Background: The rapid adoption of Large Language Models (LLMs) in software engineering, such as in customer service chatbots, has brought substantial benefits but also significant risks. Biased, inappropriate, or harmful responses may occur when integrating LLMs as commercial off-the-shelf (COTS) components into user-facing applications. Aims: This paper aims to present emerging results from an exploratory study that compares commercial guardrail frameworks, assessing their effectiveness in filtering inappropriate content during chat interactions. Method: We empirically evaluated three guardrail frameworks LLM Guard, Llama Guard, and OpenAI Moderation - using two datasets containing toxic and offensive content. Results: The frameworks achieved high accuracy for one dataset (over 90%) but underperformed in other metrics, indicating that toxic or dangerous content could still reach users in certain scenarios, such as deployment in chatbots. Conclusions: These findings$h$ighlight$t$he$n$eed$f$or further improvements in guardrail frameworks and provide insights for researchers and practitioners to support the selection of appropriate solutions for enhancing harm mitigation in LLMbased applications.

This research paper describes a controlled quasi-experiment conducted with 24 distance-learning undergraduate students on teaching introduction concepts of Experimentation in Software Engineering (ESE) with different materials and procedures. The use of experimentation is a de facto way of obtaining data to statistically answer assumptions or concerns, especially in recent decades in the Software Engineering area. The ESE teaching-learning process in undergraduate and graduate programs still lacks evidence of its effective contribution. For instance, no study explores different ways of giving classes, such as traditional in-person or distance education (online). Our focus in this research is answering the following research question: “Is there any difference in applying three different sets of materials in teaching and learning introduction concepts of experimentation in Software Engineering for distance-learning undergraduate students?”. We used three different combinations of materials during learning (i.e., only slides, only recorded class, and both). We conducted activities with such materials in a higher education institution. Despite being incipient, the results demonstrate that in each group/profile of enrolled students, the teaching-learning process mediated by hybrid content (i.e., both slides and recorded classes) had no significant difference concerning students who studied only with recorded classes or slides. The results were obtained using descriptive statistics and inferential (hypothesis) tests. Our contributions here offer evidence-based statistical analysis for ESE education through distance learning, thereby enriching current academic research regarding distance education and facilitating knowledge dissemination in the field.

[Background:] Experimentation in Software Engineering (ESE) has increased in the last several years as a way to provide evidence by using statistical techniques. These techniques contribute to an auditable and reliable body of knowledge towards evolving a given topic. Therefore, teaching ESE becomes an essential task while disseminating and establishing an experimental culture to both academia and industry. [Aims:] In this paper we seek to understand how ESE has been taught including contents, materials used, strategies employed, and evaluation method applied. We also provide a research agenda on the subject. [Method:] We conducted a web-based questionnaire survey with 31 instructors who teach ESE. [Results:] We see several aspects, such as: ESE is mainly taught in an exclusive course; most of the materials used to teach ESE concepts are from third parties and some of them themselves; most people do not use any type of license; the core materials used to define the content of the courses are papers and books; learning management systems are most often used to share materials with students; key learning practices used are active learning, project-based learning, and problem-based learning; and most instructors assess their students with experimental projects and seminars. Based on such results, we provide and discuss a research agenda to improve teaching of ESE. [Conclusions:] This survey provides results towards planning a research agenda to improve teaching of ESE, thus benefiting instructors, researchers, and practitioners.

Open Science can be seen as a movement that has been spread out by the scientific community of all areas. In this movement, practices that seek to facilitate the sharing of research artifacts are considered. Possible artifacts include articles, data, scripts, and processes. In this paper, we present and discuss the results of a survey on open science carried out in the context of the State University of Maringá (UEM) in Brazil. Such a survey is aimed at investigating the degree of knowledge about open science from lecturers who supervise Master’s degree students and PhD candidates. The university has currently 54 graduate programs, distributed in different centers, encompassing almost 900 lecturers. We collected data using a web questionnaire with 22 questions. In total, 90 lecturers answered our survey. Results show that a significant subset of respondents never heard about open science, whereas the complementary subset barely dealt with the open science principles, tools or license types. We then provide in this paper a set of assumptions on several open science-related subjects. In addition, this paper might be used to guide any other university to measure the degree level of open science knowledge and to provide a plan to inspire the institutionalization of such an extremely relevant scientific topic.

Literature faces the lack of studies relating which characteristics of the Scrum framework are adapted. Understanding such variations is useful for prospective software development projects and guiding teams at conducting Scrum customizations. We aimed at identifying how the Scrum framework has been adapted to the context of Agile software development projects and how adaptations might be represented to aid researchers and practitioner at analyzing Scrum processes deployed or to be deployed. We carried out a systematic mapping study in five electronic sources, 11 journals and 15 conferences/workshops. We submitted the 281 returned studies to various filters, which resulted in 50 studies with data extracted, analyzed, and quality evaluated. SMS provides a panorama on the Scrum characteristics adapted to roles, events, and artifacts. We decided to adopt feature models for hierarchically accommodating found Scrum adaptations as it supports adaptations in the form of variability. We evaluated the resulting feature model with practitioners from different companies in the perspective of Perceived Usefulness and Perceived Ease of Use considering the Technology Acceptance Model (TAM). Therefore, we demonstrated the produced feature model aids users to better visualize and understand the documented Scrum adaptations. The panorama on Scrum adaptations and the problems during Scrum adoption are discussed to providing a means to practically understand and tailor (configure) such adaptations. Such adaptations are an essential source of information on the variety of Scrum elements, thus researchers and practitioners may take the results of this work as a guide to understand how different adaptations occur in different contexts during software development. In addition, the conceived feature model is an important asset to guide such users at selecting Scrum characteristics and respective adaptations to perform. The feature model also promotes reuse of knowledge gathered up from several different information sources.

