One of the key challenges in constructing a Bayesian network BN is defining the node probability tables (NPT). For large-scale BN, learning NPT through domain experts knowledge elicitation is unfeasible. Previous works proposed solutions to this problem using the concept of ranked nodes; however, they have limited modeling capabilities or rely on BN experts to apply them, reducing their applicability. In this paper, we present an expert system based on production rules to define NPTs with the purpose of enabling the definition of NPTs by experts with no ranked nodes-specific knowledge. To create the rules, we elicited data from an expert in ranked nodes. To validate our approach, we executed an experiment with a BN already published in the literature to verify if, with our approach, a practitioner can achieve the same or better configuration for the NPTs. We used the Brier score to assess the NPTs accuracy and evaluated the results with the Wilcoxon test. All the Wilcoxon tests executed rejected the null hypotheses that stated that the Brier scores for the original NPTs method were the same as the new NPTs. By using our solution, a practitioner can accurately define NPTs without understanding the concept of ranked nodes.
Software metrics have a fundamental role in the process of software quality management. However, in most cases, they are only used to quantify attributes, not supporting decision-making during the software life cycle. To support decision-making, it is necessary to give them by defining thresholds. In the literature, several approaches have been proposed with this purpose. On the other hand, most of them do not consider context factors such as the domain. Given this, in this paper, we evaluate if context factors influence the definition of thresholds for software metrics. Our work is restricted to Chidamber and Kemerer metrics, due to availability of data. We conducted an empirical study composed of two quasi-experiments. Each quasi-experiment uses an approach presented in the literature to define thresholds for software metrics, with the defined thresholds as the dependent variable. As the factor, we used a variable with two possible treatments: to consider the context or not. To define context, we used factors presented in the literature. As the objects of study, we used the source code of fifteen Java-based open-source projects. For measurement purposes, we used the six original Chidamber and Kemerer metrics. For both quasi-experiments, the accuracy of the definition of thresholds improved by considering the context. Therefore, we concluded that context factors influence the definition of the threshold for Chidamber and Kemerer metrics, which is an indicator that it influences other software metrics.
Context: Companies must make a paradigm shift in which both short- and long-term value aspects are employed to guide their decision-making. Such need is pressing in innovative industries, such as ICT, and is the core of Value-based Software Engineering (VBSE). Objective: This paper details three case studies where value estimation models using Bayesian Network (BN) were built and validated. These estimation models were based upon value-based decisions made by key stakeholders in the contexts of feature selection, test cases execution prioritization, and user interfaces design selection. Methods: All three case studies were carried out according to a Framework called VALUE — improVing decision-mAking reLating to software-intensive prodUcts and sErvices development. This framework includes a mixed-methods approach, comprising several steps to build and validate company-specific value estimation models. Such a building process uses as input data key stakeholders’ decisions (gathered using the Value tool), plus additional input from key stakeholders. Results: Three value estimation BN models were built and validated, and the feedback received from the participating stakeholders was very positive. Conclusions: We detail the building and validation of three value estimation BN models, using a combination of data from past decision-making meetings and also input from key stakeholders.
Software metrics are essential resources in software enterprises. They can be used to support decision-making and, consequently, reduce costs, improve the productivity of the team and the quality of products delivered. On the other hand, this is only possible if the metrics are valid. Although there are studies related to software metrics validity, none present a solution to represent the uncertainties of the metrics selected to measure the attributes of the entities. In this paper, we present a process to build Bayesian networks to represent the uncertainties of software metrics-based models. The proposed solution is composed of two activities and focuses on the selection and validation of metrics to construct the Bayesian networks. We validated the model with simulated scenarios. Given the successful results, we concluded that the proposed solution is promising. This paper complements the state of the art by showing how to complement a popular metric selection technique, GQM, with information to model uncertainties of the metrics using the concepts of metric validation and Bayesian networks.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.