General Report on the 1st ESSENCE Autumn School by Pavlos Andreadis
Alan Bundy (University of Edinburgh)
Reformation: A Domain-Independent Algorithm for Theory Repair
We describe and invite discussion on work in progress on reformation, a new algorithm for the automated repair of faulty logical theories. A fault is revealed by a reasoning failure: either the proof of a false theorem or the failure to prove a true conjecture. Repair suggestions are systematically extracted via analysis of the attempted unification of two formulae. These suggestions will either block an unwanted but successful unification or unblock a wanted but failed unification attempt. In contrast to traditional belief revision and abduction mechanisms, the repairs are to the language of the theory as well as to the deletion or addition of axioms.
Alan Bundy is Professor of Automated Reasoning in the School of Informatics at the University of Edinburgh. His research interests include: the automation of mathematical reasoning, with applications to reasoning about the correctness of computer software and hardware; and the automatic construction, analysis and evolution of representations of knowledge, also called ontologies. His research combines artificial intelligence with theoretical computer science and applies this to practical problems in the development and maintenance of computing systems. He is the author of 272 publications and has held 60 research grants.
He is a fellow of several academic societies, including the Royal Society, the Royal Society of Edinburgh and the Royal Academy of Engineers. The major awards for his research include the IJCAI Research Excellence Award (2007) and the CADE Herbrand Award (2007). He was awarded a CBE in 2012. He was Head of Informatics at Edinburgh (1998-2001) and a member of: the Hewlett-Packard Research Board (1989-91); the ITEC Foresight Panel (1994-96) and both the 2001 and 2008 Computer Science RAE panels (1999-2001, 2005-8). He was the founding Convener of UKCRC (2000-5), a member of the Scottish Science Advisory Committee (2008-12) and a Vice President and Trustee of the British Computer Society with special responsibility for the Academy of Computing (2010-12).
More about Alan Bundy: http://homepages.inf.ed.ac.uk/bundy/
Pavel Shvaiko (Informatica Trentina)
Ontology matching is the task of finding relationships between different ontologies. In semantic web practice it is very important for mediating queries across data sources expressed in different ontologies or for interlinking open data, for instance. Ontology matching has benefited from years of very active research. The goal of this tutorial is to present ontology matching in an inclusive framework and to show by example how this is instantiated in tools for matching and manipulating alignments between ontologies. It also aims at presenting extensions of this framework towards more intricate questions (scalability, reasoning, involving other resources and people): these will also be discussed in a general way and illustrated on practical examples. This tutorial is targeted at people needing to involve ontology matching in their works, at practitioners who want to concretely learn how to start with ontology matching, and at students who are starting research involving ontology matching.
Videos (ESSENCE members only)
Pavel Shvaiko, PhD, is a programme manager, and head of the Culture & Tourism area at Informatica Trentina. He has provided various consulting services, co-authored and co-edited a number of books, contributed to, and published in various international journals and conferences in the fields of Semantic Web, Artificial Intelligence, and Information Systems. He coordinated/participated in a number of European and national industrial projects (e.g., Vivi Fiemme or Your Trentino, which provide a mobile platform for an enhanced tourist experience in the context of big events, such as World Nordic Ski championship of 2013 and Winter Universiade 2013). His specialties include: strategic consulting, innovation management, research and business development with topics involving semantic heterogeneity management.
More about Pavel Shvaiko: http://disi.unitn.it/~pavel/
David Schlangen (University of Bielefeld)
Dialogue Semantics and Pragmatics
The predominant way in which humans create shared meaning is via verbal interaction. This holds both for the acquisition of concepts during infancy as well as for everyday interactions. In this tutorial, I will sketch the basics of how linguistics studies verbal interaction. There is a fundamental puzzle that the study of interaction needs to address: A belief in the narrow sense can only be "in" one head, but yet in the wider sense, we can talk about interlocutors, after interacting, coming to "share beliefs". The need to deal with the uncertainty that is inherent in this process of communication, that is, of making something common, is what structures dialogue, as we will see.
In the second part of the tutorial, we will look at one particular approach to studying this problem, namely by building artificial agents that can take part in dialogue and create some semblance of shared meaning (and ideally, do something useful).
