a framework for transforming archaeological databases to ontological datasets

23
CAA 2010 A Framework for transforming archaeological databases to ontological datasets Monika Solanki [email protected] Department of Computer Science Joint work with Yi Hong Department of Computer Science Lin Foxhall, Alessandro Quercia School of Archaeology and Ancient History University of Leicester, UK April 8, 2010 Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Upload: monika-solanki

Post on 11-May-2015

300 views

Category:

Technology


2 download

TRANSCRIPT

Page 1: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

A Framework for transforming archaeologicaldatabases to ontological datasets

Monika [email protected]

Department of Computer Science

Joint work withYi Hong

Department of Computer Science

Lin Foxhall, Alessandro QuerciaSchool of Archaeology and Ancient History

University of Leicester, UK

April 8, 2010Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 2: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

From RDBMs to Ontological datasets

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 3: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Tracing Networks

Archaeologists study a wide range of material objects.

By tracking them at every stage of their production,distribution, use, and consumption across a largegeographical region, over a long time period, they cantrace the links between the people who made, used, andtaught others to make them.

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 4: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Tracing Networks

Archaeologists study a wide range of material objects.By tracking them at every stage of their production,distribution, use, and consumption across a largegeographical region, over a long time period, they cantrace the links between the people who made, used, andtaught others to make them.

Pertinent QuestionsHow have individuals or groups of individuals learnt how toorganise themselves?

Why did some prosper while others collapsed?

What are the dynamics of power, influence and theexchange of knowledge?

In what kinds of contexts does innovation appear?

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 5: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Tracing Networks

www.tracingnetworks.ac.uk

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 6: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Tracing Networks: The Semantic web perspective

Build the links through their datasets

www.tracingnetworks.ac.ukMonika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 7: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Tracing Networks: Loomweights

Example dataset for this talk: Loomweight dataset

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 8: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Motivation

Conventional mapping frameworks

provide scripting languages to facilitate the mapping.

apply simplistic mapping rules.

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 9: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Motivation

Realistic scenarios:

the association between columns and properties is farmore complex than a simple one-to-one correspondence.

domain specific schemas to be used for mapping havebeen extended from standard vocabularies or those usedelsewhere.

Loomweights: the ontological instances conform to adomain specific schema, e.g., CIDOC-CRM.

several ontology schemas are used and the data needs tobe suitably mapped to more than one property.

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 10: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

The Loomweights Dataset

The column diameter in theRDB table cannot be mappedas a datatype property.

To specify a relationshipbetween diameter and theconcept Loomweight, createintermediate instances ofCIDOC-CRM concepts.

Instances to be contextuallyrelated to each other toensure loomweights areassigned correct diametervalues.

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 11: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Transformation Framework

ORM Reverse Engineering.

ECA Rule-based Transformation.

Ontology Instance Generation.

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 12: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

ORM Reverse Engineering

Is used to extract the existing tables, columns, relationships(inc. primary key, foreign key, join etc) and index from aRDBMS to object-oriented data structures or “classes”.

Records in the table can be instantiated as data objectswhich can be easily manipulated and processed usingOOP techniques.

In our proposed approach, the Hibernate ORM ReverseEngineering tool is used to convert database records intoJava objects.

ORM Reverse Engineering → ECA Rule-based Transformation → Ontology Instance Generation

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 13: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

ECA Rule-based Transformation

An ECA-based (Event-Condition-Action) textualtransformation language DOTL − Database OntologyTransformation Language.

The fundamental construct of a DOTL transformation ruleis of form:

On Event if Condition Do Action

ORM Reverse Engineering → ECA Rule-based Transformation → Ontology Instance Generation

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 14: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

ECA Rule-based Transformation: DOTL

A basic DOTL rule consists of three parts:

The event part specifies the triggers of the transformationrule,

The condition part is a logical expression, which checksthe pre-condition of the action to be carried out. Thedefault conditions is “if undefined”.

The action part usually consists of a series of creation ofnew ontology instances, properties and othercorresponding operations.

ORM Reverse Engineering → ECA Rule-based Transformation → Ontology Instance Generation

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 15: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

DOTL rules: Loomweights Database

ORM Reverse Engineering → ECA Rule-based Transformation → Ontology Instance Generation

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 16: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Ontology Instance Generation

The Transfomer builder component is used to build Javasource code based on the pre-defined DOTL rules.

Finally the Transformer Engine component compiles andexecutes the Java code to generate RDF/OWL instances.

The framework exports all data to the RDF store or aspersistent RDF files.

ORM Reverse Engineering → ECA Rule-based Transformation → Ontology Instance Generation

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 17: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Prototype

Java implementation.

The open source Hibernate Reverse Engineeringframework for object/relational mapping.

DOTL Editor plugin for Eclipse: contains an integrated Javacode generator implemented in Xpand.

Formalised EBNF grammar of DOTL defined in Xtext.

Metamodel of the language is described using the EMF(Eclipse Modeling Framework).

Protégé-OWL API for the generation of RDF/OWLinstances.

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 18: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Implementation layers

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 19: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Closely related work

D2RQ, Virtuoso: a declarative language to describemappings between relational database schemata andOWL/RDFS ontologies.

R2O: XML-based language for the transformation.

STAR

TRANSLATION

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 20: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Summary

A transformation framework for migrating large volumes ofarchaeological data stored in RDBs to ontology based datasets on the Semantic Web.

The ECA-based scripting language DOTL, which allowsthe specification of complex transformation rules from dataobjects to ontologies.

A motivating example of the loomweights datasets basedon the CIDOC-CRM ontology schema as a case study.

A prototype implementation that illustrates ourmethodology.

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 21: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Future work

Refining the grammar and semantics to enhance theexpressiveness of DOTL to improve the usability of thesystem.

Implement a user-friendly graphical modeling environmentfor the language in GMF (Graphical Modeling Framework)to allow easy creation and editing of transformation rules.

Expose the datasets as LOD.

Semantic search and reasoning techniques on the LODsas methodologies to “trace” the links between the artifacts.

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 22: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

The tracing network LOD cloud

Monika Solanki A Framework for transforming archaeological databases to ontological datasets

Page 23: A Framework for transforming archaeological databases to ontological datasets

CAA 2010

Thanks!!!

paper at:http://www.tracingnetworks.ac.uk/publications/CAA2010/paper.pdfslides at:http://www.tracingnetworks.ac.uk/publications/CAA2010/slides.pdf

Monika Solanki A Framework for transforming archaeological databases to ontological datasets