Hi Rupert, yes I had already reported our discussion internally but it's
always better to discuss it on stanbol-dev.
On 3/3/12 10:13 AM, Rupert Westenthaler wrote:
As the goal is to build the first release candidate Mo-Du next week the
intension here is to find the best way on how to include the RefactorEngine in
the full launcher. For this the idea was to
* keep the Refactor engine in the full launcher. This engine is an important
feature of Apache Stanbol and should be available by default
* remove the default configuration of the Engine that uses the SEO (Search
Engine Optimization) recipe. This means the the Refactor engine is not active
in the default configuration (similar to the KeywordExtractionEngine).
* keep the SEO recipe in the default configuration. This ensures that users
that want to try this use case do not need to load/init the recipe.
this was the content of your recent commit, right? Personally it is fine
with me
* set the default values (the values in the @Property annotations) so that they
work for the SEO recipe. This requires Users that want to test/use the SEO use
case to only go to the configuration tab of the OSGI Web Console click the [+]
of the RefactorEngine and than [OK].
configuration tab of the OSGI Web Console click the [+] of the RefactorEngine
and than [OK]
+1
Longer Term plan (after the 0.9 Release)
---------
Alessandro was mentioning some points about adding native support for Clerezza
graphs for the Rule component. Alessandro/Alberto maybe you can add more
information here
Andrea should correct me if I'm wrong, but I'm told there's work under
the hood on exporting Stanbol rules as ConstructQuery Clerezza objects
natively and avoiding to go through the ordeal of Jena rule conversion.
I had also the Idea to try using an IndexMGraph in-memory graph instead of the
the file-based Jena TDB. This could considerable boost performance. However one
would need validate the memory requirements.
As I told you, I do need to measure the memory footprints of many
implementations of ours. I've checked the Oracle VisualVM you mentioned,
but if there is some cool profiler in Apache I'd be happy to know. We'll
return on this anyway.
In the meantime I've just started to use the IndexedMGraph for in-memory
work in OntoNet.
Build an other Usecase that is more simple than SEO with the goal to include it in
the default configuration. SEO includes>60 rather complex rule definition. That
is cool for showing the power of this component. For the default configuration I
would like to have a simple Usecase that just need ~5 Rules. This could also
include a 5min Tutorial how to create this 5 Rules and a 15min Tutorial that
extends this default configuration by some additional Rules to include an other
feature. Maybe the combination of IPTC, rNews, TopicEngine and RefactorEngine
could be such an Example.
It could, but on an educated guess the IPTC/rNews combination might
require a bit more than 5 rules, maybe just a subset.
best
Alessandro
On 02.03.2012, at 17:20, Alberto Musetti wrote:
Hi Rupert, all
I'm working on it, but i'm proceeding slowly because i cannot reproduce the
error.
i'm sorry
Alberto
Il giorno 02/mar/2012, alle ore 00:20, Rupert Westenthaler ha scritto:
Hi Alberto
I spent some time to look into "Build failed in Jenkins: stanbol-trunk-1.6 #772"
The most interesting file is the error.log file of the Stanbol
instance used for the integration test.
https://builds.apache.org/job/stanbol-trunk-1.6/ws/trunk/integration-tests/target/launchdir/sling/logs/error.log
For the first view calls to the enhancer everything looks fine (~5sec
for the Refactor engine).
But later - after time code "01.03.2012 19:49:21.04" (use this to
search in the file) something starts to went wrong with Jena TDB.
[Thread-36]>> seo_refactoring and
[Thread-12]>> looks like a Jena TDB demon
are the only active one (looks a little like a deadlock).
You can safely ignore the [DataFileTrackingDaemon] this checks only
every 5sec for Resources of the DataFIleProvider.
I have really no idea what is going on here. Maybe you can make more
sense of this
best
Rupert
On Thu, Mar 1, 2012 at 3:30 PM, Alberto Musetti<[email protected]> wrote:
Hi Rupert, all
Il giorno 29/feb/2012, alle ore 19:14, Rupert Westenthaler ha scritto:
Hi Alberto
On 29.02.2012, at 18:48, Alberto Musetti wrote:
Hi all,
I would like to add the refactor engine in enhancer/bundlelist.
It will be activated in the launcher full and full-war,
but not in stable because there isn't rules and ontonet.
May I add the refactor engine?
I think the full launcher should include all engines that are managed by the
Stanbol Community. So in principle a +1 from my side.
Can I ask two questions:
(1) What would that mean for the memory footprint. I usually run Stanbol with
-Xmx512m. Would that still work with the refactor engine?
Yes, it would still work.
Concerning performance and memory usage we are adding full support to Clerezza
in the Refactor with spending effort and memory in graph tansformations.
(2) Would the engine be active by default? Do you plan to ship the full
launcher with a default configuration or would a user need to manually
configure an instance?
There is a default configuration running at bundle start-up,
i.e., the refactoring for the SEO demo.
Best,
Alberto
best
Rupert
--
| Rupert Westenthaler [email protected]
| Bodenlehenstraße 11 ++43-699-11108907
| A-5500 Bischofshofen
--
M.Sc. Alessandro Adamou
Alma Mater Studiorum - Università di Bologna
Department of Computer Science
Mura Anteo Zamboni 7, 40127 Bologna - Italy
Semantic Technology Laboratory (STLab)
Institute for Cognitive Science and Technology (ISTC)
National Research Council (CNR)
Via Nomentana 56, 00161 Rome - Italy
"As for the charges against me, I am unconcerned. I am beyond their timid, lying
morality, and so I am beyond caring."
(Col. Walter E. Kurtz)
Not sent from my iSnobTechDevice
