Mahlow State of the Art in Computational Morphology

Workshop on Systems and Frameworks for Computational Morphology, SFCM 2009, Zurich, Switzerland, September 4, 2009, Proceedings
1. Auflage 2009
ISBN: 978-3-642-04131-0
Verlag: Springer
Format: PDF
Kopierschutz: Adobe DRM (»Systemvoraussetzungen)

E-Book, Englisch, 147 Seiten

This book constitutes the refereed proceedings of the Workshop on Systems and Frameworks for Computational Morphology, SFCM 2009, held in Zürich, Switzerland, in September 2009. The workshop had three main goals: to stimulate discussion among reseachers and developers and to offer an up-to-date overview of available systems for German morphology which provide deep analyses and are suitable for generating specific word forms, to stimulate discussion among developers of general frameworks that can be used to implement morphological components for several languages, to discuss aspects of evaluation of morphology systems and possible future competitions or tasks, such as a new edition of the Morpholympics.

Autoren/Hrsg.

Mahlow, Cerstin

Inhaltsverzeichnis

1;Preface;5
2;Organization;9
3;Table of Contents;11
4;Using Ranked Semirings for Representing Morphology Automata;12
4.1;Introduction;12
4.2;Definitions;14
4.3;Ranked Semirings;15
4.4;Experiments;19
4.5;Conclusion and Further Work;20
4.6;References;20
5;JSLIM – Computational Morphology in the Framework of the SLIM Theory of Language;21
5.1;Introduction;21
5.2;Foundations;22
5.2.1;LAG and SLIM;22
5.2.2;The Algebraic Description of a JSLIM Grammar;23
5.3;Applied Techniques;24
5.3.1;Undirected Programming and Declarative Syntax;25
5.3.2;Indirection;25
5.3.3;Common Subtree Sharing;27
5.4;The JSLIM Grammar Files;29
5.4.1;The Lexicon;29
5.4.2;The Allomorph Method;31
5.4.3;Allo Rules;31
5.4.4;Allo Lexicon;31
5.4.5;Combi Rules;32
5.4.6;Variables;33
5.5;Word Form Generation;33
5.5.1;Surface Generation;33
5.5.2;Generation of a Paradigm;34
5.5.3;Conditions for the Approach toWork;35
5.5.4;Generation of a SingleWord Form;36
5.6;Conclusion;37
5.7;References;37
6;HFST Tools for Morphology – An Efficient Open-Source Package for Construction of Morphological Analyzers;39
6.1;Introduction;39
6.2;HFST-LEXC;41
6.2.1;Efficient Tokenizing of a Sub-lexicon Entry;42
6.2.2;Efficient Union of Sub-lexicon Entries;42
6.2.3;Efficient Implementation of Morphotax over Sublexicons;42
6.3;HFST-TWOLC;44
6.3.1;An Example Grammar;45
6.3.2;Compiling the Rules and Resolving Rule-Conflicts;48
6.4;HFST-COMPOSE-INTERSECT;51
6.5;Full-ScaleMorphological Analyzers Using HFST Morphological Tools;51
6.6;Performance Evaluation;52
6.7;Discussion and Future Research;54
6.7.1;HFST-LexC Performance;55
6.7.2;HFST-TwolC Performance;55
6.7.3;Parallel Rules vs. Cascaded Rules;55
6.7.4;The Other-Symbol;56
6.7.5;Future Directions;56
6.8;Conclusions;57
6.9;References;57
7;fsm2 – A Scripting Language for Creating Weighted Finite-State Morphologies;59
7.1;Introduction;59
7.2;Formal Background;59
7.3;Featuresoffsm2;61
7.3.1;Semirings;61
7.3.2;Weighted Operations;62
7.4;Creating Morphological Analysers with fsm2;64
7.4.1;Symbol Signatures;64
7.4.2;Lexicons;65
7.4.3;Grammars;66
7.4.4;Replacement Rules;68
7.4.5;Scripts and Functions;69
