The XMark dataset abstractions
On XMark datasets, we obtain two different schemas: one for the wDAG-based abstractions and one for the wPR ones. With the wDAG scoring methods, five entities are reported with a lot of relationships. One can notice that the “text” entity is not fully relevant. This is because this collection is involved in a cycle and thus keeps the data weight from its leaves without transmitting it above. For the wPR-based abstractions, we obtain 5 highly relevant entities (item, open and closed auctions, persons and bidders) together with their several relationships.