SWAT (Semantic Web Authoring Tool)

Lead Research Organisation: University of Manchester
Department Name: Computer Science

Abstract

During the last decade the Semantic Web community has established basic standards for representing data and the conceptual systems (ontologies) through which they are defined. However, encoding information in these formalisms (OWL, RDF) remains a technically difficult task. Widespread adoption of these technologies (with their important potential benefits) would be facilitated if transparent interfaces to the technical formalisms were available.The project aims to show that metadata in OWL and RDF can be viewed and authored through computer-generated presentations in natural languages (e.g., English). The crucial step theoretically will be to develop a model for systemmatically mapping logical concepts and relations to phrase patterns in natural language. The practical challenge will be to develop a tool through which ontology developers can specify this mapping, without deploying deep knowledge of ontologies or grammars. This tool will draw on existing wide-coverage linguistic resources, so that developers can select from a range of pre-coded patterns rather than having to define new ones. If successful, the project would provide an innovative solution to an urgent and commercially relevant problem (as shown by the letters from our collaborators). The main partners are leading UK experts in the theory and practical application of ontologies (Manchester University), and the design of easily-used tools for knowledge-editing based on generated text (Open University).
 
Description We found that natural language definitions for a class or concept in an ontology could be generated using only the logical description of that class. We took the approach of a natural language generator that was general to any ontology, rather than one that was crafted to a particular domain of interest or style of logical expression. In doing so, we made some compromises in quality of natural language, in favour of this generic approach. The "paragraphs" of generated natural language are unordered and, while each sentence is of reasonable quality, the whole paragraph is not necessarily so readable. To address this issue, we used Rhetorical structure theory to map types of logical expression to rhetorical roles and thus were able to impose an ordering on the sentences in the generated paragraphs that made them much more acceptable. We applied the software developed in this project to ontologies from both biology and health and in English and mandarin.
Exploitation Route Now we can automatically generate coherent natural language from any logic based OWL ontology, we can also create natural language definitions for ontology classes and these can be used by both ontology developers and ontology users for easier access, comprehension and checking of the ontology's content.
Sectors Digital/Communication/Information Technologies (including Software),Pharmaceuticals and Medical Biotechnology,Other
 
Description Impact Acceleration Account Concept and Feasibility Study Scheme AVanT: Automotive Vocabularies and Terminologies 'Project'
Amount £30,000 (GBP)
Funding ID IAA 033 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 02/2014 
End 08/2014
 
Description Impact Acceleration Account Relationship Incubator Scheme: SnapOn Semantics project
Amount £4,356 (GBP)
Funding ID IAA 017 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 04/2013 
End 09/2013
 
Description Knowledge Transfer Partnership: Telematicus and SnapOn
Amount £170,977 (GBP)
Funding ID Partnership ID: 1013855 
Organisation Innovate UK 
Sector Public
Country United Kingdom
Start 08/2015 
End 08/2017
 
Description What If
Amount £311,211 (GBP)
Funding ID EP/J014176/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 09/2012 
End 06/2016
 
Description SWAT (Semantic Web Authoring Tool) 
Organisation EMBL European Bioinformatics Institute (EMBL - EBI)
Country United Kingdom 
Sector Academic/University 
PI Contribution Collaboration with James Malone at European BioInformatics Institute, Cambridge, UK on two papers
Start Year 2010
 
Description SWAT (Semantic Web Authoring Tool) 
Organisation University of Aberdeen
Country United Kingdom 
Sector Academic/University 
PI Contribution Richard Power invited to join the advisory board of the EPSRC-funded WhatIf project (University of Aberdeen and Manchester EP/J014176/1).
Start Year 2012