Text as data: The Evolution of the UN Security Council Debates (1995-2017)

Project team:

Mirco Schönfeld (TU Munich), Steffen Eckhard (Uni Konstanz), Ronny Patz (LMU Munich), Hilde van Meegdenburg (Leiden University)

Objectives:

The research project created a dataset of 65,000 speeches from UN Security Council meetings taking place between 1995-2017. The team extracted the speech data from 5,000 meeting transcripts, and the dataset includes metadata regarding each speakers’ affiliation, the position of the speech in a sequence of speeches in a meeting, and the date of the speech.

Schönfeld, M., Eckhard, S., Patz, R., & van Meegdenburg, H. (2019). The UN Security Council Debates Dataset. Harvard dataverse: https://doi.org/10.7910/DVN/KGVSYH

Publications:

Schönfeld, M., Eckhard, S., Patz, R., & van Meegdenburg, H. (2019). The UN Security Council debates 1995-2017. https://arxiv.org/abs/1906.10969.

The dataset introduction paper offers a detailed description of the creation of the dataset as well as some basic statistics. After contextualizing the dataset in recent research on the UNSC, the paper presents descriptive statistics on UNSC meetings and speeches that characterize the period covered by the dataset. Data highlight the extensive presence of the UN bureaucracy in UNSC meetings as well as an emerging trend towards more lengthy open UNSC debates. These open debates cover key issues that have emerged only during the period that is covered by the dataset, for example the debates relating to Women, Peace and Security or Climate-related Disasters.

Schönfeld, M., Eckhard, S., Patz, R., & Meegdenburg, H. v. (2018). Discursive Landscapes and Unsupervised Topic Modeling in IR: A Validation of Text-As-Data Approaches through a New Corpus of UN Security Council Speeches on Afghanistan. Paper presented at the ECPR General Conference (23-25 August 2018), Hamburg. Available via http://arxiv.org/abs/1810.05572.

The paper applies unsupervised topic modelling to the UNSC debates on Afghanistan. The aim is to establish the validity of the LDA-based topic modelling. The paper sets-up two tests using mixed-method approaches. Firstly, we evaluate the identified topics by assessing whether they conform with previous research on the development of the situation in Afghanistan. Secondly, we use network analysis to study the underlying social structures of what we will call 'speaker-topic relations' to see whether they correspondent to know divisions and coalitions in the UNSC. In both cases we find that the unsupervised LDA indeed provides valid and valuable outputs. The dataset can be accessed here (https://doi.org/10.7910/DVN/OM9RG8)

Eckhard, S., Patz, R., Schönfeld, M., & van Meegdenburg, H. (2021). International bureaucrats in the UN Security Council debates: A speaker-topic network analysis. Journal of European Public Policy, 1-20. https://doi.org/10.1080/13501763.2021.1998194

Our article combines natural language processing and network analysis to study the role of UN-bureaucracy in UNSC focusing on the debates on Afghanistan (1995-2017). We observed speaker position, topic introduction, and topic evolution and complemented the analysis with an illustrative case study on one “bureaucratic topic”. Our findings show that despite their general impartiality, UN Secretariat members sometimes acted as autonomous speechmakers, shaping the debate even in venues such as the UNSC where bureaucratic agency seems unlikely.