Automatic Construction of Aramaic-Hebrew Translation Lexicon

What it is about

The authors build an Aramaic-Hebrew translation lexicon from the parallel Targum Onkelos and Targum Jonathan (Aramaic translations of the Pentateuch and the Prophets, set beside the Hebrew). An IBM model 4 alignment (GIZA++) gives an initial lexicon of 29,335 source terms. For each term, the candidate translations are clustered by string similarity, and a candidate is kept only when its cluster, or a cluster of similar terms, supports it. Five string measures are tested (Levenshtein, Hamming, Jaccard, Jaro and a word-level match on the rarest letters) for three pair types: Hebrew-Hebrew, Aramaic-Aramaic and Aramaic-Hebrew. On 108 sampled source terms (287 annotated pairs) the filtered lexicon reaches F1 0.93 and accuracy 0.89, against 0.88 and 0.79 for the unfiltered IBM output, and 0.82 and 0.70 for a filter on the alignment probability. Jaro similarity with thresholds 0.67, 0.82 and 0.78 works best for the three pair types.

Ideas

• I1. Jaro similarity with the paper's thresholds (0.67 Hebrew-Hebrew, 0.82 Aramaic-Aramaic, 0.78 Aramaic-Hebrew) as a deterministic filter for candidate gloss pairs. Test it in The lexicon of Biblical Aramaic on the Onkelos gloss pairs, against a hand-checked sample.
• I2. The word-level measure: represent each word by its two rarest letters, in order, after dropping prefixes and matres lectionis, and require an exact match. This links Aramaic spelling variants to one lemma in The lexicon of Biblical Aramaic.
• I3. The paper's error analysis is a ready regression list: 34% of its wrong rejections were single-form words, and 66% came from low similarity scores on pairs such as variant forms of one verb. Keep these as cases that any new filter must not lose.

For Mova

I1 and I2 go into The lexicon of Biblical Aramaic. I1 and I3 go into languages/arc-x-targum/ as a test on Targum glosses. Nothing is built from the data; the pairs are used only for measurement, since the licence is non-commercial.

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@inproceedings{liebeskind2020automatic,
  title = {{Automatic Construction of Aramaic-Hebrew Translation Lexicon}},
  author = {Liebeskind, Chaya and Liebeskind, Shmuel},
  booktitle = {Proceedings of LT4HALA 2020 - 1st Workshop on Language Technologies for Historical and Ancient Languages},
  year = {2020},
  url = {https://aclanthology.org/2020.lt4hala-1.2/}
}

