Tibetan NLP
A research direction in natural language processing focused on the Tibetan language, belonging to the low-resource language NLP domain. It encompasses the full pipeline of corpus construction, word segmentation, speech recognition, and large language model development for Tibetan.
Timeline (last 90 days)
The survey covers 140 Tibetan NLP publications from 2004 to 2026 in the ACL Anthology and DBLP databases.
84% of Tibetan NLP papers were published after 2018.
The development of Tibetan NLP exhibits a bottom-up characteristic of gradual expansion along the NLP pipeline, starting from corpus construction and word segmentation, extending to text processing, speech recognition, and downstream applications, and ultimately reaching the large language model stage.
Chinese researchers are listed as first authors on 87% of Tibetan NLP papers and cover all 25 research areas identified in the survey.
The survey is claimed to be the first peer-reviewed systematic bibliometric survey covering the development trajectory of Tibetan NLP.
Only 29% of Tibetan NLP papers have made their data, code, or models publicly available.
The survey authors propose that the full-pipeline research paradigm of Tibetan NLP can serve as a reference model for other low-resource languages in the Global South.
The survey points out shortcomings in Tibetan NLP, including insufficient dialect coverage, absence of responsible NLP practices, lack of native and comparable evaluation systems, low cross-community visibility, and limited resource openness.
The survey authors recommend promoting open science, establishing shared standards, and encouraging substantive participation from language communities in the era of large models.
All Facts (9)
The survey covers 140 Tibetan NLP publications from 2004 to 2026 in the ACL Anthology and DBLP databases.
50%Unverified84% of Tibetan NLP papers were published after 2018.
50%UnverifiedThe development of Tibetan NLP exhibits a bottom-up characteristic of gradual expansion along the NLP pipeline, starting from corpus construction and word segmentation, extending to text processing, speech recognition, and downstream applications, and ultimately reaching the large language model stage.
50%UnverifiedChinese researchers are listed as first authors on 87% of Tibetan NLP papers and cover all 25 research areas identified in the survey.
50%UnverifiedThe survey is claimed to be the first peer-reviewed systematic bibliometric survey covering the development trajectory of Tibetan NLP.
50%UnverifiedOnly 29% of Tibetan NLP papers have made their data, code, or models publicly available.
50%UnverifiedThe survey authors propose that the full-pipeline research paradigm of Tibetan NLP can serve as a reference model for other low-resource languages in the Global South.
50%UnverifiedThe survey points out shortcomings in Tibetan NLP, including insufficient dialect coverage, absence of responsible NLP practices, lack of native and comparable evaluation systems, low cross-community visibility, and limited resource openness.
50%UnverifiedThe survey authors recommend promoting open science, establishing shared standards, and encouraging substantive participation from language communities in the era of large models.
50%