The data within collections from all Digital Humanities fields must be trustworthy. To this end, both provenance and change-tracking systems are needed. This contribution offers a systematic review of the metadata representation models for provenance in RDF, focusing on the problem of modelling conjectures in humanistic data.