The proliferation of mobile devices and advances in geo-positioning technologies has fueled the growth of location-based applications, systems and services. Many location-based applications have now gained high popularity and permeated the daily activities of mobile users. This has led to a huge amount of geo-location data generated on a daily basis, which draws significant interests in analyzing and mining ubiquitous location data, especially trajectories of mobile objects moving in road networks (MO trajectories). Mobile trajectories are complex spatio-temporal sequences of location points with varying sample sizes and varying lengths. Mining interesting patterns from large collection of complex MO trajectories presents interesting challenges and opportunities which can reveal valuable insights to the studies of human mobility in many perspectives. This dissertation research contributes original ideas and innovative techniques for mining complex trajectories from whole trajectories, from subtrajectories of significant characteristics, and from semantic location sequences within large-scale datasets of MO trajectories. Concretely, the first unique contribution of this dissertation is the development of NEAT, a three-phase road-network aware trajectory clustering framework to organize MO subtrajectories into spatial clusters representing highly dense and highly continuous traffic flows in a road network. Compared with existing trajectory clustering approaches, NEAT yields highly accurate clustering results and runs orders of magnitude faster by smartly utilizing traffic locality with respect to physical constraints of the road network, traffic flows among consecutive road segments and flow-based density of mobile traffic as well as road network based distances. The second original contribution of this dissertation is the design and development of TraceMob, a methodical and high performance framework for clustering whole trajectories of mobile objects. To our best knowledge, this is the first whole trajectory clustering system for MO trajectories in road networks. The core idea of TraceMob is to develop a road-network aware transformation algorithm that can map complex trajectories of varying lengths from a road network space into a multidimensional data space while preserving the relative distances between complex trajectories in the transformed metric space. The third novel contribution is the design and implementation of a fast and effective trajectory pattern mining algorithm TrajPod. TrajPod can extract the complete set of frequent trajectory patterns from large-scale trajectory datasets by utilizing space-efficient data structures and locality-aware spatial and temporal correlations for computational efficiency. A comprehensive performance study shows that TrajPod outperforms existing sequential pattern mining algorithms by an order of magnitude.

BRILHANTE, Igo Ramalho. Recommending places based on the wisdom-of-the-crowd. 2016. 164 f. Tese (Doutorado em Ciência da Computação)-Universidade Federal do Ceará, Fortaleza, 2016. / Submitted by Anderson Silva Pereira (anderson.pereiraaa@gmail.com) on 2017-06-08T21:15:36Z No. of bitstreams: 1 2016_tese_irbrilhante.pdf: 14886146 bytes, checksum: 6613aa522f50b0c6b1733926b9d1cd5d (MD5) / Approved for entry into archive by Rocilda Sales (rocilda@ufc.br) on 2017-06-09T11:12:35Z (GMT) No. of bitstreams: 1 2016_tese_irbrilhante.pdf: 14886146 bytes, checksum: 6613aa522f50b0c6b1733926b9d1cd5d (MD5) / Made available in DSpace on 2017-06-09T11:12:35Z (GMT). No. of bitstreams: 1 2016_tese_irbrilhante.pdf: 14886146 bytes, checksum: 6613aa522f50b0c6b1733926b9d1cd5d (MD5) Previous issue date: 2016 / The collective opinion of a great number of users, popularly known as wisdom of the crowd, has been seen as powerful tool for solving problems. As suggested by Surowiecki in his books, large groups of people are now considered smarter than an elite few, regardless of how brilliant at solving problems or coming to wise decisions they are. This phenomenon together with the availability of a huge amount of data on the Web has propitiated the development of solutions which employ the wisdom-of-the-crowd to solve a variety of problems in different domains, such as recommender systems, social networks and combinatorial problems. The vast majority of data on the Web has been generated in the last few years by billions of users around the globe using their mobile devices and web applications, mainly on social networks. This information carries astonishing details of daily activities ranging from urban mobility and tourism behavior, to emotions and interests. The largest social network nowadays is Facebook, which in December 2015 had incredible 1.31 billion mobile active users, 4.5 billion “likes” generated daily. In addition, every 60 seconds 510 comments are posted, 293,000 statuses are updated, and 136,000 photos are uploaded. This flood of data has brought great opportunities to discover individual and