A Wiki-based dataset of military operations with novel strategic technologies (MONSTr)

Abstract

Research on strategies and force employment in modern warfare is prolific, but siloed. While some examine boots on the ground, others focus on aerial bombing or unpiloted platforms. Consequently, most studies focus on the effects of one approach, seldom considering it in lieu of or conjunction with others. Furthermore, there is less knowledge on the origins and implementations of these strategic choices analyzed in isolation. The primary reason for these gaps lies in data limitations. This paper introduces a comprehensive dataset on the universe of United States military operations from 1989 to 2021 from a single source: Wikipedia. Using automated extraction techniques on its two structured knowledge databases–Wikidata and DBpedia–we uncover information about individual operations within nearly every post-1989 military intervention described in existing academic datasets. The data we introduce offers unprecedented coverage and granularity that enables analysis of myriad factors associated with when, where, and how the United States employs military force. We describe the data collection process, demonstrate its contents and validity, and discuss its potential applications to existing theories about force employment and strategy in war.

La investigación en materia de estrategias y estructuras de fuerza en la guerra moderna es prolífica, pero se realiza de forma aislada. Mientras que algunos se centran en las tropas terrestres, otros se centran más en los bombardeos aéreos o las plataformas no pilotadas. En consecuencia, la mayoría de los estudios se centran en los efectos de un solo enfoque, y rara vez lo consideran en lugar de otros enfoques o la posibilidad de considerarlo junto con otros enfoques. Además, existe un menor conocimiento sobre los orígenes y las formas de llevar a cabo estas opciones estratégicas si se analizan de forma aislada. La razón principal de estas brechas radica en las limitaciones de los datos. Este artículo presenta un conjunto de datos exhaustivo acerca del universo de las intervenciones militares de los Estados Unidos entre 1989 y 2021 obtenidas de una sola fuente: Wikipedia. Utilizando técnicas de extracción automatizada sobre sus dos bases de datos de conocimiento estructurado, Wikidata y DBpedia, obtenemos información sobre casi todas las intervenciones militares posteriores a 1989 descritas en conjuntos de datos académicos existentes, además de sobre 425 operaciones adicionales. Los datos que presentamos ofrecen una cobertura y un nivel de detalle sin precedentes, de forma que permiten el análisis de innumerables factores asociados con el cómo, el cuándo y el dónde lleva a cabo Estados Unidos intervenciones militares. Describimos el proceso de recopilación de datos, demostramos su contenido y su validez, y debatimos sus posibles aplicaciones a las teorías existentes sobre el diseño de la estructura de la fuerza y sobre la estrategia en la guerra.

La recherche sur les stratégies et les structures des forces de la guerre moderne est prolifique, mais cloisonnée. Quand certains travaux s’intéressent aux militaires sur le terrain, d’autres se concentrent sur les bombardements aériens ou les plateformes sans pilote. Par conséquent, la plupart des études se focalisent sur les effets d’une approche, mais la conjuguent ou la confrontent rarement à d’autres. En outre, nous en savons moins sur les origines et les mises en œuvre de ces choix stratégiques analysés séparément. Ces lacunes s’expliquent principalement par la limitation des données. Cet article présente un ensemble de données exhaustif sur l’univers des interventions militaires des États-Unis de 1989 à 2021 à partir d’une seule source: Wikipédia. À l’aide de techniques d’extraction automatique appliquées à ses deux bases de données de connaissances structurées (Wikidata et DBpedia), nous découvrons des informations concernant pratiquement toutes les interventions militaires postérieures à 1989 décrites dans des ensembles de données académiques existants, mais aussi 425 opérations supplémentaires. Les données que nous présentons proposent une couverture et une granularité sans précédent qui permettent d’analyser une myriade de facteurs relatifs à la manière, le moment et le lieu des interventions militaires américaines. Nous décrivons le processus de collecte de données, démontrons son contenu et sa validité, avant de discuter des applications potentielles aux théories existantes concernant la conception de la structure des forces et la stratégie militaire.

Keywords:

• Dataset
• military intervention
• use of force

Acknowledgments

For feedback on earlier drafts, the authors thank Christa Almonte, Jonathan Caverley, Rex Douglass, Erik Gartzke, Nadiya Kostyuk, Kendrick Kuo, Ashley Leeds, Nicole Letourneau, Erik Lin-Greenberg, Timothy MacDonald, Sara Plana, Jeff Randall, Tom Scherer, Ryan Shandler, Erin Staine-Pyne, Alison Thompson, and Sanne Verschuren. Excellent research assistance was provided by Thomas Brailey, Zoe Coutlakis, Monster Gannon-Letourneau, Allison Lilley, Amanda Madany, Christie Marquez, David McCrum, Jordan Merkel, Matthew Miltimore, Dominick Nguyen, Peyton Olszowka, Cole Reynolds, Caitlen Rodriguez, Yiyi Sun, Qitao Wu, and Lisa Yen.

Data Availability Statement

The dataset is available at http://military-operations.com/ and the replication code at https://github.com/jandresgannon/MONSTr.

Notes

1 For instance, Allen and Martinez Machain (2019, 549) remark that “[i]deally, the unit of analysis for this project would be at the campaign level. Unfortunately, while most modern Western militaries operate in terms of campaigns, data for all countries over time was [sic] not available at the campaign level.”

2 Recent exceptions that focus on use include Martinez Machain (2015), Post (2019), and Gannon (2022).

3 These levels are generalized for conceptual clarity, but boundaries between them can be blurry in reality. In defining the levels of warfare, US military Joint Publication 3-0 acknowledges that “there are no fixed limits or boundaries between these levels, but they help commanders visualize a logical arrangement of operations, allocate resources, and assign tasks to appropriate commands” (Chairman of the Joint Chiefs of Staff 2018, I-12).

4 Thank goodness “primordial clam chowder” didn’t win.

5 Please go read the entertaining opening of this short article, which we concede is better than ours.

6 Contrary to many course syllabi, Wikipedia vandalism is not as common as alleged: 81% of uncited claims are flagged by bots using automated vandalism detection within 24 h and fixed within 3 h (Tramullas, Garrido-Picazo, and Sánchez-Casabón 2016).

7 Inclusion criteria differ by page type. All observations in MONSTr fall under “events,” for which inclusion criteria is detailed at Wikipedia:Notability (events).

8 For example, Operation Vigilant Sentinel is an instance of “conflict” while Operation Northern Watch is an instance of “military operation.” For a more thorough evaluation of the scope and relationships of conflict labels in Wikipedia, see Douglass et al. (2022a).

9 The amount of missiles fired or troops deployed was of initial interest, but accurate data, especially concerning ordnance, proved challenging.

10 In cases where the exact location was unknown, additional sources were consulted to identify the location with the most precisely reliable location. Those sources and methods are documented in the dataset.

11 The many good works exploring each determinant are too many to list here.

12 Since (1) the dependent variables span the same range and (2) models stem from the same sample and feature the same explanators and specifications, we are on safe statistical ground to compare across models.

Additional information

Funding

This research was sponsored by the Office of Naval Research Grant N00014-14-1-0071, the APSA Centennial Center, Eurasia Group Foundation, and the UCSD Undergraduate Research Apprenticeship Program (URAP). Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the author and do not necessarily reflect the view of the Office of Naval Research.