Automatic Detection of Complex Quotation Patterns in Aggadic Literature

TL;DR

The paper tackles automatic detection of biblical quotations in Rabbinic literature, with a focus on short, paraphrased, and structurally embedded instances. It introduces ACT, a three-stage pipeline combining morphology-aware alignment, context-sensitive preprocessing, and a novel quotation enrichment/inference stage to classify patterns such as Simple, Wave, Echo, and Compound. Empirical results show ACT-QE achieves a high F1 of 0.91, outperforming baselines and manual annotations, while ablations reveal the critical role of the enrichment stage for complex patterns. These findings support robust intertextual analysis and genre classification in Aggadic literature, with data and methods enabling broader digital humanities applications.

Abstract

This paper presents ACT (Allocate Connections between Texts), a novel three-stage algorithm for the automatic detection of biblical quotations in Rabbinic literature. Unlike existing text reuse frameworks that struggle with short, paraphrased, or structurally embedded quotations, ACT combines a morphology-aware alignment algorithm with a context-sensitive enrichment stage that identifies complex citation patterns such as "Wave" and "Echo" quotations. Our approach was evaluated against leading systems, including Dicta, Passim, Text-Matcher, as well as human-annotated critical editions. We further assessed three ACT configurations to isolate the contribution of each component. Results demonstrate that the full ACT pipeline (ACT-QE) outperforms all baselines, achieving an F1 score of 0.91, with superior Recall (0.89) and Precision (0.94). Notably, ACT-2, which lacks stylistic enrichment, achieves higher Recall (0.90) but suffers in Precision, while ACT-3, using longer n-grams, offers a tradeoff between coverage and specificity. In addition to improving quotation detection, ACT's ability to classify stylistic patterns across corpora opens new avenues for genre classification and intertextual analysis. This work contributes to digital humanities and computational philology by addressing the methodological gap between exhaustive machine-based detection and human editorial judgment. ACT lays a foundation for broader applications in historical textual analysis, especially in morphologically rich and citation-dense traditions like Aggadic literature.

Figures (6)

• Figure 1: Structure of the Different Quotation Styles
• Figure 2: "Wave" Style Quotation, Albek edition of Bereshit Rabba, 3 7
• Figure 3: "Echo" Style Quotation, Albek edition of Bereshit Rabba, 3,3. The phenomenon head is marked in green, while the phenomenon tail is marked in blue.
• Figure 4: "Mixture of different quotation styles, Vachman edition of New Midrash, Vaetchanan 6
• Figure 5: Block diagram of ACT’s three-stage pipeline for detecting biblical quotations: (1) *Preprocessing* normalizes both corpus and examined texts; (2) *Candidate Detection* identifies and aligns potential quotations using a sliding window and a positional inverted index; (3) *Quotation Inference* ranks, filters, and classifies quotations into four styles. Colored arrows indicate the flow of text data (green), index lookup (blue), and quotation inference (gold).