[Paper Review] Topology Analysis of International Networks Based on Debates in the United Nations

Stefano Gurciullo, Slava Mikhaylov|arXiv (Cornell University)|Jul 29, 2017

Topological and Geometric Data Analysis3 references7 citations

TL;DR

This paper proposes a novel method to infer ideological and epistemic communities among UN member states by analyzing semantic patterns in speeches from the UN General Debate (1970–2014). Using LDA for topic modeling, information-theoretic similarity (normalized mutual information), and the map equation framework, it constructs and analyzes semantic networks, revealing a clear Cold War-era Soviet Bloc community and significant topological shifts during the fall of the Soviet Union.

ABSTRACT

In complex, high dimensional and unstructured data it is often difficult to extract meaningful patterns. This is especially the case when dealing with textual data. Recent studies in machine learning, information theory and network science have developed several novel instruments to extract the semantics of unstructured data, and harness it to build a network of relations. Such approaches serve as an efficient tool for dimensionality reduction and pattern detection. This paper applies semantic network science to extract ideological proximity in the international arena, by focusing on the data from General Debates in the UN General Assembly on the topics of high salience to international community. UN General Debate corpus (UNGDC) covers all high-level debates in the UN General Assembly from 1970 to 2014, covering all UN member states. The research proceeds in three main steps. First, Latent Dirichlet Allocation (LDA) is used to extract the topics of the UN speeches, and therefore semantic information. Each country is then assigned a vector specifying the exposure to each of the topics identified. This intermediate output is then used in to construct a network of countries based on information theoretical metrics where the links capture similar vectorial patterns in the topic distributions. Topology of the networks is then analyzed through network properties like density, path length and clustering. Finally, we identify specific topological features of our networks using the map equation framework to detect communities in our networks of countries.

Motivation & Objective

• To identify ideological and epistemic communities among UN member states using textual data from the General Debate.
• To address the gap in network science by modeling relationships between actual political actors rather than just semantic concepts.
• To detect structural changes in international ideological networks over time, particularly around major geopolitical transitions.
• To validate the method by testing whether it can recover known historical groupings, such as the Soviet Bloc during the Cold War.
• To provide a reproducible, data-driven framework for studying ideological proximity in international relations.

Proposed method

• Latent Dirichlet Allocation (LDA) is applied to extract 8 policy-relevant topics per year from UN speeches, generating a topic probability vector for each country.
• Normalized mutual information (NMI) is used as a similarity metric between country-specific topic vectors to construct weighted country networks.
• Network topology is analyzed via standard metrics including density, average path length, and clustering coefficient to detect structural changes over time.
• The map equation framework with the Infomap algorithm is used to detect communities in the semantic networks, identifying groups of countries with similar ideological profiles.
• The UN General Debate Corpus (UNGDC) from 1970 to 2014 is used, covering all UN member states and publicly available on Harvard Dataverse.
• Temporal analysis of network properties and community structures allows detection of phase transitions, such as the post-Cold War realignment.

Experimental results

Research questions

• RQ1 Can semantic network science detect meaningful ideological communities among UN member states based on their speech content?
• RQ2 How do the topological properties of international networks derived from UN speeches change over time, especially during major geopolitical events?
• RQ3 Does the method successfully recover historically known groupings, such as the Soviet Bloc during the Cold War?
• RQ4 To what extent do information-theoretic similarity measures capture ideological proximity better than simpler textual similarity methods?
• RQ5 How stable and interpretable are the detected communities across different years, and what do they reveal about shifting global alliances?

Key findings

• The network density, average path length, and clustering coefficient show significant structural shifts during the late 1980s and early 1990s, coinciding with the fall of the Soviet Union.
• The map equation framework successfully identifies a distinct community of Soviet Bloc countries during the Cold War, providing partial validation of the method.
• The number of detected communities varies over time, with a notable increase in complexity post-1990, reflecting the diversification of global political discourse.
• Countries such as the USA, UK, and France are consistently placed in central, high-connectivity positions in the semantic networks, indicating their ideological centrality.
• The method reveals that ideological alignment is not solely based on geography or economic status, but is shaped by shared policy priorities and rhetorical framing in multilateral diplomacy.
• The semantic networks show that countries like China, Russia, and the USA form increasingly cohesive clusters in the 2000s, reflecting rising strategic alignment in key global issues.