INVITED TALKS
Khurshid Ahmad
Trinity College Dublin
Artificial Ontologies and Real Thoughts: Populating the Semantic Web?
Abstract
The discovery of semantics by artificial intelligence researchers in the mid 20th century will serve as an exemplar of what converts feel about an extant faith that the converts have discovered for themselves. The latter-day AI researchers carried the enthusiastic confidence of the early converts: Sowa and Schank and Wilks and Spärck-Jones and McCarthy and Minsky. Broad and nebulous terms coined by these early pioneers like frames, conceptual graphs, semantic preference, inheritance and circumscription gave way to explicit conceptualizations, terminological logics, and a whole host of web acronyms– OIL, DAML and come to mind here.
The original mission of AI (c. 1960’s)was to have an overarching view of knowledge, thought and cognition. This view was expected to inform the development of intelligent systems but led to systems that could only solve ‘toy problems’. The revised mission (c. 1980’s) of was to have a narrow application driven view of knowledge, thought and cognition and this view led to the development systems that can solve problems and learn. Now, one of the tasks in hand for the AI community is to help in the categorisation and indexation of documents in the growing repositories of documents comprising text and images. An index comprises key words (and now key images as well) and the possible relationships between the words. A number of the relationships can be animated using representation schemes - some schemes are based in formal logic whilst others involve networks that link frame-like structures. If one restricts ones consideration to a restricted or specialised domain of knowledge, then the animation helps to understand the ontological commitment of the (specialist) domain community. This trace, this ontological commitment or a trace of specialist knowledge, is usually called, domain ontology. Typically, this domain ontology is produced in conjunction with domain experts and through the use of introspection. However, there is a good body of literature that attempts to derive the ontological commitments from specialist documents. The question here is this: does the trace of ontological commitments provide a good grounding for building a robust index or not?
I will talk about how to construct such an index by exploiting what we know about how a text is created by its authors using linguistic devices like lexical repetition and collocation for instance. I will draw from my own experience of building such indices for domains as diverse as cultural anthropology, nuclear physics, behavioural economics and finance, philosophy of science, radiation safety and sewer engineering.
Michael Mateas
University of California, Santa Cruz
Expressive Intelligence: Artificial Intelligence, Games and New Media
Abstract
Artificial intelligence methods open up new possibilities in art and entertainment, enabling the creation of believable characters with rich personalities and emotions, interactive story systems that incorporate player interaction into the construction of dynamic plots, and interactive installations and sculptural works that are able to perceive and respond to the human environment. At the same time as AI opens up new fields of artistic expression, AI-based art itself becomes a fundamental research agenda, posing and answering novel research questions which would not be raised unless doing AI research in the context of art and entertainment. I call this agenda, in which AI research and art mutually inform each other, Expressive AI. These ideas will be illustrated by looking at several current and past projects, including the interactive drama Facade. As a new game genre, interactive drama involves socially and emotionally charged interaction with characters in the context of a dynamically evolving plot.
Manuela Veloso
Carnegie Mellon University
Learning to Select Team Strategies in Finite-Timed Zero-Sum Games
Abstract
Games, by definition, offer the challenge of the presence of an opponent, to which a playing strategy should respond. In finite-timed zero-sum games, the strategy should enable to win the game within a limited playing time. Motivated by robot soccer, in this talk, we will present several approaches towards learning to select team strategies in such finite-timed zero-sum games. We will introduce an adaptive playbook approach with implicit opponent modeling, in which multiple team strategies are represented as variable weighted plays. We will discuss different plays as a function of different game situations and opponents. In conclusion, we will present an MDP-based learning algorithm to reason in particular about current score and game time left. Through extensive simulated empirical studies, we will demonstrate the effectiveness of the learning approach. In addition, the talk will include illustrative examples from robot soccer. The major part of this work is in conjunction with my PhD student Colin McMillen.
Trinity College Dublin
Artificial Ontologies and Real Thoughts: Populating the Semantic Web?
Abstract
The discovery of semantics by artificial intelligence researchers in the mid 20th century will serve as an exemplar of what converts feel about an extant faith that the converts have discovered for themselves. The latter-day AI researchers carried the enthusiastic confidence of the early converts: Sowa and Schank and Wilks and Spärck-Jones and McCarthy and Minsky. Broad and nebulous terms coined by these early pioneers like frames, conceptual graphs, semantic preference, inheritance and circumscription gave way to explicit conceptualizations, terminological logics, and a whole host of web acronyms– OIL, DAML and come to mind here.
The original mission of AI (c. 1960’s)was to have an overarching view of knowledge, thought and cognition. This view was expected to inform the development of intelligent systems but led to systems that could only solve ‘toy problems’. The revised mission (c. 1980’s) of was to have a narrow application driven view of knowledge, thought and cognition and this view led to the development systems that can solve problems and learn. Now, one of the tasks in hand for the AI community is to help in the categorisation and indexation of documents in the growing repositories of documents comprising text and images. An index comprises key words (and now key images as well) and the possible relationships between the words. A number of the relationships can be animated using representation schemes - some schemes are based in formal logic whilst others involve networks that link frame-like structures. If one restricts ones consideration to a restricted or specialised domain of knowledge, then the animation helps to understand the ontological commitment of the (specialist) domain community. This trace, this ontological commitment or a trace of specialist knowledge, is usually called, domain ontology. Typically, this domain ontology is produced in conjunction with domain experts and through the use of introspection. However, there is a good body of literature that attempts to derive the ontological commitments from specialist documents. The question here is this: does the trace of ontological commitments provide a good grounding for building a robust index or not?
I will talk about how to construct such an index by exploiting what we know about how a text is created by its authors using linguistic devices like lexical repetition and collocation for instance. I will draw from my own experience of building such indices for domains as diverse as cultural anthropology, nuclear physics, behavioural economics and finance, philosophy of science, radiation safety and sewer engineering.
Michael Mateas
University of California, Santa Cruz
Expressive Intelligence: Artificial Intelligence, Games and New Media
Abstract
Artificial intelligence methods open up new possibilities in art and entertainment, enabling the creation of believable characters with rich personalities and emotions, interactive story systems that incorporate player interaction into the construction of dynamic plots, and interactive installations and sculptural works that are able to perceive and respond to the human environment. At the same time as AI opens up new fields of artistic expression, AI-based art itself becomes a fundamental research agenda, posing and answering novel research questions which would not be raised unless doing AI research in the context of art and entertainment. I call this agenda, in which AI research and art mutually inform each other, Expressive AI. These ideas will be illustrated by looking at several current and past projects, including the interactive drama Facade. As a new game genre, interactive drama involves socially and emotionally charged interaction with characters in the context of a dynamically evolving plot.
Manuela Veloso
Carnegie Mellon University
Learning to Select Team Strategies in Finite-Timed Zero-Sum Games
Abstract
Games, by definition, offer the challenge of the presence of an opponent, to which a playing strategy should respond. In finite-timed zero-sum games, the strategy should enable to win the game within a limited playing time. Motivated by robot soccer, in this talk, we will present several approaches towards learning to select team strategies in such finite-timed zero-sum games. We will introduce an adaptive playbook approach with implicit opponent modeling, in which multiple team strategies are represented as variable weighted plays. We will discuss different plays as a function of different game situations and opponents. In conclusion, we will present an MDP-based learning algorithm to reason in particular about current score and game time left. Through extensive simulated empirical studies, we will demonstrate the effectiveness of the learning approach. In addition, the talk will include illustrative examples from robot soccer. The major part of this work is in conjunction with my PhD student Colin McMillen.