Using LLMs to uncover hidden patterns in the contestation of
religious authorities across a corpus of medieval inquisition
records, 1243–1522

k 2026

Using LLMs to uncover hidden patterns in the contestation of religious authorities across a corpus of medieval inquisition records, 1243–1522

ZBÍRAL, David; Zoltán BRYS; Robert Laurence John SHAW a Gideon Jozua KOTZÉ

Základní údaje

Originální název

Using LLMs to uncover hidden patterns in the contestation of religious authorities across a corpus of medieval inquisition records, 1243–1522

Autoři

ZBÍRAL, David; Zoltán BRYS; Robert Laurence John SHAW a Gideon Jozua KOTZÉ

Vydání

Digital Humanities in Nordic & Baltic Countries 2026, Aarhus (Dánsko), 2026

Další údaje

Jazyk

angličtina

Typ výsledku

Prezentace na konferencích

Stát vydavatele

Dánsko

Utajení

není předmětem státního či obchodního tajemství

Označené pro přenos do RIV

Organizační jednotka

Filozofická fakulta

Klíčová slova česky

contestation of religious authorities; religious dissent; information extraction; large language models; cultural history

Příznaky

Mezinárodní význam, Recenzováno

Změněno: 16. 3. 2026 08:23, prof. PhDr. David Zbíral, Ph.D.

Anotace

V originále

The direct contestation of religious authorities was not a dominant topic of medieval inquisition records, contrary to both common expectations and the usual extent of the notion of heresy, which has, from 1075 onwards, also encompassed any sustained disobedience to ecclesiastical authority. Most of the medieval inquisitorial documentation in fact focuses on other transgressions, such as illicit meetings, rituals, doctrinal heterodoxy, or indirect forms of challenging ecclesiastical authority and its salvific power, such as the preference of dissident ministers over those affiliated to the church, and the criticism of sacraments (Arnold 2001: 152; Biller 2001: 314; Sackville 2011: 118–9). Nevertheless, heresy trial records contain valuable information on direct and explicit contestation of the authority of the church, papacy, religious orders, and priests; only it is overshadowed by other topics and dispersed across a large body of evidence. The volume of this textual material certainly exceeds the possibilities of unassisted close reading. In addition, the diversity of expressions of religious authority contestation defies traditional corpus query methods, especially if we are not interested in the mere presence or absence of the topic itself (e.g., in inquisitorial questions about it, or in reports about other – potentially repeating – suspects), but in whether deponents themselves confessed to have held or to had previously held such opinions. Recent methodological work shows that large language models (LLMs) can support the systematic analysis of large quantities of historical textual data by automating core analytical tasks such as classification and information extraction. This development enables researchers to address well-specified research questions at scale (Barros et al. 2025). Applied studies of LLM integration demonstrate how LLMs can streamline common analytical steps such as document classification and extraction of relevant content (Fischer and Biemann 2024). At the same time, this literature also emphasizes that methodological rigour requires careful specification of prompts and evaluation metrics (Stewart and Sinha 2025).

Návaznosti

101000442, interní kód MU

Název: Networks of Dissent: Computational Modelling of Dissident and Inquisitorial Cultures in Medieval Europe (Akronym: DISSINET)

Investor: Evropská unie, Networks of Dissent: Computational Modelling of Dissident and Inquisitorial Cultures in Medieval Europe, ERC (Excellent Science)

Citovat

ZBÍRAL, David; Zoltán BRYS; Robert Laurence John SHAW a Gideon Jozua KOTZÉ. Using LLMs to uncover hidden patterns in the contestation of religious authorities across a corpus of medieval inquisition records, 1243–1522. In Digital Humanities in Nordic & Baltic Countries 2026, Aarhus (Dánsko). 2026.

@proceedings{2560877,
   author = {Zbíral, David and Brys, Zoltán and Shaw, Robert Laurence John and Kotzé, Gideon Jozua},
   booktitle = {Digital Humanities in Nordic & Baltic Countries 2026, Aarhus (Dánsko)},
   language = {eng},
   title = {Using LLMs to uncover hidden patterns in the contestation of religious authorities across a corpus of medieval inquisition records, 1243–1522},
   year = {2026}
}

TY  - CONF
ID  - 2560877
AU  - Zbíral, David - Brys, Zoltán - Shaw, Robert Laurence John - Kotzé, Gideon Jozua
PY  - 2026
TI  - Using LLMs to uncover hidden patterns in the contestation of religious authorities across a corpus of medieval inquisition records, 1243–1522
N2  - The direct contestation of religious authorities was not a dominant topic of medieval inquisition records, contrary to both common expectations and the usual extent of the notion of heresy, which has, from 1075 onwards, also encompassed any sustained disobedience to ecclesiastical authority. Most of the medieval inquisitorial documentation in fact focuses on other transgressions, such as illicit meetings, rituals, doctrinal heterodoxy, or indirect forms of challenging ecclesiastical authority and its salvific power, such as the preference of dissident ministers over those affiliated to the church, and the criticism of sacraments (Arnold 2001: 152; Biller 2001: 314; Sackville 2011: 118–9). Nevertheless, heresy trial records contain valuable information on direct and explicit contestation of the authority of the church, papacy, religious orders, and priests; only it is overshadowed by other topics and dispersed across a large body of evidence. The volume of this textual material certainly exceeds the possibilities of unassisted close reading. In addition, the diversity of expressions of religious authority contestation defies traditional corpus query methods, especially if we are not interested in the mere presence or absence of the topic itself (e.g., in inquisitorial questions about it, or in reports about other – potentially repeating – suspects), but in whether deponents themselves confessed to have held or to had previously held such opinions. Recent methodological work shows that large language models (LLMs) can support the systematic analysis of large quantities of historical textual data by automating core analytical tasks such as classification and information extraction. This development enables researchers to address well-specified research questions at scale (Barros et al. 2025). Applied studies of LLM integration demonstrate how LLMs can streamline common analytical steps such as document classification and extraction of relevant content (Fischer and Biemann 2024). At the same time, this literature also emphasizes that methodological rigour requires careful specification of prompts and evaluation metrics (Stewart and Sinha 2025).
ER  -

ZBÍRAL, David; Zoltán BRYS; Robert Laurence John SHAW a Gideon Jozua KOTZÉ. Using LLMs to uncover hidden patterns in the contestation of religious authorities across a corpus of medieval inquisition records, 1243–1522. In \textit{Digital Humanities in Nordic \&{} Baltic Countries 2026, Aarhus (Dánsko)}. 2026.

Přehled o publikaci