ontoviews – a tool for creating semantic web portals

Semantic Computing Research Group

1

http://www.cs.helsinki.fi/group/seco/

UNIVERSITY OF HELSINKI HELSINKI UNIVERSITY OF TECHNOLOGY

OntoViews – A Tool for Creating Semantic Web Portals

Eetu Mäkelä, Eero Hyvönen, Samppa Saarela and Kim Viljanen ([email protected])

Semantic Computing Research GroupHelsinki Institute for Information Technology (HIIT) & University of Helsinki, Dept. of Computer Science


2nd prize


2



The “Semantic Web” research project group (2002-2004)

• prof. Eero Hyvönen

• Miikka Junnila

• Suvi Kettula

• Eetu Mäkelä

• Samppa Saarela

• Mirva Salminen

• Ahti Syreeni

• Arttu Valo

• Kim Viljanen

Helsinki Institute for Information Technology& University of Helsinki

Helsinki, Finland

Hiroshima


3



Research Consortium

National Boardof Antiquities

Espoo CityMuseum

Lahti CityMuseum

ANTIKVARIA GROUPHelsinki University

Museum

Co-operation with:Finnish National Gallery


4



Topics

• The motivation behind OntoViews

• OntoViews from an end-user's perspective• Implementation• Summary


5



The Motivation Behind OntoViews

• Much of the semantic web content will be published using semantic portals. Such portals usually provide the end-user with two basic services:

– A search engine based on the semantics of the content

– Browsing functionality based on the semantic relations in the underlying knowledge base

• In order to realize the promises of the Semantic Web, such a portal should be:

– Usable for an end-user

– Easily integrated and extended with additional functionality

– Usable with a variety of different devices

– Adaptable to a wide variety of semantic data

– Scalable to accommodate large amounts of data

– Able to provide its functionality also to other programs as Semantic Web Services

• We wanted to create a tool to create such portals


6



Topics




7



OntoViews from an End-User's Perspective

• Search– A search engine based on the semantics of the content

– Usable for an end-user


• Browsing– Browsing functionality based on the semantic relations in the underlying

knowledge base

• Layout adaptability– Usable with a variety of different devices


8




• Search– A search engine based on the semantics of the content: Concept-based

Multi-Facet and keyword search

– Usable for an end-user: Based on the tested and true Flamenco interface

– Easily integrated and extended with additional functionality: Seamless integration in the user interface of keyword and other searches


knowledge base: Classification tree view, explicit semantic links in item view



9




• Search– A search engine based on the semantics of the content: Concept-based

Multi-Facet and keyword search

– Usable for an end-user: Based on the tested and true Flamenco interface

– Easily integrated and extended with additional functionality: Seamless integration in the user interface of keyword and other searches


knowledge base: Classification tree view, explicit semantic links in item view



10



OntoViews from an End-User's Perspective: Layout

• Layout– Usable with a variety of different devices: Different user interfaces and

functionality for different devices


11



Topics




12



Data

The Architecture of OntoViews

OntoViews-C

Ontogator Ontodella

The User

Categories

-Category Projection Rules-Semantic Linking Rules-SWI-Prolog

-Layout-User Interaction Logic-Cocoon, Java, XML, XSLT

-Multi-Facet Search-Keyword Search-Java

-Data in RDF Format

– http: text/XHTML

– http: text/RDF

– file: text/RDF

– java: text/RDF

• Comprised of three main components:– Ontodella, a prolog-based logic server

– Ontogator, a java-based multi-facet search engine

– OntoViews-C, the main Apache Cocoon -based interaction and control component


13



Implementation

• The Logic Server Ontodella– Adaptable to a wide variety of semantic data

• The Multi-Facet Search Engine Ontogator– Scalable to accommodate large amounts of data


• OntoViews-C, the Interaction and Control Component– Easily integrated and extended with additional functionality



14



The Logic Server Ontodella

• Projects view-facets from the data to be used in search utilizing defined logic rules, for example:

ontodella_view( % Place of use

'http://www.cs.helsinki.fi/seco/ns/2004/03/places#earth', % Root URI

place_sub_category, % Subcategory Rule

place_of_use_item_mapping, % Item Mapping Rule

[fi:'Käyttöpaikka', en:'Place of Use'] % Label Definition

).

place_sub_category( ParentCategory, SubCategory ) :-

SubCategoryProperty = 'http://www.cs.helsinki.fi/seco/ns/2004/03/places#isContainedBy',

rdf( SubCategory, SubCategoryProperty, ParentCategory ).

place_of_use_item_mapping( ResourceURI, CategoryURI ) :-

Relation = 'http://www.cs.helsinki.fi/seco/ns/2004/03/artifacts#usedIn',

rdf( ResourceURI, Relation, CategoryURI ).


