<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>Facilitation of Online Student Group Projects with a Support Agent</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <string-name>Janice Whatley</string-name>
          <email>J.E.Whatley@salford.ac.uk</email>
          <xref ref-type="aff" rid="aff1">1</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Martin Beer</string-name>
          <email>M.D.Beer@csc.liv.ac.uk</email>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Computing and Information</string-name>
          <xref ref-type="aff" rid="aff2">2</xref>
        </contrib>
        <aff id="aff0">
          <label>0</label>
          <institution>Department of Computer</institution>
          ,
          <addr-line>Science</addr-line>
          ,
          <institution>University of Liverpool</institution>
          ,
          <addr-line>Liverpool, L69 7ZF, 44 0151 794 3672</addr-line>
        </aff>
        <aff id="aff1">
          <label>1</label>
          <institution>Information Systems Institute, University of Salford</institution>
          ,
          <addr-line>Manchester, M5 4WT, 44 0161 295 5175</addr-line>
        </aff>
        <aff id="aff2">
          <label>2</label>
          <institution>Technology</institution>
          ,
          <addr-line>Liverpool John Moores</addr-line>
          ,
          <institution>University</institution>
          ,
          <addr-line>44 0151 2312080</addr-line>
        </aff>
      </contrib-group>
      <pub-date>
        <year>2000</year>
      </pub-date>
      <fpage>1</fpage>
      <lpage>2</lpage>
      <abstract>
        <p>As more courses are provided online, consideration must be given to providing for group working, as an integral part of the learning experience. Present online provision, based on CAL, is aimed at individual study in the main, and CSCW products are aimed at business users. We propose student support agents to help with the maintenance aims of group working, providing support in developing group relations. Our initial prototype is built in Prolog, and is based on a whiteboard architecture. In this paper we describe the design work, based on analysis of traditional face to face group project work, through to developing the prototype for testing.</p>
      </abstract>
      <kwd-group>
        <kwd>eol&gt;Agents</kwd>
        <kwd>Groupwork</kwd>
        <kwd>Online learning</kwd>
      </kwd-group>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>-</title>
      <p>INTRODUCTION
There is a drive towards making higher education accessible to a
greater number of participants than ever before. It is agreed in
some circles that modern technology should be able to provide
applications to enable students to work at their own pace or in
their own time, notably Computer Aided Learning (CAL)
packages. However, CAL packages are essentially aimed at
independent learners, and online students should be given the
opportunity to develop group working skills as well. Various
forms of Computer Mediated Communication (CMC) are being
used for online courses, to allow for discussion between students
and tutors, but their use in supporting some aspects of group
work is limited. We propose a software agent that can support
the maintenance aims of group project working, called a
Guardian Agent.</p>
      <p>
        Permission to make digital or hard copies of all or part of this work
for personal or classroom use is granted without fee provided that
copies are not made or distributed for profit or commercial
advantage and that copies bear this notice and the full citation on the
first page. To copy otherwise, or republish, to post on servers or to
redistribute to lists, requires prior specific permission and/or a fee.
2. THE ROLE OF GROUP PROJECTS IN
LEARNING
Traditional undergraduate and postgraduate courses incorporate
a mixture of individual and group learning methods, and it is
acknowledged that group learning offers several advantages to
learners [
        <xref ref-type="bibr" rid="ref4">4</xref>
        ], [
        <xref ref-type="bibr" rid="ref7">7</xref>
        ]. Students learn from each other by exchanging
ideas and resources, negotiating strategies and tasks, are able to
comment on each others’ work and contribute towards merging
and redrafting of the final written work. A key component of
undergraduate and postgraduate courses is group project work.
This is because in undertaking group projects students are given
the opportunity to develop and practice the task and the social
aspects of team working.