collective preferences, and use this information to offer services to meet people’s needs, such as recommending relevant and interesting items (e.g. news, places, movies). Furthermore, it is now possible to exploit the experiences of groups of people as a collective behavior so as to augment the experience of other. This latter illustrates the important scenario where the discovery of collective behavioral patterns, the wisdom-of-the-crowd, may enrich the experience of individual users. In this light, this thesis has the objective of taking advantage of the wisdom of the crowd in order to better understand human mobility behavior so as to achieve the final purpose of supporting users (e.g. people) by providing intelligent and effective recommendations. We accomplish this objective by following three main lines of investigation as discussed below. In the first line of investigation we conduct a study of human mobility using the wisdom-of-the-crowd, culminating in the development of an analytical framework that offers a methodology to understand how the points of interest (PoIs) in a city are related to each other on the basis of the displacement of people. We experimented our methodology by using the PoI network topology to identify new classes of points of interest based on visiting patterns, spatial displacement from one PoI to another as well as popularity of the PoIs. Important relationships between PoIs are mined by discovering communities (groups) of PoIs that are closely related to each other based on user movements, where different analytical metrics are proposed to better understand such a perspective. The second line of investigation exploits the wisdom-of-the-crowd collected through user-generated content to recommend itineraries in tourist cities. To this end, we propose an unsupervised framework, called TripBuilder, that leverages large collections of Flickr photos, as the wisdom-of-the-crowd, and points of interest from Wikipedia in order to support tourists in planning their visits to the cities. We extensively experimented our framework using real data, thus demonstrating the effectiveness and efficiency of the proposal. Based on the theoretical framework, we designed and developed a platform encompassing the main features required to create personalized sightseeing tours. This platform has received significant interest within the research community, since it is recognized as crucial to understand the needs of tourists when they are planning a visit to a new city. Consequently this led to outstanding scientific results. In the third line of investigation, we exploit the wisdom-of-the-crowd to leverage recommendations of groups of people (e.g. friends) who can enjoy an item (e.g. restaurant) together. We propose GroupFinder to address the novel user-item group formation problem aimed at recommending the best group of friends for a < user, item > pair. The proposal combines user-item relevance information with the user’s social network (ego network), while trying to balance the satisfaction of all the members of the group for the item with the intra-group relationships. Algorithmic solutions are proposed and experimented in the location-based recommendation domain by using four publicly available Location-Based Social Network (LBSN) datasets, showing that our solution is effective and outperforms strong baselines. / A opinião coletiva de um grande número de usuários, popularmente conhecida como wisdom-of-the-crowd, tem sido vista como poderosa ferramenta para resolver problemas. Como sugerido por Surowiecki em seus livros, grandes grupos de pessoas são considerados mais inteligentes do que uma elite de poucos, independentemente de quão brilhante na resolução de problemas ou tomadas de decisões sábias esses são. Este fenômeno, juntamente com a disponibilidade de uma enorme quantidade de dados na Web propiciou o desenvolvimento de soluções que empregam a sabedoria da multidão para resolver uma variedade de problemas em diferentes domínios, tais como sistemas de recomendação, redes sociais e problemas combinatoriais. A grande maioria dos dados na Web tem sido gerada nos últimos anos por bilhões de usuários em todo o mundo através de seus dispositivos móveis e aplicações web, principalmente em redes sociais. Esta informação carrega detalhes surpreendentes sobre as atividades dia ́rias, que variam da mobilidade urbana e comportamento de turismo, à emoções e interesses. Atualmente, a maior rede social é o Facebook, que em dezembro de 2015 tinha incríveis 1.31 bilhões de usuários (móveis) ativos, 4.5 bilhões de “likes” gerados diariamente. Além disso, a cada 60 segundos, 510 comentários são publicados, 293.000 status são atualizados e 136.000 fotos são enviadas. Esta inundação de dados trouxe grandes oportunidades para delinear as preferências individuais e coletivas, e usar essas informações para oferecer serviços para atender às necessidades das pessoas, como recomendar itens relevantes e interessantes (por