I would say that using separate datasets is a good idea if you have sets
of graphs that just don't belong together. The dataset as an
organisational, abstract container is an excellent idea, in my opinion.
Regards,
Daan
On 22-03-18 11:22, Mikael Pesonen wrote:
Ok seems that using many datasets is not a good idea. I had no bias and
not having any issues with speed, just wanted to see what is best way to
go.
On 21.3.2018 20:48, ajs6f wrote:
Those sure are good reasons for using named graphs. But what about
using different datasets too?
Consider that you may not be seeing such reasons because it may not
actually be as good an idea.
Here's another reason to prefer graphs: There is a standard management
HTTP API for named graphs: SPARQL Graph Store. There is no equivalent
for datasets, so each product rolls its own. That's not good for
flexibility if you have to move products.
As for performance, that will depend radically on the implementation.
Jena TIM, for example, using hashing for its indexes, so the
difference between having a lot of quads in a dataset and a few isn't
likely to be that much. Other impls will vary.
Are you sure that performance is going to be improved by separating
out datasets? (I.e. is that the measured bottleneck?) Are you now
having problems with queries accidentally querying data they shouldn't
see, and can your queries be rewritten to fix that (which might also
improve performance)? (Jena has a permissions framework that can
secure information down to the individual triple.)
ajs6f
On Mar 21, 2018, at 6:35 AM, Mikael Pesonen
<mikael.peso...@lingsoft.fi> wrote:
Those sure are good reasons for using named graphs. But what about
using different datasets too?
btw, I couldn't find info on how to run many datasets with Fuseki. is
it just one dataset per fuseki process? -loc parameter for
fuseki-server.jar?
Br
On 20.3.2018 14:22, Martynas Jusevičius wrote:
Provenance. With named graphs, it's easier to track where data came
from:
who imported it, when etc.
You can also have meta-graphs about other graphs.
Also editing and updating data. You can load named graph contents (of
smallish size) in an editor, make changes and then store a new
version in
the same graph. You probably would not want to do this with a large
default
graph.
On Tue, Mar 20, 2018 at 1:16 PM, Mikael Pesonen
<mikael.peso...@lingsoft.fi>
wrote:
Hi,
I'm using Fuseki GSP, and so far have put all data into one default
dataset and using graphs to split it.
If I'm right there would be benefits using more than one dataset
- better performance - each query is done inside a dataset so less
data =
faster query
- protection of data - can't "accidentaly" query data from other
datasets
Downsides:
- combining data from various datasets is heavier task
Is this correct? Any other things that should be considered?
Thank you
--
Lingsoft - 30 years of Leading Language Management
www.lingsoft.fi
Speech Applications - Language Management - Translation - Reader's and
Writer's Tools - Text Tools - E-books and M-books
Mikael Pesonen
System Engineer
e-mail: mikael.peso...@lingsoft.fi
Tel. +358 2 279 3300
Time zone: GMT+2
Helsinki Office
Eteläranta 10
<https://maps.google.com/?q=Etel%C3%A4ranta+10&entry=gmail&source=g>
FI-00130 Helsinki
FINLAND
Turku Office
Kauppiaskatu 5 A
<https://maps.google.com/?q=Kauppiaskatu+5+A&entry=gmail&source=g>
FI-20100 Turku
FINLAND
--
Lingsoft - 30 years of Leading Language Management
www.lingsoft.fi
Speech Applications - Language Management - Translation - Reader's
and Writer's Tools - Text Tools - E-books and M-books
Mikael Pesonen
System Engineer
e-mail: mikael.peso...@lingsoft.fi
Tel. +358 2 279 3300
Time zone: GMT+2
Helsinki Office
Eteläranta 10
FI-00130 Helsinki
FINLAND
Turku Office
Kauppiaskatu 5 A
FI-20100 Turku
FINLAND