M. Tech. (Department of Information and Communication Technology, Faculty of Applied and Computer Sciences), Vaal University of Technology. / Since their emergence in the 1990’s, recommendation systems have transformed the intelligence of both the web and humans. A pool of research papers has been published in various domains of recommendation systems. These include content based, collaborative and hybrid filtering recommendation systems. Recommendation systems suggest items to users and their principal purpose is to increase sales and recommend items that are predicted to be suitable for users. They achieve this through making calculations based on data that is available on the system. In this study, we give evidence that the research on group recommendation systems must look more carefully at the dynamics of group decision-making in order to produce technologies that will be more beneficial for groups based on the individual interests of group members while also striving to maximise satisfaction. The matrix factorization algorithm of collaborative filtering was used to make predictions and three movie recommendation for each and every individual user. The three recommendations were of three highest predicted movies above the pre-set threshold which was three. Thereafter, four virtual groups of varied sizes were formed based on four highest predicted movies of the users in the dataset. Plurality voting strategy was used to achieve this. A publicly available dataset based on Group Recommender Systems Enhanced by Social Elements, constructed by Lara Quijano from the Group of Artificial Intelligence Applications (GIGA), was used for experiments. The developed recommendation system was able to successfully make individual movie recommendations, generate virtual groups, and recommend movies to these respective groups. The system was evaluated for accuracy in making predictions and it was able to achieve 0.7027 MAE and 0.8996 RMSE. This study was able to recommend to virtual groups to enable social network group members to engage in discussions of recommended items. The study encourages members in engaging in similar activities in their respective physical locations and then discuss on social network.

Dans cette thèse, nous proposons des algorithmes de recommandation sémantique et sociale, qui recommandent un produit pour les utilisateurs qui sont connectés par un réseau de collaboration sociale. Ces algorithmes utilisent deux types d'informations : information sémantique et information sociale .L' information sémantique est basée sur la pertinence sémantique entre les utilisateurs et le produit à recommandé, tandis que l' information sociale est basée sur la position de l'utilisateur et de leur type et de la qualité des connexions entre eux dans le réseau de collaboration . Enfin, nous utilisons l'algorithme de parcoure profondeur (DFS) et l'algorithme de parcoure en largeur (BFS), pour explorer le réseau social.Utilisation de l' information sémantique et l'information sociale , dans le système de recommandation , nous aide à explorer partiellement le réseau social , ce qui nous conduit à réduire la taille des données explorées et de minimiser le temps de recherche dans le réseau.Nous appliquons nos algorithmes sur des données réelles : MovieLens et Amazon , et nous comparons la précision de la performance de nos algorithmes avec les algorithmes de recommandation classiques , comme l'algorithme de filtrage collaborative et l'algorithme hybrideNos résultats montrent un taux de précision satisfaisants , et une performance très significative du temps d'exécution et de la taille des données explorées , par rapport aux autres algorithmes de recommandation classiques .En fait , l'importance de nos algorithmes repose sur le fait que ces algorithmes explorent une très petite partie du graphe , au lieu d'explorer tout le graphe que les méthodes de recherche classiques , et encore donnent une bonne précision par rapport aux autres algorithmes de recommandation classiques . Donc , en minimisant la taille des données recherchées n'influence pas mal la précision des résultats . / In this thesis we propose semantic-social recommendation algorithms, that recommend an input item to users connected by a collaboration social network. These algorithms use two types of information: semantic information and social information.The semantic information is based on the semantic relevancy between users and the input item; while the social information is based on the users position and their type and quality of connections in the collaboration social network. Finally, we use depth-first search and breath-first search strategies to explore the graph.Using the semantic information and the social information, in the recommender system, helps us to partially explore the social network, which leads us to reduce the size of the explored data and to minimize the graph searching time.We apply our algorithms on real datasets: MovieLens and Amazon, and we compare the accuracy an the performance of our algorithms with the classical recommendation algorithms, mainly item-based collaborative filtering and hybrid recommendation.Our results show a satisfying accuracy values, and a very significant performance in execution time and in the size of explored data, compared to the classical recommendation algorithms.In fact, the importance of our algorithms relies on the fact that these algorithms explore a very small part of the graph, instead of exploring all the graph as the classical searching methods, and still give a good accuracy compared to the other classical recommendation algorithms. So, minimizing the size of searched data does not badly influence the accuracy of the results.

