Style Classification of Rabbinic Literature for Detection of Lost Midrash Tanhuma Material

What it is about

The authors classify rabbinic passages by genre (Mishnah, Midrash Halakhah, Jerusalem and Babylonian Talmud, Midrash Aggadah, Tanhuma-Yelammedenu) using 50-word blocks from Sefaria. A logistic regression over unigram to trigram counts reaches 0.867 validation accuracy, close to AlephBERT (0.879) and below BEREL (0.922); a morphology-only model reaches 0.560, against a chance rate of 0.166. The baseline is kept because its weights are readable. The classifier is then run on Yalkut Shimoni, and passages with a low score on text reuse (RWFS, 3-gram fuzzy matching, Levenshtein up to 2) are kept as candidate lost Tanhuma. Of 50 sampled candidates, a midrash expert confirmed 30 (22 cited as Yelammedenu and 8 typical Tanhuma material), about 60% precision, and precision rises with the threshold.

Ideas

I1. Use the logistic-regression weights as the explanation for a genre label: list the n-grams that pushed a passage to a class, which fits our rule that every decision says why.
I2. Keep the morphology-only model as a control: a feature set that drops the lexicon should lose accuracy; if it does not, the style signal is not what we think.
I3. Two-stage detection: a style score filtered by a text-reuse check (keep passages with few shared n-grams with known sources). The reuse check can be a deterministic index, with the threshold fixed beforehand.
I4. Sample-based precision: label a random 50 of the flagged items by hand, as they did, and report the confirmed fraction with the date.

For Mova

I1 and I2 apply to the Mishnaic and rabbinic chapter, 11. Mishnaic Hebrew (hbo-x-mishnaic), and the lexicon notes in The lexicon. I3 would be a new offline tool beside tools/langseg, not part of the parse. The Sefaria data is open in its licence as given; the classifier itself is not a Mova component yet.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.18653/v1/2022.nlp4dh-1.6

@inproceedings{tannor2022style,
  title = {{Style Classification of Rabbinic Literature for Detection of Lost Midrash Tanhuma Material}},
  author = {Tannor, Solomon and Dershowitz, Nachum and Lavee, Moshe},
  booktitle = {Proceedings of the 2nd International Workshop on Natural Language Processing for Digital Humanities},
  year = {2022},
  doi = {10.18653/v1/2022.nlp4dh-1.6},
  url = {https://doi.org/10.18653/v1/2022.nlp4dh-1.6}
}

References

6 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

Yannis Assael, Thea Sommerschield, Brendan Shillingford, Mahyar Bordbar, John Pavlopoulos, Marita Chatzipanagiotou, Ion Androutsopoulos, Jonathan Prag, and Nando de Freitas. 2022. Restoring and attributing ancient texts using deep neural networks. Nature, 603(7900):280–283. find
Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lhoest, and Alexander Rush. 2020. Transformers: State-of-the-art natural language processing. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 38–45, Online. Association for Computational Linguistics. find
Marc Bregman. 2003. The Tanhuma-Yelammedenu Literature: Studies in the Evolution of the Versions. Gorgias Press. find
Idan Dershowitz, Navot Akiva, Moshe Koppel, and Nachum Dershowitz. 2015. Computerized source criticism of biblical texts. Journal of Biblical Literature, 134(2):253–271. find
Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 4171–4186, Minneapolis, Minnesota. Association for Computational Linguistics. ↗
Ronit Nikolsky and Arnon Atzmon. 2021. Studies in the Tanhuma-Yelammedenu Literature. Brill. Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, Jake Vanderplas, Alexandre Passos, 8 A tannaitic tradition not incorporated in the Mishnah, see: “Baraita,” The Jewish Encyclopedia. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2022-tannor-style-classification-rabbinic-literature-detection/