Activity Coordination in Collaborative Learning

Environments

Carlos José M. Olguín

3,1

, Alberto B. Raposo

and Ivan Luiz M. Ricarte

UNICAMP / FEEC / DCA

Caixa Postal 6101

13083-970 – Campinas, SP – Brazil

PUC-Rio / DI / Tecgraf

R. Marquês de S. Vicente, 225

22453-900 – Rio de Janeiro, RJ – Brazil

UEM / CTC / DIN

Av. Colombo 579 – Bloco 019

87020-900 – Maringá, PR – Brazil

Abstract. In the context of computer-supported collaborative learning, discus-

sions are essential to increase the knowledge level of the members of a group.

This work proposes the modeling of the discussion activities of a study group

using an activities coordination model. Starting from the formal modeling of

the system, and using coordination mechanisms based on Petri Nets, the behav-

ior of the environment can be simulated and analyzed. These simulations allow

anticipating possible problems and help to turn interactions among students

more efficient.

1 Introduction

Studying techniques for computer-supported collaborative learning, some researchers

have been working in the construction of systems to monitor collaboration automati-

cally. Barros and Verdejo [1] defined a method to calculate a set of attributes that

characterizes the individual and group behavior. Using labeled textual contributions

and fuzzy logic the degree of collaboration of the group is calculated. Mühlenbrock

and Hoppe [2] intended to consider users' actions in a collaborative work environ-

ment as the basis for the qualitative analysis of activities. This approach makes possi-

ble to derive descriptions of group activities through a method to recognize them.

Delgado et al. [3] defined a model for monitoring the activities accomplished in-

side a study group. In that work it was proposed that the actions and the monitored

interactions were stored, as contributions, in a discussion graph. Through the analysis

of this information by a human tutor, it would be possible to infer the cognitive proc-

ess used by the student during the discussion. Extending that work, the objective of

the present work is to model discussion activities, using the coordination model based

José M. Olguín C., B. Raposo A. and Luiz M. Ricarte I. (2004).

Activity Coordination in Collaborative Learning Environments.

In Proceedings of the 1st International Workshop on Computer Supported Activity Coordination, pages 227-232

DOI: 10.5220/0002665902270232

 SciTePress

on Petri Nets (PN) proposed by Raposo and Fuks [4]. The resulting model is suscep-

tible to simulations and analyses that allow the anticipation of problems. It makes

possible the formal verification of the model, and helps to turn more efficient the

students' interactions during the discussion process. Furthermore, starting from the

formal model of the system, software components can be implemented to give sup-

port to the discussion of the group.

This work is organized as follows: Section 2 presents the guidelines of a develop-

ment methodology of collaborative learning environments that gives special attention

to the coordination of tasks that are executed in the environment. This section also

describes the model adopted to capture group discussions; Section 3 introduces as-

pects of the adopted coordination model; Section 4 presents coordination mechanisms

based on PN; and the final section presents the conclusions of the work with empha-

sis in the directions to be followed.

2 The Development Methodology and The Model of Discussion

Besides defining the requirements and pedagogic objectives of collaborative learning

environments, a development methodology of collaborative learning environments

should consider the following activities:

1) Identification of the tasks that will be executed in the environment and their in-

terdependencies;

2) Identification and modeling of the coordination mechanisms among those tasks;

3) Analysis of the model developed in (2) whose results can take back to (1);

4) When the result of the analysis (3) is satisfactory, to accomplish the project and

the construction of the environment.

In this work the considered environment is the message exchange that happens,

within a study group, when discussing about a certain subject, with the intention of

increase the knowledge level that the members of the group have about that subject.

We will concentrate our attention on activities (1), (2) and (3) of the methodology.

Our interest at this moment is the modeling of activities, through the use of a model

that considers coordination aspects among the tasks. In this paper, we consider a

model of an environment for the exchange of textual messages in a study group [3].

This model assumes that, during group activities, students use collaborative services

of the environment to generate some kind of information that others can comment or

discuss using the same services. All information or comment directly associated to the

discussion is defined as a contribution. The contribution is the basic element used to

monitor group activities. The model allows the quantification of accepted contribu-

tions and the maintenance of a discussion log, represented by a sequence of contribu-

tions.

A discussion is represented as a sequence of contributions. Contributions can be:

1) accepted or rejected; 2) substituted or reconsidered by another contribution; and 3)

doubts, questions or answers to a contribution. The registration of contributions is

organized in a discussion graph.

The discussion process of the group involves the concept of “knowledge negotia-

tion” that is an important aspect in the construction and the management of the

228

knowledge generated by the study group. This process embraces a series of collabora-

tive interactions among students, for example, to clear the meaning of some terms

used in the placement of the contributions, to discuss alternatives, etc [5].

