Relational Techniques for Storing and Querying RDF Data - Advanced Database Query Systems - page 281

3. Although their generality, there is still a clear gap between the query performance of the ( TS ) and
( BS ) encoding schemes in comparison with the tailored relational encoding scheme ( RS ) of the
RDF data. However, designing a tailored relational schema requires a detailed information about
the structure of the represented objects in the RDF dataset. Such information is not always avail-
able and designing a tailored relational schema limits the schema-free advantage of the RDF data
because any new object with a variant schema will require applying a change in the schema of the
underlying relational structure. Hence, we believe that there is still required efforts to improve the
performance of these generic relational RDF storages and reduce the query performance gap with
the tailored relational encoding schemes.
4. The property tables encoding schemes ( PS ) are trying to fill the gap between the generic encoding
schemes ( TS and BS ) and the tailored encoding schemes ( RS ). The results of our experiments show
that the ( PS ) encoding scheme can achieve a comparable query performance to the ( RS ) encoding
scheme. However, designing the schema of the property tables requires either explicit or implicit

Table 4. A comparison between the alternative relational RDF storage techniques in terms of their
storage cost

Storage Cost (in KB)

Dataset Triple Stores Binary Tables Traditional Relational Property Tables

500K 24721 32120 8175 10225

1M 48142 64214 17820 21200

2M 96251 128634 36125 43450

4M 192842 257412 73500 86200

Table 5. A comparison between the alternative relational RDF storage techniques in terms of their query
performance (in milliseconds)

1M 2M 4M

TS BS RS PS TS BS RS PS TS BS RS PS

Q1 1031 1292 606 701 1982 2208 1008 1262 3651 3807 1988 2108

Q2 1672 1511 776 1109 2982 3012 1606 1987 5402 5601 2308 3783

Q3a 982 1106 61 116 1683 1873 102 198 3022 3342 191 354

Q3b 754 883 46 76 1343 1408 87 132 2063 2203 176 218

Q3c 1106 1224 97 118 1918 2109 209 275 3602 3874 448 684

Q4 21402 21292 11876 14116 38951 37642 20192 25019 66354 64119 39964 48116

Q5 1452 1292 798 932 2754 2598 1504 1786 5011 4806 3116 35612

Q6 2042 1998 1889 2109 3981 3966 3786 4407 7011 6986 6685 8209

Q7 592 30445 412 773 1102 58556 776 1546 2004 116432 1393 2665

Q8 9013 8651 1683 1918 15932 13006 3409 3902 27611 24412 8012 8609

Q9 2502 15311 654 887 4894 26113 1309 1461 9311 37511 2204 2671

Q10 383 596 284 387 714 1117 554 708 1306 2013 1109 1507

Q11 762 514 306 398 1209 961 614 765 2111 1704 1079 1461