exemplo, notícias, lugares, filmes). Ainda, é possível explorar as experiências de grupos de pessoas como um comportamento coletivo para aumentar a experiência de outros. Este último ilustra o cenário importante onde a descoberta de padrões comportamentais coletivos, a sabedoria da multidão, pode enriquecer a experiência de usuários individuais. Neste sentido, esta tese tem o objetivo de aproveitar a sabedoria da multidão para entender melhor o comportamento da mobilidade humana de modo a alcançar o propósito final de auxiliar os usuários (por exemplo, pessoas), fornecendo recomendações inteligentes e eficazes. Alcançamos esse objetivo seguindo três linhas principais de investigação, conforme discutido abaixo. Na primeira linha de investigação, realizamos um estudo sobre a mobilidade humana usando a sabedoria da multidão, culminando no desenvolvimento de uma estrutura analítica que oferece uma metodologia para entender como os pontos de interesse (PoIs) em uma cidade estão relacionados com base no deslocamento de pessoas. Experimentamos nossa metodologia usando a topologia de rede de PoIs para identificar novas classes de pontos de interesse com base em padrões de visitas, deslocamento espacial de um PoI para outro, bem como popularidade dos mesmos. Relações importantes entre PoIs são mineradas pela descoberta de comunidades (grupos) de PoIs que estão intimamente relacionadas entre si com base nos movimentos do usuário, onde diferentes métricas analíticas são propostas para entender melhor tal perspectiva. A segunda linha de investigação explora a sabedoria da multidão coletada através de conteúdo gerado por usuários para recomendar itinerários em cidades turísticas. Para isso, propomos uma estrutura não supervisionada, chamada TripBuilder, que alavanca grandes coleções de fotos do Flickr e pontos de interesse da Wikipedia, a fim de auxiliar os turistas no planejamento de suas visitas às cidades. Experimentamos extensivamente nossa estrutura usando dados reais, demonstrando assim a eficácia e eficiência da proposta. Com base no arcabouço teórico, desenhamos e desenvolvemos uma plataforma que engloba as principais características necessárias para a realização de passeios turísticos personalizados. Esta plataforma tem recebido um interesse significativo dentro da comunidade de pesquisa, uma vez que este tem sido reconhecido como crucial para entender as necessidades dos turistas quando eles estão planejando uma visita a uma nova cidade. Consequentemente, isto levou a resultados científicos notáveis. Na terceira linha de investigação, exploramos a sabedoria da multidão para realizar recomendações de grupos de pessoas (por exemplo, amigos) que pudessem desfrutar de um determinado item (por exemplo, restaurante) em conjunto. Propomos GroupFinder para abordar o novo problema de formação de grupo de usuário-item destinado a recomendar o melhor grupo de amigos para um determinado par < usuário,item >. A proposta combina informações sobre a relevância do item para o usuário juntamente com a rede social deste (ego network), ao mesmo tempo em que tenta equilibrar a satisfação de todos os membros do grupo pelo item com as relações intra-grupais. Soluções algorítmicas são propostas e experimentadas no domínio de recomendação baseado em localização, utilizando quatro base de dados de rede sociais baseados em local (LBSN) publicamente disponíveis, mostrando que nossa solução é eficaz e supera baselines definidos.

No description available.

Online Health Communities (OHCs) have positively disrupted the modern global healthcare system as patients and caregivers are interacting online with similar peers to improve quality of their life. Social support is the pillar of OHCs and, hence, analyzing the different types of social support activities contributes to a better understanding and prediction of future user engagement in OHCs. This thesis used data from a popular OHC, called Breastcancer.org, to first classify user posts in the community into the different categories of social support using Word2Vec for language processing and six different classifiers were explored, resulting in the conclusion that Random Forest was the best approach for classification of the user posts. This exercise helped identify the different types of social support activities that users participate in and also detect the most common type of social support activity among users in the community. Thereafter, three trajectory-based methods were proposed and implemented to predict user churn (attrition) from the OHC. Comparison of the proposed trajectory-based methods with two non-trajectory-based benchmark methods helped establish that user trajectories, which represent the month-to-month change in the type of social support activity of users are effective pointers for user churn from the community. The results and findings from this thesis could help OHC managers better understand the needs of users in the community and take necessary steps to improve user retention and community management.