This paper discusses a theoretical framework for a research aimed to produce a dictionary of visual analogies used for the explanation of scientific theories, collected both from historical and contemporary sources. The artifacts will be indexed through a set of criteria and tags that will allow to navigate the contents and map correlations across time, scientific domains and types of publication. The archive will grow as an open-ended accumulation of examples, adapting the methodology for the selection and organisation of the analogies based on the new entries. A set of visualisations will be used in order to navigate the archive and make emerging patterns legible. The initial method of classification will be based on the faceted system envisioned by Luca Rosati (Rosati, 2015), in which artifacts are tagged and tags are organised according to a faceted classification. Tags will not be mutually exclusive, but they’ll act like attributes: each entry may have multiple tags, tthe number of which can grow without any limit or predetermined direction.