      </p>
      <p>
        The learning cycle by Kolb [
        <xref ref-type="bibr" rid="ref8">8</xref>
        ] gives a good summary of the
stages in student learning, and this gives a starting point for
thinking about how we approach course material design and
presentation. The main feature is that students do not learn by
simply being told facts. They need to be able to practice using
the facts, and reflect on the way they are used in order to form
connections in the brain, which can be regarded as knowledge.
Further experimentation, experience and reflection leads to
intelligence or expertise in a subject. Educated students should
be able to formulate a set of facts into information within their
own minds. If the students are able to talk about this information,
then they can be said to have knowledge of the subject, and
intelligence shows in their ability to apply the knowledge in a
variety of situations. Group projects give students an opportunity
to discuss their understanding of the subject with their peers, as
they apply the theory to practice.
      </p>
      <p>
        Experiential learning and problem-based learning have been used
to encourage applying facts and skills learned to a real life
situation, often using case studies. These give the opportunity for
students to practice and reflect on the effect in a given situation,
from which they are able to formulate patterns to determine
which situations lend themselves to particular approaches. Some
university courses have been run entirely using problem-based
learning [
        <xref ref-type="bibr" rid="ref2">2</xref>
        ]. The best results have, however, probably been
obtained where problem-based learning forms only a part of the
course content, and it is used to build upon learned facts and
skills.
3. COMPUTER SUPPORT FOR GROUP
WORKING ONLINE
Successful group working requires that the maintenance roles as
well as the task roles of the group are given attention [
        <xref ref-type="bibr" rid="ref5">5</xref>
        ]. Group
dynamics play an important role in determining how successful
the outcome of the project is, i.e. the ways in which the members
interact with each other and how this changes with time as the
group develops [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ], [
        <xref ref-type="bibr" rid="ref4">4</xref>
        ], [
        <xref ref-type="bibr" rid="ref7">7</xref>
        ]. The most successful groups are
composed of members with different personalities, so allowing
students to choose their group members, as they often do in face
to face projects, may not lead to the most successful outcomes,
as the tendency is to choose similar minded friends. At work it is
unlikely that a member will be able to choose their colleagues, so
it is important that students can work with people they do not
particularly like, and learn to be tolerant of others’ attitudes in
concentrating on the given task.
      </p>
      <p>Learning by experimentation and making mistakes and by doing,
which require immediate feedback and guidance by a tutor, are
not easily accomplished within online courses. But advances in
technology ought to enable us to design more adaptive CAL
material, which exhibit some intelligence, called intelligent
tutoring systems (ITS), and to allow for more rapid
communication of feedback, so that future online courses are an
improvement on correspondence courses.</p>
      <p>Online courses can be solitary affairs, unless Summer schools,
Saturday workshops etc. are included in the programme.
However, these reduce the self-paced element of online courses,
as students must prepare for a group tutorial on a particular date.
Management of such courses to incorporate a group project will
require considerable negotiation of availability and readiness, in
order to gather together an appropriate number of students
capable of working together for a given period of time. A
number of problems arise when adapting a conventional
approach to group projects for students working online:
•
•
•
•
•
students are working at their own pace and in their
own time, so a timetable must be imposed either by the
group itself or by the course leader;
organising conventional meetings is not possible, but a
substitute is necessary, probably aided by technology;
sharing information must be enabled by technology,
students must be able to express their opinions online,
which may require different skills;
assessment is probably not possible on an individual
basis, but a group mark may not be acceptable if
students recognise that members of a group are not
pulling their weight;
tutors may experience difficulties monitoring the
progress of groups of students.</p>
      <p>
        An important area of CSCW research is into improving the
support provided for online work. Distance learning is a growing
market and as higher education institutions strive to provide
online learning experiences, consideration should be given to
providing an interface which gives a level of support comparable
to that provided on campus based courses. Social inclusion is an
important aim of online provision, and these learners often have
the face to face groups tended to find the progress
meetings helpful and found it easiest to get started,
the online team tended not to trust each other as much
as the others, experienced less support from each other,
they also felt they learned less by working in a group
and learned less by discussing and explaining ideas to
each other.
