EMONA: Event-level Moral Opinions in News Articles

TL;DR

This work defines a new task to extract event-level moral opinions in news articles and introduces EMONA, a dataset with 400 articles, over 9k moral-labeled events, enabling both intrinsic and extrinsic evaluations. It builds baseline models for event extraction and moral classification, and demonstrates that event-level morality correlates with article ideology and sentence bias while also enabling improvement in downstream tasks through knowledge distillation. The extrinsic results show F1 gains of roughly 3.35–4.71 percentage points across article ideology, sentence bias, and event opinion identification, highlighting the practical value of fine-grained moral analysis for media bias analysis and implicit opinion detection. The work suggests promising applications in stance detection and summarization, and outlines avenues for advancing event-level moral understanding with richer context and multi-task learning.

Abstract

Most previous research on moral frames has focused on social media short texts, little work has explored moral sentiment within news articles. In news articles, authors often express their opinions or political stance through moral judgment towards events, specifically whether the event is right or wrong according to social moral rules. This paper initiates a new task to understand moral opinions towards events in news articles. We have created a new dataset, EMONA, and annotated event-level moral opinions in news articles. This dataset consists of 400 news articles containing over 10k sentences and 45k events, among which 9,613 events received moral foundation labels. Extracting event morality is a challenging task, as moral judgment towards events can be very implicit. Baseline models were built for event moral identification and classification. In addition, we also conduct extrinsic evaluations to integrate event-level moral opinions into three downstream tasks. The statistical analysis and experiments show that moral opinions of events can serve as informative features for identifying ideological bias or subjective events.

Figures (7)

• Figure 1: Moral Foundation Theory categorizes moral principles into five dimensions.
• Figure 2: Examples of moral opinions towards events in EMONA dataset.
• Figure 3: The ratio of moral events in all the annotated events, as well as the ratio of positive / negative events in moral events in different portions of EMONA dataset.
• Figure 4: Error analysis of event-level moral opinions classification
• Figure 5: Confusion matrix of ten moral labels