Federated searching system for humanities databases using automatic metadata mapping
Fuminori Kimura, Takushi Toba, Taro Tezuka, Akira Maeda · 2009
Recently, many collections and resources in libraries, museums and research institutes are digitized and opened to the public. As there are many humanities databases and each database has its own user-interface and metadata schema, it is not easy for users to find desired information. Users must input the same query in different ways for each database in order to access all related databases (Chang et al., 1999). Our goal is to construct a federated searching system for humanities databases. A federated searching system refers to the retrieval system that a user can access multiple humanities databases with only one query input. In order to realize a federated searching system for existing heterogeneous databases, a method for metadata mapping of attribute names is needed to cope with the differences in schema for each database. Therefore, we propose a method of automatic metadata mapping using metadata elements that are revised from Dublin Core Metadata Element Set (DCMI, 2008). Our proposed method consists of two preprocessing phases and four mapping phases. Figure 1 shows the flow of metadata mapping of our proposed method.