Revealing Media Bias in News Articles: NLP Techniques for Automated Frame Analysis

Description

This open access book presents an interdisciplinary approach to reveal biases in English news articles reporting on a given political event. The approach named person-oriented framing analysis identifies the coverage’s different perspectives on the event by assessing how articles portray the persons involved in the event. In contrast to prior automated approaches, the identified frames are more meaningful and substantially present in person-oriented news coverage. The book is structured in seven chapters: Chapter 1 presents a few of the severe problems caused by slanted news coverage and identifies the research gap that motivated the research described in this thesis. Chapter 2 discusses manual analysis concepts and exemplary studies from the social sciences and automated approaches, mostly from computer science and computational linguistics, to analyze and reveal media bias. This way, it identifies the strengths and weaknesses of current approaches for identifying and revealing media bias. Chapter 3 discusses the solution design space to address the identified research gap and introduces person-oriented framing analysis (PFA), a new approach to identify substantial frames and to reveal slanted news coverage. Chapters 4 and 5 detail target concept analysis and frame identification, the first and second component of PFA. Chapter 5 also introduces the first large-scale dataset and a novel model for target-dependent sentiment classification (TSC) in the news domain. Eventually, Chapter 6 introduces Newsalyze, a prototype system to reveal biases to non-expert news consumers by using the PFA approach. In the end, Chapter 7 summarizes the thesis and discusses the strengths and weaknesses of the thesis to derive ideas for future research on media bias. This book mainly targets researchers and graduate students from computer science, computational linguistics, political science, and further social sciences who want to get an overview of the relevant state of the art in the other related disciplines and understand and tackle the issue of bias from a more effective, interdisciplinary viewpoint. Preface Acknowledgments Contents 1 Introduction 1.1 Problem 1.2 Research Gap 1.3 Research Question 1.4 Thesis 1.4.1 Structure and Scientific Contributions 1.4.2 Publications 2 Media Bias Analysis 2.1 Introduction 2.2 Media Bias 2.2.1 Definitions 2.2.2 Effects of Biased News Consumption 2.2.3 Understanding Media Bias 2.2.4 Approaches in the Social Sciences to Analyze Media Bias 2.2.4.1 Content Analysis 2.2.4.2 Frame Analysis 2.2.4.3 Meta-Analysis 2.2.5 Summary 2.3 Manual and Automated Approaches to Identify Media Bias 2.3.1 Event Selection 2.3.2 Source Selection 2.3.3 Commission and Omission of Information 2.3.4 Word Choice and Labeling 2.3.5 Placement and Size Allocation 2.3.6 Picture Selection 2.3.7 Picture Explanation 2.3.8 Spin: The Vagueness of Media Bias 2.3.9 Summary 2.4 Reliability, Generalizability, and Evaluation 2.5 Key Findings 2.6 Practical View on the Research Gap: A Real-World Example 2.7 Summary of the Chapter 3 Person-Oriented Framing Analysis 3.1 Introduction 3.2 Definition of Media Bias 3.3 Discussion of the Solution Space 3.3.1 Tackling Media Bias 3.3.2 Addressing Our Research Question 3.3.3 Research Objective 3.4 Overview of the Approach 3.5 Before the Approach: Gathering News Articles 3.5.1 Method 3.5.2 Evaluation 3.5.3 Conclusion 3.6 Summary of the Chapter 4 Target Concept Analysis 4.1 Introduction 4.2 Event Extraction 4.2.1 Related Work 4.2.2 Method 4.2.2.1 Preprocessing 4.2.2.2 Phrase Extraction 4.2.2.3 Candidate Scoring 4.2.2.4 Output 4.2.2.5 Parameter Learning 4.2.3 Evaluation 4.2.4 Future Work 4.2.5 Conclusion 4.3 Context-Driven Cross-Document Coreference Resolution 4.3.1 Related Work 4.3.2 NewsWCL50: Dataset Creation 4.3.2.1 Collection of News Articles 4.3.2.2 Training Phase: Creation of the Codebook 4.3.2.3 Deductive Content Analysis 4.3.3 Method 4.3.3.1 Preprocessing and Candidate Extraction 4.3.3.2 Candidate Merging 4.3.4 Evaluation 4.3.4.1 Setup and Metrics 4.3.4.2 Baselines 4.3.4.3 Results 4.3.5 Future Work 4.3.6 Conclusion 4.4 Summary of the Chapter 5 Frame Analysis 5.1 Introduction 5.2 Exploring Person-Targeting Framing(-Effects) in News Articles 5.2.1 Related Work 5.2.2 Method 5.2.3 Exploratory Evaluation 5.2.4 Future Work 5.2.5 Conclusion 5.3 Target-Dependent Sentiment Classification 5.3.1 Related Work 5.3.2 Exploring Sentiment in News Articles 5.3.2.1 Creating an Exploratory Dataset 5.3.2.2 Annotating the Exploratory Dataset 5.3.2.3 Exploring the Characteristics of Sentiment in News Articles 5.3.2.4 Experiments and Discussion 5.3.2.5 Summary 5.3.3 NewsMTSC: Dataset Creation 5.3.3.1 Data Sources 5.3.3.2 Creation of Examples 5.3.3.3 Annotation 5.3.3.4 Consolidation 5.3.3.5 Splits and Multi-Target Examples 5.3.3.6 Quality and Characteristics 5.3.4 Method 5.3.5 Evaluation 5.3.6 Error Analysis 5.3.7 Future Work 5.3.8 Conclusion 5.4 Summary of the Chapter 6 Prototype 6.1 Introduction 6.2 Background 6.2.1 Definitions 6.2.2 Approaches 6.3 System Description 6.4 Visualizations 6.4.1 Overview 6.4.2 Article View 6.5 Study Design 6.5.1 Objectives and Questions 6.5.2 Methodology 6.5.3 Data 6.5.4 Setup and Quality 6.5.5 Overview Baselines 6.5.6 Workflow and Questions 6.6 Pre-studies 6.7 Evaluation 6.7.1 Overview 6.7.2 Article View 6.7.3 Other Findings 6.8 Future Work 6.9 Key Findings 6.10 Summary of the Chapter 7 Conclusion 7.1 Summary 7.2 Contributions 7.3 Future Work 7.3.1 Context-Driven Cross-Document Coreference Resolution 7.3.2 Political Framing and Person-Independent Biases 7.3.3 Bias Identification and Communication 7.3.4 Societal Implications Appendix A A.1 Real-World Example A.2 Context-Driven Cross-Document Coreference Resolution A.3 Prototype Evaluation A.4 Data and Source Code Downloads A.5 Publication Awards Glossary References