Democritus University of Thrace (DUTH)
Διαχείριση μεγάλου όγκου χειρογράφων σε εκδόσεις μεσαιωνικών κειμένων: ψηφιακά εργαλεία και μεθοδολογία χρήσης τους
Abstract
dc:descriptionJohn Chrysostom was one of the most influential ecclesiastical orators of Byzantium. His opera are preserved in a large number of manuscripts, amounting to approximately 20,000 copies. Previous research has shown that a significant portion of his exegetical homilies exhibits the phenomenon of double tradition, that is, the distinction between revised and unrevised manuscript recensions. However, due to the extensive volume of John Chrysostom's manuscript tradition, a systematic study of these sources has not yet been conducted, either to determine the full extent of the double tradition or, more importantly, to classify the surviving manuscripts into corresponding families based on their level of revision. This dissertation proposes the use of state-of-the-art digital tools to automate traditional philological tasks, especially in the investigation and processing of a large manuscript tradition (big data). A sample of homilies by John Chrysostom is employed as a proof of concept. The textual data are examined in two experimental phases. In the first phase, the technology of Handwritten Text Recognition (HTR) for extracting text from digitized man-uscripts is methodologically evaluated. The results of this phase lead to proposed best practices for the training and application of HTR models using the Transkribus platform. Additionally, an original transcription model for Byzantine manuscripts, developed within the framework of this dissertation, is published for the benefit of the scholarly community. In the second experimental phase, the possibility of automating the comparison of sources is investigated. Following an assessment of the available technological tools, the use of Machine Learning for textual data mining is proposed. Clustering experiments conducted with the Orange data mining library produce distinct dendrograms of manuscript groupings, enabling the differentiation of sources according to their level of revision (double tradition). Finally, based on the outcomes of both phases, the dissertation evaluates not only the reliability of the results but also the broader necessity of employing Artificial Intelligence and digital tools in traditional philological research.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Perdiki, Elpida
- Περδίκη, Ελπίδα
Subjects
dc:subject × 23- Ψηφιακές ανθρωπιστικές επιστήμες
- HTR
- Μηχανική μάθηση
- Εξόρυξη δεδομένων
- Άγιος Ιωάννης Χρυσόστομος
- Διπλή παράδοση
- Χειρόγραφη παράδοση
- Digital humanities
- Machine learning
- Data mining
- St. John Chrysostom, Chrysostomean
- Double recension
- Manuscript tradition
- Γλώσσα και Λογοτεχνία
- Ανθρωπιστικές Επιστήμες και Τέχνες
- Φιλολογία
- Άλλες Ανθρωπιστικές Επιστήμες
- Ανθρωπιστικές επιστήμες, διεπιστημονική προσέγγιση
- Languages and Literature
- Humanities and the Arts
- Philology
- Other Humanities
- Humanities, interdisciplinary
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/60990