The study of short texts in digital politics: Document aggregation for topic modeling

2025-03-07

AI Summary

Short-text political discourse—e.g., tweets—poses challenges for interpretable topic modeling due to severe sparsity and lack of contextual coherence. Method: We propose a novel document aggregation paradigm grounded in natural semantic units (e.g., legislator accounts, biographical attributes such as birthplace), moving beyond conventional random or time-window-based aggregation. This approach redefines document granularity according to inherent political entity attributes. Contribution/Results: Evaluated on over one million tweets from U.S. state legislators, account-level aggregation significantly strengthens statistical associations between inferred topics and state-level political features (e.g., party affiliation, policy orientation). The effect is robustly replicated using Wikipedia-derived birthplace-based aggregation. This study provides the first empirical evidence that document granularity critically determines interpretability in political topic modeling. It establishes a reproducible, generalizable framework for short-text political analysis, bridging computational linguistics and political science through semantically informed corpus construction.

Technology Category

• Natural Language Processing: Discourse, Pragmatics & Argument Mining
• Application Domains: Humanities & Computational Social Science
• Knowledge Representation and Reasoning: Argumentation

Application Category

• Web Mining and Content Analysis: Topic discovery and tracking
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Statistical topic modeling is widely used in political science to study text. Researchers examine documents of varying lengths, from tweets to speeches. There is ongoing debate on how document length affects the interpretability of topic models. We investigate the effects of aggregating short documents into larger ones based on natural units that partition the corpus. In our study, we analyze one million tweets by U.S. state legislators from April 2016 to September 2020. We find that for documents aggregated at the account level, topics are more associated with individual states than when using individual tweets. This finding is replicated with Wikipedia pages aggregated by birth cities, showing how document definitions can impact topic modeling results.

Problem

• Investigates effects of aggregating short documents for topic modeling
• Analyzes impact of document length on topic model interpretability
• Explores how document definitions influence topic modeling results

Innovation

• Aggregates short documents for topic modeling
• Analyzes tweets by U.S. state legislators
• Demonstrates document definition impacts results