Towards Enabling FAIR Dataspaces Using Large Language Models