Traditional hybrid recommender systems typically follow a manually created fixed prediction strategy in their decision making process. Experts usually design these static strategies as fixed combinations of different techniques. However, people&#039 / s tastes and desires are temporary and they gradually evolve. Moreover, each domain has unique characteristics, trends and unique user interests. Recent research has mostly focused on static hybridization schemes which do not change at runtime. In this thesis work, we describe an adaptive hybrid recommender system, called AdaRec that modifies its attached prediction strategy at runtime according to the performance of prediction techniques (user feedbacks). Our approach to this problem is to use adaptive prediction strategies. Experiment results with datasets show that our system outperforms naive hybrid recommender.

Thesis (Ph. D.)--Ohio State University, 2009. / Title from first page of PDF file. Includes vita. Includes bibliographical references (p. 115-119).

Orientadores: Ricardo da Silva Torres, Anderson de Rezende Rocha / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-25T19:48:27Z (GMT). No. of bitstreams: 1 Werneck_RafaeldeOliveira_M.pdf: 2395829 bytes, checksum: 8f190aeb6dbafb841d0c03f7d7099041 (MD5) Previous issue date: 2014 / Resumo: Devido ao grande crescimento do uso de tecnologias para a aquisição de dados, temos que lidar com grandes e complexos conjuntos de dados a fim de extrair conhecimento que possa auxiliar o processo de tomada de decisão em diversos domínios de aplicação. Uma solução típica para abordar esta questão se baseia na utilização de métodos de aprendizado de máquina, que são métodos computacionais que extraem conhecimento útil a partir de experiências para melhorar o desempenho de aplicações-alvo. Existem diversas bibliotecas e arcabouços na literatura que oferecem apoio à execução de experimentos de aprendizado de máquina, no entanto, alguns não são flexíveis o suficiente para poderem ser estendidos com novos métodos, além de não oferecerem mecanismos que permitam o reuso de soluções de sucesso concebidos em experimentos anteriores na ferramenta. Neste trabalho, propomos um arcabouço para automatizar experimentos de aprendizado de máquina, oferecendo um ambiente padronizado baseado em workflow, tornando mais fácil a tarefa de avaliar diferentes descritores de características, classificadores e abordagens de fusão em uma ampla gama de tarefas. Também propomos o uso de medidas de similaridade e métodos de learning-to-rank em um cenário de recomendação, para que usuários possam ter acesso a soluções alternativas envolvendo experimentos de aprendizado de máquina. Nós realizamos experimentos com quatro medidas de similaridade (Jaccard, Sorensen, Jaro-Winkler e baseada em TF-IDF) e um método de learning-to-rank (LRAR) na tarefa de recomendar workflows modelados como uma sequência de atividades. Os resultados dos experimentos mostram que a medida Jaro-Winkler obteve o melhor desempenho, com resultados comparáveis aos observados para o método LRAR. Em ambos os casos, as recomendações realizadas são promissoras, e podem ajudar usuários reais em diferentes tarefas de aprendizado de máquina / Abstract: Due to the large growth of the use of technologies for data acquisition, we have to handle large and complex data sets in order to extract knowledge that can support the decision-making process in several domains. A typical solution for addressing this issue relies on the use of machine learning methods, which are computational methods that extract useful knowledge from experience to improve performance of target applications. There are several libraries and frameworks in the literature that support the execution of machine learning experiments. However, some of them are not flexible enough for being extended with novel methods and they do not support reusing of successful solutions devised in previous experiments made in the framework. In this work, we propose a framework for automating machine learning experiments that provides a workflow-based standardized environment and makes it easy to evaluate different feature descriptors, classifiers, and fusion approaches in a wide range of tasks. We also propose the use of similarity measures and learning-to-rank methods in a recommendation scenario, in which users may have access to alternative machine learning experiments. We performed experiments with four similarity measures (Jaccard, Sorensen, Jaro-Winkler, and a TF-IDF-based measure) and one learning-to-rank method (LRAR) in the task of recommending workflows modeled as a sequence of activities. Experimental results show that Jaro-Winkler yields the highest effectiveness performance with comparable results to those observed for LRAR. In both cases, the recommendations performed are very promising and might help real-world users in different daily machine learning tasks / Mestrado / Ciência da Computação / Mestre em Ciência da Computação

