In the Human Language Technology (HLT) domain, analytic results extracted from raw document sources are captured in varied models and formats due to the depth of what can be revealed and the diversity of interpretation. However, some common model and format must be followed to allow for multiple analytics to operate together in workflows and enable both the communication between analytics and the fusion of parallel or complementary results. This data integration problem is exacerbated when placing an emphasis on extracting knowledge from text, as the data model must be both adaptable and extensible to handle current and emerging content extraction capabilities and technologies. This paper describes a common interchange format and model designed to coordinate the extracted information from raw document sources in order to generate knowledge. The approach described adheres to the principles of adaptability and extensibility.

The relationship between the human mind and body is something that has occupied philosophers at least since the father of modern philosophy, René Descartes, bequeathed his notorious "dualism" to his successors. For Descartes the mind was a different "substance" compared to the body – the former was a "thinking substance" and the latter an "extended substance", and he resolved the problem of the manner in which these mutually exclusive substances "interacted" by postulating the so-called "animal spirits" – a hybrid concept, denoting something between mind and body – as mediating between them in the pineal gland at the base of the human brain. Increasingly, from the late 19th-century onwards, thinkers started questioning the validity of such dualistic thinking; in various ways philosophers such as Edmund Husserl, Martin Heidegger, Maurice Merleau-Ponty and Jean-Francois Lyotard argued that humans cannot be broken down into mutually exclusive parts, but that they comprised beings characterised by a unity-in-totality. Through many phenomenological analyses Merleau-Ponty, for example, demonstrated that, although – in the event of an injury to your leg, for example – one is able to distance oneself from your body, as if it is something alien to yourself, referring to "the pain in your leg", and so on, it is undeniable that, at a different level of awareness, "you" are in pain, and not just your leg. In short: we don't just have bodies; we "ARE our bodies". This line of thinking, which has far-reaching implications for current thinking about the differences – or the presumed similarities – between humans and artificial intelligence (AI), has been resurrected, perhaps surprisingly, by one of the most brilliant computer-scientists in the world, namely David Gelernter of Yale University in the United States – the subject of a recent article by David Von Drehle (Encounters with the Archgenius; TIME, March 7, 2016, pp.

In recent weeks, it seems like every other day I have encountered another article or media reference to robots and to our anxiety pertaining to their growing presence and role in our life. Fears have ranged from "will they take away our jobs?" to "will they dominate and enslave us?" The latest piece I've read was "Can you trust your robot?" (an ominous and paranoia-tinged title) by a robotics professor in the US. In that article, he explained why human-robot interactions lack the instinctive aspects that human-human relations naturally have, because "we do not understand each other", and more specifically "we cannot tell each other's intentions." In a number of media references at the end of last year, 2015 was identified as the year when artificial intelligence (AI) became one of our prime concerns about the future.

For some years now, the most satisfyingly passive-aggressive way of responding to a factual query on social media has been to reply with a link from the website "Let Me Google That For You". On opening the link, your pesterer sees an animation of their exact query being typed into the Google search field, the "I'm feeling lucky" box being clicked and a page showing what is almost certainly the answer to their question. It is a sadistically elaborate vehicle for a simple message: you are wasting both our time by asking a person something, when you could ask a search engine. But the search engine is hardly infallible. It is commonly assumed these days that all useful information is on the internet, but it isn't.

When the draft version of a federal encryption bill got leaked this month, the verdict in the tech community was unanimous. Critics called it ludicrous and technically illiterate--and these were the kinder assessments of the "Compliance with Court Orders Act of 2016," proposed legislation authored by the offices of Senators Diane Feinstein and Richard Burr. The encryption issue is complex and the stakes are high, as evidenced by the recent battle between Apple and the FBI. Many other technology issues that the country is grappling with these days are just as complex, controversial, and critical--witness the debates over law enforcement's use of stingrays to track mobile phones or the growing concerns around drones, self-driving cars, and 3-D printing. Yet decisions about these technical issues are being handled by luddite lawmakers who sometimes boast about not owning a cell phone or never having sent an email.

LAS VEGAS: SAS has unveiled the new architecture on which all of its future software products will be based, which will include a suite of applications including a machine learning tool. The Viya architecture can be deployed on on-premise or in the cloud, and the company's aim is to make analytics more accessible to all users, including small and medium businesses. SAS founder and chief executive Jim Goodnight announced Viya in his keynote at SAS Global Forum 2016 in Las Vegas, saying that the architecture supports fewer procedures but is more powerful thanks to an in-memory analytics capability. The new architecture has four key applications from the outset, including a machine learning tool aimed at data scientists who want to easily apply machine learning and data mining techniques to structured and unstructured data in a bid to ease the development of algorithms. SAS has also updated its Visual Statistics application, claiming that it is much faster under Viya owing to the architecture's in-memory capabilities.

Microsoft's purchase of Swiftkey for 250 million this month was preceded by Google's 582 million for DeepMind in 2014, a buy vindicated by its subsequent ability to master the ancient Chinese game of Go. In 2012, Amazon bought Cambridge-based Evi Technologies, creator of a Siri-like product that can field users' questions. But if you can't afford to buy an AI startup, algorithms are still within reach. Enterprises with the cash, data and wherewithal are kick-starting their own machine-learning efforts in a bid to gain competitive advantage. In a Computerworld article, John Dodge details the work of sector leaders that are data-rich and savvy enough to develop their own algorithms in-house.

Like the large-vocabulary speech recognition paper we looked at yesterday, today's paper has also been described as a landmark paper in the history of deep learning. The ImageNet dataset contains over 1.5 million labeled high-resolution images of objects in roughly 22,000 categories. The annual ImageNet Large-Scale Visual Recognition Challenge (ILSVRC) competition uses a subset of ImageNet with roughly 1000 images in each of 1000 categories. There are 1.2M training images, 50,000 validation images, and 150,000 testing images. For reporting error rates, a model predicts the top 5 most likely labels.

A popular method for learning from large data sets is Random Forests (see my class on the topic, in Spanish). I would like to drive a paralellism between the way they work and our political decision structures and the so called Wisdom of the crowd. Random Forests are what is called an ensemble method as they perform better than individual methods by combining their results. The individual method used in Random Forests are Decision Trees, trained from a subset of all the available data (and because of this property of operating on subsets of the data, they are a good method for applying on large datasets). More interestingly, Random Forests (as discussed in the Machine Learning article by Leo Breiman in 2001), can not only train each of their trees on a subset of the data but also use a subset of the available information (features) when training each decision node in the tree.

Today, INRIX is launching its completely redesigned GPS app, INRIX Traffic. The app uses machine learning to better plot your routes in addition to providing turn-by-turn navigation with maps from OpenStreetMap (OSM). The app uses smart algorithms to take most routing decisions out of your hands (although you can still manually adjust if you'd like to). "We designed INRIX Traffic with one specific vision: To help drivers move through their daily lives as quickly and efficiently as possible. The app uses our advanced traffic science to make even routine trips easier," said Bryan Mistele, President and CEO of INRIX.