1. Knowledge Representation and Reasoning in IKS
Speakers:
Valentina Presutti (KReS Overview)
Andrea Nuzzolese (Integrity check demo)
Enrico Daga (KReS Core demo)
Contributors: speakers + Alessandro Adamou + Elvio Bonafede + Eva Blomqvist + Aldo Gangemi

2. Outline KReS overview KReS and CMS developers?
KReS and internal validation?
KReS and WP3 requirements
Relation with the IKS reference architecture
Ongoing and future work (towards beta)
Integrity check on remote data fusion (demo)
KReS main functionalities (demo)

3. Demo by Enrico in few minutes!
KReS: main components
Ontology Network Manager (ONM)
Rules & Inference (R&I)
Semion
Release as
RESTful services
Java API
Demo by Enrico in few minutes!

4. Ontology Network Manager
IntegrityCheckScope
A scope contains (actually is) an ontology network and serves a specific purpose
DBpedia ontology
Ontologies that are not changeable, shared with the other clients
Valid Content ontologies
Ontologies that can be modified, shared with the other clients
Private session of a client e.g. for in-memory operations
DBpedia and GeoNames retrieved datasets

5. Rules and Inference Manager
Support for rules
Expressed in KReS syntax SWRL compliant
Executed as SWRL or SPARQL
Can be executed on a specified scope
Support for reusable rules through definition of recipes i.e. sets of rules
Reasoning support
Through OWLink interfaces
HermiT built-in

6. Semion Reengineering Refactoring
Transformation of non-RDF sources to RDF
Support for XML and relational DB
Refactoring
RDF to RDF based on transformation patterns (recipes)

7. KReS and CMS developers (understand)
Simple online tutorial showing examples of KReS services usage
Inspired by user stories ( by next early adopters workshop)

8. KReS and CMS developers A user story by TXT
Article authoring
Recommendation of related content items for supporting the article authoring process
User story used for demonstrating the whole stack flow
Knowledge patterns used for improving user interaction with recommended content items
Relation with NoTube research challenges
This is initial work
Goals: understanding and show added value

9. KReS and CMS developers A user story by Nuxeo
[from an thread with Olivier (Nuxeo)] We could start with integrity check for remote data fusion: CMS administrator defines that a valid entity of type Person must always have a birth date, a picture and a nationality attribute User edit document in CMS and CMS asks fise in the background to recognize occurrences of entities of type Person coming from DBPedia and freebase FISE provide associated enhancement The triples of the recognized entities are downloaded CMS Administrator integrity rules are checked Entities that do not match the rules are rejected CMS receives the entity descriptions of valid persons and displays a widget with the person attributes next to the document edition form using an ajax polling/callback.
Integrity check for remote data fusion
Prototype Demo by Andrea in a minute!
Goal: understanding and show added value

10. KReS and CMS developers A user story by Nuxeo
Lessons learnt so far
The usage of a reasoner improves recall
Additional valid content retrieved (inferred knowledge vs. asserted knowledge)
sameAs entities are included in the results
Reasoning based on linked data ontologies produces noise, hence
External datasets e.g. LOD, need some cleaning
We will provide ontology patterns for improving reasoning results on DBpedia datasets
So called “washing machines” external services could be integrated for cleaning data e.g. datatype value format
Interesting research challenges in both stories

11. KReS alpha: relation with requirements
HLR-3201
IKS ontology network
HLR-3202
Ontology network reusability, management, and maintenance
HLR-3203
Rule language
HLR-3204
Rule reusability, management, and maintenance
HLR-3205
Compliancy with standards
HLR-3206
Reasoning services
HLR-3207
Transformation, refactoring and enriching of semantic content
HLR-3208
Workflow definition, reasoning, and execution
HLR-3209
Context-dependent selection of content and system behavior
HLR-3210
User-centered explanations (trust and provenance)
HLR-3211
External data usage and publishing
HLR-3212
Adaptivity, configurability, reliability, and customizability of KR and reasoning ser- vices
All high priority REQ addressed (either fully or partially)
Most of high-priority REQ addressed (either fully or partially)
For details please see D5.2 Chapter 5

12. KReS and the IKS reference architecture

13. Validation through Internal Use Cases: AMI
KReS is going to be validated in the context of the AMI use case (T4.1-T6.1)
Currently the ongoing work focuses on the integration of the AMI architecture with the IKS stack
adapt/apply KReS services to the AMI case
Development of ontologies started in June
First versions are already available
Requirements are being refined
A new version of the ontologies will be produced in November
The idea is to use ontologies for detecting situations
Relation with SmartProducts ontologies?

14. Ongoing and future work
A new enhancement engine for FISE
KReS-related tutorials
targeted at CMS developers (first release by December)
A master course program including IKS topics (February 2011 at University of Bologna)
KReS functionality consolidation
Intensive testing
Maintenance and functionality improvements e.g. integration of SILK (discovery of links btw RDF datasets)
Driven by user stories and internal use case
Integrity check for remote data fusion (early prototype available)
Support for article authoring process
Ami Use case

15. Integrity check for remote data fusion (prototype demo)
by Andrea Nuzzolese

16. We show all steps for demonstrating the usage of KReS
Integrity check for remote data fusion (prototype demo by Andrea Nuzzolese)
We show all steps for demonstrating the usage of KReS
The integrity check function only needs the content item and the definition of valid entities (ontologies or rule)
The web interface is meant to serve demo purposes

17. Integrity check for remote data fusion (prototype demo by Andrea Nuzzolese)
A content item (text) is passed to FISE
All fise:entity-reference are selected from the result
The RDF graphs describing all fise:entity-reference are resolved and put in the scope session
The rule for valid content is defined
The inference engine is executed and the RDF datasets are enriched based on the DBPedia ontology
The integrity check is executed

18. KReS main components demo
By Enrico Daga
Notice: The web interface is meant to serve demo purposes

19. KReS core: ONM

20. KReS core: R&I rule management

21. KReS core: R&I inference engine

22. KReS: Semion refactoring

23. KReS: Semion reengineering

24. KReS services