[...] Nos principales contributions peuvent être résumées en trois points :1. la proposition d’un système de classification de documents centrés sur les entités à l’aide d’un profil d’entité et de méta critères dans le contexte de filtrage de documents. Nous avons mis en place une approche qui est indépendante des entités et qui utilise les principes du transfert de connaissances. En effet, notre approche permet l’apprentissage à partir d’un ensemble de données annotées pour un pool d’entités tout en étant capables de catégoriser des documents concernant des entités pour lesquels aucune donnée annotée n’a été fournie ;2. la proposition d’un nouveau modèle de langue diachronique pour étendre la définition de profil d’entité afin de permettre la mise à jour de celui-ci. En effet, le suivi d’une entité nommée implique de pouvoir distinguer une information déjà connue d’une information nouvelle. Le modèle de langue diachronique permet la mise à jour automatique du profil d’entité tout en minimisant le bruit apporté ;3. la proposition d’une méthode pour découvrir la popularité d’une entité afin d’améliorer la cohérence d’un modèle de classification sur tous les aspects temporels liés à une entité. Pour détecter l’importance d’un document au regard d’une entité, nous proposons d’utiliser, entre autres, des indicateurs temporels qui peuvent varier d’une entité à l’autre. Nous proposons de regrouper les entités en fonction de leur popularité sur le Web à chaque instant pour tenter d’améliorer la cohérence des modèles et ainsi augmenter les performances des classificateurs.[...] / [...] Our main contributions are:1. We propose an entity centric classification system, which helps finding documents that are related to an entity based on its profile and a set of meta criteria. We propose to use the classification result to filter out unrelated documents. This approach is entity independent and uses transfer learning principles. We trained the classification system with a set of annotated concerning a set of entities and we categorized documents that concerns other entities;2. We introduce a diachronical language model, which extends our definition of entity profile in order to add to the capability of updating an entity profile. Tracking an entity implies to distinguish between a known piece of information from a new one. This new language model enables automatic update of entity profile while minimizing the noise;3. We develop a method to detect the entity popularity in order to enhance the coherence of a classification model concerning temporal aspects. In order to detect the importance of a document regarding an entity, we propose to use temporal sensors, which may vary from an entity to another. We cluster entities sharing the same amount of popularity on the Web at each time t to enhance the coherence of classification model and thus improve classifier performances.[...]

