Vol. 34 · No. 3

Several researchers have identified the design of collaborative systems as a particularly challenging task, because it has to consider services that are not easily identifiable by software designers. The systems that support the activities of partially virtual communities (PVC) are not an exception. Typically, designers of PVC supporting systems are able to identify services that have a visual representation on the user interface, but they usually overlook those that run in the system back-end. In an attempt to help designers to consider these two types of services, this article proposes a software architecture that can be used for both, designing new PVC supporting systems and evaluating existing ones. The architecture was used in three case studies as a design guideline during the development of PVC supporting platforms, and also as an instrument to evaluate three commercial systems. Although the obtained results are still preliminary, they indicate that the architecture is able to accomplish both roles.

Self-constructed external representations are thought to be beneficial in teaching and learning, especially when embedded in peer interactions, and can positively affect the course and type of reasoning, for example by providing grounding for explanations and self-explanations, by helping to disambiguate students' mental models of phenomena, or by increasing and sharing the task focus. This paper reports on the results of research efforts investigating which conditions are advantageous in collaborative drawing activities in computer-supported learning scenarios for young students. We describe the design, technical implementation and empirical results of a study with 94 primary school students working on a collaborative drawing task in various conditions that include adaptive prompting and scripted activities. Results showed, partially in line with our expectations, that adaptive prompting as well as scripting positively affected students' learning outcomes and discourse quality. Students in both supported conditions more often critically reviewed the work of their partner and integrated ideas from their partner in their own reasoning.

The analysis and evaluation of computer-supported collaborative activities is a complex and tedious task. However, it is necessary in order to support collaborative scenarios, to scaffold the collaborative knowledge building and to evaluate the learning outcome. Various automated techniques have been proposed to minimize the workload of human evaluators and speed up the process. In this study, we propose a memory based learning model for the analysis, classification and evaluation of collaborative activities that makes use of time series techniques along with logfile analysis. We argue that the classification of collaborative sessions, with respect to their time series attributes, may be related to their qualitative aspects. Based on this rationale, we explore the use of the model under various settings. The results of the model are compared to assessments made by expert evaluators using a rating scheme. Correlation and error analyses are further conducted.

ICT tools offer the possibility to store, analyse and visualise large amounts of educational data. However, in order to make sense of these analyses, teachers need meaningful information connected to their pedagogical intentions. We envision an enriched learning-design process, which supports the integration of the issues related to monitoring in the script. This paper presents a study where the first author and a teacher iteratively co-designed two authentic learning scenarios with the aim of defining and evaluating a monitoring-aware design model and process. These two proposals, the model and the process, were positively evaluated by the participant teacher, and are the basis for the future implementation of an authoring tool that will support the proposed monitoring-aware design process.

Computer games are currently one of the computer science applications with the highest amount of users. ``Serious gaming'' approaches try to use the attraction of playing games to convey serious content in an entertaining way. This paper presents a multi-agent-architecture for collaborative, serious and casual games. We are combining serious games with casual games, as these are known to have a high potential for frequent gaming by people of various social and educational background. To be flexible concerning the learning domain an ontology-based approach has been used. Thus, new games for different knowledge domains can be created by exchanging the ontology accordingly. Furthermore, we use a ``game with a purpose'' approach to enable the computer to learn new facts about the games knowledge domain to ease the effort of adapting the basic game to the intended target domain. The presented system is capable of proposing candidates for missing relations in the underlying ontology as well as pointing out possible misconceptions – either within the ontology or by the players – using various heuristics including ``wisdom of the crowds'' methodologies. The feasibility of the presented approach and its implementation (the Matchballs game) is shown by three case studies each focusing on one aspect of the system.

Software systems usually include a limited number of important classes that a large number of other classes depend upon. These files comprise the technical core of a software system. Evolving this core is naturally difficult and requires caution, since inappropriate changes may induce side and ripple effects. Indeed, despite the rhetoric about openness in free/libre open source software, modifications to the technical core are usually accomplished by a small set of contributors known as key developers. Automatically identifying key developers is relevant for a number of reasons, such as supporting the recruitment of specialists, assigning tutors to newcomers, and estimating the longevity likelihood of projects. Moreover, little is known about how key developers evolve and about the social skills that support them in their technical tasks. In this paper, we describe a case study involving the Apache Ant project. Our goal was to identify key developers and characterize them in terms of their social activity and contributions. We conceived, implemented, and applied a method to identify key developers. Such method identified four individuals, who were the main target of our investigation. We built communication networks from mailing list data and a coordination requirements network from their development traces. The analysis of these two networks indicated that three key developers socialized more than the others. They acted as bridges connecting other developers and communicated with almost everyone they were supposed to. The other key developer showed a very distinct behavior, as he participated very rarely in the mailing list. We also analyzed key developers' contributions and found some patterns. Core commits and non-core commits were often interleaved and key developers also contributed to peripheral portions of the system. Finally, we observed that the set of key developers was indistinguishable from the set of top committers. We expect this characterization to foster the definition of key developers' profiles that take into account their social activities and contribution characteristics.