Publicação
Deep linguistic processing of portuguese noun phrases
| Resumo: | This dissertation describes the implementation of a fragment of Portuguese in a computational grammar LXGram currently being developed in the University of Lisbon. LXGram is a computational grammar for the deep linguistic processing of Portuguese. As such, it can be used to parse Portuguese sentences, producing a formal description of their meaning, or to generate Portuguese sentences from meaning representations. LXGram is developed in a platform that is specifically designed to handle such grammars the Linguistic Knowledge Builder (LKB).The LKB implements very efficient algorithms for parsing and generation. It accepts a formalism that is declarative and resorts to unification as the fundamental operation. It also employs a strict type system with multiple inheritance, which provides an elegant means of stating interesting generalizations and allows for static type checking. Several other grammars have been developed in the LKB, for other natural languages. Some of these grammars have been integrated in useful applications, like machine translation, automated e-mail responses, grammar checking and information extraction. This dissertation describes the modeling and computational implementation of a set of linguistic phenomena in LXGram. These phenomena are related to the grammatical properties and the meaning of the Portuguese noun phrase (any Portuguese expression that can appear in the contexts where personal pronouns are allowed). The implementation of these phenomena in LXGram focuses on some aspects that are not very developed in the other LKB grammars. A computational model that accounts for several interesting interactions among them is new. In the first two chapters of this dissertation we provide an introduction to the task at hand, and we describe the tools and formalism that are adopted. The three chapters that follow present the data to be covered and the solutions that were adopted. The last chapter reviews the main points of the dissertation, includes an evaluation of the resulting implementation and suggests future work. |
|---|---|
| Autores principais: | Costa, Francisco Nuno Quintiliano Mendonça Carapeto |
| Assunto: | Natural language processing Unification grammars Typed feature logics Deep linguistic processing Teses de mestrado - 2007 |
| Ano: | 2007 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso restrito |
| Instituição associada: | Universidade de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório da Universidade de Lisboa |
| Resumo: | This dissertation describes the implementation of a fragment of Portuguese in a computational grammar LXGram currently being developed in the University of Lisbon. LXGram is a computational grammar for the deep linguistic processing of Portuguese. As such, it can be used to parse Portuguese sentences, producing a formal description of their meaning, or to generate Portuguese sentences from meaning representations. LXGram is developed in a platform that is specifically designed to handle such grammars the Linguistic Knowledge Builder (LKB).The LKB implements very efficient algorithms for parsing and generation. It accepts a formalism that is declarative and resorts to unification as the fundamental operation. It also employs a strict type system with multiple inheritance, which provides an elegant means of stating interesting generalizations and allows for static type checking. Several other grammars have been developed in the LKB, for other natural languages. Some of these grammars have been integrated in useful applications, like machine translation, automated e-mail responses, grammar checking and information extraction. This dissertation describes the modeling and computational implementation of a set of linguistic phenomena in LXGram. These phenomena are related to the grammatical properties and the meaning of the Portuguese noun phrase (any Portuguese expression that can appear in the contexts where personal pronouns are allowed). The implementation of these phenomena in LXGram focuses on some aspects that are not very developed in the other LKB grammars. A computational model that accounts for several interesting interactions among them is new. In the first two chapters of this dissertation we provide an introduction to the task at hand, and we describe the tools and formalism that are adopted. The three chapters that follow present the data to be covered and the solutions that were adopted. The last chapter reviews the main points of the dissertation, includes an evaluation of the resulting implementation and suggests future work. |
|---|