Working off the initiation and experience provided through my previous NAISS project, this project shifts the focus to parliamentary data and expands the scope to the entire records from 1915 onwards. The project aims to map out and analyze Swedish geographical references in the Swedish parliamentary debate. Outlining the 'mental maps' of the period, the project will answer what places were common knowledge and/or deemed relevant to political consideration? Which were idealized? Demonized? Most often employed to make or illustrate a point? The study combines quantitative, distant readings of digitized parliamentary records with qualitative, close readings of the key speeches and debates. It employs machine learning through KB-BERT from the National Library, which performs well at a range of AI tasks. As such, the project is designed to further methodological innovation and machine-learning expertice within Swedish humanities. These methods have not been applied in this way by anyone else and half a year would allow me to process the period systematically and provide me the data from which further research will be possible.