Approximating non-linear kernels using feature maps has gained a lot of interest in recent years due to applications in reducing training and testing times of SVM classifiers and other kernel based learning algorithms. We extend this line of work and present low distortion embeddings for dot product kernels into linear Euclidean spaces. We base our results on a classical result in harmonic analysis characterizing all dot product kernels and use it to define randomized feature maps into explicit low dimensional Euclidean spaces in which the native dot product provides an approximation to the dot product kernel with high confidence.

Over the last decade, we have seen an increasing demand for robotscapable of coexisting with people. Enabling robots to operatesafely and effectively alongside human partners withinunstructured environments poses interesting research challengesthat broadly span the field of artificial intelligence. This paperpreviews some of the challenges that we faced in developing arobot "envoy" that operates for extended periods of timethroughout an office-like environment, assisting human occupantswith everyday activities that include greeting and escortingguests as well as retrieving and delivering objects. We see threeskill areas as critical for the robot to effectively perform thesetasks. The first is shared situational awareness-the robot mustinterpret its environment through a world model that is sharedwith its human partners. Secondly, the robot should act in a safe,predictable manner and be capable of intuitive interaction withpeople, through such means as natural language speech. Thirdly,the robot, which we initially treat as a rookie, shouldefficiently utilize information provided by human partners,requesting assistance when necessary and learning from suchassistance to become more competent with time.

Profanity detection is often thought to be an easy task. However, past work has shown that current, list-based systems are performing poorly. They fail to adapt to evolving profane slang, identify profane terms that have been disguised or only partially censored (e.g., @ss, f$#%) or intentionally or unintentionally misspelled (e.g., biatch, shiiiit). For these reasons, they are easy to circumvent and have very poor recall. Secondly, they are a one-size fits all solution – making assumptions that the definition, use and perceptions of profane or inappropriate holds across all contexts. In this article, we present work that attempts to move beyond list-based profanity detection systems by identifying the context in which profanity occurs. The proposed system uses a set of comments from a social news site labeled by Amazon Mechanical Turk workers for the presence of profanity. This system far surpasses the performance of list-based profanity detection techniques. The use of crowdsourcing in this task suggests an opportunity to build profanity detection systems tailored to sites and communities.

While in the past taxonomic and ontological knowledge was traditionally produced by small groups of co-located experts, today the production of such knowledge has a radically different shape and form. For example, potentially thousands of health professionals, scientists, and ontology experts will collaboratively construct, evaluate and maintain the most recent version of the International Classification of Diseases (ICD-11), a large ontology of diseases and causes of deaths managed by the World Health Organization. In this work, we present a novel web-based tool — iCAT Analytics — that allows to investigate systematically crowd-based processes in knowledge-production systems. To enable such investigation, the tool supports interactive exploration of pragmatic aspects of ontology engineering such as how a given ontology evolved and the nature of changes, discussions and interactions that took place during its production process. While iCAT Analytics was motivated by ICD-11, it could potentially be applied to any crowd-based ontology-engineering project. We give an introduction to the features of iCAT Analytics and present some insights specifically for ICD-11.

The first indication of a new outbreak is often in unstructured data (natural language) and reported openly in traditional or social media as a new `flu-like' or `malaria-like' illness weeks or months before the new pathogen is eventually isolated. We present a system for tracking these early signals globally, using natural language processing and crowdsourcing. By comparison, search-log-based approaches, while innovative and inexpensive, are often a trailing signal that follow open reports in plain language. Concentrating on discovering outbreak-related reports in big open data, we show how crowdsourced workers can create near-real-time training data for adaptive active-learning models, addressing the lack of broad coverage training data for tracking epidemics. This is well-suited to an outbreak information-flow context, where sudden bursts of information about new diseases/locations need to be manually processed quickly at short notice.

Currently, the majority of matchers are able to establish simple correspondences between entities, but are not able to provide complex alignments. Furthermore, the resulting alignments do not contain additional information on how they were extracted and formed. Not only it becomes hard to debug the alignment results, but it is also difficult to justify correspondences. We propose a method to generate complex ontology alignments that captures the semantics of matching algorithms and human-oriented ontology alignment definition processes. Through these semantics, arguments that provide an abstraction over the specificities of the alignment process are generated and used by agents to share, negotiate and combine correspondences. After the negotiation process, the resulting arguments and their relations can be visualized by humans in order to debug and understand the given correspondences.

This paper introduces SIFU, a system that recruits in real time native speakers as online volunteer tutors to help answer questions from Chinese language learners in reading news articles. SIFU integrates the strengths of two effective online language learning methods: reading online news and communicating with online native speakers. SIFU recruits volunteers from an online social network rather than recruits workers from Amazon Mechanical Turk.Initial experiments showed that the proposed approach is able to effectively recruit online volunteer tutors, adequately answer the learners' questions, and efficiently obtain an answer for the learner. Our field deployment illustrates that SIFU is very useful in assisting Chinese learners in reading Chinese news articles and online volunteer tutors are willing to help Chinese learners when they are on social network service.

This paper focuses on care plans (i.e., rough schedules) in care houses and evaluates them from the viewpoint of a deep and stable sleep which contributes to provide comfortable and healthy life for aged persons. For this purpose, this paper investigates the care plans which are basically based on the current care plans but change a small part of a schedule as an aged person desires. Through the human subject experiments in the actual care house, the following implications have been revealed: (1) the proposed care plan decreases the time of the light sleep; and (2) the proposed care plan provides the deep sleep (i.e., 9 years younger sleep in our experiment).

Do quantified and origin) and molecular aspects of our bodies like DNA can tweeting, heavily monitored and selfreporting animals, converge. Consumer genomics websites, crowdsourcing of humans, environments and food create some new biodata but also social networking over genes, together uniformity, a dangerously homogenous, objectified and with services monitoring food flows and food authenticity standardized collective or these data offer some new can create new models of research in nutrigenomics and opportunity for interaction? Are we creating new symbiotic projects related to dieting, health and relations over these data that can lead to a new sense of lifestyle choices. How to connect various scales from community or we are witnessing some depersonalization molecules to institutions and what will be the function of and objectification? How to make meaning out of large these interactions and interfaces? How to create quantities of data and how to bring user experience to data meaningful interaction across scales and large datasets?

Service-Oriented Computing (SOC) has received much interest due to its potential to tackle many adaptive system architecture issues that were previously hard to overcome by other computing paradigms. However, it has been facing great difficulty in quickly discovering and dynamically combing available Web services to satisfy given request on-demand. Most of the current researches concentrated o n the semantic model for service discovery, composition, and so on. But there are few studies concerned the intrinsic pattern and law of the service interactions and relationships. To achiev e the vision of SOC in heterogeneous and open environment, in our opinion, not only the semantics of individual Web service but also the interactions and relationships among Web services are needed to be considered seriously. In this paper, beginning with combining Semantic Web and social networking technology within SOC paradigm, we study associations between Web services, mine the relationships among services to design and build Service Network (SN), anal y z e the structural and social characteristics and complexity of SN to reveal the user interests, business requests, information and data flow and direction. In short, we would like to reassess and reconsider the SOC paradigm from the network perspective, through finding new knowledge to build new theoretical basis and approach which can be used to guide and promote the service discovery, composition, and so on, in SOC paradigm.