O avanço da tecnologia nos últimos anos permitiu a criação de Sistemas de Informação com acesso a grandes bases de dados, abrindo diversas possibilidades de aplicações. Tem-se como exemplo a Internet, onde uma enorme quantidade de dados é gerada e publicada a todo momento por usuários ao redor do mundo. Com isso, aos poucos foi surgindo a necessidade de métodos para filtrar o conteúdo disponível de forma a permitir que um usuário pudesse focar apenas nos seus interesses. Nesse contexto surgiram os Sistemas de Recomendação e as Redes Sociais, onde, mais recentemente, surgiram trabalhos que apresentam abordagens para o uso de Sistemas de Recomendação no contexto acadêmico, de forma a aumentar a produtividade de grupos de pesquisa. Também têm sido bastante exploradas formas de se utilizar informações temporais em Sistemas de Recomendação de maneira a melhorar as recomendações feitas. O presente trabalho propõe uma abordagem de recomendação de colaborações acadêmicas utilizando a técnica de Análise de Séries Temporais, buscando melhorar os resultados obtidos por trabalhos anteriores. Foi realizado um experimento offline para avaliar o desempenho da abordagem proposta em relação às abordagens anteriores e um estudo de usuários para fazer uma análise mais profunda com feedback de usuários. Foram utilizadas métricas conhecidas das áreas de Recuperação de Informação e Sistemas de Recomendação, mas alguns resultados se mostraram inferiores em comparação com as abordagens existentes; outros, porém, foram similares. Também foram utilizadas algumas métricas de avaliação focadas em Sistemas de Recomendação, e os resultados obtidos foram similares em todas as abordagens testadas. / The advance of technology in recent years made possible the creation of Information Systems with access to large databases, opening many applications possibilities. There’s the Internet, for example, where a vast amount of data is generated and published all the time by users around the world. In this sense, the need for methods to filter the available content to enable users to focus only on their interests slowly emerged. In this context, Recommender Systems and Social Networks appeared, where, recently, works reporting approaches to provide recommendations in the academic context appeared, increasing the productivity of research groups. New ways to employ temporal information in Recommender Systems to make better recommendations are also being explored. The present work proposes an approach to academic collaborations recommendation using Time Series Analysis, aiming to improve results reported on previous and current works. An offline experiment was done to evaluate the proposed approach in comparison with other works and a user study was done to make a deeper analysis from user feedback. Known metrics from the Information Retrieval and Recommender Systems fields were used, and in some cases the results obtained were lower compared to the current methods but similar in others. Some evaluation metrics from Recommender Systems were also used, and the results were similar to all approaches.

Recommender systems are an advanced form of software applications, more specifically decision-support systems, that efficiently assist the users in finding items of their interest. Recommender systems have been applied to many domains from music to e-commerce, movies to software services delivery and tourism to news by exploiting available information to predict and provide recommendations to end user. The suggestions generated by recommender systems tend to narrow down the list of items which a user may overlook due to the huge variety of similar items or users’ lack of experience in the particular domain of interest. While the performance of traditional recommender systems, which rely on relatively simpler information such as content and users’ filters, is widely accepted, their predictive capability perfomrs poorly when local context of the user and situated actions have significant role in the final decision. Therefore, acceptance and incorporation of context of the user as a significant feature and development of recommender systems utilising the premise becomes an active area of research requiring further investigation of the underlying algorithms and methodology. This thesis focuses on categorisation of contextual and non-contextual features within the domain of context-aware recommender system and their respective evaluation. Further, application of the Multilayer Perceptron Model (MLP) for generating predictions and ratings from the contextual and non-contextual features for contextual recommendations is presented with support from relevant literature and empirical evaluation. An evaluation of specifically employing artificial neural networks (ANNs) in the proposed methodology is also presented. The work emphasizes on both algorithms and methodology with three points of consideration: contextual features and ratings of particular items/movies are exploited in several representations to improve the accuracy of recommendation process using artificial neural networks (ANNs), context features are combined with user-features to further improve the accuracy of a context-aware recommender system and lastly, a combination of the item/movie features are investigated within the recommendation process. The proposed approach is evaluated on the LDOS-CoMoDa dataset and the results are compared with state-of-the-art approaches from relevant published literature.

