Recommendation systems are primarily used in e-commerce and retail to guide the user in a vast space of available items by providing personalized recommendations that fit the user's interests and need. Numerous types of recommendation systems have been introduced over the years. The most recent development in the field is the sequential recommendation system. Sequential recommenders account for the order in which the user has interacted with items to infer the user's intent, allowing them to provide recommendations accordingly. The data analytic company Siftlab AB has already developed such a recommendation system; however, its application has been limited to transaction data(data depicting only purchases). As a result, the model cannot take advantage of the predictive values of different event types. This thesis introduces a weighted multi-type technique that allows Siftlab's recommendation model to leverage page views alongside purchases in data from an interior design store. We also developed tools and techniques, such as correlation and angle separation analysis, to enhance our examination of user-item behavior. Our research findings indicate that including page view events in training hurts recall, while their inclusion in the prediction stage yields slight improvements. We discovered a rapid decline in correlation between purchases and page views as we considered page views occurring relatively further back in time. Performing a time-based correlation analysis, it became evident that there is a robust time dependency between purchases and page views. Utilizing this time dependency, we enforced a time-dependent threshold on the page views we included in the prediction stage to eliminate irrelevant page view events, further enhancing the model's predictions. We also captured seasonalities phenomena distinctive for an interior design store. Although the result of this work might only be valid for a single data set, we anticipate our work to be the first step in the right direction since the technique we introduce here can be effortlessly adapted to analyze other event types in other data, thus uncovering patterns that can further elevate the model's performance.

Reading is a fundamental skill that each person needs to develop during early childhood and continue to enhance into adulthood. While children/teenagers depend on this skill to advance academically and become educated individuals, adults are expected to acquire a certain level of proficiency in reading so that they can engage in social/civic activities and successfully participate in the workforce. A step towards assisting individuals to become lifelong readers is to provide them adequate reading selections which can cultivate their intellectual and emotional growth. Turning to (web) search engines for such reading choices can be overwhelming, given the huge volume of reading materials offered as a result of a search. An alternative is to rely on reading materials suggested by existing recommendation systems, which unfortunately are not capable of simultaneously matching the information needs, preferences, and reading abilities of individual readers. In this dissertation, we present novel recommendation strategies which identify appealing reading materials that the readers can comprehend, which in turn can motivate them to read. In accomplishing this task, we have examined used-defined data, in addition to information retrieved/inferred from reputable and freely-accessible online sources. We have incorporated the concept of “social trust” when making recommendations for advanced readers and suggested fiction books that match the reading ability of individual K-12 readers using our readability-analysis tool for books. Furthermore, we have emulated the readers' advisory service offered at school/public libraries in making recommendations for K-12 readers, which can be applied to advanced readers as well. A major contribution of our work is in the development of unsupervised recommendation strategies for advanced readers which suggest reading materials for both entertainment and learning acquisition purposes. Unlike their counterparts, these recommendation strategies are unaffected by the cold-start or long-tail problems, since they exploit user-defined data (if available) while taking advantage of alternative publicly-available metadata. Our readability-analysis tool is innovative, which can predict the readability-levels of books on-the-fly, even in the absence of excerpts from books, a task that cannot be accomplished by any of the well-known readability tools/strategies. Moreover, our multi-dimensional recommendation strategy is novel, since it simultaneously analyzes the reading abilities of K-12 readers, which books readers enjoy, why the books are appealing to them, and what subject matters the readers favor. Besides assisting K-12 readers, our recommender can be used by parents/teachers/librarians in locating reading materials to be suggested to their (K-12) children/students/patrons. We have validated the performance of each methodology presented in this dissertation using existing benchmark datasets or datasets we created for the evaluation purpose (which is another contribution we make to the research community). We have also compared the performance of our proposed methodologies with their corresponding baselines and state-of-the-art counterparts, which further verifies the correctness of the proposed methodologies.

