67a6d127d16b577d2f3348fc3a1bdcf8.ppt

- Количество слайдов: 109

Towards ISO-Space: Annotating Spatial Information in Language ISO PWI 24617 -4 Semantic annotation framework Part 4: Space James Pustejovsky (chair), Harry Bunt, Kiyong Lee, Inderjeet Mani (Proposed editors) May 31, 2009 ISO Meeting Muenzinger D 430 University of Colorado Boulder, CO

Schedule 9: 00 9: 30 - 10: 30 – 10: 45 – 12: 30 – 14: 00 – 14: 45 – 15: 00 Introductions, Member Interests Goals, Scope, and Agenda Coffee break Resources and Data Lunch Defining a specification Action Items

WG Participants • James Allen • Rusty Bobrow • Harry Bunt • David Cooper NGA • Kiyong Lee • David Mc. Donald • Annie Zaenen • Graham Katz • James Pustejovsky • Marc Verhagen • Christy Doran University of Rochester BBN Tilburg University Korea University BBN PARC Georgetown University Brandeis University MITRE

Spatial Awareness in Language The keyhole was below the doorknob. John opened the door and entered the room. The light was to the left of the door on the wall. He walked to the living room. In front of the piano was a rug. On the rug was cat. A woman was at the piano. To her left was a window, from which one could see the garden. In the corner under a lamp, was a chair on its side.

Work Item Goals and Scenarios • Building a spatial map of objects relative to each other. • Reconstructing spatial information associated with a sequence of events. • Determining object location given a verbal description. • Translating viewer-centric verbal descriptions into other relative descriptions or absolute coordinate descriptions. • Constructing a route given a route description. • Constructing a spatial model of an interior or exterior space given a verbal description. • Integrating spatial descriptions with information from other media

Spatial Awareness Requirements • Topological relations between objects • Orientation and distance between objects • Shape and size of objects • Elevation (Lat, Long values) • Recognition of spatial entities and geopolitical regions • Granularity of the relations • Aggregates and distributed objects • Understanding the dynamics of motion and processes

Desired ISO-Space Elements • Regions • Geographic, Geopolitical Places, Functional Locations • Arbitrary Locations • Entities as Spatial Objects • intrinsic orientation, dimensionality, size, shape • Path Objects • routes, lines, turns, arcs • Links • Topological relations • Dimension and Orientation • Metrics • Spatial Functions • behind the building, twenty miles from Boulder • Movements and Spatial Processes • functions from regions to regions

Applications

Spatial Markup of Images Mapping from Spatial Event Structures to RCC 8 relations

Spatial Relations in the Image Mapping from Spatial Event Structures to RCC 8 relations

Image Annotation

Image Sequence Annotation

Cross-media Annotation Nemrava, Sadlier, Buitelaar, Declerck (2008)

Spatial Information in Language • Static Spatial Relations • PP; Mary is on the chair • Verbal; The tree stands in the yard. • Dynamic Spatial Relations • PP; Mary walked to the room. • Verbal; The tree fell. • Polysemy of Spatial Prepositions • over; over the bridge, over the hill. • at, on; on the table/wall, at the computer/party. • in; in the coffee, in the cup, in the bowl.

Spatial Prepositions • • She lives near the school. There is an ice cream shop by the store. An oak tree grows next to my house. The house is between Elm Street and Maple Street. I found my pen lying among the books. The bathroom is opposite that room. Look towards the mountains.

Manner and Path Constructions

Language and Spatial Quantifiers • A policeman stands on every corner in town. • A dark cloud covered every house in town.

Spatial Relations in Motion Predicates

Motion Classes

Biking Blogs: http: //www. rideforclimate. com/journals/? cat=3 March 7, 2006 Leaving San Cristobal de las Casas four days ago, I biked with Gregg and Brooks for one more day. We climbed over the mountains, and then descended to the east, where a thick green rainforest grew up around the road. We arrived in the town of Ocosingo and we were advised the road ahead was unsafe in the afternoon or night. I did not ask if there was a fire station and went straight to a cheap hotel that I split with Gregg and Brooks. The following morning , as I was planning to ride farther that day, I left at dawn while Gregg and Brooks were still asleep. I biked 30 miles to the town of Agua Azul where I played for 4 hours in waterfalls and clean cool pools. After 4 hours, I was surprised Gregg and Brooks had not arrived. I returned to the main road to continue on, and I asked a police officer if he had seen my friends. A few miles short of Agua Azul, while Gregg and Brooks were slowly climbing a hill, two men with machetes and masks jumped out of the forest. They demanded all of Gregg and Brooks' stuff. I spent the next day with Gregg and Brooks at the ruins of Palenque, walking around ruins of an ancient Mayan city and trying to relax. The following day, Gregg, Brooks, and I threw our bikes atop a van and drove the 90 miles to the border with Guatemala, where we had to take a boat across a river.