O avanço da tecnologia nos últimos anos permitiu a criação de Sistemas de Informação com acesso a grandes bases de dados, abrindo diversas possibilidades de aplicações. Tem-se como exemplo a Internet, onde uma enorme quantidade de dados é gerada e publicada a todo momento por usuários ao redor do mundo. Com isso, aos poucos foi surgindo a necessidade de métodos para filtrar o conteúdo disponível de forma a permitir que um usuário pudesse focar apenas nos seus interesses. Nesse contexto surgiram os Sistemas de Recomendação e as Redes Sociais, onde, mais recentemente, surgiram trabalhos que apresentam abordagens para o uso de Sistemas de Recomendação no contexto acadêmico, de forma a aumentar a produtividade de grupos de pesquisa. Também têm sido bastante exploradas formas de se utilizar informações temporais em Sistemas de Recomendação de maneira a melhorar as recomendações feitas. O presente trabalho propõe uma abordagem de recomendação de colaborações acadêmicas utilizando a técnica de Análise de Séries Temporais, buscando melhorar os resultados obtidos por trabalhos anteriores. Foi realizado um experimento offline para avaliar o desempenho da abordagem proposta em relação às abordagens anteriores e um estudo de usuários para fazer uma análise mais profunda com feedback de usuários. Foram utilizadas métricas conhecidas das áreas de Recuperação de Informação e Sistemas de Recomendação, mas alguns resultados se mostraram inferiores em comparação com as abordagens existentes; outros, porém, foram similares. Também foram utilizadas algumas métricas de avaliação focadas em Sistemas de Recomendação, e os resultados obtidos foram similares em todas as abordagens testadas. / The advance of technology in recent years made possible the creation of Information Systems with access to large databases, opening many applications possibilities. There’s the Internet, for example, where a vast amount of data is generated and published all the time by users around the world. In this sense, the need for methods to filter the available content to enable users to focus only on their interests slowly emerged. In this context, Recommender Systems and Social Networks appeared, where, recently, works reporting approaches to provide recommendations in the academic context appeared, increasing the productivity of research groups. New ways to employ temporal information in Recommender Systems to make better recommendations are also being explored. The present work proposes an approach to academic collaborations recommendation using Time Series Analysis, aiming to improve results reported on previous and current works. An offline experiment was done to evaluate the proposed approach in comparison with other works and a user study was done to make a deeper analysis from user feedback. Known metrics from the Information Retrieval and Recommender Systems fields were used, and in some cases the results obtained were lower compared to the current methods but similar in others. Some evaluation metrics from Recommender Systems were also used, and the results were similar to all approaches.

O avanço da tecnologia nos últimos anos permitiu a criação de Sistemas de Informação com acesso a grandes bases de dados, abrindo diversas possibilidades de aplicações. Tem-se como exemplo a Internet, onde uma enorme quantidade de dados é gerada e publicada a todo momento por usuários ao redor do mundo. Com isso, aos poucos foi surgindo a necessidade de métodos para filtrar o conteúdo disponível de forma a permitir que um usuário pudesse focar apenas nos seus interesses. Nesse contexto surgiram os Sistemas de Recomendação e as Redes Sociais, onde, mais recentemente, surgiram trabalhos que apresentam abordagens para o uso de Sistemas de Recomendação no contexto acadêmico, de forma a aumentar a produtividade de grupos de pesquisa. Também têm sido bastante exploradas formas de se utilizar informações temporais em Sistemas de Recomendação de maneira a melhorar as recomendações feitas. O presente trabalho propõe uma abordagem de recomendação de colaborações acadêmicas utilizando a técnica de Análise de Séries Temporais, buscando melhorar os resultados obtidos por trabalhos anteriores. Foi realizado um experimento offline para avaliar o desempenho da abordagem proposta em relação às abordagens anteriores e um estudo de usuários para fazer uma análise mais profunda com feedback de usuários. Foram utilizadas métricas conhecidas das áreas de Recuperação de Informação e Sistemas de Recomendação, mas alguns resultados se mostraram inferiores em comparação com as abordagens existentes; outros, porém, foram similares. Também foram utilizadas algumas métricas de avaliação focadas em Sistemas de Recomendação, e os resultados obtidos foram similares em todas as abordagens testadas. / The advance of technology in recent years made possible the creation of Information Systems with access to large databases, opening many applications possibilities. There’s the Internet, for example, where a vast amount of data is generated and published all the time by users around the world. In this sense, the need for methods to filter the available content to enable users to focus only on their interests slowly emerged. In this context, Recommender Systems and Social Networks appeared, where, recently, works reporting approaches to provide recommendations in the academic context appeared, increasing the productivity of research groups. New ways to employ temporal information in Recommender Systems to make better recommendations are also being explored. The present work proposes an approach to academic collaborations recommendation using Time Series Analysis, aiming to improve results reported on previous and current works. An offline experiment was done to evaluate the proposed approach in comparison with other works and a user study was done to make a deeper analysis from user feedback. Known metrics from the Information Retrieval and Recommender Systems fields were used, and in some cases the results obtained were lower compared to the current methods but similar in others. Some evaluation metrics from Recommender Systems were also used, and the results were similar to all approaches.

