Do Happy Meals really make us happy? Do salads make us blue? Is cake our comfort? FoodMood is an interactive data visualisation project that gives citizens a rare opportunity to engage and reflect, acknowledge, and understand the connection between emotion, obesity and food. The project explores the opportunities presented by the data-sharing world of today’s cities using global English-language tweets about food coupled with sentiment analysis. It aims to gain a better understanding of global food consumption patterns and its impact on the daily emotional well-being of people against the backdrop of country data such as Gross Domestic Product (GDP) and obesity levels. A key finding is that tweets can be used to find a relationship between certain foods, food sentiment and obesity levels in countries. Overall FoodMood shows a majority positive sentiment towards food. Other findings, although constantly evolving, indicate trends such as: globally meat enjoys a high sentiment rating and is often tweeted about; fast-food companies dominate the food consumption landscapes of most countries’ tweets although not all of them enjoy equal sentiment ratings across countries. Ultimately, FoodMood reveals a hidden layer of meaningful digital, social, and cultural data that provide a basis for further analysis.

As the Web has become integrated into daily life, understanding how individuals spend their time online impacts domains ranging from public policy to marketing. It is difficult, however, to measure even simple aspects of browsing behavior via conventional methods---including surveys and site-level analytics---due to limitations of scale and scope. In part addressing these limitations, large-scale Web panel data are a relatively novel means for investigating patterns of Internet usage. In one of the largest studies of browsing behavior to date, we pair Web histories for 250,000 anonymized individuals with user-level demographics---including age, sex, race, education, and income---to investigate three topics. First, we examine how behavior changes as individuals spend more time online, showing that the heaviest users devote nearly twice as much of their time to social media relative to typical individuals. Second, we revisit the digital divide, finding that the frequency with which individuals turn to the Web for research, news, and healthcare is strongly related to educational background, but not as closely tied to gender and ethnicity. Finally, we demonstrate that browsing histories are a strong signal for inferring user attributes, including ethnicity and household income, a result that may be leveraged to improve ad targeting.

Adolescent cyber-bullying on social networks is a phenomenon that has received widespread attention. Recent work by sociologists has examined this phenomenon under the larger context of teenage drama and it's manifestations on social networks. Tackling cyber-bullying involves two key components – automatic detection of possible cases, and interaction strategies that encourage reflection and emotional support. Key is showing distressed teenagers that they are not alone in their plight. Conventional topic spotting and document classification into labels like "dating" or "sports" are not enough to effectively match stories for this task. In this work, we examine a corpus of 5500 stories from distressed teenagers from a major youth social network. We combine Latent Dirichlet Allocation and human interpretation of its output using principles from sociolinguistics to extract high-level themes in the stories and use them to match new stories to similar ones. A user evaluation of the story matching shows that theme-based retrieval does a better job of finding relevant and effective stories for this application than conventional approaches.

The Isabelle/PIDE platform addresses the question whether proof assistants of the LCF family are suitable as technological basis for educational tools. The traditionally strong logical foundations of systems like HOL, Coq, or Isabelle have so far been counter-balanced by somewhat inaccessible interaction via the TTY (or minor variations like the well-known Proof General / Emacs interface). Thus the fundamental question of math education tools with fully-formal background theories has often been answered negatively due to accidental weaknesses of existing proof engines. The idea of "PIDE" (which means "Prover IDE") is to integrate existing provers like Isabelle into a larger environment, that facilitates access by end-users and other tools. We use Scala to expose the proof engine in ML to the JVM world, where many user-interfaces, editor frameworks, and educational tools already exist. This shall ultimately lead to combined mathematical assistants, where the logical engine is in the background, without obstructing the view on applications of formal methods, formalized mathematics, and math education in particular.

Computer-supported learning is an increasingly important form of study since it allows for independent learning and individualized instruction. In this paper, we discuss a novel approach to developing an intelligent tutoring system for teaching textbook-style mathematical proofs. We characterize the particularities of the domain and discuss common ITS design models. Our approach is motivated by phenomena found in a corpus of tutorial dialogs that were collected in a Wizard-of-Oz experiment. We show how an intelligent tutor for textbook-style mathematical proofs can be built on top of an adapted assertion-level proof assistant by reusing representations and proof search strategies originally developed for automated and interactive theorem proving. The resulting prototype was successfully evaluated on a corpus of tutorial dialogs and yields good results.

Low-dimensional embedding, manifold learning, clustering, classification, and anomaly detection are among the most important problems in machine learning. The existing methods usually consider the case when each instance has a fixed, finite-dimensional feature representation. Here we consider a different setting. We assume that each instance corresponds to a continuous probability distribution. These distributions are unknown, but we are given some i.i.d. samples from each distribution. Our goal is to estimate the distances between these distributions and use these distances to perform low-dimensional embedding, clustering/classification, or anomaly detection for the distributions. We present estimation algorithms, describe how to apply them for machine learning tasks on distributions, and show empirical results on synthetic data, real word images, and astronomical data sets.

Bayesian priors offer a compact yet general means of incorporating domain knowledge into many learning tasks. The correctness of the Bayesian analysis and inference, however, largely depends on accuracy and correctness of these priors. PAC-Bayesian methods overcome this problem by providing bounds that hold regardless of the correctness of the prior distribution. This paper introduces the first PAC-Bayesian bound for the batch reinforcement learning problem with function approximation. We show how this bound can be used to perform model-selection in a transfer learning scenario. Our empirical results confirm that PAC-Bayesian policy evaluation is able to leverage prior distributions when they are informative and, unlike standard Bayesian RL approaches, ignore them when they are misleading.

Boosting is a popular way to derive powerful learners from simpler hypothesis classes. Following previous work (Mason et al., 1999; Friedman, 2000) on general boosting frameworks, we analyze gradient-based descent algorithms for boosting with respect to any convex objective and introduce a new measure of weak learner performance into this setting which generalizes existing work. We present the weak to strong learning guarantees for the existing gradient boosting work for strongly-smooth, strongly-convex objectives under this new measure of performance, and also demonstrate that this work fails for non-smooth objectives. To address this issue, we present new algorithms which extend this boosting approach to arbitrary convex loss functions and give corresponding weak to strong convergence results. In addition, we demonstrate experimental results that support our analysis and demonstrate the need for the new algorithms we present.

In this paper we propose two new algorithms based on biclustering analysis, which can be used at the basis of a recommender system for educational orientation of Russian School graduates. The first algorithm was designed to help students make a choice between different university faculties when some of their preferences are known. The second algorithm was developed for the special situation when nothing is known about their preferences. The final version of this recommender system will be used by Higher School of Economics.

Measurement professionals cannot come to an agreement on the definition of the term 'item fairness'. In this paper a continuous measure of item unfairness is proposed. The more the unfairness measure deviates from zero, the less fair the item is. If the measure exceeds the cutoff value, the item is identified as definitely unfair. The new approach can identify unfair items that would not be identified with conventional procedures. The results are in accord with experts' judgments on the item qualities. Since no assumptions about scores distributions and/or correlations are assumed, the method is applicable to any educational test. Its performance is illustrated through application to scores of a real test.