Hi Rupert,
Good, what you are saying confirms what I understood from documentation.
I am thinking about a scenario like this:
(1) Tika -- get document type (mandatory)
(2) for images -- find objects (optional, depends on 1)
(3) for text -- extract topics (optional, depends on 1)
(4) do something with the extracted metadata (mandatory, depends on (2) OR
(3) )
I understand that both (2) and (3) must be optional, as only one of them
will be executed for a single input doc.
I am also thinking, how to best check the "failure", i.e. if the input can
not be processed with neither (2) nor (3). Guess this needs to be checked
in the internal logic of (4) then... Is it a good practice to analyze the
metadata graph in the canEnhance() function?
Thank you again!
Best,
Michal
On Thu, Oct 13, 2016 at 11:22 AM Rupert Westenthaler <
rupert.westentha...@gmail.com> wrote:
> On Thu, Oct 13, 2016 at 11:12 AM, Michal Krajňanský
> <michal.krajnan...@gmail.com> wrote:
> > Hi Rupert,
> >
> > Thank you for the advice, this is very useful. By the name I assumed that
> > chain list is only linear, but now I see that you can specify multiple
> > dependencies of a single node.
> >
> > Anyways, I was able to make GraphChain work, with correcting the syntax
> as
> > A. Soroka suggested -- thank you as well!
> >
> > FYI, my aim is to understand how the branching works and how the
> "optional"
> > attribute behaves, when there are multiple mutually exclusive paths to
> the
> > same "sink" node (which is not optional).
>
> Optional just means that if an engine fails the execution of the chain
> is continued.
>
> Branching is only used to improve performance by executing several
> engines at the same time (if the engines do support it). A good
> example ist Entity Linking against multiple Vocabularies. If you
> include several entity linking engines in your chain they will be
> executed concurrently in case you use a WeightedChain. In case of a
> ListChain they will be executed in the order as configured in the
> list.
>
> WeightedChain "sorts" engine executions based on metadata provided by
> the engine implementations. This allows to (1) correctly sort engine
> execution and (2) also allows for parallel execution of engines of the
> same type. If one want (needs) fine grained control over when an
> engine can run you need to use the GraphChain.
>
> In a graph chain an engine depending on an optional engine will wait
> until the optional engine has completed. It will also be executed if
> this engine fails.
>
> best
> Rupert
>
> >
> > Best Regards,
> > Michal
> >
> > On Thu, Oct 13, 2016 at 10:32 AM Rupert Westenthaler <
> > rupert.westentha...@gmail.com> wrote:
> >
> >> Hi Michael,
> >>
> >> I would suggest you try to define the execution plan by using the
> >> chain list [1]. The best would be to do it via the UI provided by the
> >> Felix Webconsole. By doing so you do not need to deal with RDF syntax
> >> directly.
> >>
> >> Anyways the GraphChain is an advanced feature that allows fine grained
> >> control over chain execution. In most of the cases configuring a
> >> WeightedChain[2] is sufficient.
> >>
> >> best
> >> Rupert
> >>
> >>
> >> [1]
> >>
> https://stanbol.apache.org/docs/trunk/components/enhancer/chains/graphchain#chainlist
> >> [2]
> >>
> https://stanbol.apache.org/docs/trunk/components/enhancer/chains/weightedchain.html
> >>
> >> On Mon, Oct 10, 2016 at 3:45 PM, A. Soroka <aj...@virginia.edu> wrote:
> >> > Ah, I see. That is not at all parseable Turtle RDF. It really is just
> a
> >> list of URIs. You named the file ".ttl", but it is not proper Turtle:
> >> >
> >> > https://en.wikipedia.org/wiki/Turtle_(syntax)#Example
> >> >
> >> > There is a great deal of punctuation missing. I'm not at all sure why
> >> the example in the docs is printed that way. It won't work.
> >> >
> >> > Perhaps someone who knows more about how the documentation is
> generated
> >> can comment? Perhaps that example was meant to acquire some formatting
> >> before publication?
> >> >
> >> > ---
> >> > A. Soroka
> >> > The University of Virginia Library
> >> >
> >> >> On Oct 10, 2016, at 8:32 AM, Michal Krajňanský <
> >> michal.krajnan...@gmail.com> wrote:
> >> >>
> >> >> Dear Mr. Soroka,
> >> >>
> >> >> Sorry for the mistake, trying again -- please find the attached file.
> >> >>
> >> >> As you can find, I took over the example from Stanbol web:
> >> >>
> >>
> https://stanbol.apache.org/docs/trunk/components/enhancer/chains/executionplan.html
> >> >> and copied the prefix declarations from here:
> >> >>
> >>
> https://stanbol.apache.org/docs/trunk/components/enhancer/chains/graphchain.html
> >> >>
> >> >> Thank you for looking into this!