<p> The dramatic improvement in information and communication technology (ICT) has made an evolution in learning management systems (LMS). The rapid growth in LMSs has caused users to demand more advanced, automated, and intelligent services. CourseNetworking is a next-generation LMS adopting machine learning to add personalization, gamification, and more dynamics to the system. This work tries to come up with two recommender systems that can help improve CourseNetworking services. The first one is a social recommender system helping CourseNetworking to track user interests and give more relevant recommendations. Recently, graph neural network (GNN) techniques have been employed in social recommender systems due to their high success in graph representation learning, including social network graphs. Despite the rapid advances in recommender systems performance, dealing with the dynamic property of the social network data is one of the key challenges that is remained to be addressed. In this research, a novel method is presented that provides social recommendations by incorporating the dynamic property of social network data in a heterogeneous graph by supplementing the graph with time span nodes that are used to define users long-term and short-term preferences over time. The second service that is proposed to add to Rumi services is a hashtag recommendation system that can help users label their posts quickly resulting in improved searchability of content. In recent years, several hashtag recommendation methods are proposed and developed to speed up processing of the texts and quickly find out the critical phrases. The methods use different approaches and techniques to obtain critical information from a large amount of data. This work investigates the efficiency of unsupervised keyword extraction methods for hashtag recommendation and recommends the one with the best performance to use in a hashtag recommender system. </p>

Master of Science / Department of Computing and Information Sciences / William H. Hsu / A recommender system is a computational mechanism for information filtering, where users provide recommendations (in the form of ratings or selecting items) as inputs, which the system then aggregates and directs to appropriate recipients. With the advent of web based media and publicity methods, the age where standardized methods of publicity, sales, production and marketing strategies do not. As such, in many markets the users are given a wide range of products and information to choose which product they like, to find a way out of this recommender systems are used in a way similar to the live social scenario, that is a user tries to get reviews from friends before opting for a product in a similar way recommender system tries to be a friend who recommends the options. Most of the recommender systems currently developed solely accuracy driven, i.e., reducing the Mean Absolute Error (MAE) between the predictions of the recommender system and actual ratings of the user. This leads to various problems for recommender systems such as lack of diversity and freshness. Lack of diversity arises when the recommender system is overly focused on accuracy by recommending a set of items, in which all of the items are too similar to each other, because they are predicted to be liked by the user. Lack of freshness also arises with overly focusing on accuracy but as a limitation on the set of items recommended making it overly predictable. This thesis work is directed at addressing the issues of diversity, by developing an approach, where a threshold of accuracy (in terms of Mean Absolute Error in prediction) is maintained while trying to diversify the set of item recommendations. Here for the problem of diversity a combination of Attribute-based diversification and user preference based diversification is done. This approach is then evaluated using non-classical methods along with evaluating the base recommender algorithm to prove that diversification is indeed is possible with a mixture of collaborative and content based approach.

Master of Science / Department of Computing and Information Sciences / Doina Caragea / In recent years, the content volume and number of users of the Web have increased dramatically. This large amount of data has caused an information overload problem, which hinders the ability of a user to find the relevant data at the right time. Therefore, the primary task of recommendation systems is to analyze data in order to offer users suggestions for similar data. Recommendations which use the core content are known as content-based recommendation or content filtering, and recommendations which utilize directly the user feedback are known as collaborative filtering. This thesis presents the design, implementation, testing, and evaluation of a recommender system within the recipe domain, where various approaches for producing recommendations are utilized. More specifically, this thesis discusses approaches derived from basic recommendation algorithms, but customized to take advantage of specific data available in the {\it recipe} domain. The proposed approaches for recommending recipes make use of recipe ingredients and reviews. We first build ingredient vectors for both recipes and users (based on recipes they have rated highly), and recommend new recipes to users based on the similarity between user and recipe ingredient vectors. Similarly, we build recipe and user vectors based on recipe review text, and recommend new recipes based on the similarity between user and recipe review vectors. At last, we study a hybrid approach, where both ingredients and reviews are used together. Our proposed approaches are tested over an existing dataset crawled from recipes.com. Experimental results show that the recipe ingredients are more informative than the review text for making recommendations. Furthermore, when using ingredients and reviews together, the results are better than using just the reviews, but worse than using just the ingredients, suggesting that to make use of reviews, the review vocabulary needs better filtering.