The dynamics of the discussion model shows the need to coordinate the activities.

In the next section is presented the coordination model of activities used in this work.

3 The Coordination Model - Interdependencies

In our context a collaborative activity is defined as a set of interdependent tasks, exe-

cuted by several actors in order to reach a common goal. In the example of our work,

students that are part of the discussion group execute a series of interdependent tasks

(for example, to make contributions, to decide about the acceptance or not of contri-

butions as valid knowledge, etc.) with the objective of increasing the knowledge level

on the discussed subject.

The interdependency concept is fundamental in the coordination theory. It is pos-

sible to characterize different types of interdependencies and identify the coordination

mechanisms that manage them, creating a set of interdependencies and related coor-

dination mechanisms capable to encompassing a wide range of collaborative applica-

tions [6].

3.1 The Basic Temporal Interdependencies

Temporal interdependencies establish the relative order of execution among a pair of

tasks. The set of temporal interdependencies of the proposed model is based on the

temporal relations defined by Allen. He proved that there is a set of primitive and

mutually exclusive relations that could be applied over time intervals [7]. A time

interval is characterized by two events, which in turn are associated to time instants.

The first event is the beginning of the interval A, denoted i

. The other event is the

ending of the same interval, denoted f

,. According to Allen, a set of seven primitive

relations may maintain temporal information on any pair of time intervals A and B (A

equals B, A starts B, A finishes B, A meets B, A overlaps B, A during B and A before

B). Based on these relations a group of axioms is defined to create a temporal logic.

The fact of being applied over time intervals (and not over time instants) made the

above relations suited for task coordination purposes, because tasks are generally

non-instantaneous operations. The adaptation of Allen’s primitives to the context of

collaborative activities considers that any task T will take some time to be executed.

Allen's temporal logic is defined in a context where it is essential to have proper-

ties such as the definition of a minimum set of basic relations, the mutual exclusion

between these relations and the possibility to get conclusions starting from them.

However, temporal interdependencies among collaborative tasks are inserted in a

different context. The important fact here is the management of the interdependencies

and the appropriate understanding of the collaborative activity.

A limitation of Allen's relations is that they are merely descriptive, in other words,

they don't express causal or functional relations among the intervals [8]. For all these

229

reasons, it was necessary to do some adaptations to Allen's basic relations. The goal

of the proposed extensions is to offer a larger set of possibilities to create coordina-

tion mechanisms that may control many different situations.

3.2 Active and Passive Interdependencies

The merely descriptive characteristic of Allen's temporal relations allows different

interpretations for the interdependencies. For instance, consider that two tasks, Ta and

Tb, are related by the interdependency Ta equals Tb. In the coordination context, this

interdependency can be interpreted in two different ways. In the first case, called

active interpretation, this relation express that the beginning of a task should begin

the other; and that the end of one of the tasks should conclude the other task. The

second possible interpretation for any coordination mechanism is called passive in-

terpretation. In this case, the coordination mechanism expresses a group of conditions

that should be obeyed to take the activity to the end.

To deal with active and passive interpretations, two operators were defined: en-

ables and forces. The operator enables represents the passive interpretation, while

forces represents the active interpretation. These operations can be applied in the

initial and final moments of each interdependent task. Additionally, these extreme

points have two states: ready and concluded, indicating, respectively, that the task is

ready to begin (or to finish) and that the task already began (or finished).

In order to exemplify the passive and active interpretations of the temporal inter-

dependencies, consider the situation in which two students, A and B, have to vote for

the acceptance or rejection of a contribution. We will denominate Tva and Tvb the

tasks of voting associated to students A and B, with initial and final points i

, i

, f

and f

respectively. Imagine now that in our environment the voting process begins

in a synchronous way. In this case, the interdependency associated to the tasks is Tva

starts Tvb that can be extended in several interpretations, such as:

(ready) enables i

AND i

(ready) enables i

– this declaration indicates the

passive situation in which the tasks will only begin their execution when both are

ready (i.e., Tvb will only be able to begin when Tva is ready to begin, and vice-versa),

but none will force the execution of the other.

(ready) forces i

– in this situation, when Tva is ready to begin, Tvb will be

forced to begin, indicating an active interdependency of the type “master/slave” (in

the same way, Tvb could be considered the master task if i

(ready) forces i

AND i

(ready) enables i

– Tva is the master task, since it

forces the beginning of task Tvb, but Tva will only begin when Tvb is ready.

4 Coordination Mechanisms based on Petri Nets

Coordination mechanisms based on PN were developed to coordinate the set of inter-

dependencies presented. The formal modeling allows the designer to anticipate and

test the behavior of the environment. PN support models at different levels of abstrac-

tion and are appropriate for simulation and formal verification.

