Merging and Enriching DCAT Feeds to Improve Discoverability of Datasets

摘要

Data Catalog Vocabulary (DCAT) is a W3C specification to describe datasets published on the Web. However, these catalogs are not easily discoverable based on a user's needs. In this paper, we introduce the Node.js module 'dcat-merger' which allows a user agent to download and semantically merge different DCAT feeds from the Web into one DCAT feed, which can be republished. Merging the input feeds is followed by enriching them. Besides determining the subjects of the datasets, using DBpedia Spotlight, two extensions were built: one categorizes the datasets according to a taxonomy, and the other adds spatial properties to the datasets. These extensions require the use of information available in DBpedia's SPARQL endpoint. However, public SPARQL endpoints often suffer from low availability, its Triple Pattern Fragments alternative is used. However, the need for DCAT Merger sparks the discussion for more high level functionality to improve a catalog's discoverability.