All data scientists know the importance of good and unambiguous definitions of data dimensions, crucial to all phases of data analysis. However, semantics is often left implicit in the data, the semantic resources used to create the data are not easily accessible, or available in non-standard formats, non (easily) machine-readable – all factors hampering the possibility of reusing data in information systems or integrating it with other datasets and ultimately limiting the interoperability of data. This paper presents recommendations to engage agrifood sciences in a necessary transition to leverage data production, sharing and reuse and the adoption of the « Findable, Accessible, Interoperable, and Reusable » (FAIR) data principles. They deal with the following data related tasks: search, information extraction, data models, data integration and automated reasoning. [Caracciolo, C, et al. 2020. 39 Hints to Facilitate the Use of Semantics for Data on Agriculture and Nutrition. Data Science Journal, 19: 47, pp. 1–12. DOI: https://doi.org/10.5334/dsj-2020-047]