Making Ukiyo-e Easier to Discover:  A Recommender System for Digital Archives

1. Abstract

Ukiyo-e is a kind of woodblock print that has high artistic and research value. It is preserved by many digital archives (DAs), such as the Art Research Center Ukiyo-e Portal Database (ARC-UDB) of Ritsumeikan University. ARC-UDB is mainly built for the experts of humanities fields. In this research, to meet the potential needs of the expert users who will browse or explore ARC-UDB, we propose a recommender system. The proposed recommender system utilizes an existing link prediction model, which exploits the graph-like datasets of ARC-UDB as the input of the recommendation algorithm. We optimize the format of input of the link prediction model, to the format that is suitable for ARC-UDB datasets. From the results, we find that the proposed method is effective for the task. This recommender system could also be applied to other DAs that are with graph-like dataset structures.