O número de atividades disponibilizadas pelos sistemas gerenciadores de workflows científicos é grande, o que exige dos cientistas conhecerem muitas delas para aproveitar a capacidade de reutilização desses sistemas. Para minimizar este problema, a literatura apresenta algumas técnicas para recomendar atividades durante a construção de workflows científicos. Este projeto especificou e desenvolveu um sistema de recomendação de atividades híbrido, considerando informação sobre frequência, entrada e saídas das atividades, e anotações ontológicas para recomendar. Além disso, neste projeto é apresentada uma modelagem da recomendação de atividades como um problema de classificação e regressão, usando para isso cinco classificadores; cinco regressores; um classificador SVM composto, o qual usa o resultado dos outros classificadores e regressores para recomendar; e um ensemble de classificadores Rotation Forest. A técnica proposta foi comparada com as outras técnicas da literatura e com os classificadores e regressores, por meio da validação cruzada em 10 subconjuntos, apresentando como resultado uma recomendação mais precisa, com medida MRR ao menos 70% maior do que as obtidas pelas outras técnicas / The number of activities provided by scientific workflow management systems is large, which requires scientists to know many of them to take advantage of the reusability of these systems. To minimize this problem, the literature presents some techniques to recommend activities during the scientific workflow construction. This project specified and developed a hybrid activity recommendation system considering information on frequency, input and outputs of activities and ontological annotations. Additionally, this project presents a modeling of activities recommendation as a classification problem, tested using 5 classifiers; 5 regressors; a SVM classifier, which uses the results of other classifiers and regressors to recommend; and Rotation Forest , an ensemble of classifiers. The proposed technique was compared to other related techniques and to classifiers and regressors, using 10-fold-cross-validation, achieving a MRR at least 70% greater than those obtained by other techniques

Com o uso da tecnologia de Identificação por Radiofrequência (RFID), arquiteturas heterogêneas de processadores e as novas tendências da TV Digital e televisão via rede IP (IPTV) foi desenvolvido um sistema para montar, em tempo real, em forma automática, uma programação televisiva personalizada, baseada no perfil do grupo de usuários de um determinado televisor. Aplicações de vídeo sob demanda (VoD), IPTV e TV Digital permitem que cada telespectador possa assistir aos programas a qualquer momento, e assim construir sua grade de programação personalizada. Com um sistema de RFID é possível identificar as pessoas que se encontram próximas ao televisor. Com essas tecnologias unidas a um subsistema de análise de perfil, junto com os dados fornecidos pelos telespectadores no momento da contratação do serviço, e uma interface (middleware) para gerenciar os dados, é possível configurar um sistema que escolhe automaticamente quais programas e quais comerciais serão apresentados no aparelho de TV. Essa escolha é baseada no perfil dos telespectadores presentes naquele momento à frente da televisão e dos programas disponíveis naquele instante. As etiquetas (tags) de RFID usadas para o levantamento da audiência foram aparelhos celulares equipados com tecnologia Bluetooth, que possibilitam a identificação simultânea dos telespectadores via rádio. O algoritmo de recomendação é híbrido, possuindo componentes baseados em conteúdo e componentes colaborativos. O uso dos novos processadores heterogêneos exigiu o desenvolvimento de algoritmos paralelos que utilizam instruções do tipo SIMD, aceleradores e GPUs. Os sistemas que existem no momento (2011) nesta área, se limitam à identificação dos usuários mediante a digitação usando o controle remoto da TV e só identificam uma pessoa de cada vez. O uso de tecnologia por rádio, proposto nesta pesquisa, permite a identificação de várias pessoas simultaneamente, exigindo o desenvolvimento de padrões de um sistema completo baseado em grupos de perfis diferentes. A arquitetura do sistema elaborado está baseada no processador Cell BE e nas arquiteturas CPU+GPU, de forma que o tempo de execução do algoritmo fosse minimizado. / Merging together Radiofrequency identification (RFID), heterogeneous architectures of processors and new tendencies of the Digital TV (DTV) and television through IP network (IPTV), a system to create, automatically and in real-time, a personalized TV program schedule, based on the group of people profile next to a TV. Video-on-Demand (VoD) applications, IPTV and DTV allow each person to watch a chosen program at any moment and to its personalized programming guide. The RFID system allows the identification of the people next to the TV. This technology used with a profile analysis subsystem accessing a database of people preferences, and a middleware to manage the data, it is possible to set a system that automatically chooses with TV shows and with TV ads will be presented in the TV. This selection is based on the profile of the people next to the TV in that instant and on the available programs. The RFID tags used to detect the audience were the mobile phones equipped with Bluetooth, which allows the identification of its owner wirelessly. The recommendation algorithm is hybrid, containing collaborative and content-based components. The new heterogeneous processors demanded the development of parallel algorithms that use SIMD instruction, accelerators and GPUs. The systems that were available in the moment of this research (2011) were limited to the identification through login using remote control, one person by time. The use of RFID technology, proposed in this research, enables the simultaneous identification of many people at a time, demanding the development standards for group profiles recommendation. The systems architectures will be based on Cell BE processor and the conjunct CPU+GPU, focusing in the reduction of the algorithm execution time.

