Processing e-Resources Usage Data using LLM Agents
| Thesis title in Czech: | Zpracování dat o využívání e-zdrojů pomocí agentů založených na velkých jazykových modelech |
|---|---|
| Thesis title in English: | Processing e-Resources Usage Data using LLM Agents |
| Key words: | velké jazykové modely|multi-agentní systém|COUNTER Metrics |
| English key words: | large language models|multi-agent system|COUNTER Metrics |
| Academic year of topic announcement: | 2024/2025 |
| Thesis type: | diploma thesis |
| Thesis language: | angličtina |
| Department: | Department of Distributed and Dependable Systems (32-KDSS) |
| Supervisor: | doc. RNDr. Jan Kofroň, Ph.D. |
| Author: | hidden - assigned and confirmed by the Study Dept. |
| Date of registration: | 28.03.2025 |
| Date of assignment: | 02.04.2025 |
| Confirmed by Study dept. on: | 02.04.2025 |
| Date and time of defence: | 09.09.2025 09:00 |
| Date of electronic submission: | 29.04.2025 |
| Date of submission of printed version: | 17.07.2025 |
| Date of proceeded defence: | 09.09.2025 |
| Opponents: | doc. Mgr. Martin Pilát, Ph.D. |
| Guidelines |
| COUNTER [1] is the standard for publishing data on e-resources, but many publishers do not adhere to it, making their data difficult to use for both users and SaaS platforms. CELUS [2] is a SaaS solution designed to process such data, providing users with all relevant information in one place. However, handling non-COUNTER data remains a challenge due to inconsistencies in how publishers report their statistics.
The goal of this thesis is to develop a proof-of-concept implementation for automating the processing of non-COUNTER data using LLMs [5,6,7]. Different approaches and various LLM models, both proprietary and open-source, should be compared. Additionally, a multi-agent system converting the non-COUNTER data to COUNTER data should be implemented, along with a simple user (web) interface. |
| References |
| [1] COUNTER standard: https://www.countermetrics.org/
[2] CELUS: https://www.celus.net/ [3] LLM agents library: https://openai.github.io/openai-agents-python/ [4] LLM management library: https://ollama.com/ [5] Gemma3: https://ai.google.dev/gemma [6] Qwen2.5-Coder Series: https://qwenlm.github.io/blog/qwen2.5-coder-family/ [7] Mistral Small 3: https://mistral.ai/news/mistral-small-3 |