Twitter is one of today’s largest and most popular social networks. The users of the service generate huge amounts of data each day and rely heavily on the service helping them find interesting tweets in short time. The concept of hashtags aids in this practice but relies on the users choosing to include the correct and commonly used hashtags for the topic of their tweet. Hashtag recommendation has been a target of research before with varying results. This thesis proposes a method taking the location of the users into account when making recommen- dations. The method generated improved results over just using similar tweets as a basis for recommendation. Various factors like the handling of different variations of vocabulary in the tweets, how many tweets the suggestions can be picked from and how the combination of similarity and geographic ranking should function could affect the result. This leads to the conclusion that geographic data can be used to improve hashtag suggestions, but a different approach in handling similarity and alternative combinations of similarity and geographic ranking could cause another result. / Twitter är ett av nutidens största och populäraste sociala nätverk. Tjänstens användare producerar stora mängder data varje dag och förväntar sig att tjänsten ska kunna hjälpa dem att hitta intressanta tweets snabbt. Därmed finns konceptet med hashtags, men detta förutsätter att användare väljer att inkludera vanligt förekommande hashtags som på ett korrekt sätt avspeglar innehållet i tweeten. Automatisk rekommendation av hashtags har därmed varit ett populärt forskningsämne de senaste åren, med varierande resultat. Denna studie undersöker en rekommendationsmetod som väger in användarens geografiska position för att rekommendera så passande hashtags som möjligt. Resultaten visar att denna metod generellt rekommenderar mer passande hashtags än metoder som enbart rekommenderar hashtags genom att analysera likhet mellan tweets. Olika faktorer så som hanterandet av olika varianter av vokabulär, hur många tweets som metoden kan föreslå hashtags från samt hur kombinationen av rekommendation baserat på likhet och geografiskt position ska fungera, kan samtidigt påverka resultaten. Detta leder till slutsatsen att geografisk data kan användas för att förbättra hashtagrekommendation, men att ett annorlunda tillvägagångsätt i att hantera likhet och alternativa kombinationer av likhetsrangordning och geografisk rangordning kan leda till ett annorlunda resultat.

Online social networks are experiencing an explosive growth in recent years in both the number of users and the amount of information shared. The users join these social networks to connect with each other, share, find content and disseminate information by sending short text messages in near realtime. As a result of the growth of social networks, the users are often experiencing information overload since they interact with many other users and read ever increasing content volume. Thus, finding the "matching" users and content is one of the key challenges for social networks sites. Recommendation systems have been proposed to help users cope with information overload by predicting the items that a user may be interested in. The users' preferences are shaped by personal interests. At the same time, users are affected by their surroundings, as determined by their geographically located communities. Accordingly, our approach takes into account both personal interests and local communities. We first propose a new dynamic recommendation system model that provides better customized content to the user. That is, the model provides the user with the most important tweets according to his individual interests. We then analyze how changes in the surrounding environment can affect the user's experience. Specifically, we study how changes in the geographical community preferences can affect the individual user's interests. These community preferences are generally reflected in the localized trending topics. Consequently, we present TrendFusion, an innovative model that analyzes the trends propagation, predicts the localized diffusion of trends in social networks and recommends the most interesting trends to the user. Our performance evaluation demonstrate the effectiveness of the proposed recommendation system and shows that it improves the precision and recall of identifying important tweets by up to 36% and 80%, respectively. Results also show that TrendFusion accurately predicts places in which a trend will appear, with 98% recall and 80% precision. / Ph. D.

For maximal market penetration, popular mobile applications are typically supported on all major platforms, including Android and iOS. Despite the vast differences in the look-and-feel of major mobile platforms, applications running on these platforms in essence provide the same core functionality. As an application is maintained and evolved, programmers need to replicate the resulting changes on all the supported platforms, a tedious and error-prone programming process. Commercial automated source-to-source translation tools prove inadequate due to the structural and idiomatic differences in how functionalities are expressed across major platforms. In this thesis, we present a new approach---Native-2-Native---that automatically synthesizes code for a mobile application to make use of native resources on one platform, based on the equivalent program transformations performed on another platform. First, the programmer modifies a mobile application's Android version to make use of some native resource, with a plugin capturing code changes. Based on the changes, the system then parameterizes a web search query over popular programming resources (e.g., Google Code, StackOverflow, etc.), to discover equivalent iOS code blocks with the closest similarity to the programmer-written Android code. The discovered iOS code block is then presented to the programmer as an automatically synthesized Swift source file to further fine-tune and subsequently integrate in the mobile application's iOS version. Our evaluation, enhancing mobile applications to make use of common native resources, shows that the presented approach can correctly synthesize more than 86% of Swift code for the subject applications' iOS versions. / Master of Science

