Analysing Timelines of National Histories Across Wikipedia Editions: A Comparative Computational Approach

Anna Samoilenko, Florian Lemmerich, Katrin Weller, Maria Zens, Markus Strohmaier

Historical narratives are fundamentally subjective, shaped by the cultural and political contexts of their authors. While traditional historiography relies on qualitative analysis, the digital age allows for a computational approach to global memory. This study uses Wikipedia—a multilingual, crowdsourced repository—to investigate how different language communities prioritize historical events across 193 UN member states. By treating year mentions as proxies for historical significance, the researchers map the "focal points" of national histories and identify systemic biases in our collective digital record.

Methodology

The researchers developed a language-independent pipeline to extract historical timelines. The process begins with identifying four-digit numbers between 1000 and 2016 within "History of [Country]" articles and their immediate outlinks across 30 Wikipedia language editions.

Figure 1: Extraction of historical timelines from Wikipedia articles, illustrated using the Slovenian entry for Portugal.

To distinguish years from other numerical data (e.g., population statistics), the team validated a sample using human coders, achieving high accuracy. To identify true "focal points," they employed a randomization (urn) model to establish a baseline distribution. Decades that significantly exceeded this baseline were assigned high z-scores. Finally, Jensen-Shannon (J-S) divergence was used to measure the narrative distance between different language editions, and hierarchical clustering grouped countries with similar temporal emphases.

Data and Scale

The study analyzed 30 of the largest Wikipedia editions using Latin or Cyrillic scripts. The resulting dataset comprised approximately 17 million date mentions retrieved from over 773,000 articles, providing a massive cross-section of global historical interest.

Key Findings

The analysis revealed several structural patterns in how history is recorded digitally.

The Recency Bias

A dominant "floating gap" exists in social memory. Across all languages, 60% to 80% of all date mentions occur after 1800. This suggests a high-resolution "communicative memory" for recent generations, while older history transitions into a lower-resolution "cultural memory."

Figure 2: Distribution of collected dates across (a) 30 language editions and (b) countries with the highest total date counts, highlighting the sharp increase in mentions toward the present day.

Eurocentrism and Visibility

European histories feature focal points distributed relatively evenly across the last millennium. In contrast, many African and South American nations remain computationally "invisible" until the era of European colonization, reflecting a Eurocentric structural bias in the source material.

Geopolitical Clustering and Consensus

Countries cluster not just by geography, but by shared political milestones. For instance, former Soviet states exhibit highly synchronized timelines due to their shared 20th-century history. Despite cultural differences, there is a surprisingly high level of inter-lingual consensus. Major peaks in timelines almost universally correspond to large-scale violent conflicts, such as the Napoleonic Wars and the World Wars.

Limitations and Conclusion

While year mentions provide a powerful quantitative lens, they are a reductionist unit of analysis that lacks semantic depth. The study's focus on major European-native languages also potentially overlooks local perspectives from regions using different numeral systems or smaller Wikipedia editions. Future research aims to integrate entity analysis (people and locations) to create a richer, more nuanced map of global history. This work demonstrates that while Wikipedia is a collaborative effort, it remains a mirror of the systemic biases and shared traumas of the modern world.

Summary updated

Abstract

Portrayals of history are never complete, and each description inherently exhibits a specific viewpoint and emphasis. In this paper, we aim to automatically identify such differences by computing timelines and detecting temporal focal points of written history across languages on Wikipedia. In particular, we study articles related to the history of all UN member states and compare them in 30 language editions. We develop a computational approach that allows to identify focal points quantitatively, and find that Wikipedia narratives about national histories (i) are skewed towards more recent events (recency bias) and (ii) are distributed unevenly across the continents with significant focus on the history of European countries (Eurocentric bias). We also establish that national historical timelines vary across language editions, although average interlingual consensus is rather high. We hope that this paper provides a starting point for a broader computational analysis of written history on Wikipedia and elsewhere.
BibTeX
@inproceedings{samoilenko2017analysing,
  author = {Anna Samoilenko and Florian Lemmerich and Katrin Weller and Maria Zens and Markus Strohmaier},
  title = {{Analysing Timelines of National Histories Across Wikipedia Editions: A Comparative Computational Approach}},
  booktitle = {ICWSM},
  year = {2017},
}
Figure