Transcript
Page 1: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

Chapter Copyright 2009 Digital Enterprise Research Institute. All rights reserved.

Digital Enterprise Research Institute www.deri.ie

Enabling Case-Based Reasoning on the Web of Data

(How to create a Web of Experience)

Benjamin Heitmann, Conor Hayes

Digital Enterprise Research Institute (DERI),National University of Ireland, Galway

Funded by Science Foundation Ireland under Grant No. SFI/08/CE/I1380 (Líon-2)

Page 2: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

Digital Enterprise Research Institute www.deri.ie

[email protected]

slide of 9

Motivation

characterisation of current CBR approaches: data storage is domain and use-case specific

no common data model

challenges: limited interoperability (“data silos”)

no reuse of cases or knowledge containers

data acquisition is expensive

the Web of Data can provide: 1. new sources of experiential data

2. standard way to publish and link experiential data

3. common data model for CBR interoperability

4. opportunity to establish CBR as a standard reasoning paradigm

2

Page 3: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

Digital Enterprise Research Institute www.deri.ie

[email protected]

slide of 9

Overview:

1. related work in the CBR domain: existing approaches for CBR interoperability

2. introduction to the Web of Data: main concepts and principles

current sources for experiential data

3. applying the CBR methodology

to the Web of Data: lightweight CBR vocabulary

example and process for constructing a case base

3

Page 4: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

Digital Enterprise Research Institute www.deri.ie

[email protected]

slide of 9

Related work: CBR interoperability

Case-Based Mark-Up Language (CBML), XML based: rigid CBR vocabulary, hard to customise for new domain.

hard to convert domain data, lack of real data.

CaseML (RDF based): rigid CBR vocabulary

requires a-priori knowledge of external sources

C-OWL (RDF based, extends OWL): formalisation of distributed reasoning for CBR using rules

common shortcomings: no reuse of domain semantics for cases no reuse by linking of case fragments

high overhead of transforming of external data into case data

4

Page 5: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

Digital Enterprise Research Institute www.deri.ie

[email protected]

slide of 9

Background: The Web of Data

the Web of Data provides: structured data, collaboratively

created, about object centred sociality domain knowledge through

ontologies (e.g. DBpedia ontology) cross-domain links between sources

Linked Data principles:

1. use URIs “for everything”

2. allow HTTP access to all URIs

3. when accessing a URI, provide relevant data in RDF

4. include links to URIs from third parties (background knowledge)

Linked Data can be very noisy, so CBR is well suited as a reasoning paradigm

5

(a) July 2007 (b) April 2008 (c) Sep 2009 (d) July 2009

Page 6: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

Digital Enterprise Research Institute www.deri.ie

[email protected]

slide of 9

Sources of experiential data from the Web of Data

DBpedia provides cross-domain links

social web sites: Live Journal

MySpace

Facebook & Open Graph API

Yelp reviews

broadcasters & news: BBC program catalogue

New York Times subject headings

search engines providing access to this data: Google and Yahoo

Sindice

6

example of structured data from Wikipedia, demonstrating the Linked Data principles

foaf:name

http://beck.com

"Beck"

foaf:homepage

dbpedia:Beck

Friend of a Friend (FOAF) vocabulary:

social relationships and information

dbpedia-owl:birthPlace dbpedia:Los_Angeles

DBPedia ontology

dbpprop:genre dbpedia:Anti-folk

DBPedia properties

skos:subject category:Anti-folk_musicians

Simple Knowledge Organisation System (SKOS): vocabulary for knowledge organisation

owl:sameAs fbase:Beck Web Ontology Language (OWL):links to identical resourcesopencyc:en/

Beck_MusicalPerformer

Page 7: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

cbr:CaseBase

cbr:Case

cbr:Solution

cbr:has_casebase

cbr:has_solution

ex:UserProfiles

deri:Heitmann

amazon:RiverOfGods

myspace:Björk

cbr:has_casebase

foaf:interest

foaf:interest

rdf:type

rdf:type

CBR vocabularyExample CBR Case Base

Sources:DBPedia,

Amazon Reviews via Google RDFa,MySpace via DBTune

amazon:GravitysRainbow

myspace:BobDylan

foaf:interest

foaf:interest

cbr:has_casebase

deri:Hayes

Digital Enterprise Research Institute www.deri.ie

[email protected]

slide of 9

CBR vocabulary for the Web of Data

modelling decisions: lightweight approach

intentional simplicity

reuse of existing domain semantics and vocabularies

flexible mapping of cases to entities

not fixed to domain or use case

focus on vocabulary and case knowledge

7

Page 8: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

Digital Enterprise Research Institute www.deri.ie

[email protected]

slide of 9

Process for constructing a case base

Step 1: discovering and aggregating data use search engine or custom crawler to discover data

8

Step 2: conversion of external data transform different RDF serialisations (RDFa, RDF/XML, XHTML) to cases in RDF

Step 3: authoring and curating of case base select relevant cases manually or automatically, via application logic

Page 9: Enabling Case-Based Reasoning  on the Web of Data (How to create a Web of Experience)

Digital Enterprise Research Institute www.deri.ie

[email protected]

slide of 9

Summary: towards a Web of Experience

our simple example illustrates the future potential

towards a Web of Experience: publish experiential data in RDF

link it to the Web of Data

use cases: mining experiences from structured, user generated content.

open recommender systems

distributed CBR

9


Recommended