Com o uso da tecnologia de Identificação por Radiofrequência (RFID), arquiteturas heterogêneas de processadores e as novas tendências da TV Digital e televisão via rede IP (IPTV) foi desenvolvido um sistema para montar, em tempo real, em forma automática, uma programação televisiva personalizada, baseada no perfil do grupo de usuários de um determinado televisor. Aplicações de vídeo sob demanda (VoD), IPTV e TV Digital permitem que cada telespectador possa assistir aos programas a qualquer momento, e assim construir sua grade de programação personalizada. Com um sistema de RFID é possível identificar as pessoas que se encontram próximas ao televisor. Com essas tecnologias unidas a um subsistema de análise de perfil, junto com os dados fornecidos pelos telespectadores no momento da contratação do serviço, e uma interface (middleware) para gerenciar os dados, é possível configurar um sistema que escolhe automaticamente quais programas e quais comerciais serão apresentados no aparelho de TV. Essa escolha é baseada no perfil dos telespectadores presentes naquele momento à frente da televisão e dos programas disponíveis naquele instante. As etiquetas (tags) de RFID usadas para o levantamento da audiência foram aparelhos celulares equipados com tecnologia Bluetooth, que possibilitam a identificação simultânea dos telespectadores via rádio. O algoritmo de recomendação é híbrido, possuindo componentes baseados em conteúdo e componentes colaborativos. O uso dos novos processadores heterogêneos exigiu o desenvolvimento de algoritmos paralelos que utilizam instruções do tipo SIMD, aceleradores e GPUs. Os sistemas que existem no momento (2011) nesta área, se limitam à identificação dos usuários mediante a digitação usando o controle remoto da TV e só identificam uma pessoa de cada vez. O uso de tecnologia por rádio, proposto nesta pesquisa, permite a identificação de várias pessoas simultaneamente, exigindo o desenvolvimento de padrões de um sistema completo baseado em grupos de perfis diferentes. A arquitetura do sistema elaborado está baseada no processador Cell BE e nas arquiteturas CPU+GPU, de forma que o tempo de execução do algoritmo fosse minimizado. / Merging together Radiofrequency identification (RFID), heterogeneous architectures of processors and new tendencies of the Digital TV (DTV) and television through IP network (IPTV), a system to create, automatically and in real-time, a personalized TV program schedule, based on the group of people profile next to a TV. Video-on-Demand (VoD) applications, IPTV and DTV allow each person to watch a chosen program at any moment and to its personalized programming guide. The RFID system allows the identification of the people next to the TV. This technology used with a profile analysis subsystem accessing a database of people preferences, and a middleware to manage the data, it is possible to set a system that automatically chooses with TV shows and with TV ads will be presented in the TV. This selection is based on the profile of the people next to the TV in that instant and on the available programs. The RFID tags used to detect the audience were the mobile phones equipped with Bluetooth, which allows the identification of its owner wirelessly. The recommendation algorithm is hybrid, containing collaborative and content-based components. The new heterogeneous processors demanded the development of parallel algorithms that use SIMD instruction, accelerators and GPUs. The systems that were available in the moment of this research (2011) were limited to the identification through login using remote control, one person by time. The use of RFID technology, proposed in this research, enables the simultaneous identification of many people at a time, demanding the development standards for group profiles recommendation. The systems architectures will be based on Cell BE processor and the conjunct CPU+GPU, focusing in the reduction of the algorithm execution time.