Military Campaigns: http: //www. militaryhistoryonline. com/wwii/dieppe/ In the early morning hours of August 19, 1942, a fleet of up to 250 ships supported by 68 R. A. F squadrons was carrying a force of 6, 000 -plus men across the English Channel towards the areas that surrounded the city of Dieppe, a port town located in the Pays de Caux region of northeastern France. They would be transported to their target beaches with anticipation from their superiors of accomplishing a series of damaging blows to the German fortifications in Dieppe as well as in the towns, villages, and open areas surrounding Dieppe. But as the minutes approached 0400 hours (4 a. m. ), a German convoy approaching from the north would be the first blow to unravel the entire operation. Nine hours later, the convoy would return to its homeports in defeat with high casualty counts. In the years after D-Day up to the present, historians and military officials agreed that Jubilee, as the operation was codenamed, was one of the Allies' greatest military blunders of World War II. The truth is, Jubilee was turning into a blundering operation long before it was executed, due to a series of planning mistakes, miscalculations, and changes made in the weeks and months leading up to August 19.

Walking Tours: http: //www. leeds. gov. uk/About_Leeds/History/Historic_walks_in_Leeds_city_centre. aspx Leeds Bridge to Temple Mills historic walk This walk begins at Leeds Bridge which crosses the river Aire. Walk down Briggate and you will pass Queen's Court which was a merchant's house. On the other side of the road is a Leeds landmark, Dyson's. This was a famous jewellers and is now a restaurant. Go up Boar Lane to Holy Trinity Church and the Griffin Hotel. There is a Civic Trust plaque where a medieval gate used to stop people going into the Park of Leeds. Mill Hill chapel facing onto City Square is where Joseph Priestley was minister. He is famous for discovering oxygen. There is a statue of him in City Square along with John Harrison and James Watt. There also statues of 'Morn' and 'Even' and in the centre a large statue of the Black Prince. Go along Neville Street and under the Dark Arches. You can see Bondman Dam which took water towards a number of mills where, during the middle ages, the people of Leeds had their corn ground into flour. Next is the Leeds Liverpool canal and this was finished in 1816 and is 127 miles long. Before crossing over Victoria Bridge is a large building which used to be a granary warehouse. On the other side of Water Lane is a development of offices and flats where the Round Foundry was. Machinery and steam engines were developed here. From Globe Road you can see two towers. The larger one, known as the Giotto Tower (1899), is based on a tower at the Florence Cathedral (1334) and was a ventilation shaft. The smaller tower on the left is a chimney (1864) which is a copy of a 12 th century Lamberti tower in Verona. In the distance are rows of terraced houses in Holbeck. This walk finishes at Temple Mills where flax was spun from 1817 -30.

DAML-Space (Hobbs et al) Rusty Bobrow, Murray Burke, Dan Connolly, Dejing Dou, George Ferguson, Andrew Gordon, Pete Haglich, Pat Hayes, Adam Pease, Steve Reed, Richard Waldinger The Semantic Web requires common ontologies with wide acceptance and use. DAML-S: an ontology of services Development began February 2001 About a dozen people in inner circle Some people have explored using it Institutional status at W 3 C Version 0. 9 just released DAML-Time: a temporal ontology Development began February 2002 Most work by 1 -3 people Abstract theory 90% complete Mapping between DAML-Time and Time. ML One site “about to” use it Want to build on this experience for a spatial ontology.

Aims A widely available ontology of geographical and other spatial properties and relations Provide convenient markup and query capabilities for spatial information in Web resources Adequate abstract coverage of most spatial applications (not necessarily efficient) Link with special purpose reasoning engines for spatial theories and large-scale GIS databases Link with various ontological resources (e. g. , Open. Cyc, SUMO, . . . ) and annotation schemes Link with various standards for geographical information (Open. GIS, GML, . . . )

Structure of Effort Cohn etc SUMO Abstract Theory of Space (FOL) Complete or Partial Realization in DAML / OWL / Rule. ML /. . . Hayes & Chaudhri Open. Cyc NLP Extraction Existing Standards Techniques Annotation Standards

Some Principles Delimiting the effort: Not a theory of physical objects, properties of materials, qualitative physics Link with numerical computation, don’t axiomatize it Link with large geographical DBs, don’t duplicate them Navigate past controversial issues by Keeping silent on issue Provide easily exercised options Use textbook logic for abstract theory; DAML/OWL-ize predicate and function declarations Provide simple, useful entry subontologies

