We define a new formal grammatical system called a link grammar . A sequence of words is in the language of a link grammar if there is a way to draw links between words in such a way that (1) the local requirements of each word are satisfied, (2) the links do not cross, and (3) the words form a connected graph. We have encoded English grammar into such a system, and written a program (based on new algorithms) for efficiently parsing with a link grammar. The formalism is lexical and makes no explicit use of constituents and categories. The breadth of English phenomena that our system handles is quite large. A number of sophisticated and new techniques were used to allow efficient parsing of this very complex grammar. Our program is written in C, and the entire system may be obtained via anonymous ftp. Several other researchers have begun to use link grammars in their own research. 1 Introduction Most sentences of most natural languages have the property that if arcs are drawn connecti...

We present theoretical algorithms for sorting and searching multikey data, and derive from them practical C implementations for applications in which keys are character strings. The sorting algorithm blends Quicksort and radix sort; it is competitive with the best known C sort codes. The searching algorithm blends tries and binary search trees; it is faster than hashing and other commonly used search methods. The basic ideas behind the algorithms date back at least to the 1960s, but their practical utility has been overlooked. We also present extensions to more complex string problems, such as partial-match searching. 1. Introduction Section 2 briefly reviews Hoare's [9] Quicksort and binary search trees. We emphasize a well-known isomorphism relating the two, and summarize other basic facts. The multikey algorithms and data structures are presented in Section 3. Multikey Quicksort orders a set of n vectors with k components each. Like regular Quicksort, it partitions its input into...

Computer Scrabble programs have achieved a level of performance that exceeds that of the strongest human players. MAVEN was the first program to demonstrate this against human opposition. Scrabble is a game of imperfect information with a large branching factor. The techniques successfully applied in two-player games such as chess do not work here. MAVEN combines a selective move generator, simulations of likely game scenarios, and the B # algorithm to produce a world-championship-caliber Scrabble-playing program.

The construction of minimal acyclic deterministic partial finite automata to represent large natural language vocabularies is described. Applications of such automata include: spelling checkers and advisers, multilanguage dictionaries, thesauri, minimal perfect hashing and text compression. Part of this research was supported by a grant awarded by the Brazilian National Council for Scientific and Technological Development (CNPq) to the second author. Authors' Address: Cl'audio L. Lucchesi and Tomasz Kowaltowski, Department of Computer Science, University of Campinas, Caixa Postal 6065, 13081 Campinas, SP, Brazil. E-mail: lucchesi@dcc.unicamp.br and tomasz@dcc.unicamp.br. 1 Introduction The use of finite automata (see for instance [5]) to represent sets of words is a well established technique. Perhaps the most traditional application is found in compiler construction where such automata can be used to model and implement efficient lexical analyzers (see [1]). Applications of finit...

In the 40 years since Arthur Samuel's 1960 Advances in Computers chapter, enormous progress has been made in developing programs to play games of skill at a level comparable to, and in some cases beyond, what the best humans can achieve. In Samuel's time, it would have seemed unlikely that only a scant 40 years would be needed to develop programs that play world-class backgammon, checkers, chess, Othello, and Scrabble. These remarkable achievements are the result of a better understanding of the problems being solved, major algorithmic insights, and tremendous advances in hardware technology. Computer games research is one of the major success stories of articial intelligence. This chapter can be viewed as a successor to Samuel's work. A review of the scientic advances made in developing computer games is given. These ideas are the ingredients required for a successful program. Case studies for the games of backgammon, bridge, checkers, chess, Othello, poker, and Scrabb...

Abstract not found

this paper attempts to recognize entire words, but should it fail, it attempts to complete the word by consulting a lexicon. Words with identical beginnings are usually morphologically related. The system selects a similar word which fits the apparent size of the input. Even if the wrong form of the word is chosen, the selection of a related word is preferable to no result or a clearly incorrect one. Both static and dynamic approaches to the recognition could benefit from such word ending postulation.

this document 3.6.1. Here, we consider a more informal and empirical dimension of variation that has great impact in the development of parsers and language models: how much of the required predictive and evidential power belongs to the grammar itself and how much resides in the search procedure controlling the use of the grammar. Choices along this dimension often involve philosophical disagreements on whether language is fundamentally governed by an innate system of rules (the rationalist position most closely identified with Chomsky) or rather a system of statistical regularities, associations and constructions derived by learning (the empiricist position informing much work in statistical language modeling). But they also relate to different choices with respect to the coverage/precision tradeoff. At one end of the spectrum, which is often associated with empiricist work, extremely unconstraining grammars are controlled by search evaluation functions automatically learned from language data. An extreme example are finite-state n-gram grammars, in 3.6 Sentence Modeling and Parsing 135 which states encode information on the last n \Gamma 1 observed words, have been used with practical success in speech recognition (Jelinek, Mercer, et al., 1992). In these grammars every sequence of words is considered a possible sentence, but probabilities are assigned to state transitions to model the relative likelihoods of different strings. As we will see, the association of probabilities to transitions is a useful technique in a wide range of grammatical settings. While n-gram grammars have proven very useful for language modeling, derivation steps do not correspond in any direct way to possible meaningful relations in the sentence, for instance, those between a main verb and i...

This paper presents a faster algorithm that uses a GADDAG, a finite automaton that avoids the non-deterministic prefix generation of the DAWG algorithm by encoding a bidirectional path starting from each letter of each word in the lexicon. For a typical lexicon, the GADDAG is nearly five times larger than the DAWG, but generates moves more than twice as fast. This time/space trade-off is justified not only by the decreasing cost of computer memory, but also by the extensive use of move-generation in the analysis of board positions used by Gordon in the probabilistic search for the most appropriate play in a given position within realistic time constraints

A trie structure can immediately determine whether a desired key is in a given key set or not, and can find its longest match easily... this paper. Insertion and deletion operations, as well as key retrieval for this double-trie, are presented. The efficiency of this method is shown by the results of simulations for various key sets.