This dissertation studies economic issues in the digital economy with a specific focus on the economic aspects of how firms acquire and use consumer data. Chapter 1 empirically studies the drivers of digital attention in the space of social media applications. In order to do so I conduct an experiment where I comprehensively monitor how participants spend their time on digital services and use parental control software to shut off access to either their Instagram or YouTube. I characterize how participants substitute their time during and after the restrictions. I provide an interpretation of the substitution during the restriction period that allows me to conclude that relevant market definitions may be broader than those currently considered by regulatory authorities, but that the substantial diversion towards non-digital activities indicates significant market power from the perspective of consumers for Instagram and YouTube. I then use the results on substitution after the restriction period to motivate a discrete choice model of time usage with inertia and, using the estimates from this model, conduct merger assessments between social media applications. I find that the inertia channel is important for justifying blocking mergers, which I use to argue that currently debated policies aimed at curbing digital addiction are important not only just in their own right but also from an antitrust perspective and, in particular, as a potential policy tool for promoting competition in these markets. More broadly, my paper highlights the utility of product unavailability experiments for demand and merger analysis of digital goods. I thank Maayan Malter for working together with me on collecting the data for this paper. Chapter 2 then studies the next step in consumer data collection process – the extent to which a firm can collect a consumer’s data depends on privacy preferences and the set of available privacy tools. This chapter studies the impact of the General Data Protection Regulation on the ability of a data-intensive intermediary to collect and use consumer data. We find that the opt-in requirement of GDPR resulted in 12.5% drop in the intermediary-observed consumers, but the remaining consumers are trackable for a longer period of time. These findings are consistent with privacy-conscious consumers substituting away from less efficient privacy protection (e.g, cookie deletion) to explicit opt out—a process that would make opt-in consumers more predictable. Consistent with this hypothesis, the average value of the remaining consumers to advertisers has increased, offsetting some of the losses from consumer opt-outs. This chapter is jointly authored with Yeon-Koo Che and Tobias Salz. Chapter 3 and Chapter 4 make up the third portion of the dissertation that studies one of the most prominent uses of consumer data in the digital economy – recommendation systems. This chapter is a combination of several papers studying the economic impact of these systems. The first paper is a joint paper with Duarte Gonçalves which studies a model of strategic interaction between producers and a monopolist platform that employs a recommendation system. We characterize the consumer welfare implications of the platform’s entry into the production market. The platform’s entry induces the platform to bias recommendations to steer consumers towards its own goods, which leads to equilibrium investment adjustments by the producers and lower consumer welfare. Further, we find that a policy separating recommendation and production is not always welfare improving. Our results highlight the ability of integrated recommender systems to foreclose competition on online platforms. The second paper turns towards understanding how such systems impact consumer choices and is joint with Duarte Gonçalves and Shan Sikdar. In this paper we study a model of user decision-making in the context of recommender systems via numerical simulation. Our model provides an explanation for the findings of Nguyen et. al (2014), where, in environments where recommender systems are typically deployed, users consume increasingly similar items over time even without recommendation. We find that recommendation alleviates these natural filter-bubble effects, but that it also leads to an increase in homogeneity across users, resulting in a trade-off between homogenizing across-user consumption and diversifying within-user consumption. Finally, we discuss how our model highlights the importance of collecting data on user beliefs and their evolution over time both to design better recommendations and to further understand their impact.

The amount of video content online will nearly triple in quantity by 2021 compared to 2016. The implementation of sophisticated filters is of paramount importance to manage this information flow. The research question of this thesis asks to what extent it is possible to generate personal recommendations, based on the data that news videos implies. The objective is to evaluate how different recommender systems compare to complete random, each other and how they are received by users in a test environment. This study was performed during the spring of 2018, and explore four different algorithms. These recommender systems include a content-based, a collaborative-filter, a hybrid model and a popularity model as a baseline. The dataset originates from a news media startup called Newstag, who provide video news on a global scale. The data is sparse and includes implicit feedback only. Three offline experiments and a user test were performed. The metric that guided the algorithms offline performance was their recall at 5 and 10, due to the fact that the top list of recommended items are of most interest. A comparison was done on different amounts of meta-data included during training. Another test explored respective algorithms performance as the density of the data increased. In the user test, a mean opinion score was calculated based on the quality of recommendations that each of the algorithms generated for the test subjects. The user test also included randomly sampled news videos to compare with as a baseline. The results indicate that for this specific setting and data set, the content-based recommender system performed best in both the recall at five and ten, as well as in the user test. All of the algorithms outperformed the random baseline. / Mängden video som finns tillgänglig på internet förväntas att tredubblas år 2021 jämfört med 2016. Detta innebär ett behov av sofistikerade filter för att kunna hantera detta informationsflöde. Detta examensarbete ämnar att svara på till vilken grad det går att generera personliga rekommendationer baserat på det data som nyhetsvideo innebär. Syftet är att utvärdera och jämföra olika rekommendationssystem och hur de står sig i ett användartest. Studien utfördes under våren 2018 och utvärderar fyra olika algoritmer. Dessa olika rekommendationssystem innefattar tekniker som content-based, collaborative-filter, hybrid och en popularitetsmodell används som basvärde. Det dataset som används är glest och har endast implicita attribut. Tre experiment utförs samt ett användartest. Mätpunkten för algoritmernas prestanda utgjordes av recall at 5 och recall at 10, dvs. att man mäter hur väl algoritmerna lyckas generera värdefulla rekommendationer i en topp-fem respektive topp-10-lista av videoklipp. Detta då det är av intresse att ha de mest relevanta videorna högst upp i sin lista av resultat. En jämförelse gjordes mellan olika mängd metadata som inkluderades vid träning. Ett annat test gick ut på att utforska hur algoritmerna presterar då datasetet blir mindre glest. I användartestet användes en utvärderingsmetod kallad mean-opinion-score och denna räknades ut per algoritm genom att testanvändare gav betyg på respektive rekommendation, baserat på hur intressant videon var för dem. Användartestet inkluderade även slumpmässigt generade videos för att kunna jämföras i form av basvärde. Resultaten indikerar, för detta dataset, att algoritmen content-based presterar bäst både med hänsyn till recall at 5 & 10 samt den totala poängen i användartestet. Alla algoritmer presterade bättre än slumpen.

