Presentation on theme: "23. April 1998 First Agile Meeting 1998 1 Aspects of Interoperability Specifications For Interoperability: Formalising Spatial Relations 'In', 'Auf' and."— Presentation transcript:
23. April 1998 First Agile Meeting 1998 1 Aspects of Interoperability Specifications For Interoperability: Formalising Spatial Relations 'In', 'Auf' and 'An' and the Corresponding Image Schemata Container, Surface and 'Link Andrew U. Frank Dept. of Geoinformation Technical University Vienna firstname.lastname@example.org
23. April 1998 First Agile Meeting 1998 2 Overview Interoperability What are interoperability research questions? Semantics as the main problem Case study: spatial relations Research methodology
23. April 1998 First Agile Meeting 1998 3 Interoperability Interoperability means working together, e.g., languages. Not a new topic for GIS cooperation between agencies dealing with space is a must; they currently work together. Interoperability is facilitated by human intervention.
23. April 1998 First Agile Meeting 1998 4 Interoperability by data exchange Previous discussion: data exchange (with extensive discussion of format changes and standardization) Disadvantage of data exchange: updates are not propagated
23. April 1998 First Agile Meeting 1998 5 Technical Solution Access over the net to the live database of another agency. Made possible by oubiquous net access Technically a communication is established between client and server. Two programs – running usually on different machines – cooperate in solving a GIS problem. Exchange of requests and answers: language to formulate both: SQL extensions for spatial data (for requests) interpretation of the data delivered
23. April 1998 First Agile Meeting 1998 6 Metaphor for Interoperable GIS connect an appliance to the power network
23. April 1998 First Agile Meeting 1998 7 Superficial level of observation the plug matches the outlet Interoperability = matching interface? Issue: Standardization of Plugs
23. April 1998 First Agile Meeting 1998 8 Standards needed for Interoperability Plugs must be standardized, but their specification is arbitrary, as demonstrated by the variety of plugs one encounters in the world. Standardization of plugs is an important question; but not a research issue!
23. April 1998 First Agile Meeting 1998 9 Interoperability means co-operation The interface must allow the exchange between systems. Expectation and offer must meet. Expectation of appliance: 110 Volt, AC 60 Hz Expectation of power supplier: resistance of 100 Ohm (draws max. 1.1 A) The standardization of these values is not a research issue; other values were possible.
23. April 1998 First Agile Meeting 1998 10 What has science contributed? A theory of electric power; the interface is described in terms of physical measures; electric tension, measured in Volt electric force, resistance scales of measurement with the corresponding units. Interoperability requires abstract description of interfaces: without a specification of 110 V and the corresponding measurement methods user interfaces would become interoperable only through testing.
23. April 1998 First Agile Meeting 1998 11 Need: Formal semantics, for example for measured values in geography and social sciences: Description of an observable quantity, Measurement method Unit values Repeatable abstract description which can be compared between cooperating partners.
23. April 1998 First Agile Meeting 1998 12 Interoperability research Mostly not research questions, but questions of standardization; not in need of academic contributions. Quality requirements for Research Interoperability standardization requires clean solutions from the research community: Comprehensive, Consistent Consent
23. April 1998 First Agile Meeting 1998 13 Problem for Geographic Information: GIS interoperability requires the exchange of Geographic Information. Description of information means description of semantics. Two major issues, closely related: What do data describe – semantics of data How to describe data – meta data Deep issue (and not novel): What does a word mean? A rose by any other name, would smell as sweet.
23. April 1998 First Agile Meeting 1998 14 Definition of Semantics: The West Peak is 1140 ft high How should I interpret this sentence? What should I do?
23. April 1998 First Agile Meeting 1998 15 Description of Semantics Natural language is ambiguous and depends on human interpretation, which varies between people. Interoperability was a problem before computers ! Formal Language descriptions required Natural language must be translated into a formal language, which can be translated in computer code alternative: write code directly, which then becomes the (unreadable) formal definition.
23. April 1998 First Agile Meeting 1998 16 Requirements for the Language: assess expressive power and precision of definition of semantics: computationally complete descriptive, not procedural referentially transparent strongly typed object oriented, polymorphic higher order functions
23. April 1998 First Agile Meeting 1998 17 Restricted user community Reduced answer for GIS: Exchange of semantics in a limited user community. Address restricted problems: Need restriction of problem size to tackle it. For GIS we need not approach the problem of full natural language understanding: I like boxing or Time flies like an arrow.
23. April 1998 First Agile Meeting 1998 18 For GIS needed: Semantics of spatial relations To capture semantics means to formalize. Example: 9-intersection Examples: Egenhofer: topology, cardinal directions and distances, configuration of points – How to approach the other spatial relations, as expressed in natural languages. (This is a topic worked on since 1986, the Las Navas meeting, and carried through several NCGIA and Varenius initiatives, starting with I2.)
