Detecting Bias in Textual Sources

This chapter introduces a novel combination of topic modeling and sentiment analysis to summarize the study’s documents and explore questions about their themes and biases. The resulting topic models visualize the ways in which the authors perceived and presented the histories of Algeria’s governors. In the case of the two French authors, the expected anti-Arab or anti-Turkish sentiments were absent, but a latent anti-Semitic sentiment appeared in all three French-language accounts. The hierarchical structure of the topics provides an overview of authorial focus, while the more specific topics hint at the richness of the region’s history. Pairing topic modeling with other methods uncovers the stories of lesser-known individuals, such as women and Jews, as well as biases inherent in their histories.