TU Darmstadt / ULB / TUprints

Validating Topic Modeling as a Method of Analyzing Sujet and Theme

Schröter, Julian ; Du, Keli (2023)
Validating Topic Modeling as a Method of Analyzing Sujet and Theme.
In: Journal of Computational Literary Studies, 2022, 1 (1)
doi: 10.26083/tuprints-00023256
Article, Secondary publication, Publisher's Version

[img] Text
jcls-91-schröter.pdf
Copyright Information: CC BY 4.0 International - Creative Commons, Attribution.

Download (334kB)
[img] Text
jcls-91-schröter.xml
Copyright Information: CC BY 4.0 International - Creative Commons, Attribution.

Download (99kB)
Item Type: Article
Type of entry: Secondary publication
Title: Validating Topic Modeling as a Method of Analyzing Sujet and Theme
Language: English
Date: 21 February 2023
Place of Publication: Darmstadt
Year of primary publication: 2022
Journal or Publication Title: Journal of Computational Literary Studies
Volume of the journal: 1
Issue Number: 1
Collation: 20 Seiten
DOI: 10.26083/tuprints-00023256
Corresponding Links:
Origin: Secondary publication from TUjournals
Abstract:

In Computational Literary Studies (CLS), several procedures for thematic analysis have been adapted from NLP and Computer Science. Among these procedures, topic modeling is the most prominent and popular technique. We maintain, however, that this procedure is used only in the context of exploration up to date, but not in the context of justification. When we seek to prove assumptions concerning the correlation between genres, methods of computational text analysis have to be set up in research environments of justification, i.e. in environments of hypothesis testing. We provide a holistic model of validation and conceptual disambiguation of the notion of aboutness as sujet, fabula, and theme, and discuss essential methodological requirements for hypothesis-based analysis. As we maintain that validation has to be performed for individual tasks respectively, we shall perform empirical validation of topic modeling based on a new corpus of German novellas and comprehensive annotations and draw hypothetical generalizations on the applicability of topic modeling for analyzing aboutness in the domain of narrative fiction.

Uncontrolled Keywords: sujet, theme, validation, topic modeling, content
Status: Publisher's Version
URN: urn:nbn:de:tuda-tuprints-232568
Additional Information:

Urspr. Konferenzveröffentlichung/Originally conference publication: 1st Annual Conference of Computational Literary Studies, 01.-02.06.2022, Darmstadt, Germany

Classification DDC: 800 Literature > 800 Literature, rhetoric and criticism
Divisions: 02 Department of History and Social Science > Institut für Sprach- und Literaturwissenschaft > Digital Philology – Modern German Literary Studies
Date Deposited: 21 Feb 2023 10:23
Last Modified: 22 Jul 2024 08:18
URI: https://tuprints.ulb.tu-darmstadt.de/id/eprint/23256
PPN:
Export:
Actions (login required)
View Item View Item