Andrew Diniz da Costa

JAAF+T: a framework to implement self-adaptive agents that apply self-test

Appropriate implementation of self-adaptive software systems able not only to check the needs for the adaptations and perform them but also to ensure their compliance with new environment requirements is still an open issue. Therefore, this paper proposes an extension to the Java self-Adaptive Agent Framework (JAAF) in order to apply the self-test concept. This framework allows for the creation of self-adaptive agents based on a process composed of a set of four main activities (monitor, analyze, plan and execute). In this paper we extend the process and framework by including the test activity that will check the adapted behavior before its execution. The applicability of the proposed process is demonstrated by a case study where a system responsible for generating susceptibility maps, i.e., maps that show locations with landslides risks in a given area, searches to adapt its behavior and checks the adaptations before using them.

Art Competition: Agent Designs to Handle Negotiation Challenges

Multi-Agent Systems are societies in which autonomous and heterogeneous entities can work together to achieve similar or different goals. However, it is often a difficult task to perceive when an agent can be trusted to perform a specific negotiation. In other words, to establish its reputation. For that purpose, a number of models and strategies have been studied and proposed in the literature. The AAMAS Agent Reputation Trust (ART) Testbed competition has been created to make it possible to compare different strategies. This is achieved through an environment for agent-based simulation games. In the second and third edition of ART that took place in 2007 and 2008, new models and strategies emerged that incorporated important research contributions to the field. In this paper, we provide the description of the main set of concerns that we took into consideration to create competitive strategies for such versions of ART.

Multidisciplinary groups learning to develop mobile applications from the challenge based learning methodology

For years traditional learning approaches have been applied in schools and universities, where teachers are considered the knowledge center, leaving students as secondary actors of learning processes. In order to investigate new learning methodologies a set of new approaches have been tried. Challenge Based Learning (CBL) is one of these proposals, created with the essential principles of a 21st century learning environment in mind. Since 2014, the Pontifical Catholic University of Rio de Janeiro (PUC-Rio) have applied such methodology in the context whose multidisciplinary students from different courses learn to develop mobile apps, and work together looking for solutions that can impact lives. This paper presents how CBL was applied for two student groups, mentioning learned lessons and changes performed in the learning process from these lessons.

RSA-MBT: a test tool for generating test artifacts based on models

Model-Based Testing (MBT) has attracted the attention of many industries and, hence, it has provided several approaches reported in the literature. The Software Engineering Lab (LES) at the Pontifical Catholic University of Rio de Janeiro has worked extensively on coordinating and carrying out tests of large-scale software systems developed (for web and desktop) for different domains (e.g. petroleum, e-commerce, etc). Based on this experience, an LES test group created a new test modeling language called UML Testing Profile for Coordination (UTP-C) to model relevant test data. However, to use the advantages of the new modeling, an appropriate test-modeling tool became necessary. Thus, this paper presents the RSA-MBT, a new plug-in developed for the Rational Software Architecture (RSA) tool, to generate a set of test artifacts from UTP-C diagrams.

Improving test models for large scale industrial systems: an inquisitive study

Although documentation of software tests is becoming increasingly important, there is little knowledge on whether modeling languages and tools are effective in industrial projects. Recent reports have pointed out that test modeling techniques might be barely used by software developers due to their inability to cover test concepts relevant in real-life large applications. This paper reports an inquisitive multi-phase study aimed at revealing test-relevant concepts not supported by modeling languages. The study encompassed several questionnaire responses and interviews with developers, and observational analyses run over two years in large-scale software projects. Various test concepts were brought forth and they fall in three categories: (i) test cases and software evolution, (ii) interdependencies between test cases, and (iii) categorization and grouping of test cases. Finally, the relevance of the identified test concepts is discussed in terms of an industrial system for inventory and supply control of petroleum products.