The following tables are the summarized final table. Please the detailed five test result in each system pages.

Query FedX(ms) SPLENDID(ms)ANAPSID(ms)ADERIS(ms)
q1 95454764890525647
q2 43321XXX
q3 109426XXX
q4 107310XXX
q5 801519XXX
q3s 106972340205XX
q5s 11216134779XX

"X" for SPLENDID: time out after returned part of the results.

"X" for ANAPSID: no result with message such as "query2 0.00762486457825 0.00762581825256 0.007625818252560.007625 81825256 0 0"

"X" for ADERIS: parser error. however no real syntax error exists.

SPARQL 1.1 for FedX

Query FedX(ms)
q1' 106689
q1" 106345
q3' 105644
q3" 106062
q4' 19935
q5' 143797

SPARQL1.0 for FedX

Query FedX(ms)
q1 9545
q2 43321
q3 109426
q4 107310
q5 801519

q1 becomes slow; q3 shows no significant change;

For q2,q4 and q5, queries of SAPRQL 1.1 are far faster;

As for SPARQL 1.1, query q1’,q3’, and q4’ show no much influence, by changing the order of service clauses; q2’ changes much, and q2” returns less results.(40 vs. 20)

In most cases, by adding endpoints information could improve the performance.

In some cases, it could take worse affect, such as q2”,longer time and less results