We propose a novel deep metric learning method. Differently from many works in this area, we define a novel latent space obtained through an autoencoder. The new space, namely S-space, is divided into different regions describing positions where pairs of objects are similar/dissimilar. We locate makers to identify these regions and estimate the similarities between objects through a kernel-based Cauchy distribution to measure the markers’ distance and the new data representation. In our approach, we simultaneously estimate the markers’ position in the S-space and represent the objects in the same space. Moreover, we propose a new regularization function to prevent similar markers from collapsing altogether. Our method emphasizes the group property (separability) while preserving instance representativity. We present evidence that our proposal can represent complex spaces, for instance, when groups of similar objects are located in disjoint regions. We compare our proposal to nine different distance metric learning approaches (four of them are based on deep learning) on 28 real-world heterogeneous datasets. According to the four quantitative metrics used, our method overcomes all of the nine strategies from the literature.

Curiosity is a natural trait of human behavior. When we take into account the time we spend consuming content online, it is expected that at least a fraction of that time was driven by curious behavior. Aiming at understanding how curiosity drives online information consumption, we here propose a model that captures user curiosity relying on several stimulus metrics. Our model relies on the well-established Wundt's curve from psychology and is based on metrics capturing Novelty, Complexity and Uncertainty as key stimuli driving one's curiosity. As a case study, we apply our model on a dataset of online music consumption from LastFM. We found that there are four main types of user behaviors in terms of how the curiosity stimulus metrics drive the user accesses to online music. These are characterized based on the diversity in the songs, artists and musical genres accessed.

How does the presence of an audience influence the social interaction with a conversational system in a physical space? To answer this question, we analyzed data from an art exhibit where visitors interacted in natural language with three chatbots representing characters from a book. We performed two studies to explore the influence of audiences. In Study 1, we did fieldwork cross-analyzing the reported perception of the social interaction, the audience conditions (visitor is alone, visitor is observed by acquaintances and/or strangers), and control variables such as the visitor's familiarity with the book and gender. In Study 2, we analyzed over 5,000 conversation logs and video recordings, identifying dialogue patterns and how they correlated with the audience conditions. Some significant effects were found, suggesting that conversational systems in physical spaces should be designed based on whether other people observe the user or not.

We here focus on the task of learning Granger causality matrices for multivariate point processes. In order to accomplish this task, our work is the first to explore the use of Wold processes. By doing so, we are able to develop asymptotically fast MCMC learning algorithms. With $N$ being the total number of events and $K$ the number of processes, our learning algorithm has a $O(N(\,\log(N)\,+\,\log(K)))$ cost per iteration. This is much faster than the $O(N^3\,K^2)$ or $O(K^3)$ for the state of the art. Our approach, called GrangerBusca, is validated on nine datasets. This is an advance in relation to most prior efforts which focus mostly on subsets of the Memetracker data. Regarding accuracy, GrangerBusca is three times more accurate (in Precision@10) than the state of the art for the commonly explored subsets Memetracker. Due to GrangerBusca's much lower training complexity, our approach is the only one able to train models for larger, full, sets of data.

In modern web, users contact lots of services, identified by the domain name of the server. The temporal sequence and transitions of visited domains form a trajectory of the user on the web. In this work, we analyze 4 weeks of such trajectories, extracted from logs collected in our university network, and mine them via big data and machine learning methodologies to extract the interests of users. Our goal is to create a model of such trajectories and find similarities so to observe peculiarity of users' browsing. Thanks to the model, we propose a methodology to automatically group together the trajectories of single users and/or communities into highly descriptive environments which in turn allow the analyst to identify the topic of interest. We propose an automatic way to highlight differences in terms of popularity and content of environments. Lastly, we analyze the transition among environments, showing how people in smaller communities, e.g., in the same department, have a much more homogeneous behaviour than people at large, e.g., in the university.

How much do visual aspects influence the perception of users about whether they are conversing with a human being or a machine in a mobile-chat environment? This paper describes a study on the influence of typefaces using a blind Turing test-inspired approach. The study consisted of two user experiments. First, three different typefaces (OCR, Georgia, Helvetica) and three neutral dialogues between a human and a financial adviser were shown to participants. The second experiment applied the same study design but OCR font was substituted by Bradley font. For each of our two independent experiments, participants were shown three dialogue transcriptions and three typefaces counterbalanced. For each dialogue typeface pair, participants had to classify adviser conversations as human or chatbot-like. The results showed that machine-like typefaces biased users towards perceiving the adviser as machines but, unexpectedly, handwritten-like typefaces had not the opposite effect. Those effects were, however, influenced by the familiarity of the user to artificial intelligence and other participants' characteristics.

Which song will Smith listen to next? Which restaurant will Alice go to tomorrow? Which product will John click next? These applications have in common the prediction of user trajectories that are in a constant state of flux over a hidden network (e.g. website links, geographic location). Moreover, what users are doing now may be unrelated to what they will be doing in an hour from now. Mindful of these challenges we propose TribeFlow, a method designed to cope with the complex challenges of learning personalized predictive models of non-stationary, transient, and time-heterogeneous user trajectories. TribeFlow is a general method that can perform next product recommendation, next song recommendation, next location prediction, and general arbitrary-length user trajectory prediction without domain-specific knowledge. TribeFlow is more accurate and up to 413x faster than top competitors.

