A multi-viewpoint spectrum paradigm for inter-actor relationship analysis in non-social textual corpora: The case of the UN General Debate Corpus
Abstract
Purpose
This paper presents a new semi-automatic methodology for identifying inter-actor relationships by discerning viewpoints in non-social, political textual corpora. Although previous research has successfully discerned viewpoints, biases, and affiliations based on textual features, the task of relationship analysis in the absence of interactional data remains unaddressed.
Design/methodology/approach
We introduce a new paradigm for topic representation as a dynamic, continuous, multi-viewpoint spectrum based on the representation of viewpoints as vectors that capture common topical themes. As a proof of concept, we applied this paradigm to scrutinize the inter-state relationships reflected in the speeches of the UN General Assembly Debate Corpus (UNGDC).
Findings
The proposed paradigm effectively identifies discursive trends in UNGDC. Our analysis reveals common attitudes towards the topic and their prominence among different groups of actors and facilitates the analysis of relationships between actors through a quantitative representation of viewpoint similarity. The method also successfully captured temporal shifts in viewpoints and overall discourse trends, correlating with major geopolitical events.
Research limitations
One limitation of this study is the method’s sensitivity to data sparsity, which can skew viewpoint representations in cases of low topic involvement.
Practical implications
The proposed paradigm can be utilized by scholars in political science and other domains as a tool for semi-automated unsupervised textual analysis of various non-social textual sources, enabling the discovery of latent relationships between actors and the modeling of viewpoints in complex topics.
Originality/value
This study presents a novel framework for unsupervised semi-automatic textual analysis of relationships in non-social corpora through a new approach for the representation of viewpoints as thematic vectors.
© 2025 Efrat Miller, Maayan Zhitomirsky-Geffet, Mor Mitrani, published by Chinese Academy of Sciences, National Science Library
This work is licensed under the Creative Commons Attribution 4.0 License.