Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
156 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

E2MoCase: A Dataset for Emotional, Event and Moral Observations in News Articles on High-impact Legal Cases (2409.09001v1)

Published 13 Sep 2024 in cs.CL, cs.AI, cs.CY, cs.DL, and physics.soc-ph

Abstract: The way media reports on legal cases can significantly shape public opinion, often embedding subtle biases that influence societal views on justice and morality. Analyzing these biases requires a holistic approach that captures the emotional tone, moral framing, and specific events within the narratives. In this work we introduce E2MoCase, a novel dataset designed to facilitate the integrated analysis of emotions, moral values, and events within legal narratives and media coverage. By leveraging advanced models for emotion detection, moral value identification, and event extraction, E2MoCase offers a multi-dimensional perspective on how legal cases are portrayed in news articles.

Summary

  • The paper introduces E2MoCase, a novel dataset linking legal case news with emotional, event, and moral annotations.
  • It details a systematic curation and semi-automatic annotation process using advanced NLP models like MoralBERT and EmoLLaMA.
  • Results reveal significant media bias patterns, underlining the role of multidimensional analysis in legal narrative research.

E2MoCase: An Analytical Resource for Media Narratives of Legal Cases

The paper "E2MoCase: A Dataset for Emotional, Event and Moral Observations in News Articles on High-impact Legal Cases" presents a comprehensive resource aimed at analyzing the intricate dynamics of media narratives concerning legal cases. This paper, conducted by Greco et al., emphasizes the importance of understanding how media portrayals influence public perception by embedding emotional tones, moral considerations, and specific events within their narratives. The presented dataset, E2MoCase, serves as a tool for researchers exploring these multidimensional aspects within news articles covering legal matters, combining the facets of emotion, morality, and events for an integrative analysis.

E2MoCase was constructed through a systematic process involving case selection, news retrieval, and annotation with the intention of providing robust support for various NLP tasks. The paper highlights the interplay between emotions, moral values, and events within media narratives, which is crucial for analyzing biases and their impacts on public discourse. Using advanced NLP models, such as MoralBERT, EmoLLaMA, and GoLLiE, E2MoCase has been annotated semi-automatically to enhance the dataset's reliability and applicability.

The paper also explores the dataset's construction methodology and validation, providing thorough insights into its potential impact on bias detection and the development of AI models sensitive to the framing of legal issues. By leveraging rich media data obtained through the Swissdox API, the authors curated and annotated a significant corpus of news articles, with a focus on legal cases that have garnered media attention and presented cultural biases. The selected legal cases span various biases, including religious, media, political, gender, and racial, asserting the comprehensive nature of the situations addressed.

Analyses conducted within the paper demonstrate key insights, such as the prevalence of negative emotions and the predominance of certain moral values like cheating, betrayal, and subversion, which correlate significantly with these emotions. The authors provide a statistical interpretation of the dataset, showcasing interrelations among different annotations, further emphasizing the dataset's complexity and the intricacies of the narratives it encapsulates.

Validation of E2MoCase is performed through comparative experiments using state-of-the-art models on existing benchmark datasets. This validation showed that models trained on E2MoCase are competitive with those trained on human-annotated datasets, establishing confidence in the automatic annotation process employed. Furthermore, the dataset's multifaceted nature implies its suitability for varied applications, including enhancing the understanding of media bias and training AI models that capture nuanced narratives in legal media coverage.

The implications of E2MoCase extend beyond theoretical research into practical applications for AI in media analysis, offering a comprehensive framework for studying the portrayal of legal cases and societal perceptions therein. The paper lays the groundwork for future expansions, including incorporating a broader array of languages for global discourse analysis, and linking media content with official legal documents to provide a richer context for narrative interpretation.

In conclusion, E2MoCase stands as a significant advancement in the analysis of media narratives encompassing legal cases, revealing the deep-seated biases and shaping AI's role in societal discourses. Researchers in the field are well-equipped with a resource that integrates complex narrative dimensions, supporting the development of equitable technologies in AI-mediated media analyses.

Youtube Logo Streamline Icon: https://streamlinehq.com