and creativity stimulating approaches, quick and
accurately, but also to build new knowledge and        efficient learning, etc. The information handling
comprehend better the subject domain in hand.
support method behind AIMS reflects all those            organization. It allows instructors to efficiently
challenges.                                              reuse their course materials, as well as be aware
                                                         of the way the students will process them.
3.1 AIMS Objectives
                                                         4 AIMS Informational Support
The main goal behind AIMS is to improve the              Approach
usability and maintenance of information in web-
based educational environments by presenting an
integrated working environment for both                  On the basis of extensive review and analysis of
instructors and students, which facilitates the          web-based education and its needs for
process of preparing and taking on-line courses in       information support of both learners and
an intelligent and efficient way. We envisage the        instructors we developed an approach for
management of information as concerning two              information handling implying some ideas of
mutually complementing processes: these of               artificial intelligence, learning theories, and
retrieving the right information and visualizing it in   human-computer factors. The approach tackles
the right way. Thus the focus of the work is on          the issues of information analysis, organization,
information search, retrieval, and effective             retrieval, and presentation (visual and
presentation to the user. An important educational       instructional). It is domain independent and
constraint is that we consider supporting learners       involves building domain ontology and specifying a
to deal efficiently with information only within a       course structure. It enables the instructors to
specific subject domain and course in respect to         analyze, classify, organize, and present
certain educational goals (tasks).                       information items optimally to the learners while
     AIMS targets to support the knowledge               focusing on a specific course targets. The result is
crystallization task, which is characterized by the      a set of precisely defined and structured
use of large amounts of heterogeneous                    information units that are easy updateable,
information, ill-structured problem solving, and a       reusable, combinable, and extendible for different
well-defined task. This point of view is related to      course objectives, environments, and situations.
externalization rather than visualization, stressing     The approach provides the learners with an easy
on the cognitive role of interactive visual              access to and an attractive visual presentation of
representations [3]. The process of knowledge            information related directly to the course tasks as
crystallization involves getting insights about data     well as with a conceptual view of the whole
relevant to the task in hand, so that the user will      subject domain.
be able to fulfil it and achieve better results.              The significant aspects of the proposed
     AIMS is a web-based information handling            approach include:
support tool for both learners and instructors. It is    • Building domain ontology and using it as a
aimed at representing and supporting the relations            basis for defining course structures and for
between a course, its subject domain, and relevant            implementing more efficient information
documents. It supports the synergy between                    searches.
problem solving and knowledge building. The              • Powerful         visualization  and     graphical
learners can discover information and build                   navigation of the domain and information
knowledge thus preparing the problem solving                  search results.
process for performing their course tasks. AIMS          • User modeling.
is aimed also at supporting the instructors by           • Agent-based architecture.
providing them with an integrated authoring
environment for subject domain representation,           4.1 Information Structuring            Based     on
course structure building, and learning materials        Domain Ontology
                                                          library documents are related to domain terms and
One of the milestones of the proposed approach is         through them - to the course tasks and topics.
building domain ontology and using it for
information structuring and classification. On a          4.2 Information Visualization
conceptual level we define three m models for
representing the information content: domain              Information visualization refers to methods and
model, course model, and library model. The               techniques       for    user-oriented    graphical
domain model defines the subject domain                   visualization and organization of information
ontology and is represented as a concept map [4]          objects. In AIMS information visualization targets
of domain concepts and links between them. The            the presentation of the subject domain structure
link types are based on the generic selection of          and the overlaying structures of the course and
link types defined in [5]. There could be a number        user models [6, 7]. As part of the supporting
of instances of the domain model, each describing         information search functionality AIMS also
a different subject domain within AIMS. The               proposes combined visualization of the search
course model defines the structure of a course            results. We use concept maps as a basic
and includes course topics, course tasks and              mechanism for information structuring and visual
course evaluation items. Since the main goal of           presentation of the subject domain, since CMs
AIMS is to provide a task-based information               furnish simple and intuitive visual form of
support, a series of tasks are associated with each       knowledge representation. The same technique is
course. Course tasks are pre-defined and                  used for presenting the course components, such
correspond to course assignments the student is           as tasks and subtasks and the user model.
supposed to perform. The main idea is that the            Information search results are also partly
system uses a task to constrain the information           presented with this graphical approach. The result
provided to the learner in support of his or her          documents are still organized in the traditional
efforts to complete this task. The system keeps           way of textual list-based presentation, but are
track of the status of all course tasks - which           related to a result concept map of domain terms
have been successfully completed by the user,             used in those documents. Some meta-data is
which are not attempted or have been started but          provided in order to support a better and quicker
not completed yet. The tasks are represented in           overall search result comprehension. Meta-data in
terms of domain concepts and contain some                 the form of short descriptive information and
additional information, such as task description,         structured tags is also used in order to describe
task status, etc.        Course evaluation items          the context of the domain terms and links.