M. Tech. Information Networks / The growth of the internet has made it easy for people to exchange information resulting in the abundance of information commonly referred to as information overload. It causes retailers to fail to make adequate sales since the customers are swamped with a lot of options and choices. To lessen this problem retailers have begun to find it useful to make use of algorithmic approaches to determine which content to show consumers. These algorithmic approaches are known as recommender systems. Collaborative Filtering recommender systems suggest items to users based on other users reported prior experience with those items. These systems are, however, vulnerable to shilling attacks since they are highly dependent on outside sources of information. Shilling is a process in which syndicating users can connive to promote or demote a certain item, where malicious users benefit from introducing biased ratings. It is, however, critical that shilling detection systems are implemented to detect, warn and shut down shilling attacks within minutes. Modern patented shilling detection systems employ: (a) classification methods, (b) statistical methods, and (c) rules and threshold values defined by shilling detection analysts, using their knowledge of valid shilling cases and the false alarm rate as guidance. The goal of this dissertation is to determine a context for, and assess the performance of Meta-Learning techniques that can be integrated in the shilling detection process.

The amount of information generated and maintained by information systems and their users leads to the increasingly important concern of information overload. Personalized systems have thus emerged to help provide more relevant information and services to the user. In particular, recommender systems appeared in the mid 1990's and have since then generated a growing interest in both industry and academia. Besides, context-aware systems have been developed to model, capture and interpret information about the user's situation, generally in dynamic and heterogeneous environments. Decision support systems like Business Intelligence (BI) platforms also face usability challenges as the amount of information available to knowledge workers grows. Remarkably, we observe that only a small part of personalization and recommendation techniques have been used in the context of data warehouses and analysis tools. Therefore, our work aims at exploring synergies of recommender systems and context-aware systems to develop personalization and recommendation scenarios suited in a BI environment. In response to this, we develop in our work an open and modular situation management platform using a graph-based situation model. Besides, dynamic aspects are crucial to deal with context data which is inherently time-dependent. We thus define two types of active components to enable dynamic maintenance of situation graphs, activation rules and operators. In response to events which can describe users' interactions, activation rules - defined using the event-condition-action framework - are evaluated thanks to queries on underlying graphs, to eventually trigger appropriate operators. These platform and framework allow us to develop and support various recommendation and personalization scenarios. Importantly, we design a re-usable personalized query expansion component, using semantics of multi-dimensional models and usage statistics from repositories of BI documents like reports or dashboards. This component is an important part of another experimentation we realized, Text-To-Query. This system dynamically generates multi-dimensional queries to illustrate a text and support the knowledge worker in the analysis or enrichment of documents she is manipulating. Besides, we also illustrate the integration and usage of our graph repository and situation management frameworks in an open and extensible federated search project, to provide background knowledge management and personalization.

Recommender systems automate the process of recommending products and services to customers based on various types of data including customer demographics, product features, and, most importantly, previous interactions between customers and products (e.g., purchasing, rating, and catalog browsing). Despite significant research progress and growing acceptance in real-world applications, two major challenges remain to be addressed to implement effective e-commerce recommendation applications. The first challenge is concerned with making recommendations based on sparse transaction data. The second challenge is the lack of a unified framework to integrate multiple types of input data and recommendation approaches.This dissertation investigates graph-based algorithms to address these two problems. The proposed approach is centered on consumer-product graphs that represent sales transactions as links connecting consumer and product nodes. In order to address the sparsity problem, I investigate the network spreading activation algorithms and a newly proposed link analysis algorithm motivated by ideas from Web graph analysis techniques. Experimental results with several e-commerce datasets indicated that both classes of algorithms outperform a wide range of existing collaborative filtering algorithms, especially under sparse data. Two graph-based models that enhance the simple consumer-product graph were proposed to provide unified recommendation frameworks. The first model, a two-layer graph model, enhances the consumer-product graph by incorporating the consumer/product attribute information as consumer and product similarity links. The second model is based on probabilistic relational models (PRMs) developed in the relational learning literature. It is demonstrated with e-commerce datasets that the proposed frameworks not only conceptually unify many of the existing recommendation approaches but also allow the exploitation of a wider range of data patterns in an integrated manner, leading to improved recommendation performance.In addition to the recommendation algorithm design research, this dissertation also employs the random graph theory to study the topological characteristics of consumer-product graphs and the fundamental mechanisms that generate the sales transaction data. This research represents the early step towards a meta-level analysis framework for validating the fundamental assumptions made by different recommendation algorithms regarding the consumer-product interaction generation process and thus supporting systematic recommendation model/algorithm selection and evaluation.