Claudia Oliveira

Constrained atomic term: widening the reach of rule templates in transformation based learning

Within the framework of Transformation Based Learning (TBL), the rule template is one of the most important elements in the learning process. This paper presents a new model for TBL templates, in which the basic unit, denominated here as an atomic term (AT), encodes a variable sized window and a test that precedes the capture of a feature’s value. A case study of Portuguese NP identification is described and the experimental results obtained are presented.

Interactive Query Expansion in a Meta-search Engine

In this article we describe a method for refining an initial set of search results, obtained through a meta-search engine, based on relevance feedback. The idea is to interactively obtain from the user a subset of relevant documents in an ongoing query, thereby providing a sample of the related vocabulary. Terms acquired in this way are combined with the terms initially in the query, in order to improve retrieval precision. In our method the user is also asked to select a subset of irrelevant documents, so that terms may be combined negatively in the query. A model of compatible architectures, in which the method can be implemented, is presented. An instance of such model, the system Web Query Reformulator (WQR), is described, with some of its performance results.

A lexicon-based stemming procedure

This paper describes a stemming technique that depends principally on a target languages lexicon, organised as an automaton of word strings. The clear distinction between the lexicon and the procedure itself allows the stemmer to be customised for any language with little or even no changes to the programs source code.

A set of NP-Extraction rules for portuguese: defining, learning and pruning

This paper presents a set of rules for extracting noun phrases from Portuguese texts. We describe how this set was gradually obtained, starting from a machine learned set of transformation rules that was manually reviewed. The noun phrases extracted by these transformations were given as input to another learner that synthesized rules for breaking up complex noun phrases into simpler ones. The results of these processes applied to a Brazilian Portuguese corpus are evaluated.

Corpus-Based compositionality

This paper introduces another mode of semantic compositionality, particularly regarding verbal Multi-word Expressions (MWE). Our hypothesis is that with the increase in similarity between the paragraphs containing a certain MWE and its separate atomic lexical items, the compositionality between these items inside the MWE also increases. The results are consistent with this hypothesis.

Classes de palavras e etiquetagem na Lingüística Computacional

A categorizacao da palavra de acordo com tracos que a posicionam dentro do sistema linguistico e um elemento formal subjacente a qualquer descricao gramatical. Na Linguistica Computacional, etiquetagem consiste na atribuicao de categorias a porcoes do texto. O objetivo desse artigo e discutir, no contexto da Linguistica Computacional, a procedencia da informacao linguistica nos conjuntos de etiquetas de POS – do ingles part of speech . Ao longo da discussao evidenciamos a relevância da participacao do linguista na compilacao teoricamente bem fundamentada dos conjuntos de etiquetas da pratica do Processamento de Linguagem Natural (PLN). Direcionamos nosso olhar, especificamente, para fenomenos relacionados a anotacao por classes de palavras, mas que tem recebido um tratamento secundario por parte da linguistica - como as formas nominais do verbo, notadamente o participio, as palavras denotativas e o aposto. Palavras-chave: conjunto de etiquetas, participio, aposto, palavras denotativas, linguistica computacional, PLN.