With the exponential growth of data in all areas of our lives, there is an increasing need of developing new approaches for effective data management. Namely, in the field of Data Mining (DM) and Knowledge Discovery in Databases (KDD), scientists often invest a lot of time and resources for collec- ting data that has already been acquired. In that context, by publishing open and FAIR (Findable, Accessible, Interoperable, Reusable) data, researchers could reuse data that was previously collected, preprocessed and stored. Motivated by this, we conducted extensive review on current approaches, data repositories and semantic technologies used for annotation, storage and querying of datasets for the domain of machine learning (ML) and data mining. Finally, we identify the limitations of the existing repositories of datasets and propose a design of a semantic data repository that adheres to FAIR principles for data management and stewardship.