Fernando Antônio Asevedo Nóbrega

General Purpose Word Sense Disambiguation Methods for Nouns in Portuguese

Word Sense Disambiguation (WSD) aims at determining the appropriate sense of a word in a particular context. Although it is a highly relevant task for Natural Language Processing, there are few works for Portuguese, which are tailored to specific applications, such as translation and information retrieval. In this work, we report our investigation of some general purpose WSD methods for nouns in Portuguese, tackling two additional challenges: using Princeton Wordnet (for English) as the sense repository and applying/customizing a WSD method for multi-document applications, which, to the best of our knowledge, has not been addressed before. In this paper, we also report our efforts on building a sense annotated corpus (for nouns, only), which was used for evaluating the investigated WSD methods.

Alignment-Based Sentence Position Policy in a News Corpus for Multi-document Summarization

This paper presents an empirical investigation of sentence position relevance in a corpus of news texts for generating abstractive multi-document summaries. Differently from previous work, we propose to use text-summary alignment information to compute sentence relevance.

Investigating Machine Learning Approaches for Sentence Compression in Different Application Contexts for Portuguese

Sentence compression aims to produce a shorter version of an input sentence and it is very useful for many Natural Language applications. However, investigations in this field are frequently task focused and for English language. In this paper, we report machine learning approaches to compress sentences in Portuguese. We analyze different application contexts and the available features. Our experiments produce good results, outperforming some previously investigated approaches.

A Qualitative Analysis of a Corpus of Opinion Summaries based on Aspects

Aspect-based opinion summarization is the task of automatically generating a summary for some aspects of a specific topic from a set of opinions. In most cases, to evaluate the quality of the automatic summaries, it is necessary to have a reference corpus of human summaries to analyze how similar they are. The scarcity of corpora in that task has been a limiting factor for many research works. In this paper, we introduce OpiSums-PT, a corpus of extractive and abstractive summaries of opinions written in Brazilian Portuguese. We use this corpus to analyze how similar human summaries are and how people take into account the issues of aspect coverage and sentiment orientation to generate manual summaries. The results of these analyses show that human summaries are diversified and people generate summaries only for some aspects, keeping the overall sentiment orientation with little variation.

Update summarization: building from scratch for Portuguese and comparing to English

Update summarization aims at automatically producing a summary for a collection of texts for a reader that has already read some previous texts about the subject of interest. It is a challenging task, since it not only brings the demands from the summarization area (as producing informative, coherent, and cohesive summaries) but also includes the issue of finding relevant new/updated content. In this paper, we report a comprehensive investigation of update summarization methods for the Portuguese language, for which there are few initiatives. We also propose new methods that combine some summarization strategies and enrich a traditional method with linguistic knowledge (subtopics), producing better results and advancing the state of the art. More than this, we present a reference dataset for Portuguese, so far inexistent, and establish an experiment setup in the area in order to foster future research. To confirm some of our summarization results, we run experiments in a well-known benchmark dataset for English language and show that our methods still do well.Update summarization aims at automatically producing a summary for a collection of texts for a reader that has already read some previous texts about the subject of interest. It is a challenging task, since it not only brings the demands from the summarization area (as producing informative, coherent, and cohesive summaries) but also includes the issue of finding relevant new/updated content. In this paper, we report a comprehensive investigation of update summarization methods for the Portuguese language, for which there are few initiatives. We also propose new methods that combine some summarization strategies and enrich a traditional method with linguistic knowledge (subtopics), producing better results and advancing the state of the art. More than this, we present a reference dataset for Portuguese, so far inexistent, and establish an experiment setup in the area in order to foster future research. To confirm some of our summarization results, we run experiments in a well-known benchmark dataset for English language and show that our methods still do well.