@incollection{ehrhardt_omission_2021,

title = {Omission of Information: Identifying Political Slant via an Analysis of Co-occurring Entities},

author = {Jonas Ehrhardt and Timo Spinde and Ali Vardasbi and Felix Hamborg},

url = {https://epub.uni-regensburg.de/44939/},

year  = {2021},

date = {2021-01-01},

booktitle = {Information between Data and Knowledge},

volume = {74},

pages = {80–93},

publisher = {Werner Hülsbusch},

address = {Glückstadt},

series = {Schriften zur Informationswissenschaft},

abstract = {Due to the strong impact the news has on society, the detection and analysis of bias within the media are important topics. Most approaches to bias detection focus on linguistic forms of bias or the evaluation and tracing of sources. In this paper, we present an approach that analyzes co-occurrences of entities across articles of different news outlets to indicate a strong but difficult to detect form of bias: bias by omission of information. Specifically, we present and evaluate different methods of identifying entity co-occurrences and then use the best performing method, reference entity detection, to analyze the coverage of nine major US news outlets over one year. We set a low performing but transparent baseline, which is able to identify a news outlet?s affiliation towards a political orientation. Our approach employing reference entity selection, i. e., analyzing how often one entity co-occurs with others across a set of documents, yields an F1-score of F1 = 0.51 compared to F1 = 0.20 of the TF-IDF baseline.},

keywords = {bias by omission, co-occurrences, media bias, news articles},

pubstate = {published},

tppubtype = {incollection}

}