Teaching approaches supported by Information and Communication Technologies, such as E-Learning, can provide valuable information from Learning Management Systems that generate real-time data about student interactions online. These data might be relevant to fit predictive models that can support educational institutions in making decisions and improving educational strategies. Based on this, several studies have proposed machine learning-based approaches to analyze students' academic performance, in particular, by using multilevel regression models that capture contextual effects and provide an analytical structure that supports the researcher in analyzing the model's behavior concerning the impact factors. Although the contributions on this topic have been quite significant, this article proposes an analytical approach that combines cluster analysis with a multilevel logistic regression model for the E-learning context. In this sense, the cluster analysis provided the formation of non-directly observable groups that proved to be significant in composing the multilevel structure of the logistic regression model. The study results support the researcher in analyzing the behavior of impact factors on academic performance considering the observed clusters. The analysis of the model's predictive capacity was also discussed, as well as considerations to direct future work.

This paper presents a systematic review of the use of multilevel models for the analysis and prediction of school dropout. Several studies were carried out in this theme, but there are still challenges to be addressed. There are many different applications of multilevel modeling for school dropouts, which makes it difficult to synthesize the main contributions and advances in the area. The lack of a holistic view makes it difficult to understand the main advances and research gaps. To shed some light on this scenario, this literature review covered the most investigated factors at the student and school levels, such as demographic, socioeconomic, family background, and student’s academic performance variables; the main educational environments in which multilevel models were used for the analysis or prediction of school dropout, such as high school/secondary education, and higher education; and the main multilevel models used in these researches, such as the multilevel logistic regression, and the multilevel linear regression. In addition, we also investigated whether the authors used multivariate exploratory techniques or other artificial intelligence techniques to support the fitting and interpretation of the modeling process.

This research full paper presents a systematic mapping on quality attributes and quality models for Open Educational Resources (OER). The OER movement is part of a greater trend towards openness in education. They are represented by teaching materials that are available at no cost to the community to retain, reuse, review, remix, and redistribute. Given the wide availability of OER, how can one guarantee its quality? Motivated by this issue, some initiatives have proposed quality attributes or quality models to design, evaluate, and improve the quality of OER. However, to the best of our knowledge, a complete and detailed overview of how quality is being treated for OER is lacking. Besides, the literature lacks consensus on which quality attributes and quality models are the most relevant and which ones could be more suitable for OER. Also, there is an absence of a comprehensive analysis of the approaches used to establish quality attributes for OER. For this purpose, we carried out a systematic mapping addressing the following research questions: i) what are the quality models or quality attributes proposed for OER?; ii) how have these quality models or quality attributes been established for OER?; iii) how have quality models or quality attributes for OER been evaluated?. We identified quality models for different OER contexts, such as OCW and MOOC, in addition to quality attributes that are important for OER openness characteristics (i.e., retain, reuse, remix, revise, and redistribute). In order to contribute to the community, we presented discussions and highlighted some points and directions for future research.

Previous studies mention that students have a hard time learning introductory programming courses. Several studies have already been conducted to understand the problems concerning programming learning. Some investigations managed to classify the problems, they are: some students are unable to obtain a concrete understanding computer programming concepts; some students are not able to apply programming concepts in the construction of programs; some students have no motivation to learn the subject; some students cannot understand the programs already implemented, and some have difficulties in factoring and refactoring programs. To support the learning of these courses, researchers in the field have established teaching support mechanisms, which can be used by teachers when teaching content or by students to train and get prepare for assessments. In particular, some studies are proposing open educational resources. However, even if there is a strong interest in establishing these open resources, the existing educational resources do not consider the difficulties that the students may have in these courses (i.e., resources that increase students' motivation but do not support students who have difficulties learning programming concepts). Still, it is necessary to consider the previous knowledge that each student may have when starting a course. Some students may have some programming knowledge, while others may not. It is also necessary to consider that students have their preferences for learning materials when they are studying (i.e., some prefer video lessons, while others prefer slides). In this sense, this paper addresses the establishment and feasibility study of an open educational resource dedicated to teaching programming, which is specialized for students who have difficulties in learning and applying concepts, understanding programs, factoring and refactoring their programs and/or do not have the motivation to program. This open educational resource was planned considering some definitions related to the students' particularities (i.e., previous knowledge, preferences for teaching materials and programming difficulties), with the function of identifying them and adapting to these particularities. We describe how we have developed the educational resource, how we plan and conduct preliminary evaluations and, in the end, we raise directions for next steps.

