Machine translation and the information soup : third Conference of the Association for Machine Translation in the Americas, AMTA '98, Langhorne, PA, USA, October 28-31, 1998 : proceedings

This book constitutes the proceedings of the Third Conference of the Association for Machine Translation in the Americas, AMTA'98, held in Langhorne, PA, USA, in November 1998. The book presents 43 revised full papers carefully reviewed and selected for inclusion in the proceedings together with nine systems descriptions. Among the topics covered are lexicon extraction, quality and robustness in MT, MT systems, parsing, cross-language information retrieval, lexical semantics, statistical methods, MT and the Internet, data mining, reuse of translated terms, multilingual information systems, etc. Machine Translation and the Information Soup Foreword Tutorial Descriptions Panel Descriptions Table of Contents A Statistical View on Bilingual Lexicon Extraction: From Parallel Corpora to Non-parallel Corpora Introduction Bilingual lexicon extraction from parallel corpora DKvec: From clean parallel corpora to noisy parallel corpora Bilingual lexicon extraction from non-parallel corpora Comparable corpora Convec: Moving on to non-parallel corpora The algorithm Finding the TF of words Mapping context words using bilingual dictionary Finding the IDF of words Similarity measures Confidence Experimental Results Discussion of results Conclusion Acknowledgement References Empirical Methods for MT Lexicon Development Introduction How Can Existing Methods Help? Translation Lexicon Construction Methods Formalization of the Question of Possible Translations Method A: The One-to-One Assumption Method B: An Explicit Noise Model Method C: Pre-Existing Word Classes Experiments Non-Compositional Compounds Integration of Manual and Automatic Methods Conclusion Bibliography A Modular Approach to Spoken Language Translation for Large Domains Introduction The C-STAR Interchange-Format Modular Semantic Grammars Integration of Multiple Sub-domain Grammars Cross Domain and Shared Grammars Disambiguation with Statistical Domain Knowledge Current and Future Work Acknowledgements References Enhancing Automatic Acquisition of the Thematic Structure in a Large-Scale Lexicon for Mandarin Chinese Introduction Thematic Structure: Grids Verb Selection Pairing Verbs and Thematic Grids English glosses Automatic Modification of Candidate Grids Results Conclusions and Future Work Acknowledgments References Ordering Translation Templates by Assigning Confidence Factors Introduction Methods for Assigning Confidence Factors Method for Assigning Confidence Factors to Facts Method for Assigning Confidence Factors to Rules Method for Assigning Confidence Factors to Rule Combinations Translation Process by Using Confidence Factors Performance Results Conclusion and Future Work References Quality and Robustness in MT - A Balancing Act Introduction Supporting Various Inputs Transfer Adapting Transfer Rules to Low-Quality Input Example Ranking of Translations Outlook References Parallel Strands: A Preliminary Investigation into Mining the Web for Bilingual Text Introduction The STRAND Architecture Candidate Generation Candidate Evaluation Language-Dependent Filtering Evaluation Future Work Acknowledgments References An English-to-Turkish Interlingual MT System Introduction Turkish The Architecture of the System The Mapping System Sentence Generation System Morphological Generation System Interface Results and Example Translations Issues and Problems Issues Related to the Differences between English and Turkish Issues Related to the Interlingua Issues Related to the Generation and Mapping Systems Future Work Acknowledgments References Rapid Prototyping of Domain-Specific Machine Translation Systems Introduction Special MT Requirements Overview of the System Two Parsers The Core Transfer Component Generation Outlook Acknowledgments Bibliography Time-Constrained Machine Translation What is a Time-Constrained Application? Consequences for Translation Objectives Consequences for System Design Fidelity, Grammaticality, Robustness Real-Time Translation Consequences for System Evaluation Conclusion References An Evaluation of the Multi-engine MT Architecture The Multi-engine MT Architecture Experiment Design: Translations Experiment Design: Evaluation Selection of Translations Method of Evaluation Evaluators Domain Results Initial Statistics Inter-source Comparison Inter-evaluator Comparison A More Informative Statistic Conclusion References An Ontology-Based Approach to Parsing Turkish Sentences Introduction Motivation Ontology