This paper presents a preliminary investigation into automatic approaches that rely solely on content-based features to compute indicators such as the "risk indicator", which aims to provide a measure of the extent to which risk is present/evoked in a set of informative resources. We built a dataset comprising English newspaper articles, labeled by ten experts in Social Sciences and Humanities using a three-level scale to denote the "degree" of risk suggested by each article. The study reports on experimental results obtained by considering different instantiations of the indicator, specifically exploiting normalized term frequencies and the Concept Mover Distance.
Exploring Approaches for Measuring Risk in the News
Di Buccio E.
;Neresini F.
2025
Abstract
This paper presents a preliminary investigation into automatic approaches that rely solely on content-based features to compute indicators such as the "risk indicator", which aims to provide a measure of the extent to which risk is present/evoked in a set of informative resources. We built a dataset comprising English newspaper articles, labeled by ten experts in Social Sciences and Humanities using a three-level scale to denote the "degree" of risk suggested by each article. The study reports on experimental results obtained by considering different instantiations of the indicator, specifically exploiting normalized term frequencies and the Concept Mover Distance.
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