Conversational agents have been established to support educational practices, solving students' questions, recommending teaching materials, increasing the motivation of students who have little interest in certain content, among others. Although many studies are discussing the establishing of conversational agents, there is no global understanding of what has been investigated in this area. There are some previous secondary studies that have attempted to map the conversational agents used for educational purposes (i.e., pedagogical conversational agents). However, these studies are limited to include primary studies published in specific sets of conferences and journals. Therefore, an overview of the current state of the art associated with pedagogical conversational agents has not been produced yet. This research paper aims to contribute to the theme, through the presentation of planning, conduction, and results analysis of a systematic mapping on pedagogical conversational agents. We address the following questions: (i) in which areas have conversation agents been investigated?; (ii) at which education levels are conversational agents used?; (iii) which mechanisms do conversational agents use to interact with students?. We also present some discussions about research opportunities that can be explored in future works. This study aims to contribute to the topic of pedagogical conversational agents, helping researchers to have a deeper understanding of what has already been done and to guide their research to subjects not explored yet.

This abstract is related to a work in progress and the category is research. The continuous search for methods to spur students' motivation and concentration at the diverse array of academic study fields has brought the need for innovations in the educational environment, in this context, it emerges the gamification concept. This work aims to identify the concept of gamification and its applicability specifically geared towards Engineering courses. To achieve this goal, a bibliographic review was carried out in two stages. The first one looked for studies that addressed the definition of techniques based on games in general, their relation with the teaching process and possibilities of use in educational environments, whether virtual or not. In a second moment, the work listed several job prospects of gamification directed specifically to the Engineering context. For this purpose, several repositories were mapped and 38 scientific papers were selected. Through the mapping of the literature it was possible to obtain a knowledge base on the subject and, from this, to trace the potentialities not yet fully explored in which one can glimpse a field for the continuity of the research.

This Research to Practice Full Paper introduces an approach for the early identification of students at risk of dropping out of their distance learning courses. Students dropout in university courses have long been a major issue leading to social, academic and financial impacts. Predictive modelling analysis has been used as a tool to identify at an early stage probable cases of dropout. In distance learning, this type of analysis is made possible mostly because of the increasing adoption rate of Virtual Learning Environments (VLEs) where data on the student academic activities can be continuously recorded. This paper discusses the design and validation of a Learning Analytics system for early identification of students at risk of dropping out. The case study presented relies on data collected from two postgraduate courses as part of Brazils Open University. The methodology for the system design and validation comprises the steps to build an intelligent data pipeline. Jupyter Notebooks have been d used as the data science analysis environment in order to create data pipelines and have their performance evaluated. Results obtained from the validation of models built out of eight machine learning techniques show an average accuracy of 84% among the set of ML techniques tested. The highest accuracy is delivered by the Extra Trees classifier with 88%. Logistic Regression performed the worst performance with an accuracy of 79%.

This abstract refers to a full paper in the research category. This paper presents a Systematic Review (SR) that aims at characterising the current scenario of Educational Recommendation Systems (ERS). ERS has recently attracted the attention of researchers in developing methods that contribute to the educational environment. Recommendation systems capable of searching and selecting relevant contents are one of the core technological resources offered to the education sector. Such systems which were initially designed for commercial applications (e.g. e-commerce) have now been used in educational application-driven research. For instance, a teacher might rethink methodologies and practises tailored to a set of pedagogical needs. These resources are enablers for the tools development that contribute to better educational practices. The goal of this SR is to provide an analysis of the current research efforts in the field of educational recommendation systems. With the string search of ”educational support” AND “recommendation system” used in four academic databases (Brazil and International) 2016 and 2018. The SR returned 439 papers of which 54 were selected for further analysis. The results show that the profile-based filtering technique for recommendation is the most used, occurring 27.78% of the studies. In 44.44% of these studies, the target audience has been the students. Furthermore, the usual evaluation method was experimental achieving 41% of the selected papers. The results demonstrate applications in learning situations as the recommendation of digital resources, related to learning styles and based on ability and performance.

Requirements extraction is an important element of the software development process. One of the most used techniques for requirements extraction is the interview. Initiatives to support the training and technical training of computing students in this area are being proposed, such as the development of support mechanisms. These initiatives are proposed by the fact that computing students are graduating with limited practical knowledge in requirements extraction. In parallel, chatterbots have been investigated as tools with the capacity to support the training of students from different areas of knowledge, since the main characteristic is verbal conversational behavior. In medicine, for example, they can take on the role of a sick patient to train students to extract information about the patient's symptoms. One subject that has been explored in the context of educational chatterbots is context awareness, so that the chatterbot can present the right information for the right user. These surveys start from the premise that not every student has the same knowledge as their peers on the subject. Thus, in this research work in full paper we describe a chatterbot that offers support to Software Engineering Education, focusing mainly on the requirements extraction, which assumes the role of a stakeholder. A prototype of a chatterbot that is sensitive to student's context is presented, as well as preliminary results on the impact of this support mechanism in Software Engineering Education.