Computational Processing of the Portuguese Language: 6th International Workshop, PROPOR 2003, Faro, Portugal, June 26-27, 2003. Proceedings (Lecture Notes in Computer Science, 2721)

✍ Scribed by Nuno J. Mamede (editor), Jorge Baptista (editor), Isabel Trancoso (editor), Maria das Gracas Volpe Nunes (editor)

• Publisher: Springer
• Year: 2003
• Tongue: English
• Leaves: 282
• Category: Library

✦ Synopsis

Since 1993, PROPOR Workshops have become an important forum for re- archers involved in the Computational Processing of Portuguese, both written and spoken. This PROPOR Workshop follows previous workshops held in 1993 (Lisboa, Portugal), 1996 (Curitiba, Brazil), 1998 (Porto Alegre, Brazil), 1999 ´ (Evora, Portugal) and 2000 (Atibaia, Brazil). The workshop has increasingly contributed to bring together researchers and industry partners from both sides of the Atlantic. The constitution of an international program committee and the adoption of high-standard referee procedures demonstrate the steady de- lopment of the ?eld and of its scienti?c community. This can also be seen in the realization of the satellite workshop AVALON, which constitutes the ?rst evaluation campaign of Portuguese NLP systems. Each one of the 64 submitted papers received a careful, triple blind-review by the program committee. All those who contributed are mentioned in the following pages. The reviewing process led to the selection of 41 papers for oral presentation, 24 regular papers and 17 short papers, which are published in this volume. Theworkshopandthisbookwerestructuredaroundtheeightfollowingmain topics: (i) speech analysis and recognition; (ii) speech synthesis; (iii) pragmatics, discourse, semantics, syntax, and the lexicon; (iv) tools, resources, and appli- tions; (v) dialogue systems; (vi) summarization and information extraction; and (vii) evaluation.

