Lotfi A. Zadeh
University of California, Berkeley, USA
Title: Semantics of Natural Languages — A Constructive Critique
Abstract: Mechanization of natural language understanding (NLU) has long been, and continues to be one of the principal objectives of AI. Semantics of natural languages has a position of centrality in natural language understanding.
Semantics of natural languages is associated with an enormous literature. In large measure, the literature is based on classical, Aristotelian bivalent logic. What is widely unrecognized within the linguistics and computational linguistics communities is that bivalent logic is not well-suited to serve as the logic of natural languages. Imprecision and partiality of truth are pervasive in natural languages. Imprecision and partiality of truth are not tolerated in bivalent logic.
Imprecision of natural languages is rooted in imprecision of perceptions. Basically, a natural language is a system for describing perceptions. Perceptions are intrinsically imprecise, reflecting the bounded ability of human sensory organs, and ultimately the brain, to resolve detail and store information. Imprecision of perceptions is passed on to natural language.
It may be argued, as I have been doing for some time, that theories of natural language should be based on fuzzy logic. In fuzzy logic, as in natural language, everything is or is allowed to be a matter of degree. It is this affinity that contributes to transparency and ease-of-understanding of fuzzy-logic-based theories of natural language.
The point of departure in fuzzy-logic-based semantics of natural language is the concept of a proposition. A proposition p, is viewed as an answer to a question of the form q: What is the value, R, of a variable, X, which is explicit or implicit in p. For example, the proposition p: Vera is middle-aged, may be viewed as an answer to the question, q: What is Vera’s age? In general, q is not unique. However, usually a particular q is suggested by the context. Equivalently, p may be viewed as a constraint, with X and R being the constrained variable and the constraining relation, respectively. Thus, p may be equated to what is referred to as a generalized constraint, p = X isr R, in which r is the modality of the constraint, that is, the way in which R constrains X. Alternatively, p may be viewed as an assignment statement which assigns to X a value, R. In fuzzy-logic-based semantics, the generalized constraint X isr R is viewed as the meaning of p.
The move from a traditional definition of proposition to one based on the concept of a generalized constraint is merely a first step toward a wide-ranging reformulation of traditional approaches to semantics of natural languages. The move may appear to be unnatural to a classically trained linguist. But in reality the move is a necessity when viewed against the backdrop of mechanization of natural language understanding and computation with information described in natural language. The principal features of the fuzzy-logic-based approach to semantics of natural languages are highlighted and illustrated by examples.
Biographie: LOTFI A. ZADEH is a Professor in the Graduate School, Computer Science Division, Department of EECS, University of California, Berkeley. In addition, he is serving as the Director of BISC (Berkeley Initiative in Soft Computing).
Lotfi Zadeh is an alumnus of the University of Tehran, MIT and Columbia University. He held visiting appointments at the Institute for Advanced Study, Princeton, NJ; MIT, Cambridge, MA; IBM Research Laboratory, San Jose, CA; AI Center, SRI International, Menlo Park, CA; and the Center for the Study of Language and Information, Stanford University. His earlier work was concerned in the main with systems analysis, decision analysis and information systems. His current research is focused on fuzzy logic, computing with words and soft computing, which is a coalition of fuzzy logic, neurocomputing, evolutionary computing, probabilistic computing and parts of machine learning.
Lotfi Zadeh is a Fellow of the IEEE, AAAS, ACM, AAAI, and IFSA. He is a member of the National Academy of Engineering and a Foreign Member of the Russian Academy of Natural Sciences, the Finnish Academy of Sciences, the Polish Academy of Sciences, Korean Academy of Science & Technology and the Bulgarian Academy of Sciences. He is a recipient of the IEEE Education Medal, the IEEE Richard W. Hamming Medal, the IEEE Medal of Honor, the ASME Rufus Oldenburger Medal, the B. Bolzano Medal of the Czech Academy of Sciences, the Kampe de Feriet Medal, the AACC Richard E. Bellman Control Heritage Award, the Grigore Moisil Prize, the Honda Prize, the Okawa Prize, the AIM Information Science Award, the IEEE-SMC J. P. Wohl Career Achievement Award, the SOFT Scientific Contribution Memorial Award of the Japan Society for Fuzzy Theory, the IEEE Millennium Medal, the ACM 2001 Allen Newell Award, the Norbert Wiener Award of the IEEE Systems, Man and Cybernetics Society, Civitate Honoris Causa by Budapest Tech (BT) Polytechnical Institution, Budapest, Hungary, the V. Kaufmann Prize, International Association for Fuzzy-Set Management and Economy (SIGEF), the Nicolaus Copernicus Medal of the Polish Academy of Sciences, the J. Keith Brimacombe IPMM Award, the Silicon Valley Engineering Hall of Fame, the Heinz Nixdorf MuseumsForum Wall of Fame, other awards and twenty-six honorary doctorates. He has published extensively on a wide variety of subjects relating to the conception, design and analysis of information/intelligent systems, and is serving on the editorial boards of over sixty journals.
Christian Bizer
Free University of Berlin, Germany
Title: The Web of Linked Data
Abstract: The World Wide Web is a global information space based on the idea to set hyperlinks between documents. In a similar fashion, Linked Data technologies provide for setting data links between records in distinct databases and thus connect these databases into open data spaces that provide for the discovery of related data by following links between databases. Linked Data technologies have been adopted by an increasing number of Web data providers over the last three years, leading to the creation of a global data space containing billions of assertions, the Web of Linked Data. In his talk, Professor Christian Bizer will introduce the principle ideas behind Linked Data. Afterwards, he will give an overview of techniques and tools for publishing Linked Data on the Web and explain the current state-of-the-art in applications that consume Linked Data from the Web.
