From Complex Sentences to a Formal Semantic Representation using Syntactic Text Simplification and Open Information Extraction

✍ Scribed by Christina Niklaus

• Publisher: Springer Vieweg
• Year: 2022
• Tongue: English
• Leaves: 340
• Category: Library

✦ Synopsis

This work presents a discourse-aware Text Simplification approach that splits and rephrases complex English sentences within the semantic context in which they occur. Based on a linguistically grounded transformation stage, complex sentences are transformed into shorter utterances with a simple canonical structure that can be easily analyzed by downstream applications. To avoid breaking down the input into a disjointed sequence of statements that is difficult to interpret, the author incorporates the semantic context between the split propositions in the form of hierarchical structures and semantic relationships, thus generating a novel representation of complex assertions that puts a semantic layer on top of the simplified sentences. In a second step, she leverages the semantic hierarchy of minimal propositions to improve the performance of Open IE frameworks. She shows that such systems benefit in two dimensions. First, the canonical structure of the simplified sentences facilitatesthe extraction of relational tuples, leading to an improved precision and recall of the extracted relations. Second, the semantic hierarchy can be leveraged to enrich the output of existing Open IE approaches with additional meta-information, resulting in a novel lightweight semantic representation for complex text data in the form of normalized and context-preserving relational tuples.

✦ Table of Contents

Acknowledgements
Abstract
Contents
Acronyms
List of Figures
List of Tables
Part I Background
1 Introduction
1.1 Semantic Hierarchy of Minimal Propositions through Discourse-Aware Sentence Splitting
1.2 Formal Semantic Representation through Open Information Extraction
1.3 Problems in Syntactic Text Simplification and Open Information Extraction
1.4 Research Questions
1.5 Contributions
1.6 Outline
1.7 Associated Publications
2 Related Work
2.1 Text Simplification
2.1.1 Approaches to Simplification
2.1.2 Data Resources for Simplification
2.2 Text Coherence
2.2.1 Discourse-level Syntactic Simplification
2.2.2 Discourse Parsing
2.3 Open Information Extraction
2.3.1 Approaches to Open Information Extraction
2.3.2 Benchmarking Open Information Extraction Approaches
2.4 Meaning Representations
Part II Discourse-Aware Sentence Splitting
3 Introduction
4 Subtask 1: Splitting into Minimal Propositions
4.1 Property of Minimality
4.1.1 Minimality on the Syntactic Level
4.1.2 Minimality on the Semantic Level
4.2 Splitting Procedure
4.3 Execution Order of the Transformation Patterns
5 Subtask 2: Establishing a Semantic Hierarchy
5.1 Constituency Type Classification
5.2 Rhetorical Relation Identification
6 Transformation Patterns
6.1 Clausal Disembedding
6.1.1 Coordinate Clauses
6.1.2 Adverbial Clauses
6.1.3 Relative Clauses
6.1.4 Reported Speech
6.2 Phrasal Disembedding
6.2.1 Coordinate Verb Phrases
6.2.2 Coordinate Noun Phrase Lists
6.2.3 Participial Phrases
6.2.4 Appositive Phrases
6.2.5 Prepositional Phrases
6.2.6 Adjectival and Adverbial Phrases
6.2.7 Lead Noun Phrases
7 Transformation Process
7.1 Data Model: Linked Proposition Tree
7.2 Transformation Algorithm
7.3 Transformation Example
8 Sentence Splitting Corpus
9 Summary
Part III Open Information Extraction
10 Introduction
11 Subtask 3: Extracting Semantically Typed Relational Tuples
11.1 Enriching State-of-the-Art Open Information Extraction Approaches with Semantic Information
11.2 Reference Implementation Graphene
12 Generating a Formal Semantic Representation
12.1 Lightweight Semantic Representation for Open Information Extraction
12.2 Human Readable Representation
12.3 Machine Readable Representation
13 Summary
Part IV Evaluation
14 Experimental Setup
14.1 Subtask 1: Splitting into Minimal Propositions
14.1.1 Datasets
14.1.2 Baselines
14.1.3 Automatic Metrics
14.1.4 Manual Analyses
14.2 Subtask 2: Establishing a Semantic Hierarchy
14.2.1 Automatic Metrics
14.2.2 Manual Analysis
14.3 Subtask 3: Extracting Relations and their Arguments
14.3.1 Baselines
14.3.2 Benchmarks
14.3.3 Comparative Analysis of the Outputs
14.3.4 Manual Analyses
14.3.5 Analysis of the Lightweight Semantic Representation of Relational Tuples
14.4 Sentence Splitting Corpus
14.4.1 Automatic Metrics
14.4.2 Manual Analysis
15 Results and Discussion
15.1 Subtask 1: Splitting into Minimal Propositions
15.1.1 Automatic Metrics
15.1.2 Manual Analyses
15.2 Subtask 2: Establishing a Semantic Hierarchy
15.2.1 Automatic Metrics
15.2.2 Manual Analysis
15.3 Subtask 3: Extracting Relations and their Arguments
15.3.1 Performance of the Reference Open Information Extraction Implementation Graphene
15.3.2 Sentence Splitting as a Pre-processing Step
15.4 Sentence Splitting Corpus
15.4.1 Automatic Metrics
15.4.2 Manual Analysis
Part V Conclusion
16 Summary and Contributions
16.1 Chapter I
16.2 Chapter II
16.3 Chapter III
16.4 Chapter IV
16.5 Chapter V
17 Discussion and Conclusions
17.1 Semantic Hierarchy of Minimal Propositions
17.1.1 Subtask 1: Splitting into Minimal Propositions
17.1.2 Subtask 2: Establishing a Semantic Hierarchy
17.2 Semantically Typed Relational Tuples
17.2.1 Subtask 3: Extracting Relations and their Arguments
17.2.2 Summary
18 Limitations and Future Research Directions
Bibliography