An Empirical Study on Readers' Intuition of the Political Stances in News Headlines
Abstract:
Prior research shows when reading news articles, most Americans solely read the headlines with less than half continuing to read the full story. However, it is unclear to what extent the readers can appreciate the framing of headlines, particularly when the news sources are unrevealed. In this work, we combine political science with natural language processing in order to analyze the effects of implicit bias within news headlines on the public. We create a dataset annotated by U.S. voters discerning their sentiment towards news headlines
which is leveraged by clustering and classification methods in order to understand the behaviors of the readers and how they are impacted by news headlines. Through our analysis, we find that a reader's interpretation of headlines is not always consistent with the self-identified political affiliation and that implicit bias affects both writer and reader.
Committee:
- Seongkook Heo, Committee Chair, (CS/SEAS/UVA)
- Yangfeng Ji, Advisor, (CS/SEAS/UVA)
- Afsaneh Doryab (SYS, CS/SEAS/UVA)