and Meaning Representation Methodology Semantic Analysis TMR Construction Conclusion References Monolingual Translator Workstation Introduction Translation Methods Preparation and Morphological Analysis of the Source Text Records of the Morphological Analysis Disambiguation Stages of Disambiguation Dictionaries Direct Dictionary Inverse Dictionary Accumulation with Inversion Selection Construction of Selection Menus Example of Selection Menu Selection Process by the User Weighting and Resolving of Ambiguities Learning Learning Short Contexts Target Text Generation Completing Syntactical and Semantic Information Morphological Generation Summary Conclusion of the Translation Process Results Conclusions References Fast Document Translation for Cross-Language Information Retrieval Introduction The TREC-6 CLIR Task Document Translation IR System Description Results and Conclusions Acknowlegements References Machine Translation in Context Introduction Language Management Process Model Controlled Authoring and Translation Deployment References EasyEnglish: Addressing Structural Ambiguity Introduction Exploring the Parse Nonfinite Clauses Coordination Double Passives Conclusion References Multiple-Subject Constructions in the Multilingual MT-System CAT2 Introduction A Brief Overview of the CAT2 System Lexically Motivated MSCs Idiomatic Expressions ́Psyche`-Adjectives and ́Exist`-Verbs Semantically Motivated MSCs Conclusion Acknowledgement References A Multilingual Procedure for Dictionary-Based Sentence Alignment Introduction The 1-to-1 Algorithm The N-to-1 Algorithm Our Scoring System The Distance Parameter Lexical Acquisition Results Spanish/English Microsoft Help Text The Japanese/English System Concluding Remarks References Taxonomy and Lexical Semantics - From the Perspective of Machine Readable Dictionary Introduction Problems in deriving taxonomy from the MRD Genus term and word sense disambiguation Covert categories from overly general definitions Bilingual taxonomy Deriving taxonomy from MRD Genus disambiguation - from tangling web to bush Discovery of covert categories - from bush to tree Bilingual taxonomy from bilingual MRD Challenges to the MRD derived taxonomy Natural language understanding and machine translation Machine translation Bilingual taxonomy for text and translation alignment Conclusion Acknowledgments References Can Simultaneous Interpretation Help Machine Translation? Introduction The Leaf: Research and Practice in Simultaneous Interpretation Anecdotal Observation Empirical Findings Explanatory Models Previous Work in Applying SI to MT The Thread Back to MT: Looking Upward from the Leaf Unanswered Questions: Looking Downward from the Leaf Conclusion References Sentence Analysis Using a Concept Lattice Introduction Related Work Sentence Analysis General Approach Finding Candidate Sentences Selecting the Best Neighbour Additional Considerations Conclusion Future Directions References Evaluating Language Technologies: The MULTIDOC Approach to Taming the Knowledge Soup Introduction Quality Requirements Analysis Workflow Quality Requirements Documentation Quality Requirements Multilinguality Quality Requirements MULTIDOC LT Components and LT Quality Language Technology Components Language Technology Quality MULTIDOC Evaluation Methodology Conceptual Framework Diagnostic Evaluation MULTIDOC Evaluation Process Quality Requirement Definition Evaluation Preparation Evaluation Procedure Conclusion and Perspectives Acknowledgements References Integrating Query Translation and Document Translation in a Cross-Language Information Retrieval System Introduction Multilingual Information System Four-Layer Multilingual Information System (MLIS) Bilingual English-Chinese Information System for WWW Query Translation Selection Strategies Experiments and Evaluations Search Engines Document Translation Analysis Module Transfer Module Synthesis Module Evaluation for Multilingual Information System Conclusion References When Stålhandske Becomes Steelglove: A Corpus Based Study of Names in Parallel Texts Introduction Parallel Corpora What is a Name? Translate or Transcribe? The Retrieval of Names in a Parallel Corpus The Method for Automatic Extraction The Linking of Names in a Parallel Corpus The Method for Translation Study Specification of Name Translations Conclusions References SYSTRAN on AltaVista: A User Study on Real-Time Machine Translation on the Internet Introduction: Going Live on the Internet User Feedback First Reaction to the Translation Service Acceptance of Machine Translation User Profile Country of Origin Languages Translated and Requested Type