Slides part 1 (PDF)
Slides part 2 (PDF)
David Schlangen is a professor of Applied Computational Linguistics at Bielefeld University. His main research interest is in language use and verbal interaction. He studies instances of such language use in (more or less) natural settings (but in the lab), deriving from these observations formal models, which he also often implements in computer systems. The goal of the modelling work is to contribute to the understanding of the phenomenon "language use", as well as, in the case of the computer implementations, to build more natural and easy to use interfaces for interacting, or even cooperating, with computer systems.
He teaches classes on "text technology" (basically, computational linguistics with some elements of computing for the humanities) and more generally on speech and language technology (mostly dialogue systems), corpus linguistics, and semantics and pragmatics. Besides this work on dialogue, he also maintains an interest in discourse processing in general, among other things in automatically analysing sentiment in texts.
More about David Schlangen: http://www.uni-bielefeld.de/(en)/lili/personen/dschlangen/
Simon Kirby (University of Edinburgh)
Iterated Learning: induction, cultural evolution, and the origins of linguistic structure
Abstract: Iterated learning describes the process whereby an individual learns their behaviour by exposure to another individual’s behaviour, who themselves learnt it in the same way. It can be seen as a key mechanism of cultural evolution. In this workshop, I will review various methods for understanding how behaviour is shaped by the iterated learning process: computational agent-based simulations; mathematical modelling; and laboratory experiments in humans and non-human animals. I will show how this framework has been used to explain the origins of linguistic structure - specifically, the fundamental “design features” of human language (e.g. compositionality, duality of patterning, recursion). We will discuss how cultural evolution might be considered alongside biological evolution in explanations of language origins, and the extent to which modelling iterated learning can solve some of the puzzles surrounding human uniqueness.
Videos (ESSENCE members only)
Simon Kirby is a professor of Language Evolution at the University of Edinburgh. His work concerns the origin and evolution of language, and the unique ways that culture and biology interact in our species. He has pioneered a new approach to understanding cultural evolution of behaviours such as language which is called Iterated Learning. A number of research groups around the world - in addition to his own - are now studying Iterated Learning using techniques as diverse as mathematical modelling, computational simulation, and psychological experiments. His view is that a complete understanding of human nature requires an account of the complex interactions between individual learning, cultural transmission and biological evolution in human populations.
More about Simon Kirby: http://www.lel.ed.ac.uk/~simon/
Matthijs Spaan (Delft University of Technology), Frans Oliehoek (University of Amsterdam) and Stefan Witwicki (École Polytechnique Fédérale de Lausanne)
Decision-theoretic approaches to planning, coordination and communication in multiagent systems
Decision making is an important skill of autonomous agents, but, in many real-world systems, this task is complicated by uncertainty about the effects of actions and limited sensing capabilities. In particular, we will be concerned with planning problems that optimize how an agent should act given a model of its environment and its task. As agents often do not exist in isolation, attention will be given to the problem of decision making under uncertainty with multiple, interacting agents. Key issues here are how agents should coordinate and whether, what, how and when agents should communicate with each other.
In this tutorial, we will build on the Markov decision process (MDP) and its extensions, such as the multiagent MDP and the partially observable MDP, to formalize such settings. We will treat models ranging from no-communication to full synchronizing communication at every time step, and will discuss a small number of basic solution methods. The extensions to MDPs and POMDPs that we will cover allow a team of agents to coordinate under a variety of different assumptions about what and when agents communicate.
Videos (ESSENCE members only)
Matthijs Spaan is an assistant professor at the Algorithmics group, Delft University of Technology, Delft, The Netherlands. His research focuses on decision making under uncertainty for single agents (such as robots) as well as multiagent systems. A major goal of Artificial Intelligence is designing agents: systems that perceive their environment and execute actions. In particular, a fundamental question is how to build intelligent agents. When uncertainty and many agents are involved, this question is particularly challenging and has not yet been answered in a satisfactory way.
Uncertainty manifests itself in various forms when computing plans for agents, in particular in real-world scenarios involving robots. For an agent in isolation, planning under uncertainty in acting and sensing has been studied using decision-theoretic models like Partially Observable Markov Decision Processes (POMDPs). However, single-agent centralized methods do not suffice for large-scale multiagent systems, for which he studies multiagent extensions such as the decentralized POMDP (Dec-POMDP) model.