correspond to different forms of student                       The combination of CM information
evaluation over the timeline of the course, such as       structuring and visualization contributes to the
instructor's tests, student's self-evaluation of his or   problem solving support for more efficient learner
her knowledge of domain terms, etc. They are              performance within the context of a course.
used on a later stage for a run-time evaluation of        AIMS gives the learners freedom to create their
the current learner's knowledge on a specific             own learning and problem solving paths, while
course topic and for consecutive adaptation of the        browsing within the subject domain. This is
system's behavior and information presentation.           realized by providing two parallel views on the
The library model provides means for                      knowledge - conceptual perspective (subject
maintaining a whole collection of information             domain knowledge) and teaching perspective
related to different courses and domains. Its main        (task/course knowledge). The visualization and
variables are course materials and domain related         domain navigation facilities help students to get
documents together with their characteristics. The        orientated within the subject domain and to build
                                                          up their own understanding and conceptual
associations. It supports their visual thinking and     term, etc. It is an overlay of the domain, course,
imagination.                                            and library models thus it is represented in terms
                                                        of domain concepts, course tasks, and library
4.3 Task-based Search Strategy                          documents. AIMS uses UM to provide user-
                                                        oriented support for information searching in the
     The proposed informational support approach        context of an educational task, as well as for
involves a task-based search strategy, which links      refining and adapting the search results to user’s
the course tasks to the search algorithm and uses       interests and course tasks. It is an open model,
the domain ontology definition as a classification      that is, built jointly by the user and the system.
scheme. The CM-based classification facilitates         The user can enter information about his or her
dynamic adaptation of the system to changes in          knowledge of or interest to a specific term on
the domain. It is directly linked to the hierarchical   system's request or on his own initiative, while
course structure, and in addition, correlates           browsing the conceptual presentation of the
efficiently with the user model. The use of             subject domain or in the process of searching for
conceptual representation of the domain                 documents.
knowledge and information structure provides                 The main system’s source of information
methods for automatic manipulation, which               about the user model is the user behavior. The
contribute to the overall search efficiency. In         system monitors user performance and collects
AIMS we propose the DoCS (Document                      information from sources like search expression
Classification and Search) model, which is              entries, course task dialogues, search result scan
keyword-based and makes use of the relations            and usage and browsing the domain concept map.
between documents and domain terms,                     On the basis of this information the system makes
documents and course tasks, domain terms and            its conclusions about the user's knowledge of and
document keywords, while considering synonyms           interest to terms and documents. Maintaining the
and other word forms of the domain terms [7,8].         user model involves incorporating new facts,
     The task-based search strategy provides for        checking that they are consistent with previously
course-oriented results and prevents overloading        learned or inferred facts and resolving any
the learner with information non-related to the         discrepancies and contradictions.
present instructional tasks. It applies automatic
reformulation of the search query to fit the            4.5 Agent-based Architecture
limitations set by the current course task.
Documents are retrieved and ranked in the               We propose agent-based system architecture,
context of their task relevancyAIMS provides            which provides a team of agents mediating the
visual and textual ways for search query refining       process of information provision to the individual
by the user, based on the already produced search       user. The collective agent’s behavior results in
results and subject domain representation.              adaptive system's behavior. Agents have their
                                                        tasks and strategies. Tasks define agents’
4.4 User Modeling                                       responsibilities and communication means and
                                                        entities within the system, and strategies define
User modeling provides grounds for system               methods and rules for performing actions related
adaptiveness. In AIMS, the user model (UM)              to the overall informational support task. Among
refers to the learner and is related to his or her      agents' tasks are maintaining domain, course,
activities in the on-line course environment, such      library, and user models, performing search
as performing a course task, browsing the domain,       activities and presenting the final results to the
searching for a term, visiting documents related to     user.     Agents'    co-operation    abilities  to
a term, requesting for terms related to a specified     communicate with each other and change their
behavior depending on the user’s action, to                    AIMS is developed as a Java client-server
exchange messages and to act even when there is           application, where the applet part acts as a client
no clear user request for an action, contribute to        and the servlet is resident on the server. Java
the overall system efficiency. The agents perform         appears to be a promising technology for such
their tasks by working in parallel as separate but        implementations since it provides platform
coordinated processes.                                    independence and efficient integration with on-line
    An agent can initialize and execute                   course environments. The server part administers
commands given by other agents or invoked by a            information maintenance, i.e., updates of
change in the environment. Currently, the                 documents, courses, and domains and controls the
incoming information and outgoing information             user profiling functions. The servlet ensures the
(through the commands CommandRequest and                  access to the server via the Internet. The servlet
CommandResponse) are Java objects but they                accepts a query sent by the applet, decodes it and
can be easily transformed in an agent-based               sends it to server, then waits for an answer and
message language like KQML or text queries.               sends it back to the applet on the client side. All
    Each agent is responsible for its own                 the persistency information is kept on the server.
