Inhalt

•  
  Introduction
•  
   Life Cycle of Knowledge in Natural Language Processing
•  
   Knowledge Induction from Text
•  
   Attribute Knowledge in Knowledge Consumers and Knowledge Creators
•  
   Attribute Meaning
•  
   Thesis Overview
•  
  Foundations of Distributional Semantics
•  
   Distributional Hypothesis
•  
   Meaning Representation in Distributional Semantic Models
•  
  Variants of Distributional Semantic Models
•  
   Conceptual and Notational Foundations
•  
   Structured vs. Unstructured Models
•  
   Syntagmatic vs. Paradigmatic Models
•  
   First-order vs. Second-order Models
•  
   Meaning Representation beyond the Word Level
•  
  Related Work
•  
   Adjective Classification
•  
   Attribute Learning
•  
   Structured Models in Distributional Semantics
•  
   Topic Models in Distributional Semantics
•  
   Distributional Models of Phrase Meaning
•  
   Distributional Enrichment of Structured Models
•  
  Distributional Models of Attribute Meaning
•  
   Research Questions
•  
   Identifying Attribute-denoting Adjectives
•  
   Compositional Representations of Attribute Meaning in Adjective-Noun Phrases
•  
   Distributional Enrichment
•  
   Contributions of this Thesis
•  
  Classification of Adjective Types for Attribute Learning
•  
  Corpus Annotation and Analysis
•  
   Classification Scheme
•  
   Annotation Process
•  
   Agreement Figures
•  
   Re-Analysis: Binary Classification Scheme
•  
   Class Volatility
•  
  Automatic Type-based Classification of Adjectives
•  
   Features for Classification
•  
   Heuristic Generation of Training Instances from Seeds
•  
   Data Set Construction
•  
   Experimental Evaluation
•  
   Discussion
•  
   Summary
•  
  Attribute Selection from Adjective-Noun Phrases: Models and Parameters
•  
  Foundations of Structured Distributional Models for Attribute Selection
•  
   Attribute-based Distributional Representations of Adjective and Noun meaning
•  
   Vector Composition Functions
•  
   Attribute Selection Functions
•  
  Pattern-based Distributional Model
•  
   Lexico-syntactic Patterns for Attribute Acquisition
•  
   Model Parameters
•  
   Distributional Attribute Models based on Weakly Supervised Topic Models
•  
   Background: Probabilistic Topic Models
•  
   Integrating Latent Topics into Distributional Attribute Models
•  
   Summary
•  
  Attribute Selection: Experimental Evaluation
•  
  Construction of Labeled Data Sets
•  
   Core Attributes Gold Standard
•  
   Large-scale Gold Standard
•  
   Summary of Data Sets
•  
  Evaluation of the Pattern-based Attribute Model
•  
   Experiment 1: Attribute Selection from Adjective Vectors
•  
   Experiment 2: Attribute Selection from Noun Vectors
•  
   Experiment 3: Attribute Selection from Phrase Vectors
•  
   Discussion
•  
   Evaluation of Topic-based Attribute Models
•  
   Experiment 4: Topic-based Attribute Selection on Core Attributes
•  
   Smoothing Power
•  
   Experiment 5: Large-scale Attribute Selection
•  
   Re-Training on Confined Subsets of Attributes
•  
   Discussion
•  
   Summary
•  
  Explaining C-LDA Performance in Large-scale Attribute Selection
•  
   Explanatory Variables
•  
   Semantic Features
•  
   Morphological Features
•  
   Ambiguity Features
•  
   Frequency Features
•  
   Uncertainty Features
•  
   Vector Quality Features
•  
   Compositionality in C-LDA
•  
  Linear Regression of C-LDA Performance at the Intersection of Word and Phrase Meaning
•  
   Foundations of Linear Regression Modelling
•  
   Phrase Level: Least Squares Regression of Phrase Vector Quality
•  
   ``Zooming in'': Regression of Word Vector Quality
•  
   Compositional Processes: Linking Word and Phrase Level
•  
   Major Findings and Discussion
•  
   Options for Enhancing C-LDA Performance
•  
   Summary
•  
   Distributional Enrichment: Improving Structured Vector Representations
•  
   General Idea and Overview
•  
   Auxiliary Distributional Models
•  
   Benchmarking First- and Second-order Auxiliary Models for Attribute-preserving Carrier Selection
•  
   Benchmark Results
•  
   Distributional Enrichment for Attribute Selection
•  
   Paradigmatic Distributional Enrichment
•  
   Syntagmatic Distributional Enrichment
•  
   Joint Distributional Enrichment of Adjective and Noun Vectors
•  
  Experiment 6: Large-scale Attribute Selection after Distributional Enrichment
•  
   Experimental Settings
•  
   Experimental Results
•  
   Evaluation on Test Set
•  
   Discussion
•  
   Summary
•  
  Conclusions
•  
   Contributions of this Thesis
•  
   Conclusions and Perspectives
•  
  Different Attribute Inventories
•  
   Core Attributes
•  
   Property Attributes
•  
   Measurable Attributes
•  
   WebChild Attributes
•  
   Large-scale Attribute Data Set
•  
  Annotation Instructions for Acquisition of HeiPLAS Gold Standard
•  
   Background and Task Definition
•  
   Functionality of the User Interface
•  
  Classification Guidelines
•  
   General Instructions
•  
   Classification Test
•  
   ``Compositionality Puzzles'': Examples from HeiPLAS Development Data