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

A variety of statistical methods for noun compound analysis are implemented and compared. The results support two main conclusions. First, the use of conceptual association not only enables a broad coverage, but also improves the accuracy. Second, an analysis model based on dependency grammar is substantially more accurate than one based on deepest constituents, even though the latter is more preva- lent in the literature.

I declare that this dissertation has not been submitted as an exercise for a degree at this or any other university and that is entirely my own work. I agree that the Library may lend or copy this dissertation on request.

We present a brief overview of the main challenges in understanding the semantics of noun compounds and consider some known methods. We introduce a new task to be part of SemEval-2010: the interpretation of noun compounds using paraphrasing verbs and prepositions. The task is meant to provide a standard testbed for future research on noun compound semantics. It should also promote paraphrase-based approaches to the problem, which can benefit many NLP applications. 1

Many methods are available for computing semantic similarity between individual words, but certain NLP tasks require the comparison of word pairs. This paper presents a kernel-based framework for application to relational reasoning tasks of this kind. The model presented here combines information about two distinct types of word pair similarity: lexical similarity and relational similarity. We present an efficient and flexible technique for implementing relational similarity and show the effectiveness of combining lexical and relational models by demonstrating state-ofthe-art results on a compound noun interpretation task. 1

Abstract not found

Every year new noun-noun combinations enter the English language and become common parlance; compounds like "notebook computer" and "soccer mom". But, why is one pair of words chosen rather than another pair ? For example,why do we not use "patio-tile computer" and "sports mother" ? Clearly, many factors influence the process. We concentrate on the cognitive factor of informativeness; namely, that a novel combination should convey its meaning unambiguously. Costello & Keane (1996) have shown that some classes of concept promote ambiguity (or polysemy) in novel nounnoun compounds; artifact and superordinate terms promote polysemy whereas natural-kind and basic-level terms do not. Here we show that the topology of these conceptual classes in a large corpus of familiar compounds indicates that they constrain the compounds that appear in a language. Introduction Each year new nominal compounds enter the English language and extend the everyday vocabulary we have to describe our world; rec...

: Our project focuses on the calculability of the semantics of nominal compounds. Our goal is to design a general model, based on domain-free lexical information, in order to exhibit and implement the principles of nominal compound interpretation. This model is based upon a precise semantic characterization of nominal constituents, which relates nouns to the predicative information that must be identified to retrieve the underlying relation of the compound. The predicate is deduced from the morpho-syntactic and semantic features of the nouns and its argument structure is used to characterize the roles of each constituent. We describe our model of interpretation of English compounds and evaluate it from the results of a program that implements this general framework. We suggest solutions to enrich this model and to adapt it to the characteristics of the compounds of a specialized corpus through the extraction of specific semantic information. Key-words: Natural Language Processing, Nom...

: Our project focuses on the calculability of the semantics of nominal compounds. Our goal is to design a general model, based on domain-free lexical information, in order to exhibit and implement the principles of nominal compound interpretation. This model is based upon a precise semantic characterization of nominal constituents, which relates nouns to the predicative information that must be identified to retrieve the underlying relation of the compound. The predicate is deduced from the morpho-syntactic and semantic features of the nouns and its argument structure is used to characterize the roles of each constituent. We describe our model of interpretation of English compounds and evaluate it from the results of a program that implements this general framework. We suggest solutions to enrich this model and to adapt it to the characteristics of the compounds of a specialized corpus through the extraction of specific semantic information. Key-words: Natural Language Processing, Nom...