Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

A Comparison of Language Processing Models in Political Analysis : Evidence from Sweden

Fredén, Annika LU orcid ; Johansson, Moa ; Kisić-Merino, Pasko LU orcid and Saynova, Denitsa (2021) Annual Meeting of the American Political Science Association
Abstract
This study evaluates two different natural language processing tech-
niques: the normalised co-occurrence (PMI) versus neural networks. In
contrast to most previous studies, the focus is on the context of a propor-
tional representation system – Sweden –, where the parties in parliament
tend to form coalitions. We test the models by collecting data from the
national parliament (Swedish Riksdag) and using the Swedish language for
training sets and dictionaries. The tests focus primarily on the meaning
and attention that the party representatives confer to important terms,
as well as left-right ideological positioning, with an emphasis on the di-
mension of “security”. The analysis covers parliamentary... (More)
This study evaluates two different natural language processing tech-
niques: the normalised co-occurrence (PMI) versus neural networks. In
contrast to most previous studies, the focus is on the context of a propor-
tional representation system – Sweden –, where the parties in parliament
tend to form coalitions. We test the models by collecting data from the
national parliament (Swedish Riksdag) and using the Swedish language for
training sets and dictionaries. The tests focus primarily on the meaning
and attention that the party representatives confer to important terms,
as well as left-right ideological positioning, with an emphasis on the di-
mension of “security”. The analysis covers parliamentary motions from
the two main competitor parties (Moderates and Social Democrats) over
two time spans, from 1988-2009 and 2010-2020. The two models delivered
different foci of keywords, and we found that balancing pre-training and
fine-tuning was crucial to obtaining differences between parties in the neu-
ral network approach. The PMI model benefited from a larger context
window, and the neural network model (word2vec) from a smaller one.
We discuss the results in relation to future opportunities to learn about
political vocabularies and the nature of conflict in parliamentary politics. (Less)
Please use this url to cite or link to this publication:
author
; ; and
publishing date
type
Contribution to conference
publication status
published
subject
pages
17 pages
conference name
Annual Meeting of the American Political Science Association
conference dates
2021-09-30
language
English
LU publication?
no
id
e3e5aaf3-f5fc-416e-a8cf-4c9fab4c9502
alternative location
https://www.researchgate.net/profile/Annika-Freden/publication/354997332_A_Comparison_of_Language_Processing_Models_in_Political_Analysis_Evidence_from_Sweden/links/615727fa61a8f46670997c9f/A-Comparison-of-Language-Processing-Models-in-Political-Analysis-Evidence-from-Sweden.pdf
date added to LUP
2026-01-08 14:35:54
date last changed
2026-01-09 09:21:43
@misc{e3e5aaf3-f5fc-416e-a8cf-4c9fab4c9502,
  abstract     = {{This study evaluates two different natural language processing tech-<br/>niques: the normalised co-occurrence (PMI) versus neural networks. In<br/>contrast to most previous studies, the focus is on the context of a propor-<br/>tional representation system – Sweden –, where the parties in parliament<br/>tend to form coalitions. We test the models by collecting data from the<br/>national parliament (Swedish Riksdag) and using the Swedish language for<br/>training sets and dictionaries. The tests focus primarily on the meaning<br/>and attention that the party representatives confer to important terms,<br/>as well as left-right ideological positioning, with an emphasis on the di-<br/>mension of “security”. The analysis covers parliamentary motions from<br/>the two main competitor parties (Moderates and Social Democrats) over<br/>two time spans, from 1988-2009 and 2010-2020. The two models delivered<br/>different foci of keywords, and we found that balancing pre-training and<br/>fine-tuning was crucial to obtaining differences between parties in the neu-<br/>ral network approach. The PMI model benefited from a larger context<br/>window, and the neural network model (word2vec) from a smaller one.<br/>We discuss the results in relation to future opportunities to learn about<br/>political vocabularies and the nature of conflict in parliamentary politics.}},
  author       = {{Fredén, Annika and Johansson, Moa and Kisić-Merino, Pasko and Saynova, Denitsa}},
  language     = {{eng}},
  title        = {{A Comparison of Language Processing Models in Political Analysis : Evidence from Sweden}},
  url          = {{https://www.researchgate.net/profile/Annika-Freden/publication/354997332_A_Comparison_of_Language_Processing_Models_in_Political_Analysis_Evidence_from_Sweden/links/615727fa61a8f46670997c9f/A-Comparison-of-Language-Processing-Models-in-Political-Analysis-Evidence-from-Sweden.pdf}},
  year         = {{2021}},
}