of Text Translated User Utilization Assimilation Tool Dissemination Tool Communication Tool Entertainment Tool Learning Tool Legal Implications Possibilities and Challenges Acknowledgements References Making Semantic Interpretation Parser-Independent Introduction Semantic Interpretation Making Semantic Interpretation Robust Coordinating Parsing and Semantic Interpretation Summary Acknowledgments References Implementing MT in the Greek Public Sector: A Users' Survey Introduction Starting up an MT Service Center Promotional Efforts The User Profile The User Experience Incorporating User Feedback To Improve System Performance The Input: Domains And Document Type Domains Document Type Conclusion Statistical Approach for Korean Analysis: A Method Based on Structural Patterns Introduction Structural Ambiguities in Korean Sentences Structural Patterns Definition of Structural Patterns Acquisition of Structural Patterns Pattern Coherency Structural Attachment Using Pattern Coherency Experimental Results Conclusion References Twisted Pair Grammar: Support for Rapid Development of Machine Translation for Low Density Languages Introduction The MT Language Paradox Focus on Knowledge Building (KBMT) Semi-automated Knowledge Building Avoid Arbitrary Decisions Theoretical Basis for Sentence Tableau Twisted Pair Grammar Constraints on Expressiveness Algorithm for Discovering Pairs A Simple Example Categorization of Nodes in Structure Automatic Generation of MT Rules Testing and Scaling the System Conclusion Acknowledgements References A Thematic Hierarchy for Efficient Generation from Lexical-Conceptual Structure Introduction Mapping LCS Roles to AMR Relations Thematic Hierarchy Implementation The Nitrogen Generation System Implementing the Thematic Hierarchy Results Conclusion Acknowledgments References The LMT Transformational System Introduction Overview of the LMT Translation Process The Transfer Tree Data Structure The Transformational Algorithm General Syntax of Transformations Basic Operations Feature Operations Modifier List Operations Lexical Transformations Other Basic Operations Examples Conclusion References Finding the Right Words: An Analysis of Not-Translated Words in Machine Translation Introduction Previous Work Approach Results Analysis Future Work Conclusion References Predicting What MT Is Good for: User Judgments and Task Performance The Machine Translation (MT) Functional Proficiency Scale Project Developing the Task Inventory Ordering the Task Inventory The Face-to-Face Interview The Snap Judgment Exercise The Task-Specific Exercises The Rating Reasons Exercise Tasks and MT Output Properties Preparing the Test Suite Conclusion References Reusing Translated Terms to Expand a Multilingual Thesaurus Introduction The Metathesaurus Pre-processing the Input Pattern Discovery Translating Terms Experimental Results Conclusion References Spicing Up the Information Soup: Machine Translation and the Internet The Multilingual Internet Marketplace Languages on the Internet Translating for the Internet Internet Players, Language Needs and L&H Technology (Semi-)Professional End-Users and Private Consumers User Profile L&H Tools Companies and Corporations User Profile L&H Tools The Internet as a Business Medium: Content Providers User Profile L&H Tools The Internet as a Business in Itself: Service Providers/Search Engines User Profile L&H Tools What ́s Next? References Revision of Morphological Analysis Errors Through the Person Name Construction Model Introduction Extraction of person names and revision of morphological analysis errors Basic procedures Person Name Construction Model Use of morphological analysis result Collection of revision error Experiment Remarks Conclusion Acknowledgments References Lexical Choice and Syntactic Generation in a Transfer System Transformations in the New LMT English-German System Introduction Scope of Transformations Criticisms (and Defense) of Transfer Systems Types of Transformations Context-Independent Transformations Context-Dependent Transformations Transformations and Interaction with Lexical Transfer The Place of Transformations in LMT Interaction of Lexical Transfer and Transformations Examples of Context-Dependent Transformations Lexically Triggered Transformations Transformations Triggered by Semantic Category Conclusion References Translation with Finite-State Devices Introduction Word-for-Word SMT Translation Models Model 1 Model 2 Model 3 Alternative Models Finite-State Devices Finite-State SMT Transducer Zero Transducer One Transducer Two Transducer Three Transducer Four Computation of P(f, a | e) Computation of