The main topic of this paper is to find ways to improve learning in a formal Higher Education Area. In this environment, the teacher publishes or suggests contents that support learners in a given course, as supplement of classroom training. Generally, these materials are pre-stored and not changeable. These contents are typically published in learning management systems (the Moodle platform emerges as one of the main choices) or in sites created and maintained on the web by teachers themselves. These scenarios typically include a specific group of students (class) and a given period of time (semester or school year). Contents reutilization often needs replication and its update requires new edition and new submission by teachers. Normally, these systems do not allow learners to add new materials, or to edit existing ones. The paper presents our motivations, and some related concepts and works. We describe the concepts of sequencing and navigation in adaptive learning systems, followed by a short presentation of some of these systems. We then discuss the effects of social interaction on the learners’ choices. Finally, we refer some more related recommender systems and their applicability in supporting learning. One central idea from our proposal is that we believe that students with the same goals and with similar formal study time can benefit from contents' assessments made by learners that already have completed the same courses and have studied the same contents. We present a model for personalized recommendation of learning activities to learners in a formal learning context that considers two systems. In the extended content management system, learners can add new materials, select materials from teachers and from other learners, evaluate and define the time spent studying them. Based on learner profiles and a hybrid recommendation strategy, combining conditional and collaborative filtering, our second system will predict learning activities scores and offers adaptive and suitable sequencing learning contents to learners. We propose that similarities between learners can be based on their evaluation interests and their recent learning history. The recommender support subsystem aims to assist learners at each step suggesting one suitable ordered list of LOs, by decreasing order of relevance. The proposed model has been implemented in the Moodle Learning Management System (LMS), and we present the system’s architecture and design.

Memory-based algorithms are the most popular among the collaborative filtering algorithms. They use as input a table containing ratings given by users to items, known as the rating matrix. They predict the rating given by user a to an item i by computing similarities of the ratings among users or similarities of the ratings among items. In the first case Memory-Based algorithms are classified as User-based algorithms and in the second one they are labeled as Item-based algorithms. The prediction is computed using the ratings of k most similar users (or items), also know as neighbors. Memory-based algorithms are simple to understand and to program, usually provide accurate recommendation and are less sensible to data change. However, to obtain the most similar neighbors for a prediction they have to process all the data which is a serious scalability problem. Also they are sensitive to the sparsity of the input. In this work we propose an efficient and effective Item-Based that aims at diminishing the sensibility of the Memory-Based approach to both problems stated above. The algorithm is faster (almost 50%) than the traditional Item-Based algorithm while maintaining the same level of accuracy. However, in environments that have much data to predict and few to train the algorithm, the accuracy of the proposed algorithm surpass significantly that of the traditional Item-based algorithms. Our approach can also be easily adapted to be used as User-based algorithms.

Recommendation systems function as a guide, helping users to discover products of interest. There are various techniques and approaches in the literature that enable the generationofrecommendations.Thisisinterestingbecauseitemphasizesthediversityof options;ontheotherhand,itcancausedoubtthesystemdesigneraboutwhichisthebest techniquetouse.Eachoftheseapproacheshasparticularitiesanddependsonthecontext to be applied. Therefore, the decision to choose between the techniques is complex to be done manually. This work proposes an evolutionary approach for combining results of recommendation techniques (Invenire) in order to automate the choice of techniques and get fewer errors in recommendations. To evaluate the proposal, experiments were performed with a dataset from MovieLens and some Collaborative Filtering techniques. The results show that the combining methodology proposed in this paper performs better than any one collaborative filtering technique separately in the context addressed. The improvement varies from 3,6% to 118,99% depending on the technique and the experiment executed.