23. April 1998 First Agile Meeting 1998 19 Case study: in, auf, an Consider relations with respect to movement of objects; Selected ontology: table top space, blocks and similar objects (fruit, bowl…) Similar relations for geography have been studied with the same methods.
23. April 1998 First Agile Meeting 1998 20 Converse of 'auf' blocks object of movement: 'Auf' blocks the movement of the supporting object. It cannot be moved unless the object 'auf' it is removed. x 'auf' y (in scene) => blocked (move y in scene) Teller und Gläser sind auf dem Tisch. Wir müssen den Tisch zuerst abräumen, bevor wir ihn auf die andere Seite des Zimmers bringen können. Plates and glasses are on the table. We have to remove all objects from the table, before we can move it to the other side of the room.
23. April 1998 First Agile Meeting 1998 21 'In', 'an': Invariance under movement of relatum These relations are invariant under movement. If x is 'in' y and y is moved, then x is still 'in' y (and the same for 'an'). x 'in' y (in scene) => x 'in' y (in move y in scene) = True x 'an' y (in scene) => x 'an' y (in move y in scene) = True These rules will not be expressed explicitly, as they are subsumed by the 'stable world property' (nothing changes unless specifically indicated).
23. April 1998 First Agile Meeting 1998 22 'In', 'an': block movement of object 'In' and 'an' create a link between the object and the relatum which resists movement. x 'in' y (in scene) => blocked (horizontal move x in scene) Der Apfel kann nicht aus der Schale rollen, aber du kannst ihn dir herausheben. The apple cannot roll out of the bowl, but you can take it out (lift it out). x 'in' y and 'closed' y (in scene) => blocked (move x in scene) Du mußt die Büchse öffnen, dann kannst du die Würfel herausnehmen. You must open the box. Then you can take out the dice.
23. April 1998 First Agile Meeting 1998 23 'In' blocks target of movement An object cannot be moved to a target if this is already in another object. This is justified by situations as: x 'in' y (in scene) => blocked (move z into x (in scene)) Du mußt den Beutel zuerst aus der Tasche nehmen, bevor du die Münze hineingeben kannst. You must take the purse out of the pocket to put the coin in.
23. April 1998 First Agile Meeting 1998 24 Undoes a Previous Relation of Object: 'auf' 'Auf' does not restrict the movement of the object: x 'auf' y (in scene) => move x in scene Du kannst das gelbe Buch nehmen, es liegt auf dem Tisch. You can take the yellow book, it is on top of the table. A previously established relation becomes false and a new relation is established: scene2 = move x Rel y (scene1) a Auf z (in scene1) = True a Auf z (in scene2) = False a Rel y (in scene2) = True
23. April 1998 First Agile Meeting 1998 25 'An' presupposes a physical connection between the object and the relatum..Movement is restricted unless the link is broken. x 'an' y (in scene) => blocked (move x) Ich habe das Papier auf das Buch gelegt, jetzt klebt es daran. Wenn du das Papier mitnehmen willst, mußt du es sorgfältig lösen. I have put the paper on (auf) the book, now it is glued on (an). If you want to take it with you, then you have to carefully remove it.
23. April 1998 First Agile Meeting 1998 26 Research Methodology Linguistic style research Restrict to particular language Select example sentences: demonstrate an existing language mechanism (neither the only one, nor all of them; completeness is usually not achievable Investigate entailments: what follows from a sentence; what can be concluded normally. Often a need to investigate small stories
23. April 1998 First Agile Meeting 1998 27 Specific research method Restrict to a particular environment e.g. table top space investigate a group of preposition with respect to a (few) operations; restrict domain of application to very restricted area; assume polysemy to identify single relations.
23. April 1998 First Agile Meeting 1998 28 Method to formalize Algebra: The operations are functions Relations Use category theory to achieve a coherent formalization (specifically Allegories)
23. April 1998 First Agile Meeting 1998 29 What are the primitives Formal suggestions from mathematical allegory theory Cognitive science research suggest image schemata There is a hypothesis that there are a small number of fundamental concepts, which are widely used (also metaphorically). Most of them are spatial. The theory of experientialism gives a base for meaning.
23. April 1998 First Agile Meeting 1998 30 Partial List of Image Schemata Container BalanceFull-EmptyIterationCompulsion BlockageCounterforceProcessSurfaceRestraint Remove EnablementAttractionMatchingPart-Whole Mass-Count PathLinkCollectionContact Centre-PeripheryCycleSplittingMerging ObjectScaleNear-FarSuperimposition
23. April 1998 First Agile Meeting 1998 31 Conclusions Interoperability is limited by semantics. How to understand the data exchanged? Need for formal methods to describe at least subsets of terminology for limited user groups. Demonstrated a case study, how semantics of spatial preposition (for table top space, but with the same method also for geographic space) can be captured.
23. April 1998 First Agile Meeting 1998 32 GIPSIE Project EU project (DG III: Information Technology) to promote Open GIS within the GI industry to bring European Issues into the OGC process to contribute with research to the Open GIS standards