P(f | e) Translation Algorithm Better Finite-State Models Discussion References Lexical Selection for Cross-Language Applications: Combining LCS with WordNet Introduction Mono-Lingual and Cross-Lingual Validation of Structure/Content Distinction Implementation of Lexical Selection Algorithm Conclusions and Future Work Acknowledgments References Improving Translation Quality by Manipulating Sentence Length Introduction: Long Sentences in Translation Some Problems in Translating across Sentence Boundaries Sentence Splitter Module Analysis Evaluation Evaluation 1 Evaluation 2 Conclusion References Machine Translation among Languages with Transitivity Divergences Using the Causal Relation in the Interlingual Lexicon Introduction The Data The Proposal Verb Entries in Interlingua Translation Process for the Interlingual Representation Translation Process for Generation Aspectual Divergence Conclusion References A Comparative Study of Query and Document Translation for Cross-Language Information Retrieval Introduction Experiment Design Same Language Query (SLQ) Dictionary-Based Query Translation (DQT) MT-Based Query Translation (MQT) MT-Based Document Translation (MDT) Foreign Language Query (FLQ) Results Conclusions Acknowledgments References Lexicons as Gold: Mining, Embellishment, and Reuse Introduction Common Ground The LSB Project QuickMT Sifting the Sand from the Gold Future Directions and Conclusion References System Description/Demo of Alis Translation Solutions Overview Introduction Alis Translation Solutions ATS Core Technology ATS Linguistic Tools Language Identifier and Code-Set Detection Multilingual Search and Query Expansion Accent Tools Word Dictionaries Structured Languages Part of Speech Identifiers Other Utilities Professional Services System Demonstration: SYSTRAN Enterprise Introduction System Builder and Contacts System Category System Characteristics Resources Minimum System Requirements: System Description: Integrating Tools with the Translation Process North American Sales and Support System information System Characteristics Demonstration Overview System Overview Process Advantages of Integration of Translation Memory and Machine Translation Review Conclusion EMIS: A Multilingual Information System References An Open Transfer Translation Introduction The Transfer Dictionary The Translation Conclusion References TransEasy: A Chinese-English Machine Translation System Based on Hybrid Approach Introduction General MT Development Platform Algorithms Knowledge Bases Current State and Future Works Acknowledgements References Sakhr Arabic-English Computer-Aided Translation System Introduction Workbench Workbench Main Features Dictionary Dictionary Main Features Aligner Aligner Main Features Software Localisation Software Localisation Features Web Page Translation Web Page Translation Main Features System Description/Demo of Alis Translation Solutions Application: Multilingual Search and Query Expansion Introduction Multilingual Search and Query Expansion ATS Benefits for Multilingual Search and Query Expansion Translation Flow Functionality Description Environment Logos8 System Description Introduction Current Development Efforts: The New and Improved Logos8 Logos8 System Architecture LogosClient Alex Logos8 Translation Server Future Development Efforts: Logos8 Derivative Products A Terminology Management Tool An Authoring Tool for Controlling Source Usage and Sentence Structure A Smart Post-Editing Environment for Logos8 MT TM Tightly Coupled with MT Author index Machine Translation and the Information Soup! Over the past fty years, machine translation has grown from a tantalizing dream to a respectable and stable scienti c-linguistic enterprise, with users, c- mercial systems, university research, and government participation. But until very recently, MT has been performed as a relatively distinct operation, so- what isolated from other text processing. Today, this situation is changing rapidly. The explosive growth of the Web has brought multilingual text into the reach of nearly everyone with a computer. We live in a soup of information, an increasingly multilingual bouillabaisse. And to partake of this soup, we can use MT systems together with more and more tools and language processing technologies|information retrieval engines, - tomated text summarizers, and multimodal and multilingual displays. Though some of them may still be rather experimental, and though they may not quite t together well yet, it is clear that the future will o er text manipulation systems that contain all these functions, seamlessly interconnected in various ways.