Due to the rapid increase of social network resources and services, Internet users are now overwhelmed by the vast quantity of social media available. By utilizing the user’s context while consuming diverse multimedia contents, we can identify different personal preferences and settings. However, there is still a need to reinforce the recommendation process in a systematic way, with context-adaptive information. This thesis proposes a recommendation model, called HPEM, that establishes a bridge between the multimedia resources, user collaborative preferences, and the detected contextual information, including physiological parameters. The collection of contextual information and the delivery of the resulted recommendation is made possible by adapting the user’s environment using Ambient Intelligent (AmI) interfaces. Additionally, this thesis presents the potential of including a user’s biological signal and leveraging it within an adapted collaborative filtering algorithm in the recommendation process. First, the different versions of the proposed HPEM model utilize existing online social networks by incorporating social tags and rating information in ways that personalize the search for content in a particular detected context. By leveraging the social tagging, our proposed model computes the hidden preferences of users in certain contexts from other similar contexts, as well as the hidden assignment of contexts for items from other similar items. Second, we demonstrate the use of an optimization function to maximize the Mean Average Prevision (MAP) measure of the resulted recommendations. We demonstrate the feasibility of HPEM with two prototype applications that use contextual information for recommendations. Offline and online experiments have been conducted to measure the accuracy of delivering personalized recommendations, based on the user’s context; two real-world and one collected semi-synthetic datasets were used. Our evaluation results show a potential improvement to the quality of the recommendation when compared to state-of-the-art recommendation algorithms that consider contextual information. We also compare the proposed method to other algorithms, where user’s context is not used to personalize the recommendation results. Additionally, the results obtained demonstrate certain improvements on cold start situations, where relatively little information is known about a user or an item.

Submitted by Fabio Sobreira Campos da Costa (fabio.sobreira@ufpe.br) on 2017-02-21T16:47:42Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) dvsTeseBiblioteca.pdf: 6571192 bytes, checksum: eb7914e5ffef25b8f01ff92d9a60c164 (MD5) / Made available in DSpace on 2017-02-21T16:47:42Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) dvsTeseBiblioteca.pdf: 6571192 bytes, checksum: eb7914e5ffef25b8f01ff92d9a60c164 (MD5) Previous issue date: 2016-07-21 / FACEPE / Traditionally, single-domain recommender systems (SDRS) have achieved good results in recommending relevant items for users in order to solve the information overload problem. However, cross-domain recommender systems (CDRS) have emerged aiming to enhance SDRS by achieving some goals such as accuracy improvement, diversity, addressing new user and new item problems, among others. Instead of treating each domain independently, CDRS use knowledge acquired in a source domain (e.g. books) to improve the recommendation in a target domain (e.g. movies). Likewise SDRS research, collaborative filtering (CF) is considered the most popular and widely adopted approach in CDRS, because its implementation for any domain is relatively simple. In addition, its quality of recommendation is usually higher than that of content-based filtering (CBF) algorithms. In fact, the majority of the cross-domain collaborative filtering RS (CD-CFRS) can give better recommendations in comparison to single domain collaborative filtering recommender systems (SD-CFRS), leading to a higher users’ satisfaction and addressing cold-start, sparsity, and diversity problems. However, CD-CFRS may not necessarily be more accurate than SD-CFRS. On the other hand, context-aware recommender systems (CARS) deal with another relevant topic of research in the recommender systems area, aiming to improve the