Building Ontologies from Concept Maps
0 likes98 views
This document discusses using data pipelines and semantic web technologies to build ontologies from existing artifacts like concept maps and spreadsheets. It presents Python and RDFLib as useful tools for transforming raw data into RDF graphs by iterating through source data and generating ontology classes, properties, and instances. The goal is to improve productivity by automating the process of building models from common documentation formats using semantic web standards and reproducible data pipelines. A demonstration transforms a simple concept map into an RDF graph to illustrate the approach.
![Transformation Tools - Python and RDFLib
APP = Namespace('http://www.example.com/ontologies/product/')
namespace_manager = NamespaceManager(Graph())
namespace_manager.bind('app', APP, override = False)
namespace_manager.bind('owl', OWL, override = False)
graph = Graph()
graph.namespace_manager = namespace_manager
# Iterating through either rows of a spreadsheet or export of CMAP Tools…
graph.set((APP['Product'], RDF.type, OWL.Class))
graph.set((APP['Product'], RDFS.label, 'Product'))
graph.set((APP['Product'], RDFS.label, 'Product'))](https://image.slidesharecdn.com/buildingontologiesfromconceptmaps-v2-181016055401/85/Building-Ontologies-from-Concept-Maps-12-320.jpg)
Building Ontologies from Concept Maps
- 1. Building Ontologies from Concept Maps and other Artefacts Tools and Patterns Eugene Morozov Twitter: @eugenemorozov LinkedIn: https://www.linkedin.com/in/emorozov Meetup: https://meetup.com/semantic-web-london
- 2. The sky above the port... ● Software Development (mostly Java) ● Technology Consulting ● Looking for ways to improve my productivity as a developer
- 3. Semantic Web and Data Analysis ● Work on dialogue management some 10 years ago and discovering some Semantic Web ideas ● Ideas of reproducible research and data pipelines from (not so) recent Data Science Coursera course ● Finding that both can be used together on projects to improve productivity
- 4. Semantic Web ● A way to share and reuse data ● Also a way to express models
- 5. Data pipelines ● A way to reproduce research analysis
- 6. Data pipeline for software artifacts ● Using data pipelines to transform existing artifacts to working software ● Not a new idea, but available Semantic Web vocabularies and modern Web frameworks make it easier and lighter weight than ever
- 7. Inspiration ● Ben Liu, Hejie Chen, Wei He, Deriving User Interface from Ontologies: A Model-based Approach https://www.researchgate.net/publication/4205885_Deriving_user_interface_fr om_ontologies_A_model-based_approach ● Anila Sahar Butt, Armin Haller, Shepherd Liu, Lexing Xie, ActiveRaUL: Automatically Generated Web Interfaces for Creating RDF Data http://www.semantic-web-journal.net/system/files/swj549.pdf ● Krishna Sapkota, Arantza Aldea, David A Duce, Muhammad Younas, René Bañares-Alcántara, Towards semantic methodologies for automatic regulatory compliance support http://dl.acm.org/citation.cfm?id=2065021
- 8. Inspiration ● Thomas Friesendal, Design Thinking Business Analysis: Business Concept Mapping Applied https://www.amazon.co.uk/Design-Thinking-Business-Analysis-Professionals/ dp/3642328431
- 9. Practicalities of building data pipelines for models ● Transforming from Excel, concept maps and other sources ● What are the most useful transformation tools? ● What are the transformation patterns?
- 10. Sources of raw data ● CMAP Tools work well in client workshop to create concept maps ● CMAP Tools don’t work well in designing the pipelines - there is no command line export, so can’t automate end to end ● Excel is what clients often use to document existing domain models
- 11. Transformation tools ● Python with RDFLib works very well for transformation - intuitive, ready to use properties, utilities to work with lists
- 12. Transformation Tools - Python and RDFLib APP = Namespace('http://www.example.com/ontologies/product/') namespace_manager = NamespaceManager(Graph()) namespace_manager.bind('app', APP, override = False) namespace_manager.bind('owl', OWL, override = False) graph = Graph() graph.namespace_manager = namespace_manager # Iterating through either rows of a spreadsheet or export of CMAP Tools… graph.set((APP['Product'], RDF.type, OWL.Class)) graph.set((APP['Product'], RDFS.label, 'Product')) graph.set((APP['Product'], RDFS.label, 'Product'))
- 13. So what? ● Lived to tell the story ● Data pipelines combined with use of Semantic Web for modelling is a viable choice for improved productivity
- 14. Demo ● Python and RDFLib to transform a simple concept map ● Source code: https://github.com/cadmiumkitty/building-ontologies-from-concept-maps