15




• Projection adaptability has been tested with different data sets:– Museum item data and ontologies from the MuseumFinland

project– Yellow-page service provider data and ontologies from the

IWebS project

• A further Java-based development of the projection system has additionally been tested with:– Website data and categorisations from the dmoz.org directory

• Further work: Neither implementation currently supports adding or deleting of individual categories or items from the knowledge base


16




• Semantically links individual items to each other by utilizing defined logic rules, for example:

same_user( Subject, Target, Explanation ) :-

ItemToUserRelatingProperty =

'http://www.cs.helsinki.fi/group/seco/ns/2004/03/artifacts#':user,

rdf( Subject, ItemToUserRelatingProperty, User ), % the users of our current item

not( some_instance_of_unknown( User ) ), % who are not unknown

rdf( Target, ItemToUserRelatingProperty, User ), % other items used by that user

viewable( TargetURI ), % make sure they are viewable

SubjectURI \= TargetURI, % don't link the item with itself

list_labels( [User], RelLabel ), % get a description of the linking resource

Explanation=[commonResources(RelationInstanceURI), label(fi:RelLabel)].


17



Implementation

• The Logic Server Ontodella– Adaptable to a wide variety of semantic data: Facet hierarchy projection and

semantic link generation are based on extendable logic rules. The projection rules have been tested with three different data sets.






18



Implementation








19



The Multi-Facet Search Engine Ontogator

• Is a generic view-based RDF search engine– Defines and implements an RDF-based query interface

defined as an OWL ontology– Replies to queries in RDF/XML that has a fixed structure.– The query operations are based on category and item

selectors. The functionality of the engine can be extended by implementing new selector types (keyword search and geolocation search in OntoViews, for example)

• Has been tested with dmoz.org data to scale to up to 2.3 million items and 275.000 categories with search times of about 5 seconds.

• Future work: Does not yet scale well to accommodate multiple simultaneous users


20



Implementation



• The Multi-Facet Search Engine Ontogator– Scalable to accommodate large amounts of data: Ontogator has been tested

with dmoz.org data to scale to up to 2.3 million items and 275.000 categories with search times of about 5 seconds.

– Easily integrated and extended with additional functionality: The architecture allows for extensions by implementing new types of selectors (keyword, geolocation). The RDF query model and fixed structure RDF/XML result allow for easy integration.




21



Implementation









22



The Interaction and Control Component OntoViews-C

• Built on top of the Apache Cocoon architecture– The Cocoon architecture is based upon the concept of

pipelines, comprised of modular components (generators, transformers and serializers) that consume and/or produce XML

– This forces a modular, reusable and extendable design

• In OntoViews, all components produce not only XML, but valid RDF/XML. This, along with a generator for handling HTTP requests, allows for the exposition of all parts of the system as Web Services


23



Implementation






• OntoViews-C, the Interaction and Control Component– Easily integrated and extended with additional functionality: Cocoon

architecture forces a modular, reusable and extendable design. All components operate independently, consuming and/or producing RDF/XML

– Able to provide its functionality also to other programs as Semantic Web Services: All subparts of the system are available to be used via Web Services


24



Summary – OntoViews is:

• A search engine based on the semantics of the content: Concept-based Multi-Facet and keyword search

• Browsing functionality based on the semantic relations in the underlying knowledge base: Classification tree view, explicit semantic links in item view

• Usable for an end-user: Based on the tested and true Flamenco interface

• Easily integrated and extended with additional functionality: Seamless integration in the user interface of keyword and other searches, the search architecture allows for extensions, and the Cocoon control architecture forces a modular, reusable and extendable design. All components operate independently, consuming and/or producing RDF/XML.

• Usable with a variety of different devices: Different user interfaces and functionality for different devices

• Adaptable to a wide variety of semantic data: Facet hierarchy projection and semantic link generation are based on extendable logic rules. The projection rules have been tested with three different data sets.

• Scalable to accommodate large amounts of data: OntoViews has been tested with dmoz.org data to scale to up to 2.3 million items and 275.000 categories with search times of about 5 seconds.

• Able to provide its functionality also to other programs as Semantic Web Services: All subparts of the system are available to be used via Web Services

• Future work: Ideas about dynamic facet creation, further optimizations to improve scalability

• Available for download at http://www.cs.helsinki.fi/group/seco/museums/dist/

ontoviews – a tool for creating semantic web portals

Documents