We present the design of a practical context-sensitive glosser, incorporating current techniques for lightweight linguistic analysis based on large-scale lexical resources. We outline a general model for ranking the possible translations of the words and expressions that make up a text. This information can be used by a simple resource-bounded algorithm, of complexity O(n log n) in sentence length, that determines a consistent gloss of best translations. We then describe how the results of the general ranking model may be approximated using a simple heuristic prioritisation scheme. Finally we present a preliminary evaluation of the glosser's performance. Shake-and-BakeTranslation. In Constraints, Language and Computation. C.J.Rupp, M.A.Rosner and R.L.Johnson (eds.) Academic Press.4 threshold 1%, radius 12 5 threshold 4%, radius 5

This book contains an introduction to intelligent reasoning. This topic is introduced from a practical viewpoint: the author aims to familiarise the reader with the implementation of intelligent reasoning using Prolog programs. The practical nature of the book is supplemented by a wealth of theoretical foundations. Most of these are based on recent journal articles and conference papers in the computational logic and artificial intelligence fields. To underline the practical character of the book, it is accompanied by a computer disk containing the Prolog programs from the book. The book's author, Peter Flach, is an Assistant Professor at the Faculty of Economics of Tilburg University, and a Researcher at the Institute for Language Technology and Artificial Intelligence (ITK). Flach has written a fair number of papers, many of which deal with inductive logic programming. This book contains a foreword written by Professor Robert Kowalski, author of the book Logic for Problem Solving. The book consists of three main parts, dealing with Logic and Logic Programming (part I), Reasoning with Structured Knowledge (part II), and Advanced Reasoning Techniques (part III). Each of the parts contains three chapters. In part I, Logic Programming and its foundations are introduced. The opening chapter deals with the main concepts in logic programming, and introduces the basics of Prolog. Chapter two is mainly theoretical in nature, and deals with resolution theorem proving in clausal logic. Concepts like Herbrand models, resolution refutation, soundness and completeness are introduced in a gradual way, starting with relational clausal logic, and working through to full clausal logic and definite clause logic. To conclude the first part, chapter three is a thorough discussion of Prolog programming. Besides the basic concepts every Prolog programmer should know, the author also discusses more complex issues like metainterpreters. Whereas in part I the focus was on logic programming, part II takes an artificial intelligence point of view. In this part, the building-blocks of AI systems are introduced: graphs and search. Chapter four discusses graphs, and their use to represent structured knowledge in Prolog. Special treatment is given to the topic of inheritance hierarchies. Chapters five and six deal with blind and informed searching techniques. In particular, chapter five looks at the use of breadth-first and depth-first search in a logic programming environment. In the discussion of these, the author highlights completeness, optimality and efficiency of each search strategy. Chapter six has Informed Search as its topic, and focuses especially on best-first search. In this final chapter of part II, Flach also briefly deals with optimal best-first search and non-exhaustive informed search. Part III departs from the introductory nature of the first six chapters. It is devoted to advanced topics in intelligent reasoning. Chapter seven contains an interesting treatment of reasoning with natural language. Flach discusses natural lan...

We present the design of a practical context-sensitive glosser, incorporating current techniques for lightweight linguistic analysis based on large-scale lexical resources. We outline a general model for ranking the possible translations of the words and expressions that make up a text. This information can be used by a simple resource-bounded algorithm, of complexity O(n log n) in sentence length, that determines a consistent gloss of best translations. We then describe how the results of the general ranking model may be approximated using a simple heuristic prioritisation scheme. Finally we present a preliminary evaluation of the glosser's performance.