An Annotated Dataset of Errors in Premodern Greek and Baselines for Detecting Them

2024-10-14

arXiv.org

AI Summary

This study addresses the problem of identifying authentic scribal errors—naturally accumulated over centuries of manuscript copying—in extant Ancient Greek texts, to assist classical scholars in reconstructing original readings. To this end, we construct the first real-world, expert-annotated dataset of scribal errors in Ancient Greek manuscripts: 1,000 error-prone tokens were sampled using BERT-based conditional probability modeling and meticulously annotated by domain experts to distinguish scribal, typographic, and digitization errors. We propose a discriminative error detector that achieves a 5-percentage-point improvement in true positive rate on authentic scribal error detection; empirical analysis confirms that scribal errors are significantly more challenging to detect than modern technical errors. This dataset establishes the first realistic benchmark for textual error detection in ancient manuscripts, is publicly released, and advances computational philology and digital classics research.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity

Abstract

As premodern texts are passed down over centuries, errors inevitably accrue. These errors can be challenging to identify, as some have survived undetected for so long precisely because they are so elusive. While prior work has evaluated error detection methods on artificially-generated errors, we introduce the first dataset of real errors in premodern Greek, enabling the evaluation of error detection methods on errors that genuinely accumulated at some stage in the centuries-long copying process. To create this dataset, we use metrics derived from BERT conditionals to sample 1,000 words more likely to contain errors, which are then annotated and labeled by a domain expert as errors or not. We then propose and evaluate new error detection methods and find that our discriminator-based detector outperforms all other methods, improving the true positive rate for classifying real errors by 5%. We additionally observe that scribal errors are more difficult to detect than print or digitization errors. Our dataset enables the evaluation of error detection methods on real errors in premodern texts for the first time, providing a benchmark for developing more effective error detection algorithms to assist scholars in restoring premodern works.

Problem

• Detect real errors in premodern Greek texts
• Evaluate error detection methods on historical data
• Improve accuracy in classifying scribal and print errors

Innovation

• BERT conditionals for error sampling
• Discriminator-based detector improves accuracy
• Dataset enables real error evaluation