7.5;Conclusion and Further Directions of fsm2;70
7.6;References;72
8;Morphisto: Service-Oriented Open Source Morphology for German;75
8.1;Introduction;75
8.2;Design of Morphisto;76
8.3;Baseline: Test Results and Statistics;76
8.4;Improving the Morphological Analyzer;78
8.4.1;Handling Overgeneration:Which Is the Best Candidate?;78
8.5;Applications;80
8.5.1;Integration into the TextGridWorkbench;80
8.5.2;Lemmatization and Morphological Analysis;80
8.5.3;GeneratingWordforms for Elexiko;80
8.5.4;Deriving the Structure of Complex Entries;81
8.6;Technical Aspects: Morphological Analysis in the Grid;83
8.7;Summary and Outlook;85
8.8;References;86
9;Morphosyntactic and Semantic Analysis of Text: The MPRO Tagging Procedure;87
9.1;Overview;87
9.2;The MPRO Tagging Procedure;88
9.2.1;The Module LESEN;88
9.3;Morphological Analysis;89
9.3.1;The Lexical Database lexdb;89
9.3.2;The Dictionaries morpho and lemwb;90
9.3.3;Morphological Analysis Procedure;91
9.3.4;The Module LEXNP;95
9.3.5;The Module KOMPLETT;95
9.3.6;The Module KORRIGIERE;95
9.4;Conclusion;97
9.5;References;97
10;Word Manager;99
10.1;Introduction;99
10.2;Problem Analysis;99
10.3;Rule Types;101
10.3.1;IRules: Inflectional Classes;102
10.3.2;SRules: Changes in Formatives;103
10.3.3;WFRules: Word Formation Rules;104
10.3.4;Phrase Manager;106
10.3.5;Interaction of Rules;108
10.4;Development Environment;109
10.4.1;The Linguist’s Interface;110
10.4.2;The Lexicographer’s Interface;114
10.4.3;The Developer’s Interface;115
10.4.4;Evaluation of the Development Environment;116
10.5;Available Resources and Their Use;116
10.6;References;117
11;Morphological Analysis Using Linguistically Motivated Decomposition of Unknown Words;119
11.1;Introduction;119
11.1.1;Word Manager Dictionary Database;120
11.1.2;WMTrans Finite State Transducers;120
11.2;Analysis of Unknown Words through Controlled Decomposition;120
11.2.1;The Unknown Word Transducers;120
11.2.2;Defining the Set of Word Formation Rules and Affixes;121
11.2.3;Defining the Set of Admissible Base Words;122
11.2.4;Deleting Undesired Results;122
11.2.5;Results of the Analysis;123
11.3;Current Implementation of the Unknown Word Analyzers;123
11.3.1;The Finite State Components;124
11.3.2;Processing Levels;124
11.3.3;Unknown Word Products and Their Output;126
11.3.4;Recognition Rates;126
11.4;Further Steps;127
11.5;Conclusion;127
11.6;References;128
12;Corpus-Based Lexeme Ranking for Morphological Guessers;129
12.1;Introduction;129
12.2;Lexicon-Based Entry Generator;131
12.3;Corpus-Based Lexeme Ranking;132
12.3.1;Estimating Lexeme Likelihoods;133
12.4;Training and Test Data;135
12.4.1;Lexical Data for the Transducer Lexicon and Entry Generator;135
12.4.2;Data Collections forWord Counts;135
12.4.3;Test Data Collections;136
12.4.4;EvaluationMeasures, Baselines and Significance Test;137
12.5;Evaluation;138
12.5.1;Corpus-Based Lexeme Ranking ofWord Forms;139
12.5.2;Page Count-Based Lexeme Ranking ofWord Forms;140
12.5.3;Corpus-Based Lexeme Ranking of Base Forms;140
12.5.4;Significance;141
12.6;Discussion;141
12.6.1;Discussion of Results;142
12.6.2;Comparison with Similar or Related Efforts;143
12.6.3;FutureWork;144
12.7;Conclusions;144
12.8;References;145
13;Author Index;147