Multilingual Power and Ideology Identification in the Parliament: a Reference Dataset and Simple Baselines

TL;DR

This paper presents a new multilingual dataset derived from parliamentary speeches for identifying political orientation and power positions, along with baseline classification results demonstrating its utility.

Contribution

It introduces a novel, multilingual dataset from parliamentary speeches and provides baseline models for political orientation and power position classification.

Findings

Baseline classifier achieves moderate accuracy on political orientation

Dataset covers 29 parliaments, enhancing multilingual political analysis

Provides insights into speech patterns related to power and ideology

Abstract

We introduce a dataset on political orientation and power position identification. The dataset is derived from ParlaMint, a set of comparable corpora of transcribed parliamentary speeches from 29 national and regional parliaments. We introduce the dataset, provide the reasoning behind some of the choices during its creation, present statistics on the dataset, and, using a simple classifier, some baseline results on predicting political orientation on the left-to-right axis, and on power position identification, i.e., distinguishing between the speeches delivered by governing coalition party members from those of opposition party members.