Presentation is loading. Please wait.

Presentation is loading. Please wait.

So What Does it All Mean? Geospatial Semantics and Ontologies Dr Kristin Stock.

Similar presentations


Presentation on theme: "So What Does it All Mean? Geospatial Semantics and Ontologies Dr Kristin Stock."— Presentation transcript:

1 So What Does it All Mean? Geospatial Semantics and Ontologies Dr Kristin Stock

2 Exercise Write a precise definition of: Write a precise definition of: A hill or A hill or A beach A beach (choose one). Write it as if to: Write it as if to: a person who has no prior knowledge or a person who has no prior knowledge or a computer. a computer. Must allow a person who doesn’t know the concept to determine whether something they encounter is an example of that concept. Must allow a person who doesn’t know the concept to determine whether something they encounter is an example of that concept.

3 Introduction Semantics are about describing and interpreting the meaning of geographic feature types. Semantics are about describing and interpreting the meaning of geographic feature types. Needed whenever data is to be shared or exchanged. Needed whenever data is to be shared or exchanged. Especially to do with helping computers to understand meaning. Especially to do with helping computers to understand meaning.

4 Example: data supply

5

6 Lecture Outline What are semantics used for? What are semantics used for? Why do semantics differ? Why do semantics differ? Methods for representing semantics. Methods for representing semantics. What are ontologies? What are ontologies? Reasoning. Reasoning. Current research. Current research.

7 What are semantics used for? allowing users unfamiliar with data to interpret and use it correctly; allowing users unfamiliar with data to interpret and use it correctly; automatically determining semantic similarity (which can be used in integration, translation, etc.); automatically determining semantic similarity (which can be used in integration, translation, etc.); querying (intelligent querying, NLP) and querying (intelligent querying, NLP) and Web Services. Web Services.

8 Why do semantics differ? A database is an abstraction of reality created by an individual using his or her cognitive model (world view). A database is an abstraction of reality created by an individual using his or her cognitive model (world view). Cognitive models differ depending on a number of characteristics, including education, experiences, theoretical assumptions, language. Cognitive models differ depending on a number of characteristics, including education, experiences, theoretical assumptions, language. Spatial data is used by a wide range of different people AND different cognitive models  databases with different semantics. Spatial data is used by a wide range of different people AND different cognitive models  databases with different semantics.

9 Methods for representing semantics (1) Methods based on existing information: Methods based on existing information: Schema characteristics. Schema characteristics.

10 Methods for representing semantics (2) Methods using additional representations: Methods using additional representations: Semantic networks and hierarchies (e.g. synonyms); Semantic networks and hierarchies (e.g. synonyms); Frames (from KR) - complex descriptions linked to create taxonomies; Frames (from KR) - complex descriptions linked to create taxonomies; Logic and Logic and Ontologies (combine semantic networks, frames and/or logic). Ontologies (combine semantic networks, frames and/or logic).

11 Semantic Networks Nodes are concepts/classes/nouns. Nodes are concepts/classes/nouns. Links between are relationships between concepts: Links between are relationships between concepts: is-a; is-a; part-of. part-of. Simple, understandable. Simple, understandable. Not semantically rich (no properties). Not semantically rich (no properties).

12 Example: Semantic networks water body ocean river creek is-a flows-to bay valley runs-through is-part-of beach is-in

13 Example: Wordnet

14 Frames A frame is a concept. A frame is a concept. Each frame has slots that are filled with values (properties or attributes). Each frame has slots that are filled with values (properties or attributes). Hierarchical relationships may be created between frames (using a kind-of slot). Hierarchical relationships may be created between frames (using a kind-of slot). Concept frames and instance frames. Concept frames and instance frames. Richer semantics, but not precisely defined. Richer semantics, but not precisely defined.

15 Example: Frames OCEAN sovereignty averageTemp WATERBODY name is-a RIVER length CREEK intermittent? waterQuality is-a

16 Logic A collection of propositions. A collection of propositions. Use predefined symbols: Use predefined symbols: AND AND OR OR THERE EXISTS THERE EXISTS FOR EVERY… FOR EVERY… Simple, understandable, reasoning mechanisms available Simple, understandable, reasoning mechanisms available Lacks structure Lacks structure

17 Example: Logic Town Town Rule = dimension [property] Rule = dimension [property] Property can be value, range, enumerated, predicate Property can be value, range, enumerated, predicate Predicate = R1 AND R8 AND R9 AND ( R101 OR R102 ) Predicate = R1 AND R8 AND R9 AND ( R101 OR R102 ) R1 = material [EARTH] R1 = material [EARTH] R8 = requirement [DEFINITION OF POSITION AND EXTENTS] R8 = requirement [DEFINITION OF POSITION AND EXTENTS] R9 = function [ADMINISTRATION] R9 = function [ADMINISTRATION] R101 = RR447 [R1 AND R8 AND R223 AND ( R183 OR R184 ) AND R226] R101 = RR447 [R1 AND R8 AND R223 AND ( R183 OR R184 ) AND R226] R102 = PR47 [R1 AND R8 AND R223 AND ( R183 OR R184 ) AND R226 R102 = PR47 [R1 AND R8 AND R223 AND ( R183 OR R184 ) AND R226

