Austrian Parliament WordStream

Documentation | GitHub

Introduction

This document presents a visual representation of the motions put forth by different political parties in the Austrian parliament. We utilize WordStream to depict the topics that each party aimed to discuss over time.

Our analysis encompasses 7 legislative periods spanning from 1996 to the present. The data is grouped by year, with the most significant terms displayed in a larger font size. The importance of a term is determined by its increased frequency compared to the previous year. This approach aims to highlight noteworthy shifts in party sentiment, rather than displaying commonly used terms associated with a particular party. Furthermore, the height of the plot reflects the total frequency of the words that each party put forth during a specific time. However, this is not directly equivalent to the total number of motions of each party.

To enhance the visibility of topic changes, we represent multiple legislative periods within a single plot. The terms presented are derived either from the EuroVoc taxonomy, which classifies motions based on predefined categories, or from the full text of the motions themselves. EuroVoc, a multilingual thesaurus created and maintained by the Publications Office of the European Union, provides a standardized set of terms and concepts, ensuring efficient organization and retrieval of EU documents and resources across various languages. This classification system proves ideal for our purpose as it offers a clear categorization of motion topics. Conversely, utilizing the full text provides a more detailed understanding of motion content, albeit with the presence of specialized vocabulary and symbols not covered by common stop-word dictionaries.

Legislative Period XX-XXII

The Liberales Forum's participation in the parliament concluded with Legislative Period XX. In the subsequent periods, namely during the Schüssel I and II governments, the parliament consisted solely of the FPÖ, ÖVP, SPÖ, and Greens. Notably, when the SPÖ transitioned out of the government, their motion submissions increased significantly.

In addition to the EuroVoc-based topics, we also display the same periods using terms extracted from the full text of motions. To accomplish this, we parsed all the documents associated with the motion submissions, resulting in a comprehensive word list. However, it is important to note that this approach introduced numerous nonsensical words due to the presence of document designations and specialized vocabulary. Moreover, handling plural and singular forms of words posed additional challenges.

Despite these limitations, analyzing the full text provides more detailed insights into the content of the motions and the language used by the parties. We are certain that a higher quality of visualisations could be achieved by developing a more sophisticated preprocessing pipeline which uses corpus-specific stopwords, lemmatisations and named entity recognition. Nevertheless, due to the associated drawbacks of text documents, we made the decision to prioritize EuroVoc as the primary source for retrieving topics.

Legislative Period XXIII-XXV

During these periods, the parliamentary landscape witnessed the emergence of several new parties, including BZÖ, Team Stronach, and NEOS. Meanwhile, the SPÖ and ÖVP governments are reflected in the comparatively infrequent motion submissions of these parties.

Legislative Period XXVI-XXVII

In the 26th legislative period, the green party ceased to be part of the parliament, making way for the newly formed Liste Pilz/JETZT. However, during their brief time in parliament, the new party did not present a substantial number of motions. In the subsequent 27th legislative period, the green party made a return to the parliament, even securing a position in the government. The number of motions from the FPÖ, which was now an opposition party once again, skyrocketed.

All Legislative Periods

By visualizing all legislative periods we get a more complete picture about how the discussions in parliament evolved. We can also see the surge in the number of motions around 2020 during the COVID19 pandamic. Due to the big difference in the number of motions, a lot of available screenspace in earlier legislative periods is not utilised. We see this as a limitation of the WordStream approach.

Dataset

The dataset consist of motions submitted in the National Assembly of the Austrian Parliament. Included motions from the legislative periods XX to XXVII. Earlier motions do not provide EuroVoc keywords and documents are usually scanned. The data was obtained through the Open Data API by requesting the full list of motions per legislative period. Unfortunately this only provides some details for the motions. To get all data related to motions, like the documents or members of parliament who submitted them, each motion must be requested and parsed individually. The scripts we used to query and parse the data are also provided in the Github Repository. For the WordStream visualisation on the fulltext corpus of motions a simple preprocessing pipline was developed. It consists of tokenization, casefolding and stopwords removal. For stopword removal we used a standard german stopword list an added terms which appeared very frequently but didn't provide any semantic meaning.


by Bernhard Weissenbach and Matthias Matt.