3.2. Comparisons between online and face to
face working
Computer mediated communication (CMC) tools, such as
conferencing, email, discussion forums support the
communication needs for the task roles of group projects,
examples include studies of co-operative learning in a virtual
university [
        <xref ref-type="bibr" rid="ref3">3</xref>
        ] and groupwork in mathematics teaching [
        <xref ref-type="bibr" rid="ref6">6</xref>
        ].
Student support using commercial groupware products enables
communication between group members and instructors. BSCW
has been used as support for group projects and was found useful
for information sharing, offering greater flexibility in students’
face to face communication, but offered limited support for the
maintenance roles of groupwork [
        <xref ref-type="bibr" rid="ref11">11</xref>
        ].
individual and specific additional requirements, not readily
provided by Groupware products, which are specifically aimed at
business users. New technology as a means of communication
has actually changed the ways in which people communicate and
collaborate, e.g. email has meant that traditional working
boundaries have been removed as communication between lower
and higher levels of staff is now acceptable.
3.1. Experiences of working in a group online
One of the authors took part in an exercise to work in a group to
produce a report on the assessment of Key Skills online. The
group worked online over a period of two months, in which time
they used email predominantly to communicate, and
occasionally used the virtual classroom provided in Blackboard
for synchronous discussion. At the end of the exercise reflection
on the group processes was recorded, which can be summarised
as follows:
we found ourselves able to communicate, but not
necessarily understand the meaning of what was said;
it was difficult to get a concensus;
at times the communication can seem aggressive;
it is even more difficult online to agree meeting times;
we found it difficult to achieve a fair distribution of
work;
the group discussion process allowed each one to
identify one’s own perspective
we were using different platforms and some members
had difficulty opening others’ files.
•
•
•
•
•
•
•
•
•
At the same time two other groups were working on similar
reports, but mainly working face to face, with CMC used as an
adjunct. When individual opinions were sought, using a
questionnaire, after the exercise it was apparent that the online
group experienced more difficulties than the other groups,
including:
      </p>
    </sec>
    <sec id="sec-2">
      <title>Face-to-face agreement to these is in a sense</title>
      <p>binding.</p>
      <p>Again face-to-face agreement is binding.
Problems will become apparent at each
meeting, and action agreed.</p>
      <p>Some discussion essential, which may take
time. Sub-groupings are spontaneous. Final
agreement of allocation is binding.</p>
      <p>Individuals will bring hard copies of their
work to explain to each other. An individual
is responsible for collating the work.</p>
    </sec>
    <sec id="sec-3">
      <title>Online Email, video-conference, bulletin board, groupware, guardian agent.</title>
    </sec>
    <sec id="sec-4">
      <title>Non-contribution to bulletin board or lack of response to email may be recognised by group members, or by guardian agent.</title>
      <p>Email, video-conference, bulletin
board, groupware, guardian agent.
Email, video-conference, bulletin
board, groupware, guardian agent.</p>
    </sec>
    <sec id="sec-5">
      <title>Email, video-conference, bulletin board, groupware, guardian agent.</title>
    </sec>
    <sec id="sec-6">
      <title>Email, video-conference, bulletin</title>
      <p>board, groupware, written post or fax,
guardian agent.</p>
      <p>In order to see how new technology can be applied to group
projects, it is necessary to analyse the stages of a group project,
to determine the particular problems encountered at the different
stages, and to determine whether new ways of working may be
possible, rather than simply adapting a current traditional project,
and where support in the form of a Guardian Agent might be
appropriate.</p>
      <p>This table identifies some of the tools that could be applied at the
different stages of the group project. It can be seen that the
meeting plays a prominent part in getting the group established,
formulating ground rules and identifying individual talents. At
face to face meetings non-contribution by members is implied by
non-attendance, also disagreement and other negative feelings
are often recognised by visual cues. Agreement between
members at a face to face meeting might be more binding than
agreement arrived at online. Hence an online project will require
a considerable amount of support, if it is to be kept on track.