information consistency and is accountable for a               There are two active applets on the client
number of commands.                                       side: one serving the learners' needs (Viewer) and
    The Agent environment consists of two                 another facilitating the instructors and experts
objects AdminServer and AgentServer.                      (Editor). They are responsible for the search
The AgentServer object is composed of a                   query processing, local information visualization
team of agents that collaborate and communic ate          and library, course, and domain content storage.
among each other while performing user’s or               Both include:
other objects' requests. AgentServer is                        • a common module responsible for
responsible for the general administration of                      transferring the queries to the server
agents, including creation, deletion, security, etc. It            (through the servlet),
receives a configuration file, defining the types of           • PseudoAgent               shells       allowing
the agents and how to initialize them. The                         visualization of the local client copy,
AgentServer also knows how to instantiate                      • interface.
them when a command/query for a task arrives.                  The Viewer sends a user log-on command
Along with each query/command goes                        and if it succeeds sends commands to get all the
identification information about the object that          information needed from the server, then stores
sends it (the name and password of the user).             and analyzes it locally, and visualizes it.
The end user does not see the agents; he or she                The Editor behaves similarly in relation to the
only sends information queries through the                storage, analysis, and visualization of the server
AgentServer.           AdminServer            checks      information. When a change is required, it sends
whether the announced user exists and whether             an update command to the responsible agent on
he or she is authorized to give such a command.           the server and then visualizes the newly updated
                                                          information. This allows several users to work in
5 AIMS Implementation                                     the same domain performing a synchronized
                                                          information update.
As we already said, AIMS is implemented as an
agent-based architecture. This architecture               6 Conclusions
supports system's intelligent behavior, which
comes as a result of agents' collaboration, learning          The AIMS system is created to support the
and information sharing.                                  students from the Faculty of Educational Science
and Technology, University of Twente,                 [4] Jonassen, D. & Marra, R. (1994) Concept
participating in the course of ‘Courseware                Mapping and Other Formalisms as Mindtools
engineering’. The intention is to integrate AIMS in       for Representing Knowledge. In Educational
the web-based course environment already                  Psychology Interactive, Readings in
created for this course. As AIMS is designed to           Educational Psychology, William G. Huitt
be a general course support tool, there are               (Ed.).
possibilities foreseen to facilitate importing of
                                                      [5] Lambiotte, J. G., Dansereau, D. F., Cross, D.
different subject domains and to create different         R., & Reynolds, S. B. (1989) Multirelational
courses in the same subject domain.
                                                          semantic maps. Educational Psychology
    Several pilot experiments have been                   Review, 1(4), 331-367.
performed for evaluating AIMS browser and
editor GUI functionality, user-friendliness, and      [6] Aroyo, L. & Dicheva, D. (2000) Conceptual
learning effectiveness. Students from the                 Visualisation in a Task-Based Information
‘Graphical User Interface Development’ course             Support     System.   ED-MEDIA      2000,
in Eindhoven Technical University, the Antwerp            Montreal, Canada (accepted).
University, and the 'Courseware Enginnering'          [7] Aroyo, L. & Dicheva, D. (1999) Information
course at the Faculty of Educational Science and          Retrieval and Visualisation within the Context
Technology, University of Twente, the                     of an Agent-based Information Management
Netherlands, took part in the experiments. The            System. In Proceeding of ED-MEDIA’99,
purpose of these formative evaluations was to             Seattle, Washington, 195-200.
validate the proposed AIMS approach and to
identify any problematic aspects in order to          [8] Aroyo, L., Diana, I. & Dicheva, D. (1998)
improve it.                                               Agents to Make Your Information
                                                          Meaningful and Visible: An Agent-based
Acknowledgements                                          Visual Information Management System. In
                                                          Proceedings of WebNet’98 Conference,
We would like to thank our student Ivan Velev for         Orlando, USA.
his implementation efforts in realizing the AIMS
prototype .


[1] Andriessen J., Sandberg J. (1999) Where is
    Education Heading and How About AI?,
    International    Journal     of    Artificial
    Intelligence in Education, 10, 130-150.
[2] Van Joolingen, W. (1999) Cognitive tools for
    discovery learning, International Journal of
    Artificial Intelligence in Education, 10, 385-
[3] Tweede, L. (1997) Characterising Interactive
    Externalisation. Proceedings of CHI’97,
    ACM Conference on Human Factors in
    Computing Systems , Atlanta, pp. 375-382.

To top