Data-driven computing is becoming a new paradigm in several scientific fields with a tremendous impact on new technologies. In solid mechanics, the availability of large volumes of data through modern experimental techniques is enabling machine learning methods to open new perspectives in material modelling. In the framework of the research activities of Spoke 7 “Materials & Molecular Sciences” of the National Centre for HPC, Big Data and Quantum Computing, the present PhD project will be focused on the development and implementation of unsupervised data-driven techniques to automatically discover constitutive laws hidden in large volumes of data. Unsupervised approaches require no stress data. Therefore, the methods will have to rely only on full-field displacements and global reaction forces data. The project aims at developing these approaches for materials with complex behaviour, such as rate-dependent, shape memory, temperature-responsive, etc.