✦ Table of Contents

Computational Processing of the Portuguese Language
Preface
Organization
Table of Contents
Devoicing Measures of European Portuguese Fricatives
Introduction
Design and Recording of a Corpus of Portuguese Fricatives
Method for Segmentation and Annotation
Measures of Devoicing
Manual Criterion for Devoicing
Automatic Criterion for Devoicing
Results of Devoicing Analysis Using the Manual Criterion
Evaluation of the Automatic Devoicing Criterion
Conclusions
References
AUDIMUS.media: A Broadcast News Speech Recognition System for the European Portuguese Language
Introduction
BN Corpus
Pilot Corpus
Speech Recognition Corpus
AUDIMUS.MEDIA Recognition System
Language Modeling
Vocabulary and Pronunciation Lexicon
Weighted Finite-State Dynamic Decoder
Alignment
Pruning
Speech Recognition Results
Concluding Remarks
References
Pitch Restoration for Robust Speech Recognition
Introduction
Noise Effect on the Baseline Spectral Normalization Domain
Proposed Noise Compensation
Experimental Results
Discussion
References
Grapheme-Phone Transcription Algorithm for a Brazilian Portuguese TTS
Introduction
The Transcription Rules
The Transcription Rules
Experimental Results
Conclusions
Improving the Accuracy of the Speech Synthesis Based Phonetic Alignment Using Multiple Acoustic Features
Introduction
Waveform Generator
Acoustic Features
Feature Normalization
Feature Selection Procedure
Frame Alignment
Results
Conclusions
Evaluation of a Segmental Durations Model for TTS
Introduction
Description of the Model
Duration Features
Neural Network
Model Evaluation
Perceptual Evaluation
Conclusion
References
From Portuguese to Mirandese: Fast Porting of a Letter-to-Sound Module Using FSTs
Introduction
Rule Formalism
Transducer Composition
The SAMPA Phonetic Alphabet for Both Languages
Transducer Approach for European Portuguese
Transducer Approach for Mirandese
FST-Based Concatenative Synthesis
Concluding Remarks
A Methodology to Analyze Homographs for a Brazilian Portuguese TTS System
Introduction
Fundamental Concepts
Hypothesis
Analysis
Nominal Constructions
Prepositional Constructions
Verbal Constructions
Experimental Results
Conclusions
Automatic Discovery of Brazilian Portuguese Letter to Phoneme Conversion Rules through Genetic Programming
Introduction
Modeling the Problem Using Genetic Programming
Experimental Results
Final Conclusions
References
Experimental Phonetics Contributions to the Portuguese Articulatory Synthesizer Development
Introduction
Phonetics Applied to Speech Processing
Multimedia Prosodic Atlas for the Romance Languages
Conclusion
A Study on the Reliability of Two Discourse Segmentation Models
Introduction
Method
Results
Conclusions
References
Reusability of Dictionaries in the Compilation of NLP Lexicons
Introduction
Preliminaries
The Thesaurus Denotations
The Reference Corpus
The TeP
The "Mining" Strategy and Pitfalls
“Mining”
Pitfalls
Three Classes of Problems
Selected Strategies
Final Remarks
References
Homonymy in Natural Language Processes: A Representation Using Pustejovsky’s Qualia Structure and Ontological Information
Introduction
The Qualia Structure
The Linguistic Phenomenon of Homonymy
Homonymy in Qualia Structure
Ontology of Concepts for Brazilian Portuguese
Homonymy in Ontological Structuring
LBK Representation Modules
The LKB
Ontological Module
Qualia Structural Module
Final Considerations and Future Perspectives
References
Using Adaptive Formalisms to Describe Context-Dependencies in Natural Language
Introduction
Illustrating Example
Conclusion
Some Regularities of Frozen Expressions in Brazilian Portuguese
Introduction
The Regularities of a Class
Conclusion
Selva: A New Syntactic Parser for Portuguese
Introduction
Related Work
The Grammar
Clause Structure
Coordination
The Pre-processor
The Parser
Evaluation and Comparison
Future Work
An Account of the Challenge of Tagging a Reference Corpus for Brazilian Portuguese
Introduction
Designing the Tagset
Criteria, Features, and Previous Work
The Current Tagset
Some Emblematic Linguistic Challenges
Current and Future Work
References
Multi-level NER for Portuguese in a CG Framework
Introduction
Previous Work
Methodological and Data Framework
Discussion of Name Categories
System Architecture and Strategies
Preprocessing
The Name Type Predictor
The CG Modules
Evaluation
Conclusion
References
HMM/MLP Hybrid Speech Recognizer for the Portuguese Telephone SpeechDat Corpus
Introduction
Database Description
Call Description
Speaker Recruitment and Characteristics
Database Annotation
ASR System Setup
Training and Test Set Definition
Acoustic Modeling
Vocabulary and Language Modeling
Experiments and Results
Conclusions
Managing Linguistic Resources and Tools
Introduction
Infrastructure
The Interface
Module Definition
Related Work
Conclusions and Future Directions
Using Morphossyntactic Information in TTS Systems: Comparing Strategies for European Portuguese
Introduction
Morphossyntactic Tagging System
Linguistic Resources
textit {Corpus}
textit {Lexica}
Experimental Results
Conclusions
Timber! Issues in Treebank Building and Use
Introduction
Annotation Schemes
Can Our Treebank Type 3 Be Turned into an Evaluation Treebank (Type 2)?
Decisions as to the Process
Decisions as to the Encoding
Águia
Kinds of Queries
Use of IMS CWB
References
A Lexicon-Based Stemming Procedure
Introduction
Stemming Methods
The Proposed Lexicon-Based Stemming Procedure
The Structure of the Lexicon
Stemming Procedure
An Experiment with Portuguese
Concluding Remarks
References
Contractions: Breaking the Tokenization-Tagging Circularity
Tokenizing-Tagging Circularity
Tokenization with Interpolated Tagging
Further Results
References
A Linguistic Approach Proposal for Mechanical Design Using Natural Language Processing
Introduction
Research Background
Preliminary Design Background
The Computational Approach in Mechanical Part Design
Correlation between Semantic/Meaning and Syntax Processing
Challenges in Structure Functional Phrases Composition
Considerations
References
Identification of Direct/Indirect Discourse in Children’s Stories
Introduction
Background
Solution
Discussion
Future Work
Curupira: A Functional Parser for Brazilian Portuguese
Introduction
General Architecture
References
ANELL: A Web System for Portuguese Corpora Annotation
Introduction
System Presentation
The Linguistic Analysis
The Annotation
Final Remarks
References
Email2Vmail – An Email Reader
Introduction
System Architecture
Homograph Disambiguation
Language Identification
Signature Analysis
Conclusions
References
A Large Speech Database for Brazilian Portuguese Spoken Language Research
Introduction
Methodology
Study of Dialects and Geographic Distribution of the Speakers
Determination of Utterance Types
Recordings
Phonetic Transcription
Conclusions and Future Work
Interpretations and Discourse Obligations in a Dialog System
Introduction
Interpretation Manager
Constructing Interpretations
Constructing Discourse Obligations
Building Domain Dependent Discourse Obligations
Using Dialogues to Access Semantic Knowledge in a Web IR System
Introduction
The Dialogue System
Web semantics
Natural Language Dialogue System
Conclusions and Future Work
Managing Dialog and Access Control in Natural Language Querying
Introduction
Motivation
Natural Language System
Dialog Management and Clarification
Access Control
Conclusions and Future Work
GistSumm: A Summarization Tool Based on a New Extractive Method
Introduction
GistSumm Description
GistSumm Premises
GistSumm Processes
Sentence Ranking
Extract Production
Evaluating GistSumm Performance
Experiment 1: Identifying the Gist Sentence
Experiment 2: Evaluating the Extracts Overall Quality
Final Remarks
References
Topic Indexing of TV Broadcast News Programs
Introduction
Topic Detection Corpus Description
Story Segmentation
Story Indexing
Segmentation Results
Indexation Results
Conclusions and Future Work
References
An Initial Proposal for Cooperative Evaluation on Information Retrieval in Portuguese
Introduction
Question Answering Evaluation
Web Search Evaluation
An Initial Proposal for Cooperative Evaluation
References
Evaluation of Finite-State Lexical Transducers of Temporal Adverbs for Lexical Analysis of Portuguese Texts*
Introduction
Some Families of Multiword Temporal Adverbs in Portuguese
Evaluating Lexical Finite-State Transducers
Methods
Results
Discussion
Lexical Coverage and Linguistic Adequacy
Final Remarks
References
Evaluating Automatically Computed Word Similarity
Introduction
Computing Word Similarity
Syntactic Contexts
Word Similarity
Lists of Words
Using Similar Words for Expanding Queries
Experiments
Concluding Remarks
Future Work
Evaluation of a Thesaurus-Based Query Expansion Technique
Introduction
Thesaurus-Based Query Expansion
Evaluation
Evaluation over a Small Corpus
Evaluation over the Internet
Concluding Remarks
Cooperatively Evaluating Portuguese Morphology
Introduction
Test Materials Creation
Test Texts
Golden List Compilation
Different Linguistic Points of View
Absence of Standard
Different Testing Points of View
Measuring
Tokenization Data
Comparison with the Golden List
Coarse-Grained Comparison of the Output for All Tokens
References
Author Index