Apesar dos processos de recuperação e filtragem de informação sempre terem usado técnicas básicas de Processamento de Linguagem Natural (PLN) no suporte à estruturação de documentos, ainda são poucas as indicações sobre os avanços relacionados à utilização de técnicas mais sofisticadas de PLN que justifiquem o custo de sua utilização nestes processos, em comparação com as abordagens tradicionais. Este trabalho investiga algumas evidências que fundamentam a hipótese de que a aplicação de métodos que utilizam conhecimento linguístico é viável, demarcando importantes contribuições para o aumento de sua eficiência em adição aos métodos estatásticos tradicionais. É proposto um modelo de representação de texto fundamentado em sintagmas nominais, cuja representatividade de seus descritores é calculada utilizando-se o conceito de evidência, apoiado em métodos estatísticos. Filtros induzidos a partir desse modelo são utilizados para classificar os documentos recuperados analisando-se a relevância implícita no perfil do usuário. O aumento da precisão (e, portanto, da eficácia) em sistemas de Recuperação de Informação, conseqüência da pós-filtragem seletiva de informações, demonstra uma clara evidência de como o uso de técnicas de PLN pode auxiliar a categorização de textos, abrindo reais possibilidades para o aprimoramento do modelo apresentado / Although Information Retrieval and Filtering tasks have always used basic Natural Language Processing (NLP) techniques for supporting document structuring, there is still space for more sophisticated NLP techniques which justify their cost when compared to the traditional approaches. This research aims to investigate some evidences that justify the hypothesis on which the use of linguistic-based methods is feasible and can bring on relevant contributions to this area. In this work noun phrases of a text are used as descriptors whose evidence is calculated by statistical methods. Filters are then induced to classify the retrieved documents by measuring their implicit relevance presupposed by an user profile. The increase of precision (efficacy) in IR systems as a consequence of the use of NLP techniques for text classification in the filtering task is an evidence of how this approach can be further explored