Nowadays people and organizations are more and more interconnected in the forms of social networks: the nodes are social entities and the links are various relationships among them. The social network theory and the methods of social network analysis (SNA) are being increasingly used to study such real-world networks in order to support knowledge management and decision making in organizations. However, most existing social network studies focus on the static topologies of networks. The dynamic network link formation process is largely ignored. This dissertation is devoted to study such dynamic network formation process to support knowledge management and decision making in networked environments. Three challenges remain to be addressed in modeling and analyzing the dynamic network link formation processes. The first challenge is about modeling the network topological changes using longitudinal network data. The second challenge is concerned with examining factors that influence formation of links among individuals in networks. The third challenge is regarding link prediction in evolving social networks. This dissertation presents four essays that address these challenges in various knowledge management domains. The first essay studies the topological changes of a major international terrorist network over a 14-year period. In addition, this paper used a simulation approach to examine this network's vulnerability to random failures, targeted attacks, and real world authorities' counterattacks. The second essay and third essay focuses on examining determinants that significantly influence the link formation processes in social networks. The second essay found that mutual acquaintance and vehicle affiliations facilitate future co-offending link formation in a real-world criminal network. The third essay found that homophily in programming language preference, and mutual are determinants for forming participation links in an online Open Source social network. The fourth essay focuses on the link prediction in evolving social networks. It proposes a novel infrastructure for describing and utilizing the discovered determinants of link formation process (i.e. semantics of social networks) in link prediction to support expert recommendation application in an Open Source developer community. It is found that the integrated mechanism outperforms either user-based or Top-N most recognized mechanism.

Coleções digitais de mídias, tanto pessoais como online, crescem rapidamente. Para que grandes quantidades de músicas sejam acessíveis à usuários, serviços populares como iTunes, Last.fm e Pandora oferecem recomendações. Essa abordagem livra usuários de lembrarem de músicas, e permite a descoberta de canções novas ou esquecidas. Mas recomendações apresentam problemas com usuários, como credibilidade e falta de controle. A motivação deste trabalho é melhorar a experiência de usuários com recomendações de música através do uso de explicações. Ao usar um sistema de recomendação, a satisfação e aprovação de usuários não depende só da eficácia do algoritmo, mas também de explicações. Pesquisas mostram que estas podem beneficiar sistemas de recomendação, aumentando a credibilidade e satisfação de usuários, ao oferecer mais transparência e formas de correção. O objetivo deste trabalho é projetar e desenvolver uma nova forma de visualização de tags, e testar sua viabilidade para explicar e filtrar recomendações de músicas. Mais precisamente, investigamos se esta visualização pode favorecer as metas de inspeção (scrutability), eficiência, eficácia e satisfação. A partir da pesquisa em necessidades de usuários para recomendações e música, a visualização Tag Strings foi projetada e desenvolvida. Tag Strings inclui tanto a interface da visualização, quanto o processo de coleta e cálculo de relevância de tags e músicas. Para a avaliação da visualização Tag Strings, dois tipos de experimentos foram construídos: a comparação entre uma lista de recomendações com Tag Strings, e a comparação entre o design de referência (baseado nos serviços Last.fm e Pandora) e Tag Strings. A construção desses dois experimentos permitiu a avaliação de Tag Strings como uma forma de explicação para recomendações de música. Os resultados dos experimentos evidenciam que a nova forma de visualização Tag Strings favorece as metas de inspeção (scrutability), eficiência, eficácia e satisfação, melhorando a usabilidade e experiência de usuários com recomendações de música. / Digital media collections, both personal and online, grow rapidly. To make large music collections available to users, popular services such as iTunes, Last.fm and Pandora offer recommendations. This approach frees users from searching for music, and allows for the discovery of new or forgotten items. But recommendations present issues such as user trust and lack of control. The motivation for this project is to improve user experience with music recommendations through explanations. While using a recommendation system, user acceptance and satisfaction depends not only on the algorithm effectiveness, but also on explanations. Research shows that recommendations benefit from explanations, increasing user trust and satisfaction by offering more transparency and scrutability. The goal of this project is to design and develop a new form of tag visualization, and test its feasibility to explain and filter music recommendations. We specifically investigate if the visualization can support the aims of scrutability, efficiency, effectiveness and satisfaction. Based on the user research and needs for music recommendation, the visualization Tag Strings was designed and developed. Tag Strings includes both the visualization interface and the process of collecting and calculation of tag and track relevancy. To evaluate the visualization Tag String, we designed two types of experiments: comparing Tag Strings with a recommendation list, and comparing Tag Strings with a design reference (based on the services Last.fm and Pandora). The design of these two experiments allowed the evaluation of Tag Strings as a form of explanation to music recommendation. The experiment results highlight that the new visualization Tag Strings favors the aims of scrutability, efficiency, effectiveness and satisfaction, improving the user experience with music recommendations.

