Text Representations and Explainability for Political Science Applications
Licentiatavhandling, 2023

This work explores the utility of natural language processing approaches for the study of political behavior by examining two main aspects - representation and explainability. We investigate how current representation approaches capture politically relevant signals in a proportional representation system. In particular we test static word embeddings trained by transfer learning. We find that some signals in the embedding spaces can be validated from domain knowledge, however, there are multiple factors affecting the performance and stability of the results, such as pre-training and frequency of terms.

Due to the complexity of current NLP techniques interactions between the model and the political scientist are limited, which can impact the utility of such modeling. Therefore, we turn to explainability and develop a novel approach for explaining a text classifier. Our method extracts relevant features for a whole prediction class and can sort those by their relevance to the political domain.

Generally, we find current NLP methods are capable of capturing some politically relevant signals from text, but more work is needed to align the two fields. We conclude that the next step in this work should focus on investigating frameworks such as hybrid models and causality, which can improve both the representation capabilities and the interaction between model and social scientist.

Political Science

Explainability

NLP

Representation

EDIT Room Analysen
Opponent: Prof. Dr. Simone Paolo Ponzetto, University of Mannheim, Germany

Författare

Denitsa Saynova

Chalmers, Data- och informationsteknik, Data Science och AI

Annika Fredén, Moa Johansson, Denitsa Saynova, Word embeddings on ideology and issues from Swedish parliamentarians' motions over time: A comparative approach

Bias and methods of AI technology studying political behaviour

Marianne och Marcus Wallenberg Stiftelse (M&MWallenbergsStiftelse), 2020-01-01 -- 2023-12-31.

Ämneskategorier

Språkteknologi (språkvetenskaplig databehandling)

Statsvetenskap (exklusive studier av offentlig förvaltning och globaliseringsstudier)

Infrastruktur

C3SE (Chalmers Centre for Computational Science and Engineering)

Utgivare

Chalmers

EDIT Room Analysen

Online

Opponent: Prof. Dr. Simone Paolo Ponzetto, University of Mannheim, Germany

Mer information

Senast uppdaterat

2023-08-21