A History of Philosophy in Colombia through Topic Modelling

TL;DR

This study addresses how Colombian academic philosophy has evolved by applying Dynamic Topic Modelling to 875 articles from Ideas y Valores spanning 1951–2022. By fitting a DTM on the corpus with $K=90$ topics and mapping topics to the PhilPapers taxonomy, the authors reconstruct the rise of Value Theory, Epistemology, and Philosophy of Science, and reveal a persistent engagement with German philosophy (Kant, Hegel, Husserl, Heidegger). The analysis finds no statistically significant decline in historical topics over time, despite editorial shifts, and demonstrates the feasibility of DTMs for regional philosophy histories. The work offers a replicable workflow for analyzing non-English philosophical corpora and motivates extending the approach to other Latin American journals and multilingual NLP pipelines.

Abstract

Data-driven approaches to philosophy have emerged as a valuable tool for studying the history of the discipline. However, most studies in this area have focused on a limited number of journals from specific regions and subfields. We expand the scope of this research by applying dynamic topic modelling techniques to explore the history of philosophy in Colombia and Latin America. Our study examines the Colombian philosophy journal Ideas y Valores, founded in 1951 and currently one of the most influential academic philosophy journals in the region. By analyzing the evolution of topics across the journal's history, we identify various trends and specific dynamics in philosophical discourse within the Colombian and Latin American context. Our findings reveal that the most prominent topics are value theory (including ethics, political philosophy, and aesthetics), epistemology, and the philosophy of science. We also trace the evolution of articles focusing on the historical and interpretive aspects of philosophical texts, and we note a notable emphasis on German philosophers such as Kant, Husserl, and Hegel on various topics throughout the journal's lifetime. Additionally, we investigate whether articles with a historical focus have decreased over time due to editorial pressures. Our analysis suggests no significant decline in such articles. Finally, we propose ideas for extending this research to other Latin American journals and suggest improvements for natural language processing workflows in non-English languages.

Figures (12)

• Figure 1: Words, topics and documents in a dynamic topic model. Original articles (a) from Ideas y Valores are processed into bag-of-words by extracting and fixing the text from either PDF or HTML sources, followed by a normalization process in which stop-words are removed and words are lemmatized (b). Using our amassed collection of articles processed into bag-of-words, we use Dynamic Topic modelling to model each document as having a proportion of 90 different topics, 8 of which are shown in (c). By topic, we refer to a distribution over words that evolves over time. In the topic Wittgenstein, the words Wittgenstein, lenguaje (language), regla (rule) and juego (game) were the most probable from 2010 onwards, with slight variations on their likelihood in previous years (d). Dynamic Topic modelling accounts for how the discussion of a certain subject evolves over time.
• Figure 2: Number of articles and average length of documents per period of 5 years. Document length is computed after stopwords have been removed.
• Figure 3: Ratio of words recognized by PySpellChecker before and after orthographical correction. A word is recognized if it is contained in PySpellChecker's default Spanish, English or German dictionaries, RAE's 2020 frequency list, or if it is contained in articles that were available in HTML format. Each point represents a single document, and its location represents how many words are recognized before and after correction against the whole document's word length.
• Figure 4: Number of articles for each main area in the PhilPapers taxonomy in absolute quantity (left) and ratio of papers (right) in each area per year between 1951 and January 2022. Since an article may belong to more than one topic, they can also belong to more than one main area. Here we ignore overlaps between main areas, which means that the sum of documents in the graph is greater than the total number of documents in our corpus. We also include a line for 1983, the year in which the journal started publishing continuously after a period of hiatus.
• Figure 5: Proportion of articles per year for each main area and largest subarea within each main area.