A Controlled Natural Language Interface for Semantic Media Wiki Using the Rabbit Language
Workshop on Controlled Natural Language CNL09 In Press (2009)
Available from eprints.ecs.soton.ac.uk
or
Abstract
Despite their potential value as collaborative knowledge editing systems, semantic wikis present a number of usability challenges. One way to respond to these challenges is to develop controlled natural language (CNL) interfaces for semantic wiki systems. In this paper, we present our efforts to develop a CNL interface for semantic wikis using the Rabbit language. We firstly present an architectural model to support the collaborative editing of community knowledge using a specific semantic wiki system, namely Semantic Media Wiki (SMW), and a specific CNL, namely Rabbit. We then describe a set of meta-model extensions to SMW that enable it to support the representation of highly expressive CNL knowledge models. Finally, we present an online prototype system based on the architectural model that enables users to create knowledge content using the Rabbit language. Future development efforts will focus on the provision of improved parsing and editing capabilities. In addition, we aim to extend the prototype system to accommodate multiple CNLs, such as Sydney OWL Syntax (SOS) and Attempto Controlled English (ACE). The development of such generic (multi-CNL) interfaces for semantic wiki systems will, we suggest, support large-scale collaborative knowledge modelling using a flexible CNL editing and presentation mechanism.
Available from eprints.ecs.soton.ac.uk
Page 1
A Controlled Natural Language Interface for Semantic Media Wiki Using the Rabbit Language
Jie Bao, Paul Smart, Dave Braines, Nigel
Shadbolt
Shadbolt
Page 2
Page 3
Advent of Web 2.0 supports
greater user participation in
the creation of Web content
Good way to generate lots
of online content
◦ e.g. Wikipedia
◦ ~3 million (English) articles
Can we enable better
exploitation of user-
generated content?
◦ retrieval, filtering, reasoning
greater user participation in
the creation of Web content
Good way to generate lots
of online content
◦ e.g. Wikipedia
◦ ~3 million (English) articles
Can we enable better
exploitation of user-
generated content?
◦ retrieval, filtering, reasoning
Page 4
The conventional web is intended for human
consumption
◦ content consists largely of natural language text,
images, video, etc.
Semantic Web seeks to make data more
amenable to automated forms of information
processing
◦ standard data model + explicit semantics
Resource Description Framework (RDF)
◦ core data model + some semantics
Web Ontology Language (OWL)
◦ more advanced semantics
◦ OWL typically used to create ontologies that describe the
conceptual structure of a specific domain of interest
consumption
◦ content consists largely of natural language text,
images, video, etc.
Semantic Web seeks to make data more
amenable to automated forms of information
processing
◦ standard data model + explicit semantics
Resource Description Framework (RDF)
◦ core data model + some semantics
Web Ontology Language (OWL)
◦ more advanced semantics
◦ OWL typically used to create ontologies that describe the
conceptual structure of a specific domain of interest
Page 5
Combination of standard data models and
explicit semantics supports:
◦ information exchange and interoperability
◦ data integration
◦ improved search and retrieval
◦ reasoning and inference
explicit semantics supports:
◦ information exchange and interoperability
◦ data integration
◦ improved search and retrieval
◦ reasoning and inference
Page 6
Shared
Repository
Multinational Planning
Teams
Military Platforms
Unmanned Vehicles
Analysts
Intelligent
Agents/Assistants
Patrols/Field Reports
Non-Military
Organizations
Remote
Sensors
Repository
Multinational Planning
Teams
Military Platforms
Unmanned Vehicles
Analysts
Intelligent
Agents/Assistants
Patrols/Field Reports
Non-Military
Organizations
Remote
Sensors
Page 7
But…
◦ limited amounts of high-quality, semantically-enriched
data available
◦ grounding in formal logic presents a usability barrier to
many individuals and organizations
◦ establishing consensus during ontology development is