The internet advances have led to an increase of data and information availability. This overload of information tends to compromise the capacity to manage and filter the available data. In the health domain, the increasing digitalization in healthcare led to a substantial rise of the recorded data. Various recommendation systems (RS) have been developed to help healthcare professionals integrate all information and make efficient and effective decisions. Here, a preliminary RS based on collaborative filtering is proposed to reduce the time that healthcare professionals spend in registering medical items consumed during patients’ hospitalization. For that purpose, the RS was built to perform suggestions of the medical items and respective quantities needed in the first day of hospitalization of a patient. Data regarding the diagnostics, surgical procedures and medical item records associated to surgeries of inpatients during a period of one year in Hospital da Luz Lisboa was filtered, restructured, and analysed (N = 5088 surgeries) for the construction of the RS. A 75-25% split of the data was considered with a 4-fold cross-validation procedure applied on the train set to tune the hyperparameters settings for the algorithm. The RS was then tested and evaluated regarding its overall performance in terms of accuracy, classification performance, and coverage. The same measures were applied to assess the quality of the recommendations for each medical specialty of the hospital. Furthermore, the trust of healthcare professionals in the RS was also assessed. A moderate overall performance was achieved (precision = 0.608, recall = 0.729, F1-Measure = 0.663, RMSE = 6.901) and the quality of the algorithm’s recommendations varied between medical specialties. Additionally, the algorithm presented higher values of precision, recall and F1-Measure in the predictions of the most frequently registered medical items in the test set, which corresponded to approximately 85% of the consumptions in the first day of hospitalization. Regarding the coverage of the RS, approximately 80% of the medical items used in the test set were never recommended by the algorithm, corresponding to only 5.57% of the consumptions. Lastly, although in the point of view of hospital’s nurses there is some trust in the RS results, several suggestions were given for further improvements of the algorithm. Despite the limitations of the RS, the observed results represent a starting point for the development of a tool that can support healthcare professionals of Hospital da Luz Lisboa in registering medical items needed during inpatients’ hospitalization.

O crescimento exponencial da informação disponível na Web torna difícil para os utilizadores a tarefa de obter a informação que pretendem e quando dela necessitam. Para ultrapassar o problema, os sítios Web têm vindo a incorporar sistemas de recomendação que, baseados no histórico de acessos, têm como objetivo maximizar a satisfação dos utilizadores, disponibilizando-lhes recomendações de alta qualidade. A complexidade do problema e a natureza distribuída da Web justificam abordagens baseadas na tecnologia dos agentes inteligentes autónomos e sistemas multiagente, permitindo combinar múltiplos algoritmos de recomendação, aumentando assim as hipóteses das recomendações sugeridas serem efetivamente do interesse do utilizador. É este o tipo de abordagem explorada pelo sistema de recomendação multiagente AMAAFWA (A Multi-Agent Approach for Web Adaptation) (Morais, 2013). Os testes realizados em modo offline mostraram que essa abordagem multiagente, baseada em agentes implementando diferentes algoritmos, apresenta um desempenho superior ao dos algoritmos considerados individualmente. O objetivo desta dissertação é adaptar e testar o sistema AMAAFWA em tempo real, com o objetivo de validar os resultados obtidos em modo offline, pelo que se procedeu à sua adaptação para funcionamento online, integrando-o num sítio Web. O sistema AMAAFWA baseia-se numa classificação implícita dos itens e os algoritmos de recomendação são baseados em memória e incrementais. Foi também criada e testada uma versão do sistema que considera uma classificação explícita dos itens por parte dos utilizadores, com o propósito de comparar o desempenho de ambos os tipos de classificação. Demonstra-se na presente dissertação que o sistema de recomendação multiagente AMAAFWA, em funcionamento online, apresenta um desempenho superior ao dos algoritmos considerados individualmente, sendo ainda capaz de melhorar a satisfação do utilizador e contribuir para o aumento do sucesso do sítio Web em que se insere. Relativamente à comparação dos tipos de classificação implícita e explícita dos itens, os resultados mostram desempenhos similares.