A oferta de produtos,informação e serviços a partir de perfis de usuários tem tornado os sistemas de recomendação cada vez mais presentes na Web, aumentando a facilidade de escolha e de permanência dos usuários nestes sistemas. Entretanto, existem otimizações a serem feitas principalmente com relação à modelagem do perfil do usuário. Geralmente, suas preferências são modeladas de modo superficial, devido à escassez das informações coletadas,como notas ou comentários, ou devido a informações indutivas que estão suscetíveis a erros. Esta dissertação propõe uma ferramenta de recomendação baseado em interações multimodais, capaz de combinar informações de usuários processadas individualmente por algoritmos de recomendação tradicionais. Nesta ferramenta desenvolveram-se quatro técnicas de combinação afim fornecer aos sistemas de recomendação, subsídios para melhoria na qualidade das predições em diversos domínios. / Providing products, information and services from user profiles has made the recommendation systems to be increasingly present, increasing the ease of selection and retention of users in Webservices. However, there are optimizations to be made in these systems mainly with respect to modeling the user profile. Generally, the preferences are modeled superficially, due to the scarcity of information collected, as notes or comments, or because of inductive information that is susceptible to errors. This work proposes are commendation tool based on multimodal interactions that combines users\' interactions, wich are processed individually by traditional recommendation algorithms. In this tool developed four combination of techniques in order to provide recommendation systems subsidies to improve the quality of predictions.

