The Database of Religious History (DRH): ontology, coding strategies and the future of cultural evolutionary analyses

ABSTRACT

Perhaps the biggest challenge facing designers of large-scale, cross-cultural databases is that of ontology, both in terms of defining the unit of analysis and the construction of an appropriate back-end architecture. These decisions are also impacted by the coding strategies adopted, envisioned users, and funding limitations. This article explores how one particular database project, the Database of Religious History (DRH), has addressed these issues, the advantages and drawbacks of the approaches adopted, and the potential of the DRH as a data resource for exploring the cultural evolution of religion.

KEYWORDS:

• Large-scale cultural databases
• big data
• cultural evolution
• evolution of religion

Acknowledgements

The authors would like to thank Rachel Spicer for her help with preparing images, our DRH postdocs and editors, and the over 500 experts who have contributed to this project. Thanks also to our generous funders, Canada's Social Sciences and Humanities Research Council (SSHRC), Templeton Religion Trust and the John Templeton Foundation, for supporting the DRH for over a decade. This paper was also greatly improved in response to thoughtful and constructive comments from an anonymous referee, Richard Sosis, and especially Andrew Buskell.

Disclosure statement

The first author is Director, the second author Associate Director, and the third author Technical Director of the Database of Religious History.

Data availability statement

All of the data discussed in this paper can be freely accessed at religiondatabase.org.

Notes

1 A summary of current DRH coverage can be found on our landing page and entries can browsed at https://religiondatabase.org/browse/.

2 The DRH was originally created as part of a Canadian Social Sciences and Humanities Research Council (SSHRC) grant to PI Slingerland in 2012, with further development funded by a grant by Templeton Religious Trust to Slingerland and Muthukrishna in 2017. Current DRH development and operations is funded by a John Templeton Foundation grant to Slingerland and Monroe from 2021 to 2024.

3 See, especially, Slingerland et al. (2020a pp. 4–5) and Watts et al. (2021, pp. 65–68).

4 Exceptions to this trend include studies that have created their own, bespoke databases, such as Sosis & Bressler, 2003.

5 In terms of technical structure, the DRH has a modular, cloud-based architecture built using industry standard open source technologies. The core of the platform is built using Django with a PostgresSQL database. The front-end user-facing pages use ReactJS and communicate with the database via the aforementioned Django application as well as through a GraphQL endpoint using Hasura. The entire project is hosted on Amazon's Web Services (AWS) platform and is fully containerized as Docker images. This architecture ensures stability and maintainability of the platform and data as we improve existing features and implement new functionality. Daily backups are made to multiple endpoints, including AWS and UBC servers, with regular full backups stored on offsite hard drives.

6 For a current list of DRH editors, see https://religiondatabase.org/landing/about/people/editors.

7 The DRH Text poll will soon be expanded into a “Text/Object” Poll in order to encompass religiously relevant object that do not include writing.

8 For current poll types and to browse the sections and individual questions, see Overview of Polls.

9 Religious Groups that are, from our perspective, overly broad, such as Stowers’ Mediterranean Religion entry, is one way in which our deference to the preferences of our experts has affected the structure of our data, a topic to which we will return below.

10 Carleton (2017) was the product of an attempt to incorporate data from Peregrine &and Ember (2001). Every entry in the DRH is tagged with its entry type and, in the case of entries drawn from external sources, the name of their source. Entries from external sources often possess their own branding (a distinct color and logo) and link to the original source. Also, analysts dubious about the value of such entries can easily, with the click of a button, exclude either Secondary Source entries or entries based on a particular external source or both.

11 The elements of the current DRH religion tagging tree can be viewed at https://religiondatabase.org/browse/tags-group/3. Experts are allowed, with editor approval, to add new tags, so the tree is always growing, and one editorial challenge faced by the DRH is maintaining consistency and coherence in the tree. This is one of the areas where top-down editorial management is required. Tag hierarchies are represented in entry set-up (during the tagging process), but experts can also pull laterally by simply searching for other relevant tags they might like to apply. Representing the entire tree in a single hierarchy is extremely challenging considering the various ways in which religions could be classified (e.g., a form of Chinese Buddhism might be a sub-tag in a tree headed by “Buddhism,” or in a tree headed by “Chinese religion”; “Buddhism” itself might be classed under “Indic religions”). Our response to this problem is allowing a plethora of tagging decision-trees and relying on the search feature of the site to pull all relevant tags in browsing or analyzing. We see the tags as capturing (with editorial oversight) the intuition of scholars about the relationship between their entry and other entries on the site; generally these intuitions are held within and between related disciplines representing top-down forms of categorization.

12 “The variables coded are different across unit categories, limiting the ability to draw comparisons between them” (p. 67). A perhaps related problem is that Watts et al. (2021) seem to conflate DRH tags (which include regions and religious traditions) with poll types (“The predefined categories of units in [the DRH] include regions, religious groups, religious places, places, and polities”; p. 67).

13 Accessed through D-PLACE, https://d-place.org/society/SCCS73 and https://d-place.org/society/SCCS74.

14 Certain extremely broad groups, with worldwide extent, were excluded.

15 For Cham Ahiér, the Group entry is supplemented with a place entry, The Ppo Romé temple-tower complex (Noseworthy, 2019), and a text entry, Dalikal Ppo Klaong Garai (Noseworthy, 2022).

16 “High god” “refers to a supernatural being who is identified by the religious group as qualitatively more important and powerful than any other supernatural being, even if not all-powerful.”

17 http://seshatdatabank.info/databrowser/moralizing-supernatural-punishment-narratives.html

18 Snapshots of the relevant variables from the two entries are accessible here (accessed March 3, 2023): https://web.archive.org/web/20230403034740/http://seshatdatabank.info/browser/CnErJin

https://web.archive.org/web/20230403034901/http://seshatdatabank.info/browser/CnNWei%2A

19 In the Group poll we have two separate questions for afterlife reward and punishment (“Supernatural punishments are meted out in the afterlife”; “Supernatural rewards are meted out in the afterlife”), with analogous questions in the Text poll. The results in Table 2 refer to the punishment version of the question in the two polls.

20 Inspired by a similar feature in Glottolog (glottolog.org; Kirby et al. 2016), we are planning to introduce a new feature to the DRH where experts can click a button on a particular poll question to see guidelines on coding (clear yes and no examples, borderline cases), as well as any discussion on the topic from previous experts.

21 For an example see the entry on Tel Arad by Kristina Shishkova (2022).

22 Responsible postdocs, external advisors and coverage plans can be viewed at Public Postdoc Progress.

Additional information

Funding

This work was funded by a grant from the John Templeton Foundation (Grant ID# 61837) to the first and second authors, “Exploring the Cultural Evolution of Religion Employing a Large-Scale, Quantitative-Qualitative Historical Database” (2021–2024).