An Evaluation Dataset for Linked Data Profiling

摘要

Since the beginning of the Linked Open Data initiative, the number of published Linked Data datasets has gradually increased. However, the reuse of datasets is hindered by a lack of descriptive and reliable metadata about the nature of the data, such as their topic coverage. Manual curation of metadata is however costly and hard to maintain, because of which we advocate a Linked Data profiling approach that will be able to automatically extract topics from datasets as metadata. One of the main challenges in developing this is the lack of evaluation data, i.e. manually curated metadata (topics) for datasets. In this paper we describe such an evaluation dataset and the framework that enabled its creation.