Biographie: Professor Christian Bizer is the head of the Web-based Systems Group at Freie Universität Berlin, Germany. The group explores technical and economic questions concerning the development of global, decentralized information environments. The results of his work include the Named Graphs data model, which was adopted into the W3C SPARQL standard, the Fresnel display vocabulary implemented by several data browsers, and the D2RQ mapping language which is widely used for mapping relational databases to the Web of Data. He initialized the Linking Open Data community project and the DBpedia project.
Vinay Chaudri
Artificial Intelligence Center, SRI, California, USA
Title: AURA: Capturing Knowledge and Answering Questions on Science Textbooks?
Abstract: AURA is an AI-motivated system with a healthy intersection with the sciences: its short-term goal is to enable domain experts to construct declarative knowledge bases (KBs) from 50 pages of a science textbook for Physics, Chemistry, and Biology in a way that another user can pose questions similar to those in an Advanced Placement (AP) exam and get answers and explanations.
In building AURA, a key question and challenge has been: How much of the knowledge in the three domains can be captured through a generic knowledge capture and reasoning capability and to what extent does it need to be specialized for each domain?
We will attempt to answer this question in the context of the overall design and implementation of AURA. We will also present some recent results on the evaluation of AURA.
Ramesh Jain
University California Irvine, USA
Title: Semantics of Events in Multimedia
Abstract: Multimedia data captures particular characteristics of the environment of interest. Multimedia semantics must relate objects and events in the dynamic world to the sensory signals captured to represent the world and the symbolic information that must be extracted from those signals. Two very important, yet often overlooked, characteristics in multimedia semantics are the dynamic nature of the world as well as different signals used to capture the world and that in most, almost all, cases more than one types of signals must be combined to really represent the world. In this chapter, we discuss the semantics of multimedia by considering some fundamental requirements and how to try to capture those. Events are at least as important as objects in modeling the dynamic universe. Since there is already some body of knowledge on object related semantics, but event semantics has been mostly overlooked, we will focus more on events in our discussion.
Biographie: Ramesh Jain is an educator, researcher, and entrepreneur. Currently he is the Donald Bren Professor in Information & Computer Sciences at University of California, Irvine. He is also a distinguished visiting professor at National University of Singapore. Earlier he served on faculty of Georgia Tech (Farmer Chaired Professor), University of California at San Diego (Founder of Visual Computing Laboratory), The university of Michigan, Ann Arbor (Founding Director of Michigan’s Artificial Intelligence Lab), Wayne State University, and Indian Institute of Technology, Kharagpur. Ramesh was the founding Editor-in-Chief of IEEE Multimedia magazine and has served and continues to serve on the editorial boards of several. He has co-authored more than 350 research papers in well-respected journals and conference proceedings. He has co-authored and co-edited several books. He is a Fellow of ACM, IEEE, AAAI, IAPR, and SPIE. His current research interests are in searching multimedia data and creating EventWebs for experiential computing.
Ramesh co-founded three companies, managed them in initial stages, and then turned them over to professional management. These companies were PRAJA in event-based business activity monitoring (acquired by Tibco); Virage for media management solutions and visual information management (a NASDAQ company acquired by Autonomy); and ImageWare for surface modeling, reverse engineering rapid prototyping, and inspection (acquired by SDRC). Currently, he serves as advisor to four start-up and established companies.
Charles Fillmore
University California Berkeley, USA
Title: Words, Grammar, and Language Understanding
Abstract: This talk will be an exploration of the kinds of linguistic knowledge speakers of a language bring to the task of trying to understand spoken or written texts. Much of what goes on when we try to make sense of what we read or what people are telling us involves details of context and shared experience between communicants, but this talk will be limited to knowledge about language as such. There’s plenty to say.
The talk will build on the story of a lexical resource known as FrameNet and will track the path its researchers have taken
(a) from discovering and recording the meanings and behavior of individual lexical units,
(b) to testing the extent to which the meanings of whole sentences can be constituted by integrating the meanings of its lexical items into the basic grammatical structure of the sentences,
(c) to exploring the vast world of idioms, phrasal patterns, and complex constructions that take us far beyond lexical meanings and simple grammatical relations, but that are absolutely necessary to complete the linguistic part of language understanding.
The general theme is what is involved in knowing a language, how the aspects of that knowledge can be discovered and recorded, and how systematic descriptions of such knowledge can enable computational processes that might lead to improvements in question answering, linguistically sophisticated search, automatic translation, and other language engineering tasks.
Biographie: Charles J. Fillmore is a retired professor Linguistics (University of California, Berkeley) who has been working since 1997 on the computational lexicography project called FrameNet (at the International Computer Science Institute, Berkeley). He earned his doctorate in Linguistics at the University of Michigan in 1960, he taught at the Ohio State University in the 60’s and at UC Berkeley since 1971, after one year (1970-71) at the Center for the Advanced Study in the Behavior Sciences at Stanford. His writings have been mainly on questions of grammar and lexical semantics. His work has touched on issues of deixis, lexicography, case grammar, frame semantics, construction grammar. Information about FrameNet can be found at http://framenet.icsi.berkeley.edu/.