Typical problems with face-to-face meetings include: making
sure that discussion is relevant, keeping to a reasonable meeting
time, ensuring all members are present and keeping to the
agenda. Email and bulletin board messages tend to be shorter and
succinct, though it is possible to be sidetracked into discussion
that is not relevant, and a member that suggests online that this is
happening would be very brave. Obviously time is a critical
factor in face-to-face meetings, but not so for online discussions
which are asynchronous, though long waits for responses from
group members may be unwelcome. Online discussions should
be controlled by an agenda with time limits set, so that all group
members know what stage in the discussion is current.
When problems arise it is often too late to do anything about
them, but if a problem can be anticipated, the project can often
get back onto schedule without wasting too much time. At
faceto-face meetings individual worries may be aired that may or
may not be significant. Problems, such as lack of skills, can often
be resolved by explanation from other group members. When the
time plan appears to be getting behind schedule it is often
possible for group members to agree to a change of plan, by
reallocating tasks. This means that a traditional time plan is
likely to be more fluid than an online one.</p>
      <p>Drawbacks with online technologies include:
• the time taken for a discussion and to reach collective
agreement;
• the time it might take to recognise potential problems;
• getting all members to agree their responsibilities;
• knowing who can do which parts of the project;
• recognising when extra help with skills is needed;
• bringing the project together.</p>
      <p>
        Some groupware products provide features which help with
these problems, but these are designed for business use, and may
not adequately help the students to acquire appropriate group
working skills. There is also a need for tutors to be able to
monitor the progress of the project. These difficulties represent
the sort of capabilities that could be offered by a group support
agent, however, a commitment to the group project by each
individual is still an essential ingredient.
4. ARCHITECTURE OF A GUARDIAN
AGENT
An agent is a self-contained, concurrently executing software
process, which encapsulates the current state in terms of
knowledge, and is able to communicate with other agents
through message passing [
        <xref ref-type="bibr" rid="ref13">13</xref>
        ]. We propose an agent to support
students, called a Guardian Agent, which works quietly in the
background on each group member’s workstation. The agent will
autonomously monitor the progress of the group project, suggest
ways in which the students can act to improve the progress of the
project and enhance the communication between members of the
group.
      </p>
      <p>Each student working on the project will have an individual
agent, operating in the background, watching progress,
measuring it against the plan and taking remedial action when
necessary. The agents belonging to each member of the group
are able to communicate with each other to exchange
information and negotiate on behalf of the students.</p>
      <p>The agent might recognise that lack of time is becoming a
problem when monitoring the progress of the project against a
plan of the work to be completed, and a variety of solutions
might be suggested. Similarly, lack of skills could be recognised
by the appearance of errors in the work, or the student asking
how to do a particular task. The individual agent will support an
individual student, but in addition this agent will communicate
with the other students’ agents to support the group as a whole to
ensure that the project is completed satisfactorily. The student
will also be free to work unaided by the agent if so desired,
though monitoring will continue unobtrusively.</p>
      <p>
        It is not proposed that such an agent will replace the tutor’s
input, but the agent will perform some of the mundane
administrative tasks, which are usually performed by the group
members during face-to-face meetings automatically, but which
do not need to be performed by the students in order to fulfil the
learning outcomes of the group project. The role of the tutor is
different for online learning [
        <xref ref-type="bibr" rid="ref10">10</xref>
        ], and support for the tutor by an
agent is an additional functionality to be pursued. Multi-agent
systems can combine different types of agents, but in our
architecture each individual agent will have a similar structure
when the project begins. Each agent will have interfacing
capabilities for communicating with its student, reasoning
capabilities for monitoring and analysing the current situation, a
knowledge base personal to its student and communication
capabilities for communicating with other students’ agents. The
architecture for our agent system is shown in Figure 1, which
shows that communication with other students’agents is by
means of a whiteboard.
