Resumen de A framework for interaction training in Global Software Development
Miguel Jiménez Monasor
Global Software Development (GSD) is an emerging paradigm in which development teams are geographically distributed throughout different countries whilst working on the same projects. Lower costs, greater access to a skilled workforce, the leveraging of time-zone effectiveness and a closer proximity to the market are some of the proven advantages that have contributed to this recent expansion. However, GSD also entails certain drawbacks caused mainly by distance and socio-cultural differences. Traditional face-to-face meetings are no longer common, thus making communication, coordination and collaboration more complex. Moreover, interactions frequently involve members who use non-native languages and whose customs and behavior may be misinterpreted. Companies consequently require their workforce to have a balanced set of socio-technical skills more than ever before. Students and practitioners therefore need to acquire not only new technical skills, but also the ability to collaborate and interact effectively in a variety of geographic, cultural and linguistic settings. This kind of training is not currently common and companies often complain about their new recruits¿ lack of preparation in GSD related skills. However, companies do not always have the resources to provide these kinds of training programs, as the teaching of this kind of skills requires appropriate tools, methods and course materials. Moreover, training these skills is not easy at universities since it necessitates providing trainees with real experiences that it may be difficult to reproduce. Some universities organize joint courses through collaboration with distant universities so that students can benefit from a wide set of knowledge and experiences, but the coordination problems of these approaches make them hard to apply. One way in which to avoid the problems outlined above, i.e., a lack of resources, coordination complexities, and poor levels of realism, is by using simulation, which has been used in various areas of Software Engineering training such as project management. However, although simulation has the potential to enable independent and controlled training, no significant proposals exist in the field of GSD training. The research question that guides this work is therefore: ¿How can we model a GSD educational framework that replicates real-world settings?¿ This thesis presents a framework (VENTURE) for interactive GSD training which is based on simulation. The framework simulates real world interactions that occur in GSD which are appropriate for use in both industrial and academic settings. In order to simulate GSD settings, trainees take on a specific role and interact with Virtual Agents, characterized by a specific culture. A special kind of Virtual Agent, called a Virtual Guide, directs the trainees and provides feedback when they interact in an inappropriate manner. These training scenarios take place by means of a chat simulator for synchronous communication training and through an email simulator for asynchronous communication training. The feedback provided by the Virtual Guide is automatically used to assess the trainees¿ performance in real time following a defined assessment method. The courses are provided via an e-learning environment that allows students to access the training sessions at any time. Moreover, a scenario designer has been implemented to support the design of new simulated interactions. New simulations will build on predefined GSD problems, including cultural and linguistic rules that represent typical interaction challenges in real settings. In order to design GSD scenarios, realistic low-level GSD problems that arise during the day-to-day work are required. However, accessing these problems or scenarios is difficult, mainly because companies are not always willing to share their knowledge and experience. In order to tackle this challenge, the framework additionally considers a Web based knowledgebase of GSD patterns and scenarios. The GSD community can, through this open social network, share their experiences, problems and scenarios in an organized manner. Simulations may therefore be designed by applying the data collected in this collaborative knowledgebase. The VENTURE framework was created by following a mixed methodology that combines both qualitative and quantitative methods. An initial systematic literature review on GSD served to focus the topic on the field of GSD education, after which a second systematic literature review was conducted in the field of GSD education. Having studied the problems of the existing educational methods, a prototype of the framework was built and evaluated by experts, leading to a first version of the framework. The principal components of VENTURE (the chat simulator and the scenario designer) were then evaluated by a group of experts. These experts in GSD and e-learning examined the design and executed a brief training simulation in order to analyze the tool¿s potential to provide GSD training, along with the tool¿s usability. The results of the survey, which were analyzed by means of a thematic analysis procedure, were mainly positive. Both evaluations served to confirm that VENTURE has the potential to fill the gap in GSD virtual team training. Moreover, indications on where improvements could be made served to develop a new version of the framework. A Field Study of the new version of the framework was conducted in order to evaluate whether learning was achieved after participation on a real GSD course based on two training simulations. This evaluation involved potential users of the framework (students, trainers, researchers) interacting with VENTURE and completing two training scenarios. A rigorous evaluation of the user experience was conducted using qualitative and quantitative data collection. Several questionnaires (including pre and post questionnaires) were administered at different stages of the process and the interactions were also tracked for reasons of analysis in order to uncover any significant differences after the simulation process took place. The results revealed that participants, after going through the evaluation process, acquired new knowledge and capabilities that would allow them to perform better in GSD settings. The participants also agreed that they were satisfied with both the learning outcomes and the user experience. This thesis provides two main contributions: the first of these is a training framework designed to take advantage of Virtual Agents in order to design accurate simulations of GSD interactions. Part of this training framework is also to automatically assess the student¿s progress. An invention disclosure has been produced, and commercialization activities are currently being addressed with regard to this contribution. The second contribution is the introduction of a new pattern model with which to acquire real world GSD problems that can be used with pedagogical purposes. Future work will consider the population of the GSD pattern based knowledgebase, along with improvements that could allow a more intelligent interaction of the Virtual Agents and facilitate the design of the simulations.