References

24 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Adler, M. and Elhadad, M. (2006). An unsupervised morpheme-based hmm for hebrew morphological disambiguation. In Proceedings of the 21st International Conference on Computational Linguistics and the 44th annual meeting of the Association for Computational Linguistics, pages 665–672. Association for Computational Linguistics. ↗ our card
2. Brown, P. F., Pietra, V. J. D., Pietra, S. A. D., and Mercer, R. L. (1993). The mathematics of statistical machine translation: Parameter estimation. Computational linguistics, 19(2):263–311. find
3. Chiang, D. (2007). Hierarchical phrase-based translation. computational linguistics, 33(2):201–228. find our card
4. HaCohen-Kerner, Y., Kass, A., and Peretz, A. (2010). Haads: A hebrew aramaic abbreviation disambiguation system. Journal of the American Society for Information Science and Technology, 61(9):1923–1932. ↗
5. HaCohen-Kerner, Y., Schweitzer, N., and Mughaz, D. (2011). Automatically identifying citations in hebrew-aramaic documents. Cybernetics and Systems: An International Journal, 42(3):180–197. ↗
6. HaCohen-Kerner, Y., Kass, A., and Peretz, A. (2013). Initialism disambiguation: Man versus machine. Journal of the American Society for Information Science and Technology, 64(10):2133–2148. ↗
7. Hamming, R. W. (1950). Error detecting and error correcting codes. The Bell system technical journal, 29(2):147– 160. ↗
8. Jaccard, P. (1901). Étude comparative de la distribution florale dans une portion des alpes et des jura. Bull Soc Vaudoise Sci Nat, 37:547–579. find
9. Jaro, M. A. (1989). Advances in record-linkage methodology as applied to matching the 1985 census of tampa, florida. Journal of the American Statistical Association, 84(406):414–420. ↗
10. Koehn, P., Och, F. J., and Marcu, D. (2003). Statistical phrase-based translation. In Proceedings of the 2003 Conference of the North American Chapter of the Association for Computational Linguistics on Human Language Technology-Volume 1, pages 48–54. Association for Computational Linguistics. ↗
11. Koppel, M. and Schweitzer, N. (2014). Measuring direct and indirect authorial influence in historical corpora. Journal of the Association for Information Science and Technology, 65(10):2138–2144. ↗
12. Levenshtein, V. I. (1966). Binary codes capable of correcting deletions, insertions, and reversals. In Soviet physics doklady, volume 10, pages 707–710. find
13. Liebeskind, C., Dagan, I., and Schler, J. (2012). Statistical thesaurus construction for a morphologically rich language. In * SEM 2012: The First Joint Conference on Lexical and Computational Semantics–Volume 1: Proceedings of the main conference and the shared task, and Volume 2: Proceedings of the Sixth International Workshop on Semantic Evaluation (SemEval 2012), pages 59– 64. find
14. Liebeskind, C., Dagan, I., and Schler, J. (2016). Semiautomatic construction of cross-period thesaurus. Journal on Computing and Cultural Heritage (JOCCH), 9(4):22. ↗
15. Liebeskind, C., Dagan, I., and Schler, J. (2019). An algorithmic scheme for statistical thesaurus construction in a morphologically rich language. Applied Artificial Intelligence, 33(6):483–496. ↗
16. Moghaz, D., Hacohen-Kerner, Y., and Gabbay, D. (2019). Text mining for evaluating authors’ birth and death years. ACM Transactions on Knowledge Discovery from Data (TKDD), 13(1):7. ↗
17. Mughaz, D., HaCohen-Kerner, Y., and Gabbay, D. (2017). Mining and using key-words and key-phrases to identify the era of an anonymous text. In Transactions on Computational Collective Intelligence XXVI, pages 119–143. Springer. ↗
18. Och, F. J. and Ney, H. (2000). A comparison of alignment models for statistical machine translation. In Proceedings of the 18th conference on Computational linguistics-Volume 2, pages 1086–1090. Association for Computational Linguistics. ↗
19. Och, F. J. and Ney, H. (2003). A systematic comparison of various statistical alignment models. Computational linguistics, 29(1):19–51. ↗
20. Och, F. J., Tillmann, C., and Ney, H. (1999). Improved alignment models for statistical machine translation. In 1999 Joint SIGDAT Conference on Empirical Methods in Natural Language Processing and Very Large Corpora. find
21. Porat, E., Koppel, M., and Shmidman, A. (2018). Identification of parallel passages across a large hebrew/aramaic corpus. Journal of Data Mining & Digital Humanities. ↗ our card
22. Snyder, B. and Barzilay, R. (2008). Unsupervised multilingual learning for morphological segmentation. In Proceedings of ACL-08: HLT, pages 737–745, Columbus, Ohio, June. Association for Computational Linguistics. find
23. Vogel, S., Ney, H., and Tillmann, C. (1996). Hmm-based word alignment in statistical translation. In Proceedings of the 16th conference on Computational linguistics-Volume 2, pages 836–841. Association for Computational Linguistics. ↗
24. Zohar, H., Liebeskind, C., Schler, J., and Dagan, I. (2013). Automatic thesaurus construction for cross generation corpus. Journal on Computing and Cultural Heritage (JOCCH), 6(1):4. ↗

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2020-liebeskind-automatic-construction-aramaic-hebrew-translation/