More about Matthijs Spaan: http://www.st.ewi.tudelft.nl/~mtjspaan/
Frans Oliehoek is a Research Scholar at the Informatics Institute of the University of Amsterdam and a Lecturer at the Department of Computer Science of the University of Liverpool. He received his Ph.D. in Computer Science (2010) and M.Sc. Artificial Intelligence (2005), both from the University of Amsterdam. From 2010-2013 he was a postdoc at MIT (CSAIL) and Maastricht University (Department of Knowledge Engineering - DKE), after which he was appointed Assistant Professor at DKE. Frans' research focuses on decision making under uncertainty, with emphasis on multiagent systems. He organized several workshops on Multiagent Sequential Decision Making Under Uncertainty at AAMAS and taught tutorials on Decision Making Under Uncertainty at AAMAS and the European Agent Systems Summer School. He received the best PC-member award at AAMAS 2012, and was awarded a prestigious NWO VENI research grant that funds his research since the beginning of 2014.
More about Frans Oliehoek: http://www.fransoliehoek.net/
Stefan Witwicki is currently a scientist at EPFL working in the intersection of planning and robotics. He obtained his PhD at the University of Michigan in 2011, where he was advised by Ed Durfee. His research interests include planning and reasoning under uncertainty, multiagent coordination and applications to robotics, smartgrids, and automated services. He is active in the AAMAS and AAAI research communities, where he has served on a number of related program and review committees, has received nominations for best paper and best dissertation awards, and has been co-organizing the annual Workshop on Multiagent Sequential Decision Making Under Uncertainty (MSDM) since 2011.
More about Stefan Witwicki: http://people.epfl.ch/stefan.witwicki
Sanjay Modgil (King's College London)
Argumentation and Dialogue
This tutorial teaches fundamental concepts in logic-based models of argumentation and dialogue, and discusses the benefits of such models for agent reasoning and communication in the presence of uncertainty and conflict. Students will first be taught the fundamentals of reasoning using logic-based argumentation. Argument game proof theories will then be presented for these reasoning models, and subsequently generalised to models of dialogue. Students will be taught fundamentals of dialogue protocols which govern the goal orientated submission of agent locutions, the contents of which implicitly define arguments that are evaluated in order to further guide dialogical interactions. The tutorial will then review the benefits of argumentation as a means for facilitating reasoning and communication between both human and/or computational agents, through enhancing the rationality of these communicative interactions.
Slides part 1 (PDF)
Slides part 2 (PDF)
Sanjay Modgil is a lecturer in the AIS (Agents and Intelligent Systems Group) at King's College London. His current work involves research on argumentation theory; in particular the ASPIC+ model of logic-based argumentation, extensions to abstract argumentation systems to accommodate argumentation over preferences and values, and metalevel argumentation. He also works on applications of argumentation to agent reasoning and communication. His previous research interests lie in the areas of default reasoning, belief revision, non-monotonic logics, modal logics, and their applications in artificial intelligence.
More about Sanjay Modgil: http://www.dcs.kcl.ac.uk/staff/smodgil/
Carles Sierra (Artificial Intelligence Research Institute, Spanish Research Council)
In modern IT-enabled societies, the human user is being assisted with an increasing number of tasks by computational communicating entities/software (usually called agents). Agents interact with and act on behalf of their human users. Their assistance could take different forms, starting with simple technical support such as email filtering, information retrieval, shopping, etc., and moving towards full delegation of more complex tasks, such as service composition for travel organization, dispute resolution in the context of divorces, labour controversies, traffic accidents, etc. To support the agents with the more complex tasks, we argue that the concept of “agreement” lies at the basis of agent communication and interaction. Interacting agents will need to base their decisions and actions on explicit agreements. Agreement Computing aims at proposing a plethora of adequate theoretical methods and applied techniques in order to allow for the design and implementation of those new generation "intelligent" communicating artefacts that will form the basis of future modern "mixed" societies populated by interconnected and mutually interacting humans and artefacts.