A Software Product Line (SPL) represents a set of systems sharing common and variable features. The varying features and respective elements (variability) enable differentiating products of a certain domain. Thus, managing variability is a crucial activity for the success of SPL engineering, especially those based on UML due to a large amount of variability representation in different diagrams. There are few experiments in the literature to evaluate and to compare UML-based variability management approaches. In this paper, we analyze a subset of such approaches: SMarty (our approach), PLUS, and Razavian and Khosravi. We empirically compared them by conducting an experiment with more than 50 participants in terms of configuring SPL products and variability traceability among class and component diagram variable elements. We also analyzed the influence of the participants knowledge on the use of each approach and the amount of material consultation required for each variability management approach. In addition, we checked whether participants comprehend traceability capabilities of approaches. Results pointed out: SMarty is as effective as other studied approaches to configuring SPL specific products; the number of consultations on each approach instructional material did not influence effectiveness; and SMarty needs more participants previous knowledge on UML to configure SPL products.

With the continuous improvement of software processes, it is possible to increase quality, to address different application domains and accelerate the development of software products. However, existing process modelling notations like the SPEM meta‐model typically do not have appropriate constructs for expressing process variability. Thus, SPEM‐based approaches such as SMartySPEM and vSPEM provide mechanisms for representing variabilities to address characteristics of different projects. Here, the authors empirically compared SMartySPEM and for variability representation in software processes models, aiming to analyze correction, time, and efficiency. The performed experiment provides preliminary evidence based on 11 participants and sample size N = 44. Participants took longer to comprehend diagrams in SMartySPEM. However, the correctness of SMartySPEM diagrams had a superior result and higher efficiency based on the median values and hypothesis tests. With regard to variability mechanisms, the diagrams modelled with SMartySPEM had a slightly lower efficiency and it took longer for modifying them, thus the correctness of the SMartySPEM diagrams had a superior result. Therefore, an initial body of knowledge indicated a positive efficiency of SMartySPEM for variability representation in process models, as well as potential improvements as the reduction in SMartySPEM diagrams complexity, which may contribute to its evolution.

Smarty is a variability management approach for UML-based software product lines. It allows the identification, representation and tracing of variabilities in several UML models by means of an UML profile, the Smarty Profile, and a systematic process, the Smarty Process, with guidelines to provide user directions for applying such a profile. The existing UML-based variability management approaches in the literature, including Smarty, do not provide empirical evidence of their effectiveness, which is an essential requirement for technology transfer to industry. Therefore, this paper presents empirical evidence of the Smarty approach at class level. In addition, this paper demonstrates how Smarty has evolved, by means of its profile and guidelines, based on the obtained results of an experiment and the subjects feedback analysis.

This paper presents a customization of a learning design approach, OULDI, to designing and implementing Software Engineering courses. We propose an iterative process for the application of the OULDI views. This process starts with a course map view and follows a series of steps that ends with the evaluation of the design reflecting on the balance of the proposed activities. A case study is presented in which two institutions were involved in the design and implementation of an experimental software engineering course. Feedback from students, designers and lecturers was collected to support the validation of the design and implementation of the course. This showed that learning design, with the process proposed here, is a feasible approach for the design of software engineering courses.

The increasing presence of mobile devices in the society has motivated the use of these gadgets in several segments. This reality has contributed to the emergence of a new and innovative modality of learning - the mobile learning, or simply, m-learning. In short, mobile learning is based on the use of a set of specific mobile applications, which possesses similar features and specific characteristics according to the learning goals. In a different but related perspective, the emerging reuse technique of Software Product Line (SPL) allows mass customization and systematic derivation of products, such as mobile learning applications. Motivated by this scenario, we have worked on the establishment of M-SPLearning, a SPL to the mobile learning applications domain. M-SPLearning has been developed throughout a proactive adoption model, according to the basics of Service-Oriented Architecture (SOA). In this paper, we discuss the main aspects of the development of M-SPLearning, focusing on the implementation phase. Also, as a case study, we illustrate two products generated by the SPL proposed. The main results obtained suggest the practical feasibility of adopting M-SPLearning in the development of mobile learning applications.

A product-line architecture (PLA) entails a design reused by a family of products sharing several features. The long-term stability of a PLA largely depends on the modularization of each feature from the design outset. As many features of a product-line crosscut the PLA decomposition, their modularity and stability are often improved with aspect-orientation according to recent studies. However, the use of this technique in proactive design of PLAs is challenging and has rarely been investigated. The problem is that information about the crosscutting nature of features is limited in this case. This paper presents a first study aimed at analyzing the feasibility of deriving stable aspect-oriented PLAs through a proactive design method. Our study relies on a product line and presents an in-depth exploratory analyse. We compare the stability of aspect-oriented and alternative candidates for proactively-designed PLAs. The results indicate that it is possible to derive aspect-oriented PLAs with superior stability based on a proactive design method. In addition, it is possible to anticipate pros and cons of PLAs with aspects when the latter are incorporated at an early design stage.

To modernize legacy software into adaptable software, a program understanding procedure is needed to study software for identifying the mechanisms that support adaptability. Though the procedure can benefit from modeling techniques, current software meta-models do not fully support particular aspects of software adaptability. The goal of this research is to develop a meta-model, which facilitate comprehending applications for adaptability by annotating a set of pre-determined adaptability factors in software models. To this end, we investigate application-level sensing and effecting mechanisms, identify the core adaptability factors, and propose a meta-model for adaptability.