We here investigate what drives the popularity of information on social media platforms. Focusing on YouTube, we seek to understand the extent to which content by itself determines a video's popularity. Using mechanical turk as experimental platform, we asked users to evaluate pairs of videos, and compared users' relative perception of the videos' content against their relative popularity reported by YouTube. We found that in most evaluations users could not reach consensus on which video had better content as their perceptions tend to be very subjective. Nevertheless, when consensus was reached, the video with preferred content almost always achieved greater popularity on YouTube, highlighting the importance of content in driving information popularity on social media.

Understanding the factors that impact the popularity dynamics of social media can drive the design of effective information services, besides providing valuable insights to content generators and online advertisers. Taking YouTube as case study, we analyze how video popularity evolves since upload, extracting popularity trends that characterize groups of videos. We also analyze the referrers that lead users to videos, correlating them, features of the video and early popularity measures with the popularity trend and total observed popularity the video will experience. Our findings provide fundamental knowledge about popularity dynamics and its implications for services such as advertising and search.

User generated content (UGC) has emerged as the predominant form of media publishing on the Web 2.0. Motivated by the large adoption of video sharing on the Web 2.0, the objective of our work is to understand and predict popularity trends (e.g, will a video be viral?) and hits (e.g, how may views will a video receive?) of user generated videos. Such knowledge is paramount to the effective design of various services including content distribution and advertising. Thus, in this paper we formalize the problem of predicting trends and hits in user generated videos. Also, we describe our research methodology on approaching this problem. To the best of knowledge, our work is novel in focusing on the problem of predicting popularity trends complementary to hits. Moreover, we intend on evaluating efficacy of our results not only based on common statistical error metrics, but also on the possible online advertising revenues our predictions can generate. After describing our proposal, we here summarize our latest findings regarding (1) uncovering common popularity trends; (2) measuring associations between UGC features and popularity trends; and (3) assessing the effectiveness of models for predicting popularity trends.

Understanding content popularity growth is of great importance to Internet service providers, content creators and online marketers. In this work, we characterize the growth patterns of video popularity on the currently most popular video sharing application, namely YouTube. Using newly provided data by the application, we analyze how the popularity of individual videos evolves since the video's upload time. Moreover, addressing a key aspect that has been mostly overlooked by previous work, we characterize the types of the referrers that most often attracted users to each video, aiming at shedding some light into the mechanisms (e.g., searching or external linking) that often drive users towards a video, and thus contribute to popularity growth. Our analyses are performed separately for three video datasets, namely, videos that appear in the YouTube top lists, videos removed from the system due to copyright violation, and videos selected according to random queries submitted to YouTube's search engine. Our results show that popularity growth patterns depend on the video dataset. In particular, copyright protected videos tend to get most of their views much earlier in their lifetimes, often exhibiting a popularity growth characterized by a viral epidemic-like propagation process. In contrast, videos in the top lists tend to experience sudden significant bursts of popularity. We also show that not only search but also other YouTube internal mechanisms play important roles to attract users to videos in all three datasets.

The growth of popularity of Web 2.0 applications greatly increased the amount of social media content available on the Internet. However, the unsupervised, user-oriented nature of this source of information, and thus, its potential lack of quality, have posed a challenge to information retrieval (IR) services. Previous work focuses mostly only on tags, although a consensus about its effectiveness as supporting information for IR services has not yet been reached. Moreover, other textual features of the Web 2.0 are generally overseen by previous research.

With the emergence of utility computing and the continuous search for reducing the cost of running information technology (IT) infrastructures, we will soon experience an important change on the way these infrastructures are assembled, configured and managed. In this paper we consider the problem of managing a hybrid high-performance computing infrastructure whose processing elements comprise in-house dedicated machines, a utility computing service provider, and idle machines from a best-effort peer-to-peer grid. This infrastructure supports the execution of both best-effort and real-time applications. Realtime applications use primarily computing power from the in- house machines and any processing power that can be attained from the best-effort grid. Extra capacity required to meet deadlines is purchased from the utility computing service provider. This extra capacity is reserved for future use through short term contracts which are negotiated with no human intervention. We take a business-driven approach for the management of this hybrid infrastructure and propose heuristics that can be used by a contract planner agent to reduce the cost of running the applications at the same time that guarantees that deadlines are met. In particular, we show that constructing an estimation for the behavior of the grid is essential for making contracts that lead to high efficiency in the use of the hybrid infrastructure.

High performance computing is currently not affordable for those users that cannot rely on having a highly qualified computing support team. To cater for these users' needs we have proposed, implemented and deployed OurGrid. OurGrid is a peer-to-peer grid middleware that supports the automatic creation of large computational grids for the execution of embarrassingly parallel applications. It has been used to support the OurGrid Community - a public free-to-join grid that is in production since December 2004. In this paper we show how the OurGrid Community has been used to support the execution of a number of applications. Further we discuss the main benefits brought up by the system and the difficulties that have been faced by the system developers and the users and managers of the OurGrid Community.