Quantitative newspaper sentiment analysis of the 2017 Battle of Marawi

Abstract

Newspapers are expected to be objective but have often been accused of being biased. In this work, we quantitatively measure the bias of Filipino newspapers by using the TextBlob sentiment analysis tool on the coverage of the 2017 Battle of Marawi in southern Philippines. We see that an overwhelming majority (71.1% to 80.0%) of the articles on Marawi had a positive slant, while only two out of 1,185 articles were perfectly neutral. However, the small polarity absolute values (0.0286 to 0.1106) and the large deviations suggest that the articles may be considered within the "neutral" range. Moreover, these results are in sharp contrast with our previous work using another sentiment analysis tool, showing that there is still much room for improvement regarding natural language processing.