In this paper we present a case study of collaborative diagram drawing involving 36 students in Computer Science. Their task was to collaboratively draw a Use Case Diagram about the scenarios provided at the begin of the experiment. As students of a Software Engineering course, they had a general knowledge of such type of diagrams and related terminology, but they were not experts and had not real and practical experiences in diagram drawing. The tools used were a synchronous collaborative drawing tool integrated with a chat tool to support communication among the participants. Moreover, the experiment has been structured following the 'think, pair, share' method. The analysis of the collaboration process outlines a twofold result: first, a significant equal participation of all the students and second, an implicit and recurrent self-regulatory behavior employed by the students to create and refine the diagram and to reach agreement about the final result.
The collaborative creation of diagrams is commonly used in brainstorming
processes, development of models, problem solving processes and, in general terms,
in the creation of shared knowledge and understanding. This activity is used also
in the educational setting to support collaborative learning. Indeed, in educational
settings the aims are different from the working setting: students are responsible for
one another’s learning as well as their own, thus, the success of one student helps
other students to be successful. Therefore, the aspects about the users’participation
and the self-regulated behaviors become particularly important: the free-riding
behavior of some students de-motivate the other students and, then, the overall team
performance decreases (
In this paper we present our evaluation of an experiment of collaborative diagram drawing in a Software Engineering course. We have used a chat and a tool (the Shared Drawing Tool, SDT) for collaboratively drawing of Use Case Diagrams. Both the tools are integrated in the CoFFEE platform, which offers also the possibility to structure the collaboration in customized phases. In the analysis of the experiment we have found that all the students participated with a similar engagement and effort, and by using effectively the tools. 2
Our experiment has been conducted in collaboration with the University of
Basilicata (
The groupware used is CoFFEE (Collaborative Face-to-Face Educational
Environment), a set of applications aiming to enhance the computer supported
collaborative learning in f2f setting. In this experiment, we have simulated a remote
condition: the students were in the same classroom but were grouped and seated
so that they could not have f2f interactions. The main applications are the
CoFFEE Controller and the CoFFEE Discusser, used in classroom respectively by the
teacher and the students. These applications offer several collaborative tools which
can be arranged together following the specific path designed preliminary by the
teacher (
The collaboration through CoFFEE has been organized following the “Think, Pair, Share” method (TPS) to encourage students participation. The students were organized in groups of four people, and the activity was structured in three steps: think, students work individually on the task to carry out; in this phase CoFFEE was configured to offer to each student his/her own istance of the SDT; pair, students work in pairs on the task to carry out; in this phase CoFFEE was configured to manage groups of 2 persons and offer them the SDT and the Chat; share, students work all together to produce a final solution; in this phase CoFFEE was configured to manage groups of 4 people and offer them the SDT and the Chat.
At each step, the results from the previous phase are copied onto the SDT workspace, so that the students can start the work of the new step on the basis of the previous one. The experiment generated 18 traces. A trace is an XML file where the Controller records all the events of a collaborative session: the chat messages, the shared drawing tool actions, clients connections and disconnections and so on. Two of the traces were corrupted so the data analysis is based on 16 traces. 3
A first study of the traces of the experiment aimed to evaluate the participation
of students in the collaborative session by using the Gini coefficient (Gc) and it
indicated that the participation among all the groups has been well balanced: the
users who chatted more, drew less, and vice-versa, with no explicit agreement about
the roles of the participants (
We analyze, here, the experiment by looking at any pattern of coordination that can be found in traces. In general terms, the chat has been used as a mean to organize and coordinate the work on the diagram; moreover the usage of the chat and the contributions on the SDT are not totally casual: indeed, they present a regular pattern. In all the traces there is an initial phase in which the students use almost only the chat. This is due to the structure “think-pair-share” of the experiment: in the initial phase the students use the chat to describe the work that they have done in the previous step. After this initial phase, we found two kind of patterns, A (5 traces) and B (11 traces), which differ in the way of usage of the tools. In fig.2 we show the cumulative sequences of contributions on the chat and SDT.
The pattern A (top graphic of fig. 2) presents sequences of frequent chat messages followed by contributions on the SDT. In these traces, then, the coordination work goes through the whole phase as micro-coordination tasks.
The pattern B (bottom graphic of fig. 2) presents sequences of macro phases of many chat messages followed by macro phases of work on the diagram. In these traces, then, there are a well defined analysis and coordination phase followed by a wide phase of implementation of the work.
In the analysis of the patterns, we paid particular attention to the step 3 because it involves all the students in the work group, so we consider it as the most meaningful; however, most traces present the same kind of pattern (A or B) both in the steps 2 and 3 (we have not considered the pattern of the step 1 because, in the think-pairshare method, it is the step where each student thinks alone). It should be noticed that, in some traces, in the step 2, when the students are organized in two groups, one group presents the pattern A and the other group presents the pattern B; however, all of these traces present pattern B in the step 3. Therefore, the pattern B prevails on the pattern A and this could explain the greater number of traces of kind B. The students employed these pattern spontaneously: the teacher did not stimulate any behavior nor action, he was just responsible to pass from a step to the next one.
The patterns employed by the students seems to influence the level of re-using of the existing diagram through the several steps. As previously described, the structure of the experiment follows the “think, pair, share” method and in each step the students had the work of the previous step available as starting point. Then, the expectation is that at the third step most part of the diagram has been created and the students should work to organize and refine the final outcome by re-using the existing draw, with a minimal numer of creation of new items in the diagram. Indeed, the level of reusing of the existing diagrams and the number of new items created in the third step seem influenced by the pattern employed by the students: the groups adopting the pattern B create fewer items (mean 2,06% respect to the total actions on the SDT) than their colleagues employing the pattern A (mean 9,81% respect to the total actions on the SDT), as shown in fig. 3. This suggests that the students employing the pattern B, during the macro phases of work coordination, are able to optimize the work better than their colleagues and are able to achieve an higher level of re-using of the existing work. 4
In the collaborative usage and creation of diagrams it is fundamental supporting users’ participation, reducing the free-riding effect and users’ idleness as well as scaffolding self-regulated behaviors. We believe that a key factor is the integration of discussion and drawing tools in a seamless environment, so that the users can switch between the tools without any overhead, and the discussion and the drawing activities can converge in a single and natural collaborative flow. Moreover, we believe that, in the learning setting, it is fundamental the possibility to embed in the groupware a structure to drive the collaboration process: this allows to adopt wellknown pedagogical strategies to enhance the students’engagement and learning performance. These ideas are supported by the analysis of the experiment that we have presented in this paper, which shows an equal participation of all the students and an effective usage of the tools to organize the work and create the diagrams.
We are aware that the conditions of our experiment (small groups, no
facilitators or modellers, similar cultural background and modelling skills among
participants) are very different from the business environment, where the modelling
activity could involve larger groups, expert modellers, stakeholders with no modelling
skills, and could require a severe check on the model quality (
Acknowledgments. The authors gratefully acknowledge the support by Andrea Guaiana during the development of Shared Drawing tool. The authors also thank the collaboration by Ugo Erra and Giuseppe Scanniello in setting up the experiment.