Abstract

The Land Matrix initiative (https://landmatrix.org) and its globalobservatory aim to provide reliable data on large-scale land acquisitions toinform debates and actions in sectors such as agriculture, extraction, orenergy in low- and middle-income countries. Although these data are recognizedin the academic world, they remain underutilized in public policy, mainly dueto the complexity of access and exploitation, which requires technicalexpertise and a good understanding of the database schema. The objective of this work is to simplify access to data from differentdatabase systems. The methods proposed in this article are evaluated using datafrom the Land Matrix. This work presents various comparisons of Large LanguageModels (LLMs) as well as combinations of LLM adaptations (Prompt Engineering,RAG, Agents) to query different database systems (GraphQL and REST queries).The experiments are reproducible, and a demonstration is available online:https://github.com/tetis-nlp/landmatrix-graphql-python.