The Joconde database is a French database, which describes about 600,000 works from French art collections. In the Data&Musée project, we process data from museums and monuments. We have chosen to model the data using a knowledge graph approach. We enrich the data of the project partners with data from other sources. In this article, we present the semantic representation that we have adopted for the Joconde database and the methods used to obtain this representation. Our semantic representation of the Joconde database is available as Open data as the SemJoconde dataset. We believe that the SemJoconde data can become useful references for work on the use of semantic techniques in the cultural field.