How do people know as much as they do with as little information as they get? The problem takes many forms; learning vocabulary from text is an especially dramatic and convenient case for research. A new general theory of acquired similarity and knowledge representation, latent semantic analysis (LSA), is presented and used to successfully simulate such learning and several other psycholinguistic phenomena. By inducing global knowledge indirectly from local co-occurrence data in a large body of representative text, LSA acquired knowledge about the full vocabulary of English at a comparable rate to schoolchildren. LSA uses no prior linguistic or perceptual similarity knowledge; it is based solely on a general mathematical learning method that achieves powerful inductive effects by extracting the right number of dimensions (e.g., 300) to represent objects and contexts. Relations to other theories, phenomena, and problems are sketched. Prologue "How much do we know at any time? Much more, or so I believe, than we know we know!" —Agatha Christie, The Moving Finger A typical American seventh grader knows the meaning of

. Currently, most approaches to retrieving textual materials from scientific databases depend on a lexical match between words in users' requests and those in or assigned to documents in a database. Because of the tremendous diversity in the words people use to describe the same document, lexical methods are necessarily incomplete and imprecise. Using the singular value decomposition (SVD), one can take advantage of the implicit higher-order structure in the association of terms with documents by determining the SVD of large sparse term by document matrices. Terms and documents represented by 200-300 of the largest singular vectors are then matched against user queries. We call this retrieval method Latent Semantic Indexing (LSI) because the subspace represents important associative relationships between terms and documents that are not evident in individual documents. LSI is a completely automatic yet intelligent indexing method, widely applicable, and a promising way to improve users...

This paper presents a new approach for resolving lexical ambiguities in one language using statistical data from a monolingual corpus of another language. This approach exploits the differences between mappings of words to senses in different languages. The paper concentrates on the problem of target word selection in machine translation, for which the approach is directly applicable. The presented algorithm identifies syntactic relationships between words, using a source language parser, and maps the alternative interpretations of these relationships to the target language, using a bilingual lexicon. The preferred senses are then selected according to statistics on lexical relations in the target language. The selection is based on a statistical model and on a constraint propagation algorithm, which handles simultaneously all ambiguities in the sentence. The method was evaluated using three sets of Hebrew and German examples and was found to be very useful for disambiguation. The paper includes a detailed comparative analysis of statistical sense disambiguation methods. 1. Introduction The resolution of lexical ambiguities in non-restricted text is one of the most difficult tasks of natural language processing. A related task in machine translation, on which we focus in this paper, is target word selection. This is the task of deciding which target language word is the most appropriate equivalent of a source language word in context. In addition to the alternatives introduced by the different word senses of the source language word, the target language may specify additional alternatives that differ mainly in their usage. Traditionally several linguistic levels were used to deal with this problem: syntactic, semantic and pragmatic. Computationally the syntactic methods...

In this paper, we present a lexical resource where WordNet synsets are annotated with Subject Field Codes. We discuss both the methodological issues we dealt with and the annotation techniques used. A quantitative analysis of the resource coverage, as well as a qualitative evaluation of the proposed annotations, are reported. 1.

The automatic disambiguation of word senses has been an interest and concern since the earliest days of computer treatment of language in the 1950's. Sense disambiguation is an “intermediate task ” (Wilks and Stevenson, 1996) which is not an end in itself, but rather is necessary at one level or another to accomplish most natural language processing tasks. It is

Abstract. In many applications of natural language processing (NLP) it is necessary to determine the likelihood of a given word combination. For example, a speech recognizer may need to determine which of the two word combinations “eat a peach ” and “eat a beach ” is more likely. Statistical NLP methods determine the likelihood of a word combination from its frequency in a training corpus. However, the nature of language is such that many word combinations are infrequent and do not occur in any given corpus. In this work we propose a method for estimating the probability of such previously unseen word combinations using available information on “most similar ” words. We describe probabilistic word association models based on distributional word similarity, and apply them to two tasks, language modeling and pseudo-word disambiguation. In the language modeling task, a similarity-based model is used to improve probability estimates for unseen bigrams in a back-off language model. The similaritybased method yields a 20 % perplexity improvement in the prediction of unseen bigrams and statistically significant reductions in speech-recognition error. We also compare four similarity-based estimation methods against back-off and maximum-likelihood estimation methods on a pseudo-word sense disambiguation task in which we controlled for both unigram and bigram frequency to avoid giving too much weight to easy-to-disambiguate high-frequency configurations. The similaritybased methods perform up to 40 % better on this particular task.

This paper proposes an algorithm for word sense disambiguation based on a vector representation of word similarity derived from lexical co-occurrence. It differs from standard approaches by allowing for as fine grained distinctions as is warranted by the information at hand, rather than supposing a fixed number of senses per word, and by allowing for more than one sense to be assigned to a given word occur-rance. The algorithm is applied to the standard vectorspace information retrieval model and an evaluation is performed over the Category B TREC-1 corpus (WSJ subcollection). Results show that this sense disambiguation algorithm improves performance by between 7o and 1o on aver-age.

Statistical language learning research takes the view that many traditional natural language processing tasks can be solved by training probabilistic models of language on a sufficient volume of training data. The design of statistical language learners therefore involves answering two questions: (i) Which of the multitude of possible language models will most accurately reflect the properties necessary to a given task? (ii) What will constitute a sufficient volume of training data? Regarding the first question, though a variety of successful models have been discovered, the space of possible designs remains largely unexplored. Regarding the second, exploration of the design space has so far proceeded without an adequate answer. The goal of this thesis is to advance the exploration of the statistical language learning design space. In pursuit of that goal, the thesis makes two main theoretical contributions: it identifies a new class of designs by providing a novel theory of statistical natural language processing, and it presents the foundations for a predictive theory of data requirements to assist in future design explorations. The first of these contributions is called the meaning distributions theory. This theory

Manual development of large subcategorised lexicons has proved difficult because predicates change behaviour between sublanguages, domains and over time. Yet access to a comprehensive subcategorization lexicon is vital for successful parsing capable of recovering predicate-argument relations, and probabilistic parsers would greatly benefit from accurate information concerning the relative likelihood of different subcategorisation frames (scfs) of a given predicate. Acquisition of subcategorization lexicons from textual corpora has recently become increasingly popular. Although this work has met with some success, resulting lexicons indicate a need for greater accuracy. One significant source of error lies in the statistical filtering used for hypothesis selection, i.e. for removing noise from automatically acquired scfs. This thesis builds on earlier work in verbal subcategorization acquisition, taking as a starting point the problem with statistical filtering. Our investigation shows that statistical filters tend to work poorly because not only is the underlying distribution zipfian, but there is also very little correlation between conditional distribution of

We consider the use of language models whose size and accuracy are intermediate between different order n-gram models.