[copie] BioLLM

BioLLM aims to develop and implement a model for extracting information from environmental assessment using LLMs (Large Language Models) specifically adapted to this task. The prototype will have to extract information deemed relevant by a community of environmental assessment stakeholders in order to create a consolidated reference database that, once analysed, can be used to report on the application of the ERC sequence.

 

If this prototype proves to be effective, the approach could easily be extended and replicated in other areas of the ecological engineering sector, for which a vast literature exists but from which it is tedious and particularly time-consuming to extract usable data (legal watch, taxonomic watch, etc.).

 

The model will be developed and used on the environmental assessment database compiled by the Occitanie Regional ERC Community, which currently includes more than 800 documents. If the model's performance is satisfactory and a relevant database is produced, this work could be the subject of the data paper.