quality of recommendations too. Different contextual information (e.g., location, time, mood, etc.) can be leveraged in order to provide recommendations that are more suitable and accurate for a user depending on his/her context. In this way, we believe that the integration of techniques developed in isolation (cross-domain and contextaware) can be useful in a variety of situations, in which recommendations can be improved by information from different sources as well as they can be refined by considering specific contextual information. In this thesis, we define a novel formulation of the recommendation problem, considering both the availability of information from different domains (source and target) and the use of contextual information. Based on this formulation, we propose the integration of cross-domain and context-aware approaches for a novel recommender system (CD-CARS). To evaluate the proposed CD-CARS, we performed experimental evaluations through two real datasets with three different contextual dimensions and three distinct domains. The results of these evaluations have showed that the use of context-aware techniques can be considered as a good approach in order to improve the cross-domain recommendation quality in comparison to traditional CD-CFRS. / Tradicionalmente, “sistemas de recomendação de domínio único” (SDRS) têm alcançado bons resultados na recomendação de itens relevantes para usuários, a fim de resolver o problema da sobrecarga de informação. Entretanto, “sistemas de recomendação de domínio cruzado” (CDRS) têm surgido visando melhorar os SDRS ao atingir alguns objetivos, tais como: “melhoria de precisão”, “melhor diversidade”, abordar os problemas de “novo usuário” e “novo item”, dentre outros. Ao invés de tratar cada domínio independentemente, CDRS usam conhecimento adquirido em um domínio fonte (e.g. livros) a fim de melhorar a recomendação em um domínio alvo (e.g. filmes). Assim como acontece na área de pesquisa sobre SDRS, a filtragem colaborativa (CF) é considerada a técnica mais popular e amplamente utilizada em CDRS, pois sua implementação para qualquer domínio é relativamente simples. Além disso, sua qualidade de recomendação é geralmente maior do que a dos algoritmos baseados em filtragem de conteúdo (CBF). De fato, a maioria dos “sistemas de recomendação de domínio cruzado” baseados em filtragem colaborativa (CD-CFRS) podem oferecer melhores recomendações em comparação a “sistemas de recomendação de domínio único” baseados em filtragem colaborativa (SD-CFRS), aumentando o nível de satisfação dos usuários e abordando problemas tais como: “início frio”, “esparsidade” e “diversidade”. Entretanto, os CD-CFRS podem não ser mais precisos do que os SD-CFRS. Por outro lado, “sistemas de recomendação sensíveis à contexto” (CARS) tratam de outro tópico relevante na área de pesquisa de sistemas de recomendação, também visando melhorar a qualidade das recomendações. Diferentes informações contextuais (e.g. localização, tempo, humor, etc.) podem ser utilizados a fim de prover recomendações que são mais adequadas e precisas para um usuário dependendo de seu contexto. Desta forma, nós acreditamos que a integração de técnicas desenvolvidas separadamente (de “domínio cruzado” e “sensíveis a contexto”) podem ser úteis em uma variedade de situações, nas quais as recomendações podem ser melhoradas a partir de informações obtidas em diferentes fontes além de refinadas considerando informações contextuais específicas. Nesta tese, nós definimos uma nova formulação do problema de recomendação, considerando tanto a disponibilidade de informações de diferentes domínios (fonte e alvo) quanto o uso de informações contextuais. Baseado nessa formulação, nós propomos a integração de abordagens de “domínio cruzado” e “sensíveis a contexto” para um novo sistema de recomendação (CD-CARS). Para avaliar o CD-CARS proposto, nós realizamos avaliações experimentais através de dois “conjuntos de dados” com três diferentes dimensões contextuais e três domínios distintos. Os resultados dessas avaliações mostraram que o uso de técnicas sensíveis a contexto pode ser considerado como uma boa abordagem a fim de melhorar a qualidade de recomendações de “domínio cruzado” em comparação às recomendações de CD-CFRS tradicionais.

