Seminar in Computational Linguistics

  • Datum: –14.30
  • Plats: Zoom:
  • Föreläsare: Ellinor Lindqvist
  • Kontaktperson: Gongbo Tang
  • Seminarium

NLP for historical texts

Through written petitions, citizens in many pre-modern and pre-democratic societies could address those in power, in order to ask for help, redress, confirmation of existing rights, or advancement to higher positions. Petitions are a rich historical source of information that could provide insight into the past and ordinary people's everyday life. In this seminar, I will talk about the petitions project; an interdisciplinary research collaboration between several departments of Uppsala University and the Swedish National Archives. I will also present my initial work with historical texts and Topic Modeling (TM). TM can be used to summarize, explore, and understand large collections of texts. Though TM could potentially help us observe and compare the textual contents of different time periods, historical data pose challenges to statistical models due to non-standardized orthography and often limited amount of data. As a final objective of my presentation, I will discuss some ideas for future work within the petitions project.