published_or_final_version / abstract / Speech and Hearing Sciences / Master / Master of Philosophy

Computer Sciences / In order to respond to increasing demand for natural language interfaces---and provide meaningful insight into user query intent---fast, scalable lexical semantic models with flexible representations are needed. Human concept organization is a rich phenomenon that has yet to be accounted for by a single coherent psychological framework: Concept generalization is captured by a mixture of prototype and exemplar models, and local taxonomic information is available through multiple overlapping organizational systems. Previous work in computational linguistics on extracting lexical semantic information from unannotated corpora does not provide adequate representational flexibility and hence fails to capture the full extent of human conceptual knowledge. In this thesis I outline a family of probabilistic models capable of capturing important aspects of the rich organizational structure found in human language that can predict contextual variation, selectional preference and feature-saliency norms to a much higher degree of accuracy than previous approaches. These models account for cross-cutting structure of concept organization---i.e. selective attention, or the notion that humans make use of different categorization systems for different kinds of generalization tasks---and can be applied to Web-scale corpora. Using these models, natural language systems will be able to infer a more comprehensive semantic relations, which in turn may yield improved systems for question answering, text classification, machine translation, and information retrieval. / text

No description available.

The thesis explores the semantics of epistemic justification discourse, a very important part of overall epistemic discourse. It embarks from a critical examination of referentialist theories to arrive at a certain nonreferential, expressivist approach to the semantics of epistemic justification discourse. That is, it criticizes the main referentialist theories and then goes on to argue for an expressivist approach on the basis of its theoretical capacity to outflank the problems referentialist theories meet. In the end, I also identify some problems for a prominent expressivist theory and, as a response to these problems, propose a novel norm-expressivist approach that seems to evade these problems. In particular, in Ch.1 I introduce what I call ‘the epistemic justification puzzle’ and then in Chs.2-4 criticize naturalistic referential theories: analytic naturalistic reductionism, synthetic naturalistic reductionism and epistemic kinds realism. In Ch.5 I criticize nonnaturalist referential theories: what I call ‘naïve’ nonnaturalism and J.McDowell’s (1994) more sophisticated quietist version of nonnaturalism. Next, in Ch.6 I introduce the semantic programme of expressivism and go on to construct a simple version of epistemic norm-expressivism (inspired by A.Gibbard (1990)) in order to explain how expressivism can easily outflank the identified problems of referentialist theories. This simple norm-expressivist theory, however, is only used as a theoretical ‘toy’ for the mere sake of motivating the possibility of expressivism, as in Ch.7 I go on to argue for a more sophisticated version of norm-expressivism: habitsendorsement expressivism. In Ch.7 I introduce a prominent expressivist theory of moral and knowledge discourses, namely, plan-reliance expressivism (credited to A.Gibbard (2003, 2008)) and extend it cover the epistemic justification discourse. I then identify some problems for plan-reliance expressivism as extended to cover justification discourse and in response to these problems propose habits-endorsement expressivism. Habits-endorsement expressivism builds on the intuition that (justified) belief-fixation is habitual and exploits the theoretical flexibility of the notion of habits in order to overcome the identified problems of plan-reliance expressivism.

The programs that are easiest to write and understand are often not the most efficient. This thesis gives methods of converting programs of the former type to those of the latter type; this involves converting definitions of algorithms given as recursion equations using high level primitives into lower level flow chart programs. The main activities involved are recursion removal (c.f. Strong), loop elimination, and the overwriting of shared structures. We have concentrated on the semantics, rather than the syntax, of the programs we are transforming and we have used techniques developed in work done on proving the correctness of programs. The transformations are done in a hierarchical manner and can be regarded as compiling a program defined in a structured manner (Dijkstra) to produce an efficient low level program that simulates it. We describe the implementation of a system that allows the user to specify algorithms in a simple set language and converts them to flow chart programs in either a bitstring or list processing language. Both of these lower languages allow the sharing of structures. The principles are applicable to other domains and we describe how our system can be applied more generally.

Vagueness is ubiquitous in natural language. It seems incompatible with classical, bivalent logic, which tells us that every statement is either true or false, and none is vaguely true. Yet we do manage to reason using vague natural language. In fact, the majority of our day-to-day reasoning involves vague terms and concepts. There is a puzzle here: how do we perform this remarkable feat of reasoning? I argue that vagueness is a kind of semantic indecision. In short, that means we cannot say exactly who is bald and who is not because we have never decided the precise meaning of the word 'bald'--there are some borderline cases in the middle, which might be bald or might not. That is a popular general strategy for addressing vagueness. Those who use it, however, do not often say what they mean by 'borderline case'. It is most frequently used in a loose way to refer to in-between items: those people who are neither clearly bald nor clearly not bald. But under that loose description, the notion of borderline cases is ambiguous, and some of its possible meanings create serious problems for semantic theories of vagueness.Here, I clarify the notion of a borderline case, so that borderline cases can be used profitably as a key element in a successful theory of vagueness. After carefully developing my account of borderline cases, I demonstrate its usefulness by proposing a theory of vagueness based upon it. My theory, vagueness as permission, explains how classical logic can be used to model even vague natural language.

In my dissertation I argue that prominent descriptivist metaethical views face a serious semantic problem. According to standard descriptivism, moral thought and discourse purports to describe some ontology of moral properties and/or relations: e.g., the term `good' purports to refer to some property or cluster of properties. Central to any such theory, then, is the recognition of certain items of ontology which, should they actually exist, would count as the referents of moral terms and concepts. And since one commonly accepted feature of moral thought and discourse is a supervenience constraint, descriptivists hold that any ontology suitable for morality would have to supervene upon non-moral ontology. But this lands descriptivists with the task of providing a semantic account capable of relating this ontology to moral terms and concepts. That is, they must explain why it is that certain items of ontology and not others would count as the referents of moral terms and concepts, in a way that is consistent with the supervenience constraint. I argue that this important explanatory task cannot be carried out. And because the problem generalizes from metaethics to all normativity, we are left with good reason to pursue alternatives to descriptivist accounts of normative semantics.

No description available.

No description available.

No description available.