The Prescriptions dataset abstractions
On the Prescriptions dataset, Abstra find two different results depending on the methods used. For the DAG score, one main entity is found. It is not classified because it has only three properties and no collection label, thus no sufficient insight. For the abstractions using the PR scoring method (regardless the boundary method), we obtain two main entities. First we obtain the collection of medicines given by the doctors (misclassified as a creative work) and then the entity with the “npi” and “provider_variables” properties. DAG scoring methods perform better on this dataset.
 wDAG score and DAG boundary |
 wDAG score and FL boundary |
 wDAG score and FLAC boundary |
 wPR score and DAG boundary |
 wPR score and FL boundary |
 wPR score and FLAC boundary |