Historicamente, as pessoas têm recorrido a amigos próximos ou experts para ajudar na tomada de decisão ou em recomendações sobre assuntos das mais diversas áreas. O crescimento da era digital nas últimas duas décadas, principalmente na web criou um overload de informação. No entanto, a nossa capacidade de avaliar as especificações de cada produto e escolher entre as enumeras alternativas existentes no mercado online, é limitada. Neste sentido, a ciência e tecnologia tem reagido adequadamente através do desenvolvimento de ferramentas para aliviar essa limitação. Os sistemas de recomendação são exemplos dessas ferramentas, que surgiram em meados dos anos 90 e têm tido muito sucesso. Os sistemas de recomendação funcionam como um atendimento personalizado. Numa situação de atendimento presencial apenas é possível apurar a pretensão do cliente após este a ter descrito. Estes sistemas visam optimizar e indicar a opção mais ajustada de acordo com o perfil do próprio cliente. Uma boa recomendação poderá ir ao encontro das pretensões de um determinado utilizador quando confrontado com uma plataforma e-commerce, o que faz com que o rácio utilizador/compra aumente. Além disso, quantos mais utilizadores satisfeitos com o sistema, mais popular e melhor o sistema se tornará, além de que será criada uma relação de proximidade entre o utilizador e o website. Existem vários métodos para gerar recomendações personalizadas para utilizadores específicos, mas o estado da arte de sistemas de recomendação são baseados na filtragem colaborativa. Estes sistema são amplamente utilizados por empresas como a Amazon, Netflix entre outras e tem obtido resultados muito significativos sem a necessidade de extrapolar as características dos artigos. Métodos ligados ao Deep Learning levaram a um elevado progresso em vários campos da Inteligência Artificial e, nos anos mais recentes, a um substancial número de propostas para melhorar sistemas de recomendação com Redes Neuronais Artificiais. Nesta dissertação, realizada na empresa Kodly Consulting, propõem-se a criação de uma solução baseada em redes neuronais recorrentes para geração de recomendações em ambientes e-commerce. Para que o sistema proposto seja facilmente integrado em qualquer aplicação e-commerce, pretende-se criar uma API que disponibiliza um conjunto de serviços úteis e fáceis de integrar.

In order to provide tailored recommendations to a user, a Recommendation System makes use of a preference elicitation process whenever a new user is registered in the system. We propose that this task is best achieved not by the classic method, in which users first express their preferences for individual items, but instead by expressing preferences for groups of items, since the classic method inefficiently converts a user’s effort into a personalized profile. We tested this idea by developing and evaluating an interactive process, where users express preferences through groups of items that are automatically generated by clustering algorithms (clustering). This strategy of recommendation can be applied to any collaborative filtering based system. We assess our process, both with offline simulation methods, using the MovieLens data set; and with an online experiment with 312 users. Our evaluation reveals pros and cons associated with moving from preference elicitation per item to preference elicitation per group. The user experiments showed that the top-N recommendations list generated by our proposal contains more itens which users may be interested than the classic method, and that a greater number of groups positively impacts prediction accuracy, at the expense of greater user effort. Furthermore, we have found that, in comparison with a baseline interface of 15 items, users are able to complete the preference elicitation process in less than half the time.

In the current scenario, distance and even face-to-face education has been using virtual learning environments as a tool. However, generally these tools do not adapt according to the student, as is the case with Moodle. In order to solve the problem for the so-called Moodle evaluation questions, this work proposes and develops a REST API to generate recommendations using Bloom’s Revised Taxonomy, collaborative user-based filtering and also an efficiency module. The recommender used data from an LC-EAD course at UFPB and the results obtained were recommendations for a target user. Therefore, a use of this service can bring benefits to students, who use suggested activities according to their profile and cognitive level, and can help in the learning and and consequently have a reduction in dropout.

The flowcharts are built based on the (NDE) which responsible for the conception of the pedagogical project of a course and may vary depending on the recommendations of NDE itself that uses criteria such as: the minimum amount of credits per period, the duration. of course, the dependencies between disciplines. The flowchart construction stage is of importance, as poorly formulated flowcharts may contain poorly distributed subjects, increasing the time of completion of the course, leading to higher financial expenses for universities. Students have uncertainties in choosing an ideal set of subjects over a period, as they do not know a priori the difficulties they will face in each of them. Given this, the need arises for the development of custom flowchart recommendation systems. The present work proposes the development of a personalized flowchart recommendation based on a recommendation algorithm by the collaborative filtering technique with the use of graphical flowchart visualization. Colors (red, yellow and green) will be used to alert the student about the difficulty level of each discipline. In this way, one can make an association between the predictions of grades and the levels of difficulties for each of them. The following parameters were used: Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and the confusion matrix analysis to test the efficiency of the recommender. The flowchart recommender was tested with a set of 298 active students from the Computer Engineering course at UFPB, making subject recommendations for all periods of the course, resulting in an RMSE grade prediction of 1.74 and the MAE of 1.33. In addition, 88.71 % accuracy, 98.52 % specificity and 55 % sensitivity were obtained. As possible results of this proposal, it is expected to reduce the risk of failure in the subjects, since the student will have, in advance, flowchart information with easy and intuitive visualization about the difficulties of the subjects through a colored flowchart, so plan your dedication better to each of them.