      </p>
      <p>Student</p>
      <p>Student</p>
      <p>Student
Guardian</p>
      <p>Agent</p>
      <p>Guardian</p>
      <p>Agent</p>
      <p>Guardian</p>
      <p>Agent</p>
    </sec>
    <sec id="sec-7">
      <title>Whiteboard</title>
      <p>
        4.1. Development of the Guardian Agent
Students undertaking group projects in a traditional setting face
several problems to a greater or lesser degree. The results of
surveys on group projects have been considered closely when
deciding what functions software agents should undertake in
supporting group working [
        <xref ref-type="bibr" rid="ref12">12</xref>
        ]. Table 2 shows the stages of a
group project and the roles we have identified for our agents.
      </p>
      <p>The process of allocating roles begins with finding out about
each other’s abilities. Students who have worked with each other
previously will have a good idea about this already, but online
students will have to explicitly describe themselves. The agent
asks its student to identify the predetermined task areas he or she
likes, is good at, dislikes and is not good at. These are filed as
Prolog facts and posted to the whiteboard, where the other agents
can access the information. One of the agents is given the role of
allocating tasks to the students and, after checking that all the
students have posted their abilities list to the whiteboard, this
agent determines which students should be allocated which task,
using the following rule:</p>
    </sec>
    <sec id="sec-8">
      <title>If studentA likes X and is able at X</title>
      <p>Then studentA should do X.</p>
      <p>A series of allocation facts will then be posted to the whiteboard.
If no student has been allocated to a particular task the agent
looks again at the facts at its disposal and using the following
rule, decides whether any student may be able to perform a task
if given appropriate training:</p>
    </sec>
    <sec id="sec-9">
      <title>If studentB likes X, but is unable at X</title>
      <p>Then studentB should be offered training in X
A series of training facts are similarly posed to the whiteboard. It
is likely that there will be some conflict to resolve, such as too
many students being allocated to the same task, or no student
allocated or offered training for a task. After informing the
students of the conflict, the agents may negotiate on behalf of
their student to solve the conflicts if the student does not choose
to negotiate unaided.</p>
      <p>
        The next planning task is to set the ground rules, identified as an
important element in successful group projects [
        <xref ref-type="bibr" rid="ref9">9</xref>
        ]. Each student
can suggest a rule, such as answering email messages within 48
hours, informing the other students in advance of
nonparticipation in a conference or when experiencing difficulties
completing a set task on time. The ground rules are likely to be
different from those for a face-to-face project, so the agent will
be able to offer help to students, by suggesting rules, and
collating the rules for agreement by all of the students.
An initial prototype has been implemented in Prolog, using the
declarative features for handling facts and rules, which can be
passed between each student’s agent and the whiteboard.
Evaluation of the prototype will give us valuable information as
to the suitability of autonomous agents for supporting students
undertaking group project work.
      </p>
      <sec id="sec-9-1">
        <title>5. CONCLUSIONS</title>
        <p>If online courses are to be regarded as an acceptable substitute
for traditional campus-based courses, an element of group
working should be incorporated into the scheme. However,
online group projects require very different ways of working,
though the learning outcomes should remain the same, for
example to prepare the students for team working, and practice
what has been learned on the course. In the future when global
online team working becomes the accepted practice at work, the
learning outcomes may need to be extended. Group workers
experience a number of problems associated with the
mainrenance aims of groupwork when working online with
limited CMC support. Agents are suggested as appropriate to
provide additional support which will overcome some of these
difficulties.</p>
        <p>In this paper we have reported on the initial design of agents to
support students undertaking group projects, showing the chosen
architecture for our prototype system. The implementation in
Prolog uses an agent tool, which can be extended for a number of
additional purposes.</p>
      </sec>
      <sec id="sec-9-2">
        <title>6. REFERENCES</title>
      </sec>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          [1]
          <string-name>
            <surname>Bion</surname>
            ,
            <given-names>W.</given-names>
          </string-name>
          , Experiences in groups.
