The marine biodiversity in Areas beyond national jurisdiction (ABNJ), encompassing approximately two-thirds of the global ocean, is persistently declining. In 2023, the agreement on the Conservation and Sustainable Use of Marine Biodiversity of Areas Beyond National Jurisdiction (BBNJ) was officially adopted. Implementing the BBNJ Agreement has the potential to effectively meet global needs for preserving marine biodiversity. Nevertheless, the implementation requires dealing with thousands of legal clauses, and the parties participating in the process lack adequate means to acquire knowledge connected to BBNJ. This paper introduces ChatBBNJ, a highly efficient question-answering system that combines a novel data engineering technique with large language models (LLMs) of Natural Language Processing (NLP). The system aims to efficiently provide stakeholders with BBNJ-related knowledge, thereby facilitating and enhancing their comprehension and involvement with the subject matter. The experimental results demonstrate that the proposed ChatBBNJ exhibits superior expertise in the BBNJ domain, outperforming baseline models in terms of precision, recall, and F1-scores. The successful deployment of the suggested system is expected to greatly assist stakeholders in acquiring BBNJ knowledge and facilitating the effective implementation of the BBNJ Agreement. Therefore, this is expected to contribute to the conservation and sustainable use of marine biodiversity in ABNJ.