The Fermi-LAT telescope has provided an unprecedented view of the GeV gamma-ray sky since its launch in 2008. The latest Fermi-LAT catalogue of Active Galactic Nuclei lists 1591 sources associated with AGN, of which 460 are classified as blazar candidates of uncertain type (BCU). The characterisation of the physical properties of these BCU sources is important for observational cosmology and fundamental physics, as these sources and their environments constitute a natural laboratory to study particle acceleration and matter/radiation interactions in extreme conditions. Of particular interest is the search for new and interesting/unusual sources that may be observable at very high energies by ground-based imaging atmospheric Cherenkov telescopes. Based on the observed gamma-ray properties, a number of machine learning techniques are being investigated to classify these sources. However, the classification of a blazar as a FSRQ or BL Lac depends on the optical spectral properties. Here we discuss the work that we have thus far undertaken to optically characterise a selection of sources as well as future plans to undertake classification to help calibrate an artificial neural network method.