Umeå universitets logga

Text analytics of frequency, context, and media portrayal of individuals in war reporting provides insights into key figures, biases, and socio-political narratives. However, due to named entity ambiguity, the number of unique individuals mentioned does not always align with the total number of PERSON entities identified in the dataset, which leads to reduced accuracy in the text analysis. To address this challenge and improve the accuracy of individual identification while ensuring a more reliable analysis of the dataset, we applied the Damerau-Levenshtein distance metric and machine learning techniques to identify and consolidate mentions of personal named entities in news coverage of the Russian-Ukrainian war in 2022. As a result, we created a comprehensive personal names dictionary containing 6,414 entries, with each entry grouping name variants that refer to the same individual.