O número de atividades disponibilizadas pelos sistemas gerenciadores de workflows científicos é grande, o que exige dos cientistas conhecerem muitas delas para aproveitar a capacidade de reutilização desses sistemas. Para minimizar este problema, a literatura apresenta algumas técnicas para recomendar atividades durante a construção de workflows científicos. Este projeto especificou e desenvolveu um sistema de recomendação de atividades híbrido, considerando informação sobre frequência, entrada e saídas das atividades, e anotações ontológicas para recomendar. Além disso, neste projeto é apresentada uma modelagem da recomendação de atividades como um problema de classificação e regressão, usando para isso cinco classificadores; cinco regressores; um classificador SVM composto, o qual usa o resultado dos outros classificadores e regressores para recomendar; e um ensemble de classificadores Rotation Forest. A técnica proposta foi comparada com as outras técnicas da literatura e com os classificadores e regressores, por meio da validação cruzada em 10 subconjuntos, apresentando como resultado uma recomendação mais precisa, com medida MRR ao menos 70% maior do que as obtidas pelas outras técnicas / The number of activities provided by scientific workflow management systems is large, which requires scientists to know many of them to take advantage of the reusability of these systems. To minimize this problem, the literature presents some techniques to recommend activities during the scientific workflow construction. This project specified and developed a hybrid activity recommendation system considering information on frequency, input and outputs of activities and ontological annotations. Additionally, this project presents a modeling of activities recommendation as a classification problem, tested using 5 classifiers; 5 regressors; a SVM classifier, which uses the results of other classifiers and regressors to recommend; and Rotation Forest , an ensemble of classifiers. The proposed technique was compared to other related techniques and to classifiers and regressors, using 10-fold-cross-validation, achieving a MRR at least 70% greater than those obtained by other techniques

This thesis proposes the design, development and evaluation of a hybrid video recommendation system. The proposed hybrid video recommendation system is based on a graph algorithm called Adsorption. Adsorption is a collaborative filtering algorithm in which relations between users are used to make recommendations. Adsorption is used to generate the base recommendation list. In order to overcome the problems that occur in pure collaborative system, content based filtering is injected. Content based filtering uses the idea of suggesting similar items that matches user preferences. In order to use content based filtering, first, the base recommendation list is updated by removing weak recommendations. Following this, item similarities of the remaining list are calculated and new items are inserted to form the final recommendations. Thus, collaborative recommendations are empowered considering item similarities. Therefore, the developed hybrid system combines both collaborative and content based approaches to produce more effective suggestions.

Recommendation systems are emerging as an important business application as the demand for personalized services in E-commerce increases. Collaborative filtering techniques are widely used for predicting a user's preference or generating a list of items to be recommended. In this thesis, we develop several new approaches for collaborative filtering based on model combination and kernel smoothing. Specifically, we start with an experimental study that compares a wide variety of CF methods under different conditions. Based on this study, we formulate a combination model similar to boosting but where the combination coefficients are functions rather than constant. In another contribution we formulate and analyze a local variation of matrix factorization. This formulation constructs multiple local matrix factorization models and then combines them into a global model. This formulation is based on the local low-rank assumption, a slightly different but more plausible assumption about the rating matrix. We apply this assumption to both rating prediction and ranking problems, with both empirical validations and theoretical analysis. We contribute with this thesis in four aspects. First, the local approaches we present significantly improve the accuracy of recommendations both in rating prediction and ranking problems. Second, with the more realistic local low-rank assumption, we fundamentally change the underlying assumption for matrix factorization-based recommendation systems. Third, we present highly efficient and scalable algorithms which take advantage of parallelism, suited for recent large scale datasets. Lastly, we provide an open source software implementing the local approaches in this thesis as well as many other recent recommendation algorithms, which can be used both in research and production.

