INVITED SPEAKERS

John Domingue

Email: j (dot) b (dot) domingue (at) open (dot) ac (dot) uk
Organization: The Knowledge Media Institute, The Open University, http://kmi.open.ac.uk/
Phone: +44 (0)1908 655014
Homepage: http://kmi.open.ac.uk/people/domingue/

John Domingue is the Deputy Director of the Knowledge Media Institute at The Open University and the President of STI International, a semantics focused networking organization with just under 50 members. He has published over 150 refereed articles in the areas of Artificial Intelligence and the Web and his current work is focused on how semantic technology can automate the management, development and use of Web services. Currently he serves as the Scientific Director of SOA4All a 13M Euro project which aims at creating a Web of billions of services. Within the Future Internet arena he helps coordinate the Future Internet Service Offer Working Group within the Future Internet Assembly – a collaboration of 100 European projects with a combined budget of over 600M Euros aiming to develop a next-generation Internet. Prof. Domingue also serves on the editorial boards for the Journal of Web Semantics and the Applied Ontology Journal.

Keynote talk - Linking Services and Linked Data

Web services based on standards such as WSDL and SOAP have had a tremendous impact in enterprise settings enabling corporations to organize internal IT systems and expose functionality to trusted collaborators. The story of services on the Web thus far has not been the same. From sites such as [1] we can see that over the last few years there has been practically no growth in the available services based on standard service technologies from the 28,000 seen a few years ago. A higher growth can be seen in Web APIs which use REST-style communications and are typically combined in mashups. Web APIs have proven popular for accessing Web 2.0 applications and platforms such as Facebook. Another trend that we have seen is the ever increasing exposure of data on the Web based on standards such as RDF(S) and SPARQL. Over the last few years this growth has led to a state where 10s of billions of Linked Data statements are now available. The takeup of this technology by governments and major Web and Media players such as Google, Facebook, Yahoo! and the BBC is leading to an emergent Web of Data: a global online resource of machine-processable statements in semantic form. In this talk I will describe work that we have been carrying out in the area which we term "Linked Services" which seeks to connect the spheres of Linked Data and Web services to form a new global computing platform.

[1] http://webservices.seekda.com/about/web_services

Yukie Nagai
Osaka University
Graduate School of Engineering
2-1 Yamada-oka, Suita, Osaka, 565-0871 Japan

Email: yukie [AT] ams.eng.osaka-u.ac.jp / yukie [AT] ieee.org
Phone: +81 6 6879 4724
Fax: +81 6 6879 4724
Homepage: http://cnr.ams.eng.osaka-u.ac.jp/~yukie/

Yukie Nagai is a Specially Appointed Associate Professor at Osaka University, Japan. She is also a Visiting Researcher at Bielefeld University, Germany. Her research interests are in the area of human-robot interaction for robot learning.

Keynote talk - The Importance of Starting Small in Robot Learning: Lessons from Human Intelligence

How should we design artificial intelligence? What clues can we obtain from human intelligence? Cognitive developmental robotics (Asada et al., 2001) has been advocated to address these questions. It puts emphasis on learning processes of cognitive functions rather than on established abilities to designmore adaptive intelligence for robots as well as to better understand human intelligence. Inspired by human development, we further suggest that starting with limited capacities ensures successful learning. Limitations in perception, action, and memory enable robots to detect more prominent information in sensory input, toefficiently explore an environment, and toacquiremore significant sensorimotor association, which all facilitate learning.

In this talk, I will present our work demonstrating the importance of perceptual development in robot learning. Our robots equipped with visual developmentlearned to acquire social cognitive abilitiessuch as joint attention and imitation. When learning to achieve joint attention, an early restriction in vision enableda robot to detect more dominant and thus important information in visual input (Nagai et al., 2006). The restrictionyieldedbetter representation of a neural network, which improved the robustness of the acquired ability. Forimitating others, a robot needs to acquire correspondence between self- and other-motion. A robot's vision starting with a lower spatiotemporal resolution enhanced the similarities between self and other, which consequently produced a property of mirror neuron system in thesensorimotor mapping (Nagai et al., 2011).I will provide further applications to discuss various roles of starting small.

M. Asadaet al., "Cognitive developmental robotics as a new paradigm for the design of humanoid robots," Robotics and Autonomous Systems, vol. 37, no. 2-3, pp. 185-193, 2001.

Y. Nagaiet al., "Learning for joint attention helped by functional development," Advanced Robotics, vol. 20, no. 10, pp. 1165-1181, 2006.

Y. Nagai et al., "Emergence of Mirror Neuron System: Immature vision leads to self-other correspondence," in Proc. of the 1st Joint IEEE International Conference on Development and Learning and on Epigenetic Robotics, 2011

Reinhard Muskens
Department of Philosophy
Tilburg University
P.O. Box 90153
NL 5000 LE Tilburg
The Netherlands

Email: r.a.muskens@uvt.nl
Phone: +31 13 466 2095
Fax: (+31 13 466 3110
Homepage: http://let.uvt.nl/general/people/rmuskens/

Reinhard Muskens is interested in the logic of ordinary language. The logicians study concepts such as truth, consequence, proof, computation and information with the help of formal languages, but it has increasingly become clear that many insights thus obtained also apply to the vernacular. His particular interest is in the question how linguistic expressions manage to carry meaning. A closely related interest is in the use of logic to model aspects of context and social interaction.

Keynote talk - Natural Logic for Natural Reasoning

What follows from what in natural language? A standard approach to answering this question is to translate the text or discourse under consideration into some kind of logic (first-order logic is a better choice than many people think) and to then check entailments with the help of a theorem prover. This is fine as far as it goes, but much more research into natural language needs to be done before we will be able to provide algorithms translating more than toy fragments of language correctly into logic, preserving the entailment relations that language users perceive.

In this talk I will sketch some desiderata for an alternative computational theory of natural language entailment and I will make some proposals. First of all, I think that a logic for natural language should be based on linguistic representations directly, it should hardly be necessary to translate language into logic. I will explain why satisfying this requirement need not bring us outside the realm of formal logic. A second wish is that the theory should be informed by the psychology of reasoning and in particular by Johnson-Laird's view of reasoning and interpretation as model search. As others have argued before, this view has important connections with tableau theorem proving and model generation. A third desideratum is that the theory should be approximative and should be amenable to gradual improvement. It should not be necessary to have a full-blown theory in place before results can be obtained. I will argue that work in Natural Logic can help satisfy this third requirement and I will give a tableau calculus in which all three threads come together.