Recommender systems support online customers by suggesting products and services of likely interest to them. Research in recommender systems is now starting to recognise the importance of multiple selection criteria and the role of customer context in improving the recommendation output. This thesis investigates the inclusion of criteria and context information in the recommendation process. Firstly, a novel technique for multi-criteria recommendation is proposed. It assumes that some selection criteria for an item (product or a service) will dominate the overall rating, and that these dominant criteria will be different for different users. Following this assumption, users are clustered based on their criteria preferences, creating a “preference lattice”. The recommendation output for a user is then based on ratings by other users from the same or nearby clusters. Secondly, a context similarity metric for context aware recommendation is presented. This metric can help improve the prediction accuracy in two ways. On the one hand, the metric can guide the aggregation of the feedback from similar context to improve the prediction accuracy. This aggregation is important because the recommendation generation based on prior feedback by similar customers reduces the quantum of feedback used, resulting in a reduction in recommendation quality. On the other hand, the value returned by the context similarity metric can also be used to indicate the importance of the context information in the prediction process for a context aware recommendation.The validation of the two proposed techniques and their applications are conducted in the service domain because the relatively high degree of user involvement attracts users to provide detailed feedback from multiple perspectives, such as from criteria and context perspectives. In particular, hotel services and web services areas are selected due to their different levels of maturity in terms of users’ feedback. For each area, this thesis proposes a different recommendation approach by combining the proposed techniques with a traditional recommendation approach. The thesis concludes with experiments conducted on the datasets from the two aforementioned areas to evaluate the proposed techniques, and to demonstrate the process and the effectiveness of the techniques-based recommendation approaches.

Pervasive computing systems aim to improve human-computer interaction, using users situation variables that define context. The boom of Internet makes growing availables items to choose, giving cost in made decision process. Affective Computing has in its goals to identify user s affective/emotional state in a computing interaction, in order to respond to it automatically. Recommendation systems help made decision selecting and suggesting items in scenarios where there are huge information volume, using, traditionally, users prefferences data. This process could be enhanced using context information (as physical, environmental or social), rising the Context-Aware Recommendation Systems. Due to emotions importance in our lives, that could be treated with Affective Computing, this work uses affective context as context variable, in recommendation process, proposing the Affective-Recommender a recommendation system that uses user s affective state to select and to suggest items. The system s model has four components: (i) detector, that identifies affective-state, using the multidimesional Pleasure, Arousal and Dominance model, and Self-Assessment Maniking instrument, that asks user to inform how he/she feels; (ii) recommender, that selects and suggests items, using a collaborative-filtering based approache, in which user s prefference to an item is his/her affective reaction to it as the affective state detected after access; (iii) application, which interacts with user, shows probable most interesting items defined by recommender, and requests affect identification when it is necessarly; and (iv) data base, that stores available items and users prefferences. As a use case, Affective-Recommender is used in a e-learning scenario, due to personalization obtained with recommendation and emotion importances in learning process. The system was implemented over Moodle LMS. To exposes its operation, a use scenario was organized, simulating recommendation process. In order to check system applicability, with students opinion about to inform how he/she feels and to receive suggestions, it was applied in three UFSM graduation courses classes, and then it were analyzed data access and the answers to a sent questionnaire. As results, it was perceived that students were able to inform how they feel, and that occured changes in their affecive state, based on accessed item, although they don t see improvements with the recommendation, due to small data available to process and showr time of application. / Sistemas de Computação Pervasiva buscam melhorar a interação humano-computador através do uso de variáveis da situação do usuário que definem o contexto. A explosão da Internet e das tecnologias de informação e comunicação torna crescente a quantidade de itens disponíveis para a escolha, impondo custo para o usuário no processo de tomada de decisão. A Computação Afetiva tem entre seus objetivos identificar o estado emocional/afetivo do usuário durante