How to do a PhD and survive it
This talk will give and up-to-date commentary on the book by the Nobel Prize winner Santiago Ramon y Cajal “Advice for a young investigator” published in 1899 and containing a deep reflexion on the values involved in scientific research. Despite being a rather old book its reading is still relevant. It touches upon many questions that young researchers put to themselves. It gives recommendations on publishing behaviour, relevance of research, theoretical vs practical research, motivation for research, ethics of research, and many others.
Carles Sierra is a professor (and vice director) at Artificial Intelligence Research Institute (IIIA) of the Spanish Research Council (CSIC). He is also an adjunct professor at the University of Technology, Sydney (UTS). His expertise is in artificial intelligence broadly, with specific interests in agents, negotiation, and electronic institutions. He has participated in more than twenty research projects funded by the European Commission and the Spanish Government, and has published more than two hundred papers in specialised conferences and scientific journals. He has received several best paper awards in the area of agents and artificial intelligence. He is an ECCAI fellow, and on the editorial board for a range of journals, including the top journal in the area of agents - Journal of Autonomous Agents and Multiagent Systems. (JAAMAS) and Artificial Intelligence (top AI journal).
More about Carles Sierra: http://www.iiia.csic.es/~sierra/public/Home.html
Vincenzo Maltese (KnowDive Research Group, University of Trento)
Linguistic and Knowledge Resources
In order to automate tasks and properly interact with their users, modern ICT applications require very accurate, up-to-date and diversity-aware knowledge resources. Examples of tasks that benefit from such resources are natural language processing (NLP), ontology and classification matching, entity and semantic search, document classification. At this purpose, in the past forty years several general-purpose and domain-specific linguistic and knowledge resources have been developed. Work spans across several disciplines and communities such as Artificial Intelligence, Cognitive Science, Linguistics, Knowledge Representation, and Knowledge Organization. We first introduce the general notions and then go through some notable examples of vocabularies, ontologies and knowledge bases exploring strengths and weaknesses.
Vincenzo Maltese is a post-doc research fellow at the University of Trento, within the KnowDive Research Group. He has published about thirty conference and journal papers. His main area of expertise spans between Knowledge Representation and Knowledge Organization. Current work is mainly devoted to methodologies and tools for the creation and maintenance of knowledge resources. He participated in several projects including InterConcept (mapping large scale Knowledge Organization Systems), LiveMemories (active digital memories of collective lives), Semantic Geo-Catalogue (extending geo-catalogues with semantic capabilities), and the LivingKnowledge EU FET project (dealing with diversity in knowledge). He is currently the project manager of the SmartSociety EU project (hybrid societies of humans and machines). He is co-author of the open source tools S-Match and GeoWordNet (http://semanticmatching.org/).
More about Vincenzo Maltese: http://disi.unitn.it/~maltese/index.html
Sara Shinton (Shinton Consulting Ltd)
Taking Control of your Research Project
A researcher’s role is to grasp a complex problem or topic, develop an approach to address it and present a novel solution. Alongside the intellectual challenge of working at the limits of understanding, there is also a significant challenge in managing your workload.
Project management principles can help you to tame the project and manage your time effectively under pressure. This workshop will take these principles and tailor them for academic research and writing.
You’ll have the opportunity to use a range of tools, helping you to:
- explore possibilities,
- make decisions,
- organise your actions,
- identify priorities,
- monitor progress,
- evaluate your work and your plan.
Since 2000 Sara has run Shinton Consulting Ltd, a researcher development company. Focusing on academic career issues and skills development she works with institutions across the UK and in Europe. Her background includes postdoctoral research (physical chemistry), careers guidance and academic development. She has been short-listed for a Times Higher Education prize on three occasions (2007, 2010 and 2013) winning in 2010 for her role in designing and facilitating the Scottish Crucible and Scottish Futures programmes.
She regularly writes about academic and researcher career development issues and has contributed chapters to a range of books, most recently “Achieving Impact in Research”. She writes regularly for the Institute of Physics on topics including career breaks and career balance, researcher employability and career change. In 2004 she wrote “What Do PhDs Do?” the first ever analysis of doctoral destinations.
Sara is active on social media, including Twitter (@sarashinton) and LinkedIn. Her website includes many resources for researchers and academic leaders (www.shintonconsulting.com).