Museums around the world have built databases with metadata about millions of objects, the people who created them, and the entities they represent. This data is stored in proprietary databases and is not readily available for use. Recently, museums embraced the Semantic Web as a means to make this data available to the world, but the experience so far shows that publishing museum data to the Linked Data cloud is di cult: the databases are large and complex, the information is richly structured and varies from museum to museum, and it is di cult to link the data to other datasets. We have been collaborating with the Smithsonian American Art Museum to create a set of tools that allow museums and other cultural heritage institutions to publish their data as Linked Open Data. In this demonstration we show the end-to-end process of starting with the original source data, modeling the data with respect to a ontology of cultural heritage data, linking the data to DBpedia, and then publishing the information as Linked Data.
Datasets. In this demonstration we will use datasets extracted from Smithsonian American Art Museum collection management database. Figure 1 shows a sample of raw data about artists. One of the challenges in mapping this particular table to RDF is that the meaning of each row is encoded in the last column of the table, which shows whether the data speci es the place of birth, place of death, or an associated place. Each row must be mapped to a di erent property in the ontology.
Mapping the Data to RDF. The rst step in the process is to model the
raw data using an ontology of cultural heritage data. We developed a tool called
Karma [
Linking to External Datasets. Once the RDF is published, the next step is to link the information about entities, including artists and locations, to the corresponding entities in other sources. In particular we link artists to people in DBpedia and locations to the corresponding locations in Geonames. The name elds of the artists, possibly augmented with other identifying information such as birth year, is sent to a reconciliation service that has a fuzzy index of DBpedia information. Figure 3 shows a user invoking the reconciliation service on the Person class, and it shows the resulting links and scores for each match to DBpedia. The interface also allows a user to verify each of the links and to drill down into the individual data sources to determine whether the links are correct. Publishing the Linked Data Once the links have been veri ed, the information is then published and made available as Linked Open Data. We created a user-friendly version of Pubby4 that describes the content and shows the images 4 http://wifo5-03.informatik.uni-mannheim.de/pubby/ and labels directly on the page to make the resulting linked data more readable. Figure 4 shows a screen shot of the published data from the same data source shown in Figure 1.
Using the Linked Data Once the linking is complete, the Linked Data can also be used to augment other sources of data. In our project with the Smithsonian, they found that one use of the Linked Data is to augment their current Web pages with the additional information available from Wikipedia and the New York Times, which follows directly from the linking of their data about artists to DBpedia. Figure 4 shows screen shots of our user friendly version of Pubby and the Smithsonian American Art Museum's Web site with the additional links generated directly from the Linked Data that we produced for them. The Linked Data can also be used to curate the Smithsonian's own database, create virtual online museums, or create new applications that build on the Linked Data.