often difficult – extensive collaboration is required
So…
◦ can we learn from Web 2.0
greater user participation
delivers lots of content
easy to use
emphasis on collaborative or, at least, collective efforts
◦ limited amounts of high-quality, semantically-enriched
data available
◦ grounding in formal logic presents a usability barrier to
many individuals and organizations
◦ establishing consensus during ontology development is
often difficult – extensive collaboration is required
So…
◦ can we learn from Web 2.0
greater user participation
delivers lots of content
easy to use
emphasis on collaborative or, at least, collective efforts
Page 8
Support multi-user
content creation
and editing via a
Web browser
interface
Encourages large-
scale participation
Easy to use
Content usually of
reasonable quality
Problems:
◦ natural language text
◦ difficult for machines
to participate in
content generation
content creation
and editing via a
Web browser
interface
Encourages large-
scale participation
Easy to use
Content usually of
reasonable quality
Problems:
◦ natural language text
◦ difficult for machines
to participate in
content generation
Page 9
Uses Wikipedia engine
Perhaps the most
popular semantic wiki
system
Supports the creation
of semantically-
enriched content
◦ uses semantic
annotations
Combines features of
conventional wiki
system with semantic
technologies
Perhaps the most
popular semantic wiki
system
Supports the creation
of semantically-
enriched content
◦ uses semantic
annotations
Combines features of
conventional wiki
system with semantic
technologies
Page 10
◦
Good basis for developing an online, collaborative
knowledge editing system whose content is both
structured and semantically-rich.
◦
Good basis for developing an online, collaborative
knowledge editing system whose content is both
structured and semantically-rich.
Page 11
Usability
◦ semantic content (esp. ontologies) difficult to create
◦ departure from Web 2.0 emphasis on ease-of-use
◦ even experienced knowledge engineers can find it
difficult to create/edit ontologies
Automatic content integration
◦ sometimes content needs to be automatically imported
without user intervention
◦ recall the case of sensor feeds
Expressivity constraints
◦ semantic wikis (including SMW) do not always support
the full range of OWL modelling formalisms and axioms
Inference constraints
◦ limited support for rule representation and inference
◦ semantic content (esp. ontologies) difficult to create
◦ departure from Web 2.0 emphasis on ease-of-use
◦ even experienced knowledge engineers can find it
difficult to create/edit ontologies
Automatic content integration
◦ sometimes content needs to be automatically imported
without user intervention
◦ recall the case of sensor feeds
Expressivity constraints
◦ semantic wikis (including SMW) do not always support
the full range of OWL modelling formalisms and axioms
Inference constraints
◦ limited support for rule representation and inference
Page 12
Usability
◦ use CNLs
◦ potential production and comprehension benefits
◦ multiple OWL-compliant CNLs are available:
e.g. Rabbit, Sydney OWL Syntax, ACE-OWL
Automatic content integration
◦ develop an RDF import mechanism for SMW
◦ support the automatic creation of wiki pages and page
content from external RDF/OWL models
Expressivity constraints
◦ extend SMW with an OWL meta-model
Inference constraints
◦ implement rule representation and inference capabilities
for SMW (reported elsewhere)
◦ use CNLs
◦ potential production and comprehension benefits
◦ multiple OWL-compliant CNLs are available:
e.g. Rabbit, Sydney OWL Syntax, ACE-OWL
Automatic content integration
◦ develop an RDF import mechanism for SMW
◦ support the automatic creation of wiki pages and page
content from external RDF/OWL models
Expressivity constraints
◦ extend SMW with an OWL meta-model
Inference constraints
◦ implement rule representation and inference capabilities
for SMW (reported elsewhere)
Page 13
Extend expressivity of SMW to provide full
support for OWL
Support the creation of ontologies and
ontology content within SMW
Explore ways to serialize SMW contents as
(multiple) CNLs
Investigate mechanisms to support wiki
content creation using (multiple) CNLs
