TWC Presentation for the ESIP Semantic Web Cluster
TWC Presentation for the ESIP Semantic Web Cluster, 4/22/2014 Experience in ontology engineering with the Global Change Information System Xiaogang (Marshall) Ma Tetherless World Constellation Rensselaer Polytechnic Institute
TWC Acknowledgements • Project: – Global Change Information System: Information Model and Semantic Application Prototypes, funded by NSF through UCAR • Collaborators: – – – – – Peter Fox (PI, TWC/RPI) Curt Tilmes (Co-PI, NASA/USGCRP) Xiaogang (Marshall) Ma (Project lead, TWC/RPI) Jin Guang Zheng (TWC/RPI) Justin Goldstein (USGCRP/UCAR) Stephan Zednik (TWC/RPI) Linyun Fu (TWC/RPI) Brian Duggan (USGCRP/UCAR) Steve Aulenbach (USGCRP/UCAR) Patrick West (TWC/RPI) 2
TWC Contents 1. 2. 3. 4. Ontologies in computer science The GCIS Ontology Experience from ontology engineering practice Additional operations and tools to refine an ontology 3
TWC 1. Ontologies in computer science • An ontology spectrum Italic text explains typical features of concepts and relationships in each ontology type (from Ma 2011, adapted from Borgo et al. , 2005; Mc. Guinness, 2003; Obrst, 2003; Uschold and Gruninger, 2004; Welty, 2002) 4
TWC A few examples following that spectrum • Catalog/Glossary – Neuendorf, K. K. E. , Mehl, J. J. P. , Jackson, J. A. , 2005. Glossary of Geology, 5 th edition. American Geological Institute: Alexandria, VA, USA, 800 pp. See latest version at: http: //www. agiweb. org/pubs/glossary/ • Taxonomy – BGS Rock Classification Scheme, see: https: //www. bgs. ac. uk/bgsrcs/ • Thesaurus – AQSIQ, 1988. GB/T 9649 -1988 The Terminology Classification Codes of Geology and Mineral Resources. General Administration of Quality Supervision, Inspection and Quarantine of P. R. China (AQSIQ). Standards Press of China, Beijing, China. 1937 pp. (In CN&EN) • Conceptual Schema – NADM Steering Committee, 2004. NADM Conceptual Model 1. 0—A conceptual model for geologic map information: U. S. Geological Survey Open-File Report 2004 -1334, North American Geologic Map Data Model (NADM) Steering Committee, Reston, VA, USA, 58 pp. See: http: //pubs. usgs. gov/of/2004/1334 • Ontologies encoded in RDF format – Semantic Web for Earth and Environmental Terminology (SWEET). See: http: //sweet. jpl. nasa. gov/ 5
TWC Another dimension of ontologies top-level ontology Specialization of domain ontology task ontology Specialization of application ontology Ontologies according to their level of dependence on a particular task or point of view (Guarino, 1997) • Top-level ontologies describe very general concepts like space, time, matter, object, event, action, etc. , which are independent of a particular problem or domain • Domain ontologies and task ontologies describe, respectively, the vocabulary related to a generic domain (e. g. , medicine) or a generic task or activity (e. g. , diagnosing) • Application ontologies describe concepts depending both on a particular domain and task, which are often specializations of both the related ontologies 6
TWC A few examples following that dimension • Top-level ontology – DOLCE: Descriptive Ontology for Linguistic and Cognitive Engineering, see: http: //www. loa. istc. cnr. it/old/DOLCE. html • Domain ontologies and Task ontologies – PROV-O: The W 3 C PROV Ontology (for represent and interchange provenance information), see: http: //www. w 3. org/TR/prov-o/ – BIBO: The Bibliographic Ontology, see: http: //bibliontology. com/ – ORG: The Organization Ontology, see: http: //www. w 3. org/TR/vocab-org/ – DCAT: The Data Catalog Vocabulary, see: http: //www. w 3. org/TR/vocab-dcat/ • Application ontology – GCIS: The GCIS Ontology, see: http: //tw. rpi. edu/web/project/gcisimsap/GCISOntology 7
TWC A few methods for ontology engineering • Ontology Design Patterns – Widely used are Content Ontology Design Patterns: small ontologies that mediate between use cases and ontology design solutions (Gangemi and Presutti, 2009) • Agile Methods for Software Engineering – Adaptive planning; evolutionary development; a time-boxed iteration; and rapid and flexible response to change (Cohen et al. , 2004) • Use case-driven iterative approach – Use cases for identifying questions, resources & methods; small team & mixed skills; a context for collaboration between computer scientists & domain scientists; review & iteration; rapid prototype (Fox and Mc. Guinness, 2008) 8
TWC The use case-driven iterative approach More details at: http: //tw. rpi. edu/web/doc/TWC_Semantic. Web. Methodology 9
