-
Notifications
You must be signed in to change notification settings - Fork 0
/
details.txt
75 lines (59 loc) · 2.78 KB
/
details.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
SPARUB = SPARQL UPDATE BENCHMARK
================================
> SPARUB: The Tyrex' benchmark dealing with the SPARQL UPDATE
recommendation of the W3C.
> Version: 1.0
Context
-------
Recently, the W3C extended the SPARQL syntax[2] --which is the RDF
query language[1]-- with a new recommendation dealing with
updates[3]. This extension provides new keywords to add of delete
triples from a database using the classic SPARQL syntax.
Such an evolution allows to build more realistic application for the
semantic web. Indeed, the dynamicity of data is often part of the
game. Thereby, instead of manually (or externally) modify the
datasets, this extension offers a way to perform these changes
internally.
In the same time, parallely to the development of SPARQL evaluators
and RDF stores, several RDF/SPARQL benchmarks have been published in
order to rank the various solutions. According to the literature, each
benchmark offers its own specifities; for instance, they are not
focusing on the same SPARQL fragment… They are mostly twofold: first a
dataset --which can be generated (usually deterministically) or
extracted from real datasources-- second a set of queries. Most of
them allow to compare evaluators and stores according to time
performance metrics and sometimes take also in account the memory/disk
footprint during the steps of computation (either loading or
querying).
Similarly to the extension of the SPARQL standard proposed by the W3C,
we propose a tool which extends the already existing benchmarks in
order to integrate updating routines in the test suites called SPARUB
(SPARQL UPDATE Benchmark).
Concept of SPARUB
-----------------
The idea of SPARUB is to extend all the existing RDF/SPARQL benchmarks
i.e. keeping the already run test suite and giving new scenarios
involving dynamicity of data. To do that, SPARUB takes as inputs an
RDF dataset (N-Triples format to be easily parsed) and an optional
list of SPARQL queries. It then returns a set of RDF files, lots of
SPARQL queries and a testing scenario (and also some logs…).
Technically, SPARUB starts by analyzing the various SPARQL queries in
order to list the used SPARQL keywords and thus having an idea of the
benchmarked fragment. It then splits the input RDF file into pieces
according (1) to statistics of the initial dataset and (2) to the
various sub-fragment that can be extracted from the general tested
one.
License
-------
This project is under the CeCILL license[4].
References
----------
[1] RDF <https://www.w3.org/TR/2004/REC-rdf-primer-20040210/>
[2] SPARQL 1.1 <https://www.w3.org/TR/sparql11-query/>
[3] SPARQL UPDATE <https://www.w3.org/TR/sparql11-update/>
[4] CeCILL license <http://www.cecill.info/index.en.html>
Contributor
-----------
Damien Graux
Tyrex Team, Inria (France)
<http://tyrex.inria.fr>