Metal conservators confronted with unknown artefacts rely on previous literature to develop treatment protocols. This search can be tedious given the dissemination of information across corpus of unstructured texts, mainly in the form of research papers and semi-structured databases of artefacts. In order to improve the search of artefacts sharing similar characteristics (metal composition and structure, conservation condition, etc.), this project proposes a hybrid search engine based on a domain ontology. Using a database populated with information resulting from comprehensive investigations of historic and archaeological artefacts, we extracted and selected key concepts and their relations through the use a various lexical analysis tools. Based on this corpus and frequency analysis, we were able to build an ontology of the domain, opening new perspective on information retrieval. Conservators are able to leverage the power of the hybrid search engine to compare their observations on a specific artefact with objects already stored in the database or with indexed research papers. Using keywords to describe corrosion forms they are confronted with, conservators can retrieve artefacts showing similar corrosion phenomena and assess the conservation condition of their artefacts, e.g. diagnosing the stability of metals or determining the location of the limit of the original surface in corrosion product crusts.
De Santo, Alessio; Vonlanthen, Yann; Rosselet, Antoine; Degrigny, Christian; and Gaspoz, Cédric, "An Ontology to Support Non-Invasive Diagnosis of Heritage Metals" (2016). MCIS 2016 Proceedings. 32.