TWC 2. The GCIS Ontology • Global Change Information System (GCIS) – An information system under development through the United States Global Change Research Program (USGCRP) that establishes data interfaces and interoperable repositories of climate and global change data which can be easily and efficiently accessed, integrated with other data sets, maintained over time and expanded as needed into the future • GCIS Ontology – An application ontology designed for representing and capturing provenance information in GCIS – Currently focusing on the third National Climate Assessment draft report (draft NCA 3) – More information: http: //tw. rpi. edu/web/project/gcisimsap/GCISOntology 10
TWC Ontology reuse: improve interoperability • • • PROV-O: DCTerms: DCType: FOAF: BIBO: ORG: SKOS: OWL: RDFS: XSD: W 3 C Provenance Ontology Dublin Core Metadata Terms Dublin Core Types Friend Of A Friend Vocabulary Bibliographic Ontology Organization Ontology Simple Knowledge Organization System Web Ontology Language Resource Description Framework RDF Schema XML Schema 11
TWC • • • PROV-O DCTerms DCType FOAF BIBO ORG SKOS OWL RDFS XSD @prefix prov: <http: //www. w 3. org/ns/prov#>. @prefix dcterms: <http: //purl. org/dc/terms/>. @prefix dctype: <http: //purl. org/dc/dcmitype/>. @prefix foaf: <http: //xmlns. com/foaf/0. 1/>. @prefix bibo: <http: //purl. org/ontology/bibo/>. @prefix org: <http: //www. w 3. org/ns/org/>. @prefix skos: <http: //www. w 3. org/2009/08/skosreference/skos. rdf#>. @prefix owl: <http: //www. w 3. org/2002/07/owl#>. @prefix rdf: <http: //www. w 3. org/1999/02/22 -rdf-syntaxns#>. @prefix rdfs: <http: //www. w 3. org/2000/01/rdf-schema#>. @prefix xsd: <http: //www. w 3. org/2001/XMLSchema#>. 12
TWC Ontology engineering: use case analysis The first use case • Title: Visit data center website of dataset used to generate a report figure • Actor and system: a reader of the draft NCA 3 on the GCIS website • Flow of interactions: A reader wishes to identify the source of the data used to produce a particular figure in the draft NCA 3. A reference to the paper in which the image contained in this figure was originally published appears in the figure caption. Clicking that reference displays a page of metadata information about the paper, including links to the datasets used in that paper. Pursuing each of those links presents a page of metadata information about the dataset, including a link back to the agency/data center web page describing the dataset in more detail and making the actual data available for order or download. 13
TWC Use case analysis: Concept map • Concept map – Graphical tool for organizing and representing knowledge (Novak and Cañas, 2008) – Often used as the first step in information models that are precursors to ontology engineering (Starr and de Oliveira, 2013) The IHMC Cmap. Tools is widely used for use case analysis in Semantic Web applications, see: http: //cmap. ihmc. us/ 14
TWC An intuitive concept map of the 1 st use case 15
TWC An intuitive concept map of the use case Classes and properties recognized from the use case 16
TWC An intuitive concept map of the use case From an intuitive model to an ontology: (1) A defined class or property should be meaningful and robust enough to meet the requirements of various use cases (2)and An properties ontology can be extended classes and properties Classes recognized fromby theadding use case recognized from new use cases through the iterative approach 17
TWC The second use case • Title: Identify roles of people in the generation of a chapter in the draft NCA 3 • Actor and system: a viewer of the GCIS website • Flow of interactions: A viewer sees that Chapter 6 (Agriculture) in the draft NCA 3 was written by a group of authors mentioned in a list. On the title page of that chapter the reader can view the role of each author, e. g. , convening lead author, lead author or contributing author, in the generation of this report chapter. • We decided to use the PROV-O ontology to describe this use case 18
TWC The three Starting Point classes in PROV-O ontology and the properties that relate them Source: http: //www. w 3. org/TR/prov-o/ 19
TWC Mapping the use case into PROV-O Author of Chapter 6 in NCA 3 is. A Writing of is. A Chapter 6 in NCA 3 20
TWC Roles of agents in an activity in PROV-O Source: http: //www. w 3. org/TR/prov-o/ 21
TWC Mapping roles of chapter authors into PROV-O is. A Author of Chapter 6 Writing of Chapter 6 in NCA 3 is. A Convening lead author Lead author is. A Contributing author 22
TWC Roles of people in the activity ‘Writing of Chapter 6’ Here only three of the eight authors of this chapter are shown. Each author had a specific role for this chapter.