A presente tese versa sobre a construção de um Sistema de Recomendação (SR), denominado RecETC (Recomendador do ETC), para auxiliar no processo de Escrita Coletiva Digital (ECD) no Editor de Texto Coletivo (ETC). O RecETC tem como propósito a recomendação de materiais nos formatos de texto, imagens e vídeos, acerca do assunto que está sendo tratado na produção textual coletiva. Para a sua construção, utilizou-se da metodologia de estudo de caso através da abordagem qualitativa e quantitativa. Para isso, esta pesquisa foi desenvolvida em seis etapas, a saber: 1) Estudo teórico sobre as temáticas de Sistemas de Recomendação e Escrita Coletiva Digital, visando aprofundar o conhecimento nas respectivas áreas e identificar trabalhos correlatos. 2) Construção de Objetos de Aprendizagem produzidos como material de apoio para os cursos de extensão. 3) Desenvolvimento da primeira versão do RecETC. 4) Aplicação da primeira versão através de um curso piloto. 5) Desenvolvimento da segunda versão do RecETC 6) Aplicação da segunda versão em curso de extensão. Os dados foram coletados por meio de questionários e analisados tendo como base a metodologia de Análise de Conteúdo, o que possibilitou a definição de três categorias: Categoria I - O ETC como ambiente de Escrita Coletiva Digital, Categoria II - Requisitos técnicos do RecETC e Categoria III - Requisitos pedagógicos do RecETC. A partir do estudo do referencial teórico, do desenvolvimento e da análise das aplicações do RecETC por meio das categorias definidas, foi possível mapear os requisitos necessários para a sua construção e responder ao problema de pesquisa. Esses foram classificados em técnicos e/ou pedagógicos visando enfatizar os aspectos de funcionamento e as contribuições educacionais do RecETC para a ECD. Além disso, foi elaborado um plano de ação para auxiliar professores e alunos na ECD com o apoio do RecETC. Por fim, os resultados indicam que o desenvolvimento do RecETC atende ao propósito desse estudo e os requisitos identificados podem servir de referência para a construção de outros SR voltados para a ECD. / The present thesis deals with the construction of a Recommendation System (SR), called RecETC (ETC Recommender), to assist in the Digital Collective Writing (ECD) process in the Collective Text Editor (ETC). RecETC purpose is to recommend materials in text, image and video formats about the subject being treated in collective textual production. For its construction, it was used the methodology of case study through the qualitative and quantitative approach. For this, this research was developed in six stages, namely: 1) Theoretical study on the topics of Recommendation Systems and Digital Collective Writing, aiming to deepen the knowledge in the respective areas and to identify related works. 2) Construction of Learning Objects produced as support material for extension courses. 3) Development of the first version of RecETC. 4) Application of the first version through a pilot course. 5) Development of the second version of RecETC 6) Application of the second version in the course of extension. The data were collected through questionnaires and analyzed based on the Content Analysis methodology, which enabled the definition of three categories: Category I - ETC as a Digital Collective Writing environment, Category II - Technical requirements of RecETC and Category III - Pedagogical requirements of RecETC. From the study of the theoretical reference, development and analysis of RecETC applications through the defined categories, it was possible to map the necessary requirements for its construction and to respond to the research problem. These were classified as technical and / or pedagogical in order to emphasize the functional aspects and educational contributions of RecETC to ECD. In addition, a plan of action was developed to assist teachers and students in ECD with the support of RecETC. Finally, the results indicate that the development of RecETC fulfills the purpose of this study and the requirements identified can serve as a reference for the construction of other SRs focused on ECD.

A busca de informações em bases de dados de instituições que possuem grande volume de dados necessita cada vez mais de processos mais eficientes para realização dessa tarefa. Problemas de grafia, idioma, sinonímia, abreviação de termos e a falta de padronização dos termos, tanto nos argumentos de busca, quanto na indexação dos documentos, interferem diretamente nos resultados. Diante disso, este estudo teve como objetivo avaliar o impacto da adição de vocabulários estruturados da área de Ciências da Saúde no Currículo Lattes, na recuperação de perfis similares de pesquisadores das áreas de Ciências Biológicas e Ciências da Saúde, utilizando técnicas de mineração de dados, expansão de consultas, modelos vetoriais de consultas e utilização de algoritmo de trigramas. Foram realizados cruzamentos de informações entre as palavras-chaves de artigos publicados registrados no Currículo Lattes e as informações contidas no Medical Subject Headings (MeSH) e nos Descritores em Ciências da Saúde (DeCS), bem como comparações entre os resultados das consultas, utilizando as palavras-chaves originais e adicionando-lhes os termos resultantes do processo de expansão de consultas. Os resultados mostram que a metodologia adotada neste estudo pode incrementar qualitativamente o universo de perfis recuperados, podendo dessa forma contribuir para a melhoria dos Sistemas de Informações do Conselho Nacional de Desenvolvimento Científico e Tecnológico - CNPq. / Information retrieval in large databases need increasingly more efficient ways for accomplishing this task. There are many problems, like spelling, language, synonym, acronyms, lack of standardization of terms, both in the search arguments, as in the indexing of documents. They directly interfere in the results. Thus, this study aimed to evaluate the impact of the addition of structured vocabularies of Health Sciences area in Lattes Database, in the recovery of similar profiles of researchers that work in Biological Sciences and Health Sciences, using Query Expansion, Data Mining procedures, Vector Models and Trigram Phrase Matching algorithm. Crosschecking keywords of articles registered in Lattes Database and Medical Subject Headings (MeSH) and Health Sciences Descriptors (DeCS) terms, as well as comparisons between the results of queries using the original keywords and adding them to query expansion terms. The results show that the methodology used in this study can qualitatively increase the set of recovered profiles, contributing to the improvement of CNPq Information Systems.