uma interação computacional, para automaticamente responder a ele. Já Sistemas de Recomendação auxiliam a tomada de decisão, selecionando e sugerindo itens em situações onde há grandes volumes de informação, tradicionalmente, utilizando as preferências dos usuários para a seleção e sugestão. Esse processo pode ser melhorado com o uso do contexto (físico, ambiental, social), surgindo os Sistemas de Recomendação Sensíveis ao Contexto. Tendo em vista a importância das emoções em nossas vidas, e a possibilidade de tratamento delas com a Computação Afetiva, este trabalho utiliza o contexto afetivo do usuário como variável da situação, durante o processo de recomendação, propondo o Affective-Recommender um sistema de recomendação que faz uso do estado afetivo do usuário para selecionar e sugerir itens. O sistema foi modelado a partir de quatro componentes: (i) detector, que identifica o estado afetivo, utilizando o modelo multidimensional Pleasure, Arousal e Dominance e o instrumento Self-Assessment Manikin, solicitando que o usuário informe como se sente; (ii) recomendador, que escolhe e sugere itens, utilizando uma abordagem baseada em filtragem colaborativa, em que a preferência de um usuário para um item é vista como sua reação estado afetivo detectado após o contato ao item; (iii) aplicação, que interage com o usuário, exibe os itens de provável maior interesse definidos pelo recomendador, e solicita que o estado seja identificado, sempre que necessário; e (iv) base de dados, que armazena os itens disponíveis para serem sugeridos e as preferências de cada usuário. Como um caso de uso e prova de conceito, o Affective-Recommender é empregado em um cenário de e-learning, devido à importância da personalização, obtida com a recomendação, e das emoções no processo de aprendizagem. O sistema foi implementado utilizando-se como base o AVEA Moodle. Para expor o funcionamento, estruturou-se um cenário de uso, simulando-se o processo de recomendação. Para verificar a aplicabilidade real do sistema, ele foi empregado em três turmas de cursos de graduação da UFSM, sendo analisados dados de acesso e aplicado um questionário para identificar as impressões do alunos quanto a informar como se sentem e receber recomendações. Como resultados, percebeu-se que os alunos conseguiram informar seus estados afetivos, e que houve uma mudança em neste estado com base no item acessado, embora não tenham vislumbrado melhorias com as recomendações, em virtude da pequena quantidade de dados disponível para processamento e do curto tempo de aplicação.

Information Filtering and Recommender Systems have been used and has been implemented in various ways from various entities since the dawn of the Internet, and state-of-the-art approaches rely on Machine Learning and Deep Learning in order to create accurate and personalized recommendations for users in a given context. These models require big amounts of data with a variety of features such as time, location and user data in order to find correlations and patterns that other classical models such as matrix factorization and collaborative filtering cannot. This thesis researches, implements and compares a variety of models with the primary focus of Machine Learning and Deep Learning for the task of music recommendation and do so successfully by representing the task of recommendation as a multi-class extreme classification task with 100 000 distinct labels. By comparing fourteen different experiments, all implemented models successfully learn features such as time, location, user features and previous listening history in order to create context-aware personalized music predictions, and solves the cold start problem by using user demographic information, where the best model being capable of capturing the intended label in its top 100 list of recommended items for more than 1/3 of the unseen data in an offine evaluation, when evaluating on randomly selected examples from the unseen following week. / Informationsfiltrering och rekommendationssystem har använts och implementeratspå flera olika sätt från olika enheter sedan gryningen avInternet, och moderna tillvägagångssätt beror påMaskininlärrning samtDjupinlärningför att kunna skapa precisa och personliga rekommendationerför användare i en given kontext. Dessa modeller kräver data i storamängder med en varians av kännetecken såsom tid, plats och användardataför att kunna hitta korrelationer samt mönster som klassiska modellersåsom matris faktorisering samt samverkande filtrering inte kan. Dettaexamensarbete forskar, implementerar och jämför en mängd av modellermed fokus påMaskininlärning samt Djupinlärning för musikrekommendationoch gör det med succé genom att representera rekommendationsproblemetsom ett extremt multi-klass klassifikationsproblem med 100000 unika klasser att välja utav. Genom att jämföra fjorton olika experiment,så lär alla modeller sig kännetäcken såsomtid, plats, användarkänneteckenoch lyssningshistorik för att kunna skapa kontextberoendepersonaliserade musikprediktioner, och löser kallstartsproblemet genomanvändning av användares demografiska kännetäcken, där den bästa modellenklarar av att fånga målklassen i sin rekommendationslista medlängd 100 för mer än 1/3 av det osedda datat under en offline evaluering,när slumpmässigt valda exempel från den osedda kommande veckanevalueras.