> >> >>
> >> >> Michal Krajnansky
> >> >>
> >> >>
> >> >> On Mon, Oct 10, 2016 at 2:23 PM A. Soroka <aj...@virginia.edu>
> wrote:
> >> >> That RDF quote didn't come through the mailing list (at least not for
> >> me). It became a list of URIs. Perhaps you can try pasting it as very
> plain
> >> text? Or you could put it in Gist or some other pastebin service?
> >> >>
> >> >>
> >> >>
> >> >> ---
> >> >>
> >> >> A. Soroka
> >> >>
> >> >> The University of Virginia Library
> >> >>
> >> >>
> >> >>
> >> >> > On Oct 10, 2016, at 5:22 AM, Michal Krajňanský <
> >> michal.krajnan...@gmail.com> wrote:
> >> >>
> >> >> >
> >> >>
> >> >> > Dear Stanbol Users,
> >> >>
> >> >> >
> >> >>
> >> >> > I have been experimenting with the system and I would appreciate
> your
> >> help
> >> >>
> >> >> > with the following problem.
> >> >>
> >> >> >
> >> >>
> >> >> > I am trying to implement a custom Graph chain like the example
> here:
> >> >>
> >> >> >
> >>
> https://stanbol.apache.org/docs/trunk/components/enhancer/chains/executionplan.html
> >> >>
> >> >> > and here:
> >> >>
> >> >> >
> >>
> https://stanbol.apache.org/docs/trunk/components/enhancer/chains/graphchain.html
> >> >>
> >> >> >
> >> >>
> >> >> > However, when trying to invoke the chain, I get the following
> >> exception:
> >> >>
> >> >> >
> >> >>
> >> >> > *org.apache.jena.riot [line: 6, col: 1 ] Undefined prefix: urn*
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> > My execution plan RDF looks like this:
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> > *@prefix xsd: <http://www.w3.org/2001/XMLSchema#
> >> >>
> >> >> > <http://www.w3.org/2001/XMLSchema#>> .@prefix rdf:
> >> >>
> >> >> > <http://www.w3.org/1999/02/22-rdf-syntax-ns#
> >> >>
> >> >> > <http://www.w3.org/1999/02/22-rdf-syntax-ns#>> .@prefix ep:
> >> >>
> >> >> > <http://stanbol.apache.org/ontology/enhancer/executionplan#
> >> >>
> >> >> > <http://stanbol.apache.org/ontology/enhancer/executionplan#>>
> >> .@prefix ehp:
> >> >>
> >> >> > <http://stanbol.apache.org/ontology/enhancementproperties#
> >> >>
> >> >> > <http://stanbol.apache.org/ontology/enhancementproperties#>>
> >> >>
> >> >> > .urn:execPlan rdf:type ep:ExecutionPlan ep:hasExecutionNode
> >> >>
> >> >> > urn:node1, urn:node2, urn:node3, urn:node4, urn:node5 ep:chain
> >> >>
> >> >> > "demoChain"urn:node1 rdf:type stanbol:ExecutionNode
> >> >>
> >> >> > ep:inExecutionPlan urn:execPlan ep:engine langIdurn:node2
> >> rdf:type
> >> >>
> >> >> > ep:ExecutionNode ep:inExecutionPlan urn:execPlan ep:dependsOn
> >> >>
> >> >> > urn:node1 ep:engine nerurn:node3 rdf:type ep:ExecutionNode
> >> >>
> >> >> > ep:inExecutionPlan urn:execPlan ep:dependsOn urn:node1
> ep:engine
> >> >>
> >> >> > dbpediaLinkingurn:node4 rdf:type ep:ExecutionNode
> >> ep:inExecutionPlan
> >> >>
> >> >> > urn:execPlan ep:dependsOn urn:node1 ep:engine
> >> >>
> >> >> > geonamesLinkingurn:node5 rdf:type ep:ExecutionNode
> >> ep:inExecutionPlan
> >> >>
> >> >> > urn:execPlan ep:engine zemanta ep:optional
> "true"^^xsd:boolean*
> >> >>
> >> >> >
> >> >>
> >> >> > Thank you in advance for your kind help.
> >> >>
> >> >> >
> >> >>
> >> >> > Best Regards,
> >> >>
> >> >> >
> >> >>
> >> >> >
> >> >>
> >> >> > Michal Krajnansky
> >> >>
> >> >>
> >> >>
> >> >> <example-chain.ttl>
> >> >
> >>
> >>
> >>
> >> --
> >> | Rupert Westenthaler rupert.westentha...@gmail.com
> >> | Bodenlehenstraße 11 ++43-699-11108907
> <+43%20699%2011108907>
> >> <+43%20699%2011108907>
> >> | A-5500 Bischofshofen
> >> | REDLINK.CO
> >>
> ..........................................................................
> >> | http://redlink.co/
> >>
>
>
>
> --
> | Rupert Westenthaler rupert.westentha...@gmail.com
> | Bodenlehenstraße 11 ++43-699-11108907
> <+43%20699%2011108907>
> | A-5500 Bischofshofen
> | REDLINK.CO
> ..........................................................................
> | http://redlink.co/
>
