8
Copyright 2007 LarKC Early Adopters Hands-on Florian Fischer STI Innsbruck

LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

  • Upload
    larkc

  • View
    601

  • Download
    0

Embed Size (px)

DESCRIPTION

The aim of the EU FP 7 Large-Scale Integrating Project LarKC is to develop the Large Knowledge Collider (LarKC, for short, pronounced “lark”), a platform for massive distributed incomplete reasoning that will remove the scalability barriers of currently existing reasoning systems for the Semantic Web. The LarKC platform is available at larkc.sourceforge.net. This is the first of two hand-ons that introduce participants to working with directly LarKC code.

Citation preview

Page 1: LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

Copyright 2007

LarKC Early Adopters

Hands-on

Florian FischerSTI Innsbruck

Page 2: LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

Introduction

• Purpose:– Show how a workflow of plug-ins fits together and

interacts– Show wrapping an existing reasoner (IRIS) (later)– Demonstrate anytime behaviour

• Rule-based reasoning in LarKC– Setup an eclipse project with prototype plug-ins– Execute some queries– Change the statements to be reasoned with– Modify reasoning

2

18/05/2009

Page 3: LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

Getting Started

• Required software– Java JDK 1.6– Eclipse

• Import LarKC eclipse projects– File menu -> Import– Choose general -> existing projects in to workspace

• Platform• ISWCDemo

3

18/05/2009

Page 4: LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

Execute a query

• Step 1: In the project ISWCDemo' run the class:– Run ISWCDemo with the first settings– set max heap size: -Xmx1024m

• If you have an internet connection, then you should see several iterations of pipeline output

• Plug-ins used:– SPARQL to triple pattern query transformer– Sindice identifier– Growing data selecter– IRIS reasoner plug-in (with no rules)

4

18/05/2009

Page 5: LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

Different Data

• Step 2: Switch to some local data– Based on http://dev.isb-sib.ch/projects/uniprot-rdf/

• Use the SimpleFileReaderIdentifier to read in local Uniprot data-sets

• Use the IRIS reasoner plug-in, but without any rules yet– More on the inner workings of the IRIS plug-in later

• Execute a query to get all triples– Two iterations of workflow– ~9800 triples

5

18/05/2009

Page 6: LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

Change Inference Rules

• Step 3: Switch on inference– RDF, L2 (custom ruleset) or RDFS– Output: More triples (~28000 for RDFS)

• Step 4: Look for sub-classes of 'pathway 402'– No inference: ~50 sub-classes

• Step 5: Look for sub-classes of 'pathway 402'– With inference (RDFS): 238 sub-classes

6

18/05/2009

Page 7: LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

Create Inference Rule

• Step 6: Sub-classes of self– Output (any entailment): Nothing!– Why?– Look for ':Class' in core.owl

• Step 7: Fix– Copy L2_entailment.rules to L2_plus.rules– Add the following statement:

triple(_iri("owl:Class"),_iri("rdfs:subClassOf"),

_iri("rdfs:Class")) :- .

– Rerun!

7

18/05/2009

Page 8: LarKC Tutorial at ISWC 2009 - First Hands-on Scenario

Summary

• You have:– Seen a handful of LarKC plug-ins– Used them together in a simple workflow– Swapped plug-ins– Experimented with a plug-in

8

18/05/2009