@inproceedings{cohen-etal-2026-scaling,
title = "Scaling Sentence Similarity for {C}lassical {T}ibetan with Automatic Annotations",
author = "Cohen, Shay and
Yang, Jingyi and
Rabinovitz, Gal and
Choden, Sonam and
Shtrosberg, Ofir and
Bajetta, Nicola and
Horin, Goody Ben and
Sund{\'e}n, Rebecca and
Drori, Omri and
Jamtsho, Sonam and
Wangchuk, Dorji and
Bar, Kfir and
Almogi, Orna and
Fine, Shai",
editor = {Hamilton, Sil and
{\"O}hman, Emily and
Hicke, Rebecca M. M. and
Bizzoni, Yuri and
Bax, Axel and
Matthews, Jacob A. and
H{\"a}m{\"a}l{\"a}inen, Mika},
booktitle = "Proceedings of the 6th International Conference on Natural Language Processing for the Digital Humanities",
month = jul,
year = "2026",
address = "San Diego, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.nlp4dh-1.15/",
pages = "150--166",
ISBN = "979-8-89176-427-9",
abstract = "Identifying intertextual parallels is central to philology, traditionally requiring labor-intensive manual analysis. While digitized historical corpora enable automated approaches using semantic sentence embeddings, training such models requires large annotated datasets, which are scarce for low-resource languages. We address this challenge by introducing a scalable automatic annotation pipeline for training semantic embedding models for Classical Tibetan. Our method combines unsupervised contrastive bootstrapping with iterative pair mining, generating silver-standard similarity labels through two complementary annotation strategies: (1) an ensemble of embedding models and rerankers, and (2) an LLM-as-a-judge committee using best{--}worst scaling. When combined with a domain-specific, gold-standard annotated dataset for sequential fine-tuning, the resulting text-similarity model achieves a state-of-the-art Spearman correlation of 0.864 on the STS task. This enables effective semantic search in Classical Tibetan and provides a framework for automatic supervision in low-resource languages used in digital humanities. We will make our code, dataset, and trained model publicly available upon publication."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="cohen-etal-2026-scaling">
<titleInfo>
<title>Scaling Sentence Similarity for Classical Tibetan with Automatic Annotations</title>
</titleInfo>
<name type="personal">
<namePart type="given">Shay</namePart>
<namePart type="family">Cohen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jingyi</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gal</namePart>
<namePart type="family">Rabinovitz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sonam</namePart>
<namePart type="family">Choden</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ofir</namePart>
<namePart type="family">Shtrosberg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nicola</namePart>
<namePart type="family">Bajetta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Goody</namePart>
<namePart type="given">Ben</namePart>
<namePart type="family">Horin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rebecca</namePart>
<namePart type="family">Sundén</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Omri</namePart>
<namePart type="family">Drori</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sonam</namePart>
<namePart type="family">Jamtsho</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dorji</namePart>
<namePart type="family">Wangchuk</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kfir</namePart>
<namePart type="family">Bar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Orna</namePart>
<namePart type="family">Almogi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shai</namePart>
<namePart type="family">Fine</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 6th International Conference on Natural Language Processing for the Digital Humanities</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sil</namePart>
<namePart type="family">Hamilton</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Emily</namePart>
<namePart type="family">Öhman</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rebecca</namePart>
<namePart type="given">M</namePart>
<namePart type="given">M</namePart>
<namePart type="family">Hicke</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuri</namePart>
<namePart type="family">Bizzoni</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Axel</namePart>
<namePart type="family">Bax</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jacob</namePart>
<namePart type="given">A</namePart>
<namePart type="family">Matthews</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mika</namePart>
<namePart type="family">Hämäläinen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, USA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-427-9</identifier>
</relatedItem>
<abstract>Identifying intertextual parallels is central to philology, traditionally requiring labor-intensive manual analysis. While digitized historical corpora enable automated approaches using semantic sentence embeddings, training such models requires large annotated datasets, which are scarce for low-resource languages. We address this challenge by introducing a scalable automatic annotation pipeline for training semantic embedding models for Classical Tibetan. Our method combines unsupervised contrastive bootstrapping with iterative pair mining, generating silver-standard similarity labels through two complementary annotation strategies: (1) an ensemble of embedding models and rerankers, and (2) an LLM-as-a-judge committee using best–worst scaling. When combined with a domain-specific, gold-standard annotated dataset for sequential fine-tuning, the resulting text-similarity model achieves a state-of-the-art Spearman correlation of 0.864 on the STS task. This enables effective semantic search in Classical Tibetan and provides a framework for automatic supervision in low-resource languages used in digital humanities. We will make our code, dataset, and trained model publicly available upon publication.</abstract>
<identifier type="citekey">cohen-etal-2026-scaling</identifier>
<location>
<url>https://aclanthology.org/2026.nlp4dh-1.15/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>150</start>
<end>166</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Scaling Sentence Similarity for Classical Tibetan with Automatic Annotations
%A Cohen, Shay
%A Yang, Jingyi
%A Rabinovitz, Gal
%A Choden, Sonam
%A Shtrosberg, Ofir
%A Bajetta, Nicola
%A Horin, Goody Ben
%A Sundén, Rebecca
%A Drori, Omri
%A Jamtsho, Sonam
%A Wangchuk, Dorji
%A Bar, Kfir
%A Almogi, Orna
%A Fine, Shai
%Y Hamilton, Sil
%Y Öhman, Emily
%Y Hicke, Rebecca M. M.
%Y Bizzoni, Yuri
%Y Bax, Axel
%Y Matthews, Jacob A.
%Y Hämäläinen, Mika
%S Proceedings of the 6th International Conference on Natural Language Processing for the Digital Humanities
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, USA
%@ 979-8-89176-427-9
%F cohen-etal-2026-scaling
%X Identifying intertextual parallels is central to philology, traditionally requiring labor-intensive manual analysis. While digitized historical corpora enable automated approaches using semantic sentence embeddings, training such models requires large annotated datasets, which are scarce for low-resource languages. We address this challenge by introducing a scalable automatic annotation pipeline for training semantic embedding models for Classical Tibetan. Our method combines unsupervised contrastive bootstrapping with iterative pair mining, generating silver-standard similarity labels through two complementary annotation strategies: (1) an ensemble of embedding models and rerankers, and (2) an LLM-as-a-judge committee using best–worst scaling. When combined with a domain-specific, gold-standard annotated dataset for sequential fine-tuning, the resulting text-similarity model achieves a state-of-the-art Spearman correlation of 0.864 on the STS task. This enables effective semantic search in Classical Tibetan and provides a framework for automatic supervision in low-resource languages used in digital humanities. We will make our code, dataset, and trained model publicly available upon publication.
%U https://aclanthology.org/2026.nlp4dh-1.15/
%P 150-166
Markdown (Informal)
[Scaling Sentence Similarity for Classical Tibetan with Automatic Annotations](https://aclanthology.org/2026.nlp4dh-1.15/) (Cohen et al., NLP4DH 2026)
ACL
- Shay Cohen, Jingyi Yang, Gal Rabinovitz, Sonam Choden, Ofir Shtrosberg, Nicola Bajetta, Goody Ben Horin, Rebecca Sundén, Omri Drori, Sonam Jamtsho, Dorji Wangchuk, Kfir Bar, Orna Almogi, and Shai Fine. 2026. Scaling Sentence Similarity for Classical Tibetan with Automatic Annotations. In Proceedings of the 6th International Conference on Natural Language Processing for the Digital Humanities, pages 150–166, San Diego, USA. Association for Computational Linguistics.