TWC Re-using existing ontologies for the GCIS ontology By such mappings we can use reasoners that are suitable for the PROV-O ontology, and thus to retrieve provenance graphs from the established GCIS 24
TWC • We have had more use case analyses to build the GCIS ontology 25
TWC 3. Experience from ontology engineering practice Informal message: Some times, a method is not a method at all. 26
TWC 3. Experience from ontology engineering practice • For human: A modeling approach – Transform the knowledge in our brains into a list of concepts and their inter-relationships – Level of details: application needs & interoperability • think about the ontology spectrum and the dimension of ontologies • For machine: An encoding approach – Record the model in a format that can be used by computers in a specific context • CSV, UML, XML, RDF/XML, Turtle, N 3, etc. 27
TWC • For human: concept map helps – Such as those in preceding slides • For machine: AVOID ontology hijacking – We should not modify classes/properties that are defined in external ontologies (e. g. , those in PROV-O, BIBO, FOAF, ORG, etc. ) • For machine: domain and range of properties – Be careful about this when reuse properties from external ontologies 28
TWC For machine: avoid ontology hijacking • For example, we can make such assertions in GCIS ontology: prov: Agent gcis: Agent rdfs: subclass. Of foaf: Agent • And we should avoid such assertions in GCIS ontology: foaf: Agent rdfs: subclass. Of prov: Agent foaf: Agent owl: equivalent. Class prov: Agent 29
TWC For machine: domain and range of properties • For example, to use prov: was. Generaged. By between an instance of gcis: Report and an instance of gcis: Report. Generation • We should assert that gcis: Report is a subclass of prov: Entity and gcis: Report. Generation is a subclass of prov: Activity : was. Generated. By Definition of : was. Generated. By in a owl: Object. Property ; the W 3 C PROV Ontology rdfs: domain : Entity ; rdfs: range : Activity ; rdfs: is. Defined. By <http: //www. w 3. org/ns/prov-o#> ; rdfs: sub. Property. Of : was. Influenced. By ; … : inverse "generated" ; : qualified. Form : Generation, : qualified. Generation. 30
TWC • After rounds of use case analysis, we had a concept map for the GCIS ontology: – http: //cmapspublic 3. ihmc. us/rid=1 MCJMLST 01 G 0 CSWH-2 YH 4/GCIS_Ontology_v 1_2. cmap • And an RDF file synchronized with the concept map, serialized in Turtle format (. ttl): – http: //escience. rpi. edu/ontology/GCISIMSAP/2/GCISOntology_v_1_2. ttl For more information about the Turtle format, see: http: //www. w 3. org/Team. Submission/turtle/ 31
TWC 4. Additional operations and tools to refine an ontology • For machine: ontology syntax check • For human: ontology documentation • Namespace prefix: brand your ontology 32
TWC For machine: ontology syntax check • There are many online tools that help check the grammar of an RDF file: – Such as the RDF Validator and Converter, see: http: //www. rdfabout. com/demo/validator/ 33
TWC For human: ontology documentation • There are several online tools that help generate an ontology document for human to read – Such as the Live OWL Documentation Environment, see: http: //www. essepuntato. it/lode See a list of similar tools at: http: //tw. rpi. edu/web/p roject/Se. SF/Working Group/Ontology. Docu mentation 34
TWC Namespace prefix: brand your ontology • For the GCIS ontology we use gcis as the namespace prefix – One can register namespace prefix and look up existing ones at: http: //prefix. cc/ 35
TWC Final output of the GCIS ontology • Ontology documentation – http: //escience. rpi. edu/ontology/GCISIMSAP/2/GCISOntology_v_1_2. htm • Concept map – http: //cmapspublic 3. ihmc. us/rid=1 MCJMLST 01 G 0 CSWH-2 YH 4/GCIS_Ontology_v 1_2. cmap • Ontology RDF serialized in Turtle format – http: //escience. rpi. edu/ontology/GCISIMSAP/2/GCISOntology_v_1_2. ttl 36
TWC See also • Ma, X. , Fox, P. , Tilmes, C. , Jacobs, K. , Waple, A. , 2014. Capturing and presenting provenance of global change information. Nature Climate Change. In Press. • Tilmes, C. , Fox, P. , Ma, X. , Mc. Guinness, D. , Privette, A. P. , Smith, A. , Waple, A. , Zednik, S. , Zheng, J. , 2013. Provenance representation for the National Climate Assessment in the Global Change Information System. IEEE Transactions on Geoscience and Remote Sensing 51 (11), 5160 -5168. • Ma, X. , Fox, P. , 2013. Recent progress on geologic time ontologies and considerations for future works. Earth Science Informatics 6 (1), 31– 46. 37
TWC Thank you! Sponsors gcis rpi max 7@rpi. edu
- Slides: 38