{"id":989,"date":"2026-09-11T16:01:59","date_gmt":"2026-09-11T13:01:59","guid":{"rendered":"https:\/\/sisu.ut.ee\/digits\/?p=989"},"modified":"2026-09-11T16:03:02","modified_gmt":"2026-09-11T13:03:02","slug":"open-calls-for-doctoral-studies","status":"publish","type":"post","link":"https:\/\/sisu.ut.ee\/digits\/open-calls-for-doctoral-studies\/","title":{"rendered":"Open Call for Doctoral Studies: Join DigiTS as a Junior Researcher!"},"content":{"rendered":"

<\/p>\n\n\n\n

\"\"
Photo: Andero Kalju<\/figcaption><\/figure>\n\n\n\n

The Centre for Digital Humanities is looking for a PhD Researcher to join the team of DigiTS to study linguistic annotation with LLM-s.<\/strong><\/p>\n\n\n\n

This specialisation belongs to the doctoral programme in Humanities<\/a>. There is 1 position<\/strong> available. It is not possible to apply with your own topic during the application period.<\/p>\n\n\n\n

Research topic open for admission<\/strong>: Linguistic Annotation with LLMs (supervisor Maciej Eder)<\/p>\n\n\n\n

Description of the topic<\/strong>:<\/p>\n\n\n\n

The PhD project is aimed at testing several Large Language Models in solving a variety of different linguistic annotation tasks. Intuitively, some of these tasks should be accomplished by the models at an acceptable accuracy level, but we have no empirical evidence to support such a hypothesis. This project will try to test, in a series of carefully designed experiments, the extent to which LLMs are capable of recognizing patterns in scenarios that they were not optimized for.<\/p>\n\n\n\n

Primarily, the datasets used in this project will be already-existing manually-annotated \u201cgold sets\u201d from a number of previous studies. Some of the chosen datasets will be relatively easy to solve by most LLMs, whereas some other tasks will be chosen specifically to make the problem difficult for the models.<\/p>\n\n\n\n

The primary methodological setup used in this project will be a series of supervised machine-learning scenarios, in which the same set will be automatically annotated, and then compared against the human-annotated \u201cgold set\u201d. A series of different models will be tested. Distilled and quantized Large Language Models will be used, adjusted through zero-shot prompting. A series of alternative prompting strategies will be conducted. Also, the novel class of reasoning models, implementing chain-of-thought output and imitating human reasoning, will be heavily tested in this project.<\/p>\n\n\n\n

Admitted doctoral researcher will work as a junior research fellow<\/strong> at the university. The estimated workload is 1.0 and the estimated time period is four years. Studies are expected to start in autumn 2026.<\/strong> The final workload and exact starting date will be fixed after the student is admitted, during work contract negotiations.<\/p>\n\n\n\n

Candidates must submit an application with other requested documents<\/strong>:<\/p>\n\n\n\n