An Approach for Generating and Semantically Enriching Dataset Profiles

摘要

The identification of appropriate datasets on the Web for a given task is still a challenge. To help matters, metadata describing datasets can be provided. It is possible to make these metadata available through a Dataset Profile (DSP). In this light, this work presents an approach which generates a DSP composed by descriptive, structural and quality metadata. The DSP is enriched by semantically referencing the provided metadata and by means of some new metadata, such as the dataset domain and some quality metadata, e.g. comprehensibility and processability. In order to evaluate the proposed approach, a prototype has been developed and some experiments have been accomplished.