In this thesis we propose semantic-social recommendation algorithms, that recommend an input item to users connected by a collaboration social network. These algorithms use two types of information: semantic information and social information.The semantic information is based on the semantic relevancy between users and the input item; while the social information is based on the users position and their type and quality of connections in the collaboration social network. Finally, we use depth-first search and breath-first search strategies to explore the graph.Using the semantic information and the social information, in the recommender system, helps us to partially explore the social network, which leads us to reduce the size of the explored data and to minimize the graph searching time.We apply our algorithms on real datasets: MovieLens and Amazon, and we compare the accuracy an the performance of our algorithms with the classical recommendation algorithms, mainly item-based collaborative filtering and hybrid recommendation.Our results show a satisfying accuracy values, and a very significant performance in execution time and in the size of explored data, compared to the classical recommendation algorithms.In fact, the importance of our algorithms relies on the fact that these algorithms explore a very small part of the graph, instead of exploring all the graph as the classical searching methods, and still give a good accuracy compared to the other classical recommendation algorithms. So, minimizing the size of searched data does not badly influence the accuracy of the results.

Orientadora: Profa. Dra. Itana Stiubiener / Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Engenharia da Informação, 2014. / A partir do uso das Tecnologias de Informação e Comunicação (TIC) no apoio ao processo de aprendizagem, as diversas ferramentas utilizadas com esta finalidade como: ambientes virtuais de aprendizagem (AVA), cursos online, portais educacionais costumam disponibilizar conteúdos e outros recursos educacionais por meio de repositórios de Objetos de Aprendizagem (OAs). São exemplos de OAs: atividades, avaliações, vídeos, dentre outros, ou seja, toda e qualquer entidade que possa ser utilizada para aprendizagem, educação ou treinamento. O uso de OAs em ambientes digitais pode enriquecer a aprendizagem dos estudantes, entretanto, a partir do aumento da disponibilidade de tais recursos, surge a necessidade de melhorar a busca e seleção desses itens nos diversos repositórios. Os estilos pessoais de aprendizagem, ou seja, o modo como os indivíduos absorvem, processam e transformam informação em conhecimento, possibilitam a recomendação personalizada de OAs, de maneira que o aluno acesse o recurso educacional mais adequado no seu processo de ensino-aprendizagem. Assim, este trabalho apresenta um sistema de recomendação de OAs ao aluno com base em seu estilo de aprendizagem. Foi desenvolvido um protótipo utilizando a técnica de recomendação baseada em utilidade, que busca, seleciona e sugere OAs de acordo com seu grau de utilidade (relevância) para o usuário, considerando seu estilo de aprendizagem. Para teste e validação da solução proposta foi realizado um experimento com turma real de graduandos para disponibilização de materiais de apoio a aulas presenciais. Constatou-se uma adesão de 85% dos alunos ao uso do ambiente e a índice de satisfação atingiu 89%, ou seja, as recomendações dos recursos educacionais foram avaliadas como úteis na maioria dos casos. Essa solução pode beneficiar estudantes, ao sugerir OAs com maior chance de atender às suas necessidades particulares. / Starting from the use of Information and Communication Technologies (ICT) to support the learning process, the various tools used for this purpose such as Learning Management Systems (LMS), online courses, educational Web portals often provide content and other educational resources through repositories of learning objects (LOs). Examples of LOs are: activities, assessments, videos, among others, ie, any entity that can be used for learning, education or training. The use of LOs in digital environments can improve the student learning, however, from the increased availability of such resources, it arises the need of improving the search and selection of these items in different repositories. Personal learning styles, that is to say, how individuals comprehend, process and transform information into knowledge, allow personalized recommending learning objects, so that the student can access the most appropriate instructional resource in their teaching-learning process. Thus, this work presents a recommendation system of LOs to students according to their learning style. A prototype was developed by using the technique utility-based recommendation, which seeks, selects and suggests LOs according to their degree of usefulness (relevance) to the user considering their learning style. In order to test and validate the proposed solution, an experiment was conducted with real undergraduate class to share supporting material to class attendances. It was verified an adherence of 85% of the students using the environment and the satisfaction rate reached 89%, in other words, the recommendations of educational resources were evaluated as useful in most cases. This solution can benefit students by suggesting LOs with a greater chance to meet your particular needs.