18 What are Ontologies? An ontology is a formal specification of a shared conceptualisation. An ontology is a formal specification of a shared conceptualisation. Based on the idea of a group or information community with a common world view. Based on the idea of a group or information community with a common world view. Formal specification = machine readable. Formal specification = machine readable. Can be: Can be: lightweight, including concepts, taxonomies, relationships between concepts and properties or lightweight, including concepts, taxonomies, relationships between concepts and properties or heavyweight, add axioms and constraints to clarify the meaning of concepts. heavyweight, add axioms and constraints to clarify the meaning of concepts.

19 Types of Ontology Language Logic. Logic. Frames. Frames. Networks. Networks. Description Logics (frame, network + logic). Description Logics (frame, network + logic). Use subsumption hierarchies Use subsumption hierarchies Including limited reasoning on those hierarchies. Including limited reasoning on those hierarchies. Tutorial/practical. Tutorial/practical.

20 Example: OWL Ontology

21 Reasoning Use semantics to perform automated reasoning. Use semantics to perform automated reasoning. Examples: Examples: If suburb in city and city in region, then suburb must be in region. If suburb in city and city in region, then suburb must be in region. If high density suburbs have high crime rates and suburb x is high density, then suburb x has a high crime rate. If high density suburbs have high crime rates and suburb x is high density, then suburb x has a high crime rate. If rivers flow to oceans, and the Trent River is a river, then the Trent River flows to the ocean. If rivers flow to oceans, and the Trent River is a river, then the Trent River flows to the ocean. Can be used for Can be used for Direct inference; Direct inference; Service chaining; Service chaining; Dynamic spatial analysis? Dynamic spatial analysis?

22 Spatial Semantics vs. the Semantics of Spatial Features: What is the Difference? 1. The semantics of geometries and spatial representations. 2. The semantics of features that have spatial representations.

23 What are semantics used for? allowing users unfamiliar with data to interpret and use it correctly; allowing users unfamiliar with data to interpret and use it correctly; automatically determining semantic similarity (which can be used in integration, translation, etc.); automatically determining semantic similarity (which can be used in integration, translation, etc.); querying (intelligent querying, NLP) and querying (intelligent querying, NLP) and Web Services. Web Services.

24 Future research directions Geo-ontology languages. Geo-ontology languages. Methods for determining semantics based on context. Methods for determining semantics based on context. Similarity assessment. Similarity assessment. Ontology learning. Ontology learning. Use of semantics with reasoning to automate processes and geospatial analysis. Use of semantics with reasoning to automate processes and geospatial analysis.

25 Tutorial (1) Download Protégé Full. load/release/full/ Download Protégé Full. load/release/full/ load/release/full/ load/release/full/ Copy from t drive to c drive Copy from t drive to c drive Double click to install – select ‘Everything’ when asked about the type of installation. Double click to install – select ‘Everything’ when asked about the type of installation. Start up the software. Start up the software. Create a new OWL Ontology. Create a new OWL Ontology.

26 Tutorial (2) Go through the tutorial at ode.org/resources/tutorials/ProtegeO WLTutorial.pdf Go through the tutorial at ode.org/resources/tutorials/ProtegeO WLTutorial.pdf ode.org/resources/tutorials/ProtegeO WLTutorial.pdf ode.org/resources/tutorials/ProtegeO WLTutorial.pdf Use a spatial domain of your choice. For example: Use a spatial domain of your choice. For example: Transportation networks; Transportation networks; Administrative areas; Administrative areas; Natural landscape features; Natural landscape features; Human settlement. Human settlement. Identify some hypothetical classes in the domain and build an ontology following the tutorial. Identify some hypothetical classes in the domain and build an ontology following the tutorial.

27 Notes In Section 4.3, Exercise 5, Step 2, use Tools > QuickOWL > Create Multiple Subclasses In Section 4.3, Exercise 5, Step 2, use Tools > QuickOWL > Create Multiple Subclasses Go to the end of Section 4.7 Go to the end of Section 4.7 Look at Code > Show RDF/XML Source Code to see the Actual OWL syntax. Look at Code > Show RDF/XML Source Code to see the Actual OWL syntax. Continue if you want to, but it is not required. Continue if you want to, but it is not required.


Download ppt "So What Does it All Mean? Geospatial Semantics and Ontologies Dr Kristin Stock."

Similar presentations


Ads by Google