SEMI-AUTOMATIC RECOGNITION OF SEMANTIC
RELATIONSHIPS IN ENGLISH TECHNICAL TEXTS

    Abstract   (front.pdf)

     
    1   Overview of the Project   (chap1.pdf)

    1.1   Automated Semantic Analysis
      1.1.1   Oracles
      1.1.2   Entities and Events
    1.2   Goals of the Project
    1.3   Semantic Analysis in TANKA
      1.3.1   DIPETT Parse Trees
      1.3.2   Three-Tiered Analysis
           Clause Level Relationships
           Case Relationships
           Noun Modifier Relationships
      1.3.3   Background Knowledge
           Linguistic Knowledge
           Semantic Knowledge
           User Knowledge
    1.4   Evaluation
      1.4.1   Test Texts
      1.4.2   Parser Evaluation
      1.4.3   User Actions
      1.4.4   User Burden
    1.5   Applications
    1.6   Organization of the Dissertation

     
    2   Clause Level Relationship Analysis   (chap2.pdf)

    2.1   Introduction
      2.1.1   Semantic Relationships
      2.1.2   Tense and Modality
    2.2   Input Structures
      2.2.1   Verb Sequence Features
      2.2.2   Logical Clausal Structure
           Coordination or Subordination
           Correlative Coordination
           Subordinator/Conjunct Correlation
    2.3   The Clause Level Relationships
      2.3.1   CLR Glossary
           Causal CLRs
           Temporal CLRs
           Conjunctive CLRs
    2.4   The CLR Marker Dictionary
    2.5   Assigning CLRs
      2.5.1   Verb Phrase Polarity and Connective Polarity
      2.5.2   Verb Phrase Tense and Modality
           Absence of Modals
           Modals in Negative Verb Phrases
           Marginal Modals
           Should as a Past Tense of Shall
      2.5.3   Using Argument Features to Choose CLRs
           CLR Preference Rules
      2.5.4   CLR Competitions
      2.5.5   Using User Assignments to Choose CLRs
           CLR Assignment Attributes
           When to Use User Assignments
           Least-General Generalization of Attribute Patterns
           Avoiding Overgeneralization
    2.6   Evaluation
      2.6.1   Diagnostic Evaluation
      2.6.2   Performance Evaluation
           System Performance
           Coverage
      2.6.3   Can CLRA Learn?
    2.7   An Example
    2.8   Chapter Summary

     
    3   Case Relationships   (chap3.pdf)

    3.1   Introduction
      3.1.1   Cases
      3.1.2   Case Theory
      3.1.3   Valency Theory
      3.1.4   Other Case Systems
    3.2   Case Markers
      3.2.1   Positional Markers
      3.2.2   Prepositional Markers
      3.2.3   Adverbial Markers
      3.2.4   Marker Order
    3.3   Case System Design
      3.3.1   The Case Marker Dictionary
      3.3.2   Evaluation Criteria
           Generality
           Completeness
           Uniqueness
      3.3.3   Using the Criteria to Guide Selection
    3.4   The Cases
      3.4.1   Case Glossary
           Participant
           Causality
           Space
           Time
           Quality
    3.5   Assigning Cases
    3.6   Evaluation
      3.6.1   Case Analyzer Evaluation
      3.6.2   Case System Evaluation
           Generality
           Completeness
           Uniqueness
    3.7   Chapter Summary

     
    4   Noun Modifier Relationship Analysis   (chap4.pdf)

    4.1   Introduction
      4.1.1   Noun Compounds
      4.1.2   Semantic Relations in Noun Phrases
      4.1.3   Recognizing Semantic Relations
    4.2   Input Structures
      4.2.1   Attributes and Head Noun Premodifiers
      4.2.2   Noun Phrase Postmodifiers
           Relative Clauses
           Appositives
           Comparison
           Prepositional Phrases
    4.3   Noun Modifier Bracketing
      4.3.1   Subphrases and Reduced Subbracketings
      4.3.2   Bracketing Models
      4.3.3   A Bracketing Algorithm
      4.3.4   Confidence in Branching Decisions
      4.3.5   User Interaction
      4.3.6   When Is "Left-Branching" "Not-Right-Branching"?
      4.3.7   A Walk through Bracketing
      4.3.8   When All Else Fails
           Redoing the Bracketing Interaction
           Ignoring Bracketing History
           Bracketing by Hand
    4.4   The Noun Modifier Relationships
      4.4.1   NMR Glossary
    4.5   The NMR Marker Dictionary
    4.6   Assigning NMRs
      4.6.1   Reduced Modifiers and Heads
      4.6.2   Modifier-Head-Marker Triples
      4.6.3   Distance between Triples
      4.6.4   The Best NMRs
           Absolute Frequency
           Relative Frequency
           Weighted Relative Frequency
      4.6.5   User Interaction
      4.6.6   Classifying Function of Premodifiers
    4.7   Evaluation
      4.7.1   Bracketer Evaluation
           System Performance
           The Effect of the Threshold
           Branching Frequencies
      4.7.2   NMRA Evaluation
           System Performance
           Coverage
           User Burden
      4.7.3   Some Difficulties
           Questionably Endocentric Compounds
           Postpositive Adjectives
           Adjectives in Paraphrases
    4.8   An Example
    4.9   Chapter Summary

     
    5   Future Work   (chap5.pdf)

    5.1   Clause-Level Relationship Analysis
      5.1.1   Evaluation
      5.1.2   Generalizing User Assignments
      5.1.3   Embedded CLR Structures
      5.1.4   Are Attribute Patterns Empirical Preference Rules?
    5.2   Case Analysis
      5.2.1   Assigning Cases One by One
      5.2.2   Aggressive Case Analysis
      5.2.3   A General Purpose Case Pattern Dictionary
    5.3   Noun Modifier Relationship Analysis
      5.3.1   Noun Modifier Bracketing
           Storing Pairs from Prepositional Phrases
           Branching Frequencies
           Insensitivity to the Bracketing Threshold
      5.3.2   Other Noun Modifier Relationships
      5.3.3   Methods for Choosing the Best NMRs
      5.3.4   Taxonomic Information from NMRs
    5.4   A Unified Set of Semantic Relationships
    5.5   Other Directions Altogether
      5.5.1   Fully Automatic Recognition of Semantic Relationships
      5.5.2   Extending HAIKU with WordNet
      5.5.3   Other Sources of Syntactic Information

     
    6   Conclusions   (chap6.pdf)

    6.1   Summary
      6.1.1   Clause Level Relationship Analysis
      6.1.2   Case Relationships
      6.1.3   Noun Modifier Relationship Analysis
    6.2   Goals Revisited
      6.2.1   Semantic Relationships
      6.2.2   Semi-Automatic Recognition of Semantic Relationships
      6.2.3   Learning
      6.2.4   Coverage
      6.2.5   User Burden
    6.3   Closing Words

     
    7   References   (chap7.pdf)

     
    Appendix I:   CLR Marker Dictionary   (appendices.pdf)

     
    Appendix II:   Case Marker Dictionary
     
    Appendix III:   NMR Marker Dictionary
(last update July 18, 1998)
kbarker@site.uottawa.ca