          <source>London: Tavistock Publications</source>
          ,
          <year>1961</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          [2]
          <string-name>
            <surname>Boud</surname>
            ,
            <given-names>D.</given-names>
          </string-name>
          and
          <string-name>
            <surname>Feletti</surname>
            ,
            <given-names>G.</given-names>
          </string-name>
          ,
          <article-title>The challenge of problem-based learning</article-title>
          , ,
          <source>Kogan Page</source>
          ,
          <year>1998</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          [3]
          <string-name>
            <surname>English</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          &amp;
          <string-name>
            <surname>Yazdani</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <article-title>Computer supported cooperative learning in a virtual university</article-title>
          . Virtual University Internet publication, http://www.mbc.co.uk/virtual-university-press/vuj.
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          [4]
          <string-name>
            <surname>Gibbs</surname>
            ,
            <given-names>G.</given-names>
          </string-name>
          ,
          <source>Learning in Teams. Oxford Centre for Staff Development</source>
          ,
          <year>1995</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          [5]
          <string-name>
            <surname>Hartley</surname>
            ,
            <given-names>P.</given-names>
          </string-name>
          , Group communication.
          <source>London: Routledge</source>
          ,
          <year>1997</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref6">
        <mixed-citation>
          [6]
          <string-name>
            <surname>Hendson</surname>
            ,
            <given-names>B.</given-names>
          </string-name>
          ,
          <article-title>Groupwork with multi-media in maths: the role of the technology and teacher</article-title>
          .
          <source>Br J of Educational Technology</source>
          , Vol
          <volume>28</volume>
          No 4,
          <fpage>pp257</fpage>
          -
          <lpage>270</lpage>
          ,
          <year>1997</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref7">
        <mixed-citation>
          [7]
          <string-name>
            <surname>Jaques</surname>
            ,
            <given-names>D.</given-names>
          </string-name>
          , Learning in Groups. London : Croom Helm,
          <year>2000</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref8">
        <mixed-citation>
          [8]
          <string-name>
            <surname>Kolb</surname>
            ,
            <given-names>D.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Experiential</surname>
            <given-names>learning</given-names>
          </string-name>
          ,
          <source>Prentice Hall</source>
          ,
          <year>1984</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref9">
        <mixed-citation>
          [9]
          <string-name>
            <given-names>O</given-names>
            <surname>'Sullivan</surname>
          </string-name>
          ,
          <string-name>
            <surname>T</surname>
          </string-name>
          &amp; Rice,
          <string-name>
            <surname>J</surname>
          </string-name>
          et al, Successful group work,
          <source>Kogan Page</source>
          ,
          <year>1996</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref10">
        <mixed-citation>
          [10]
          <string-name>
            <surname>Salmon</surname>
            ,
            <given-names>G.</given-names>
          </string-name>
          , E-Moderating, Kogan Page, London,
          <year>2000</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref11">
        <mixed-citation>
          [11]
          <string-name>
            <surname>Vliem</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Using</surname>
          </string-name>
          the Internet in university education, Ergonomics Group, University of Twente, Netherlands,
          <year>1998</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref12">
        <mixed-citation>
          [12]
          <string-name>
            <surname>Whatley</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Beer</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Staniford</surname>
            ,
            <given-names>G.</given-names>
          </string-name>
          and
          <string-name>
            <surname>Scown</surname>
            ,
            <given-names>P.</given-names>
          </string-name>
          , HCI International,
          <year>1999</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref13">
        <mixed-citation>
          [13]
          <string-name>
            <surname>Wooldridge</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          and
          <string-name>
            <surname>Jennings</surname>
            ,
            <given-names>N.</given-names>
          </string-name>
          ,
          <source>Intelligent agents: ECAI94 Workshop on agent theories</source>
          . Springer-Verlag: London.
          <year>1995</year>
          .
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>