Bayesian probabilistic modeling provides a powerful framework for building flexible models to incorporate latent structures through likelihood model and prior. When we specify a model, we make certain assumptions about the underlying data-generating process with respect to these latent structures. For example, the latent Dirichlet allocation (LDA) model assumes that when generating a document, we first select a latent topic and then select a word that often appears in the selected topic. We can uncover the latent structures conditioned on the observed data via posterior inference. In this dissertation, we apply the tools of probabilistic latent variable models and try to understand complex real-world data about music semantics and user behavior. We first look into the problem of automatic music tagging -- inferring the semantic tags (e.g., "jazz'', "piano'', "happy'', etc.) from the audio features. We treat music tagging as a matrix completion problem and apply the Poisson matrix factorization model jointly on the vector-quantized audio features and a "bag-of-tags'' representation. This approach exploits the shared latent structure between semantic tags and acoustic codewords. We present experimental results on the Million Song Dataset for both annotation and retrieval tasks, illustrating the steady improvement in performance as more data is used. We then move to the intersection between music semantics and user behavior: music recommendation. The leading performance in music recommendation is achieved by collaborative filtering methods which exploit the similarity patterns in user's listening history. We address the fundamental cold-start problem of collaborative filtering: it cannot recommend new songs that no one has listened to. We train a neural network on semantic tagging information as a content model and use it as a prior in a collaborative filtering model. The proposed system is evaluated on the Million Song Dataset and shows comparably better result than the collaborative filtering approaches, in addition to the favorable performance in the cold-start case. Finally, we focus on general recommender systems. We examine two different types of data: implicit and explicit feedback, and introduce the notion of user exposure (whether or not a user is exposed to an item) as part of the data-generating process, which is latent for implicit data and observed for explicit data. For implicit data, we propose a probabilistic matrix factorization model and infer the user exposure from data. In the language of causal analysis (Imbens and Rubin, 2015), user exposure has close connection to the assignment mechanism. We leverage this connection more directly for explicit data and develop a causal inference approach to recommender systems. We demonstrate that causal inference for recommender systems leads to improved generalization to new data. Exact posterior inference is generally intractable for latent variables models. Throughout this thesis, we will design specific inference procedure to tractably analyze the large-scale data encountered under each scenario.

Electronic mail, email, is one of the most widespread computer applications today.While email in general is very popular among its users, there are also drawbacks withemail usage: an increasing amount of messages that overwhelm users, systems that aretoo complex for naive users and at the same time do not support the needs of experiencedusers.In order to answer the main research question “Which design solutions couldimprove the situation of individual email users in a working context when it comes tocommunication and handling large numbers of incoming and stored email messages?”three studies conducted in email users’ working environment are described. The studiedorganisations are one academic research laboratory, one technical company, andone primary medical service organisation. The studies are focused on email usage,organisation of email messages, novice versus experienced users’ needs, managers’email usage, and information and communication overflow.The results indicate that the different strategies used to handle email are a matter ofa balance between advantages and disadvantages of these strategies. The choicebetween them is depending on the users’ total work situation and cannot be understoodby investigating the email communication alone.One advantage of email is the cognitive comfort it brings to its users by liberatingthem from thinking about tasks that can be solved by sending an email message, butthis advantage disappears when the sender cannot trust that the receiver will act uponthe message.Users develop their handling of email with experience and work position. Themedia that managers use to handle the increased communication that follows with ahigher position are email and meetings. One habit that do not change with position isto allow incoming messages to interrupt other work tasks, despite the asynchronousnature of email. This is particularly remarkable for managers who often complain thatthey need more uninterrupted time. The interruptions may partly be attributed to thelack of functionality in email systems to adapt the interfaces to the users’ work habits.In this case incoming messages result in a signal regardless the importance of them.Email is a part of an information and communication flow. Some users have problemshandling this flow. Overflow problems could be diminished by making senders ofmessages more aware of the receivers’ communicative situation. Email systems couldprovide feedback to senders of messages based on the receivers’ perception of his/hersituation.One of the studies indicates that it may be even more complicated to replace an oldemail system than introducing an email system for the first time in an organisation.The investment experienced users have made in the old system may be substantial.A model of time usage for organisation of email messages is also presented in orderto compare different strategies.Several design solutions are suggested with respect to folder usage, sorting emailmessages into folders, reducing the number of stored messages, and tailoring the emailsystem to the user’s work habits. / QC 20100524

University of Macau / Faculty of Science and Technology / Department of Computer and Information Science