230

In this proposal, the project of a collaborative learning environment is divided in

three hierarchical levels: workflow, coordination, and execution. In the workflow

level, the interdependencies among the tasks attributed to the actors of the environ-

ment are established (activity 1 of the development methodology presented in Section

2). The coordination level is built from the workflow level, through the expansion of

the interdependent tasks and the insertion of the corresponding coordination mecha-

nisms (activity 2 of the methodology). The model of the environment is simulated and

analyzed in this level (activity 3). The execution level deals with the execution of the

tasks in the environment (activity 4).

During the passage from the workflow to the coordination level, each task that has

interdependency with another is expanded in a sub-net, as presented in Figure 1. In

this model, events i and f (beginning and end of the task) are represented as transi-

tions, while the states ready and concluded are represented as connected places to the

respective transitions. After having triggered event i, the flow is divided in two paral-

lel flows, one that indicates that the task is in execution – i(concluded) – and another

that represents the interaction with the execution of the task in the system. The task

execution is modeled through a transition with token reservation (represented with the

letter “R”) that is a non-instantaneous transition – the tokens are removed from their

input place when the transition is triggered and only some time later is increased to

their output places, representing in this way the duration of the task.

Figure 1. PN representation of an interdependent task at the coordination level.

When considering two related tasks for interdependencies, it is necessary to con-

nect the places and the transitions of both models correctly to create the correspond-

ing coordination mechanisms. To do this, it is necessary to define how to map the

previously defined operators and parameters to the PN model. The mapping of the

forces and other operators to the PN model are presented in [4].

5 Conclusions

In this paper, a proposal for the use of an activities coordination model of a study

group was presented with the intention of turning students' interactions more effi-

cient. This model offers a certain degree of flexibility through the separation between

tasks and their interdependencies and it is adapted to deal with some interoperability

aspects, since the set of interdependencies is generic and the implementation of the

coordination mechanisms can be accomplished using any tool. This coordination

input

places

i (ready)

i (concluded)

[task in execution]

f (ready) f (concluded)

output

places

task

Task

231

model is quite appropriate to represent interactions that occur among the members of

a discussion group. Starting from the resulting model, software components can be

derived to give subsidies to the discussion of the group. The software components

will allow standardize the interactions between tasks and coordination mechanisms

associated to them in a way that does not depend on the implementation.

Although the presented coordination mechanisms appropriately represent the in-

terdependencies among tasks that compose a collaborative activity, the resulting

model is static. This characteristic of the model does not allow that alternative situa-

tions could be represented. In this sense, the continuation of the work is centered in

the extension of the presented coordination mechanisms in order to turn them more

flexible. A possibility would be the definition of new operators that will consider

concepts of fuzzy logic to deal with alternative situations.

References

1. Barros, B., Verdejo, M.: An approach to analyze collaboration when shared structured

workspaces are used for carrying out group learning processes. In: Proc. of the 9

Intl.

Conf. on Artificial Intelligence in Education, Lemans, France (1999) 449-456

2. Mühlenbrock, M., Hoppe, H.: A Collaboration Monitor for Shared Workspaces. In: Proc. of

the 10

Intl. Conf. on Artificial Intelligence in Education, San Antonio, Texas, USA (2001)

3. Delgado, A.L.N., Olguín, C.J.M., Ricarte, I.L.M.: Monitoring Learners Activities in a Col-

laborative Environment. In: Proc. of the 7

Intl. Workshop on Groupware, Darmstadt, Ger-

many (2001) 148-152

4. Raposo, A.B., Fuks, H.: Defining Task Interdependencies and Coordination Mechanisms for

Collaborative Systems. In: Blay-Fornarino, M., Pinna-Dery, A. M., Schmidt, K., Zaraté, P.

(eds.): Cooperative Systems Design. Frontiers in Artificial Intelligence and Applications,

Vol. 74. IOS PRESS, Amsterdam (2002) 88-103

5. Stahl, G.: The Model of Collaborative Knowledge-Building. In: Proc. of the 4

Intl. Conf.

of the Learning Sciences, Ann Arbor, USA (2000) 70-77

6. Malone, T.W., Crowston, K.: The Interdisciplinary Study of Coordination. ACM Computing

Surveys, vol. 26, no. 1 (1994) 87-119

7. Allen, J.F.: Towards a General Theory of Action and Time. Artificial Intelligence, vol. 23

(1984) 123-154

8. Duda, A., Keramane, C.: Structured Temporal Composition of Multimedia Data. In: Proc.

of the Intl. Workshop on Multi-Media Database Management Systems, Blue Mountain

Lake, USA (1995) 136-142

232