Develop CNL editors to support content
creation
support for OWL
Support the creation of ontologies and
ontology content within SMW
Explore ways to serialize SMW contents as
(multiple) CNLs
Investigate mechanisms to support wiki
content creation using (multiple) CNLs
Develop CNL editors to support content
creation
Page 14
Wiki Database
Form Editor
Interface
CNL Interface
CNL Editing
Interface
RDF Export
Interface
RDF Model
RDF Import Semantic Query
Interface
RDF Export
SELECT ?x
WHERE
{
?x rdf:type owl:Class
}
Form Editor
Interface
CNL Interface
CNL Editing
Interface
RDF Export
Interface
RDF Model
RDF Import Semantic Query
Interface
RDF Export
SELECT ?x
WHERE
{
?x rdf:type owl:Class
}
Page 15
Required for CNLs, RDF import, ontology
development
Use wiki templates to create OWL meta-model
extensions for SMW
Each wiki template is created using the wiki
scripting language
OWL elements (e.g. classes, subClassOf axioms)
are represented using individual wiki templates
Instances of the templates encode information
about the classes, properties and individuals in a
specific ontology
development
Use wiki templates to create OWL meta-model
extensions for SMW
Each wiki template is created using the wiki
scripting language
OWL elements (e.g. classes, subClassOf axioms)
are represented using individual wiki templates
Instances of the templates encode information
about the classes, properties and individuals in a
specific ontology
Page 16
Page 17
Page 18
Each wiki template is associated with UI
components that support the editing of data
associated with instances of the template
Multiple templates can be associated with a
wiki page to create an editing interface for
ontology elements (i.e. classes, properties
and individuals)
components that support the editing of data
associated with instances of the template
Multiple templates can be associated with a
wiki page to create an editing interface for
ontology elements (i.e. classes, properties
and individuals)
Page 19
Wiki templates are also used to generate CNL
Each wiki CNL generation template contains
embedded semantic queries to retrieve
information from the wiki database
The retrieved information is then structured
according to the syntax of the target CNL
(e.g. Rabbit) – again using wiki script
Accommodating new CNLs (e.g. ACE) requires
relatively minor changes to the wiki script
◦ future work: enable users to create/modify their
own CNL generation templates
Each wiki CNL generation template contains
embedded semantic queries to retrieve
information from the wiki database
The retrieved information is then structured
according to the syntax of the target CNL
(e.g. Rabbit) – again using wiki script
Accommodating new CNLs (e.g. ACE) requires
relatively minor changes to the wiki script
◦ future work: enable users to create/modify their
own CNL generation templates
Page 20
{{#vardefine:label|{{CNL.getLabel|{{{1}}} }} }}
{{#vardefine:super |
{{#ask: [[:{{{1|{{FULLPAGENAME}}}}}]]
|?Category= |mainlabel=-|format=list|link=none
}} }}
{{#if: {{#var:super}}
|{{#arraymap:{{#var:super}}|,|xxx|<li>Every
[[:{{{1}}}{{!}}{{#var:label}}]] is a kind of
[[:xxx|{{CNL.getLabel|xxx}}]] }}|
}}
http://tw.rpi.edu/proj/cnl/Template:CNL.Rabbit.getConceptRelationAssertions
{{#vardefine:super |
{{#ask: [[:{{{1|{{FULLPAGENAME}}}}}]]
|?Category= |mainlabel=-|format=list|link=none
}} }}
{{#if: {{#var:super}}
|{{#arraymap:{{#var:super}}|,|xxx|<li>Every
[[:{{{1}}}{{!}}{{#var:label}}]] is a kind of
[[:xxx|{{CNL.getLabel|xxx}}]] }}|
}}
http://tw.rpi.edu/proj/cnl/Template:CNL.Rabbit.getConceptRelationAssertions
Page 21
Page 22
Interface to support the creation and editing of wiki
content using CNLs
◦ light-weight integration with SMW environment
◦ language agnosticism - support for multiple CNLs
Rabbit, ACE-OWL, etc.
requires flexible representation of grammar rules
◦ constrain user input to grammatically-correct sentences
intellisense / autocompletion capabilities
◦ display, sort, filter, search asserted CNL sentences
◦ view ‘related’ sentences in other ontologies
◦ view inferred CNL sentences
display reason why sentence has been inferred
◦ provide logical consistency checking, redundancy checking
and error diagnosis
◦ speech input?
