LAW: Link-AWare Source Selection for Virtually Integrating Linked Data - Technologies and Applications of Artificial Intelligence - page 246

Fig. 4. Average times to execute the BSBM query mix

Tabl e 3. Average Number of Requests

DARQ SPLENDID FedX LAW
Q1 2 2 5 1

Q2 35 12 22 6

Q3 92,358 689 172 8

Q4 22,331 1,810 696 4

Q5 1 1 2 1

Q6 28,673 1,574 233 6

Q7 1,879 3,231 1,674 4

Q8 35,832 2,785 1,346 4

Q9 1 1 1 1

Q10 4,582 2,056 2,897 4

Q11 X 18,542 2343 4

Q12 1,478 983 486 6

7 Conclusions

In this paper we presented LAW, an approach for link-aware federated querying
over the Web of Data. LAW is based on property link graphs approximating triple
link graphs. We evaluated our approach against DARQ, SPLENDID and FedX.
The evaluation shows that by using the LAWS extension the query execution
times were improved in most of the cases. Wide experiments based on BSBM
show that LAW as an independent query engine has higher performance than
the other three federated query systems.

Acknowledgment. This work was supported by the National Natural Science
Foundation of China (no. 61370137, 61272361) and the 111 Project of Beijing
Institute of Technology.