clarke@csri.toronto.edu (Jim Clarke) (02/13/89)
(SF = Sandford Fleming Building, 10 King's College Road) (GB = Galbraith Building, 35 St. George Street) SUMMARY: COLLOQUIUM - Tues., Feb. 28, 11 a.m. in Room SF 1105 -- Daniel Grayson "Mathematica - A Program for symbolic Manipulation" AI SEMINAR - Thurs., March 2, 11 a.m. in Room SF 1105 -- Dan Fass "Collative Semantics: Main Assumptions, Features, and Implications" SYSTEMS SEMINAR - Thurs., March 2, 2 p.m. in Room GB 305 -- Weidong Chen "C-Logic of Complex Objects" THEORY SEMINAR - Thurs., March 2, 3 p.m. in Room GB 244 -- Trevor J. Smedley "A Fast Heuristic Method for Computations Involving Algebraic Numbers" ------------------------ COLLOQUIUM - Tuesday, February 28, 11 a.m. in Room SF 1105 Daniel Grayson Universite of Illinois "Mathematica - A Program for symbolic Manipulation" Mathematica is a package for performing computations, numerical and symbol- ic, and producing graphical representations of the results. I will explain how we managed to create it in two years, explain some features of its design, and give some examples of its use in mathematics. AI SEMINAR - Thursday, March 2, 11 a.m. in Room SF 1105 Dan Fass Simon Fraser University "Collative Semantics: Main Assumptions, Features, and Implications" Collative Semantics (CS) is a semantic theory for natural language process- ing (NLP) which has been implemented in a computer program called meta5. CS investigates word and sentence meaning in natural language, with exten- sions towards the meaning of whole texts. The main features of CS rest on its linguistic and theoretical assumptions. The main linguistic assump- tions are that lexical ambiguity and what I call "semantic relations" are essential to understanding word and sentence meaning. Seven kinds of seman- tic relation are distinguished: literal, metonymic, metaphorical, anomalous, redundant, inconsistent, and novel relations. Three main theoretical assumptions are made: [1] a division between knowledge and coherence (with a focus on the latter), [2] a framework of four theoretical constructs, including knowledge representation and what I call "coherence representation" (the key to modelling sentence meaning), and [3] a linguistic view of knowledge representation (the way word meaning is represented). These assumptions will be described and the main features of CS will be illustrated in the analysis of a metaphorical sentence. Some implications of CS will be discussed through a contrast with other semantic theories in NLP, notably Wilks' Preference Semantics (from which CS originally developed) and the body of work based on Schank's Conceptual Dependency. SYSTEMS SEMINAR - Thursday, March 2, 2 p.m. in Room GB 305 Weidong Chen SUNY at Stony Brook "C-Logic of Complex Objects" In this talk I present a logic, called C-logic, for the natural representa- tion and manipulation of complex objects. The development of complex objects is a response to the criticism that relational database systems (and logic programming systems) are too low-level to model complex real- world entities naturally. However, in any more complex system, we would like to retain the simple, elegant semantics that underlies relational sys- tems, which makes them easy to understand and use. C-Logic is a language that supports complex objects and has a simple first-order semantics. The language supports object identities as terms which can be constructed from constants and functions. It has a dynamic notion of types, which corresponds to the concept of active domains in databases. Multi-valued labels are inherent features of the logic. They are more general than single-valued labels and support certain pragmatically useful aspects of set manipulation. C-logic also provides a simple frame- work for exploring efficient logic deduction over complex objects. THEORY SEMINAR - Thursday, March 2, 3 p.m. in Room GB 244 Trevor J. Smedley University of Waterloo "A Fast Heuristic Method for Computations Involving Algebraic Numbers" Algorithms for doing computations involving algebraic numbers have been known for quite some time[5,8,10] and implemen- tations now exist in many computer algebra systems[1,3,9]. Many of these algorithms have been analysed and shown to run in poly- nomial time and space[6,7], but in spite of this many real prob- lems take large amounts of time and space to solve. In this talk, I will review the standard algorithms for computing with polynomials over an algebraic extension of the ration- als, and describe a heuristic method which can be used for many operations involving algebraic numbers[4]. The heuristic will not solve all instances of these problems, but it returns either with the correct result or with failure very quickly, and succeeds for a very large number of prob- lems. The heuristic method is similar to, and based on concepts in[2]. I will also describe a method for extending the heuristic to perform alge- braic number polynomial factorisation. A possible method for making the heuristic into a probabilistic algorithm is also given, as well as a method for extending the heuristic to work with multiple extension fields. -- Jim Clarke -- Dept. of Computer Science, Univ. of Toronto, Canada M5S 1A4 (416) 978-4058 clarke@csri.toronto.edu or clarke@csri.utoronto.ca or ...!{uunet, pyramid, watmath, ubc-cs}!utai!utcsri!clarke