◦ enable users to create/modify CNL input grammars?
content using CNLs
◦ light-weight integration with SMW environment
◦ language agnosticism - support for multiple CNLs
Rabbit, ACE-OWL, etc.
requires flexible representation of grammar rules
◦ constrain user input to grammatically-correct sentences
intellisense / autocompletion capabilities
◦ display, sort, filter, search asserted CNL sentences
◦ view ‘related’ sentences in other ontologies
◦ view inferred CNL sentences
display reason why sentence has been inferred
◦ provide logical consistency checking, redundancy checking
and error diagnosis
◦ speech input?
◦ enable users to create/modify CNL input grammars?
Page 23
Developed by Tobias
Kuhn at the
University of Zurich
Wiki system based
on a subset of ACE
Includes predictive
editor that
constrains user
input to ACE-
compliant sentences
Differences:
◦ underlying wiki system
◦ editing interface
◦ light-weight extensions
◦ support for multiple
CNLs
◦ customization of target
CNLs
Kuhn at the
University of Zurich
Wiki system based
on a subset of ACE
Includes predictive
editor that
constrains user
input to ACE-
compliant sentences
Differences:
◦ underlying wiki system
◦ editing interface
◦ light-weight extensions
◦ support for multiple
CNLs
◦ customization of target
CNLs
Page 24
Coalition Planning
◦ ontology-mediated collaborative planning
◦ users
brigade staff
Human Terrain Analysis
◦ cultural profiling
◦ cultural analysis and training
◦ users
cultural anthropologists, psychologists, IO/PSYOP teams,
indigenous individuals/organizations
Intelligence Gathering/Analysis
◦ social network analysis
◦ activity monitoring
◦ users
intelligence analysts, platoon leaders
◦ ontology-mediated collaborative planning
◦ users
brigade staff
Human Terrain Analysis
◦ cultural profiling
◦ cultural analysis and training
◦ users
cultural anthropologists, psychologists, IO/PSYOP teams,
indigenous individuals/organizations
Intelligence Gathering/Analysis
◦ social network analysis
◦ activity monitoring
◦ users
intelligence analysts, platoon leaders
Page 25
We have developed an OWL meta-model extension to SMW to
support the representation of OWL ontologies
We have provided a light-weight form-based interface to
support ontology editing
We have provided an RDF import mechanism to support the
import of existing ontologies
We have developed multiple CNL ‘verbalizers’ to support the
serialization of semantic wiki content to CNLs
Future work:
◦ enable users to create/customize CNL output
◦ implement wiki-based CNL editing capability
http://tw.rpi.edu/proj/cnl/Main_Page
baojie@cs.rpi.edu
support the representation of OWL ontologies
We have provided a light-weight form-based interface to
support ontology editing
We have provided an RDF import mechanism to support the
import of existing ontologies
We have developed multiple CNL ‘verbalizers’ to support the
serialization of semantic wiki content to CNLs
Future work:
◦ enable users to create/customize CNL output
◦ implement wiki-based CNL editing capability
http://tw.rpi.edu/proj/cnl/Main_Page
baojie@cs.rpi.edu
Sign up today - FREE
Mendeley saves you time finding and organizing research. Learn more
- All your research in one place
- Add and import papers easily
- Access it anywhere, anytime
Start using Mendeley in seconds!
Readership Statistics
3 Readers on Mendeley
by Discipline
33% Education
33% Social Sciences
by Academic Status
67% Ph.D. Student
33% Post Doc
by Country
67% United States
33% Cyprus