Topics SPACE TIME Topology Dimension -- Orientation & Shape -- Length, area, volume Duration Lat/long, elevation Clock & calendar Geopolitical subdivisions -- Granularity Aggregates, distributions Temporal aggregates

Topology Points, arcs, regions, volumes Closed loops and surfaces Ordering relations & “between” in arcs; directions on lines and loops Connectedness, continuity Boundaries & surfaces, interior & exterior, directed boundaries; “airspace above” Disjoint, touching, bordering, overlapping, containing regions (RCC 8); location at Holes NOT open and closed sets NOT pathological topologies

Dimension and Orientation Abstract characterization of dimension, projections on component dimensions Links w topological notions of dimension Frames of reference: earth-based, person-based, vehicle-based, force-based Relative orientations: parallel, perpendicular Cartesian vs polar coordinate systems, bearing & range Transformations between coordinate systems Degrees of freedom Qualitative trigonometry: granularities on orientations 2 1/2 dimensions: elevation as 2 nd class dimension, system mostly thought of as planar Elevation from sea level vs ground level Planar vs spherical geometry

Shape 2 D vs 3 D shapes Linking w shape descriptions in geographical databases Shape descriptors: round, tall, narrow, convex, . . . Relative shapes: rounder, sharper, . . . Same shape as, negative-shape, fits-in Symmetry Links w functionality of shape In artifacts, shape is almost always functional In natural objects, shape often has consequences ? Texture

Size Length, distance, area, and volume Precise measures Alternate descriptions of size English-metric conversions Coarse granularities: order of magnitude, half order of magnitude, implied precision, qualitative measures (large, medium, small) relative to comparison set Encoding uncertainty: bounded error, egg yolk theories Uncertainty of location vs imprecise regions

Granularity A city can be viewed as a point, a region, or a volume. How should these different perspectives be accommodated? One approach: City is an entity with 3 D, 2 D, and 0 D realizations. User can pick which one(s) to use. Build granularity considerations into spatial ontology from the beginning, not as an add-on.

Geopolitical Regions Latitude and Longitude Natural geographical regions: Land masses: continent, island, . . . Bodies of water: ocean, lake, river, . . . Terrain features: mountain, valley, forest, desert, . . . Political regions: Countries Political subdivisions: state, province, county, . . . Municipalities: city, town, village, . . . Residences and street addresses Other: Indian reservations, regulatory zones, . . .

Linkages Exploit the large amount of research on spatial representation and reasoning Open. Cyc, SUMO, Cohn, Galton, Hayes & Chaudhri, Hayes, Asher & Vieu, Egenhofer, Forbus Axiomatize best of this work in coherent fashion Link with existing large ontologies and annotation schemes SUMO, Open. Cyc Ontology should bottom out in existing standards Open. GIS, GML

Target Applications As drivers for what has to be represented Flight map system, COA planning, trafficability Travel system involving lat/longs, political divisions, weather Alexandrian Digital Library Geologic and space (NASA) applications (3 -D) Cell biology Image interpretation and description Robotics Virtual reality For some of these, we are collecting brief descriptions of the requirements for spatial representation and reasoning

Organization degree of community acceptance # of participants

Organization time to completion # of participants

Organization quality of ontology # of participants

Organization daml-spatial mailing list Web page - George Ferguson Coherent construction of abstract theories by small group of people Committee of interested persons in U. S. and Europe Email for commentary / feedback Telecons every 2 weeks to track issues/progress Presentations and discussion sessions at relevant workshops Early realizations of relevant parts of ontology in DAML Early construction of application-oriented entry subontologies

Spatial Information in Language • Static Spatial Relations • PP; Mary is on the chair • Verbal; The tree stands in the yard. • Dynamic Spatial Relations • PP; Mary walked to the room. • Verbal; The tree fell. • Polysemy of Spatial Prepositions • over; over the bridge, over the hill. • at, on; on the table/wall, at the computer/party. • in; in the coffee, in the cup, in the bowl.

Spatial Prepositions • • She lives near the school. There is an ice cream shop by the store. An oak tree grows next to my house. The house is between Elm Street and Maple Street. I found my pen lying among the books. The bathroom is opposite that room. Look towards the mountains.

Manner and Path Constructions

Language and Spatial Quantifiers • A policeman stands on every corner in town. • A dark cloud covered every house in town.

Spatial Relations in Motion Predicates

Motion Classes

Desired ISO-Space Elements • Regions • Geographic Locations • Geopolitical Places • Functional Locations • Arbitrary Locations • Entities as Spatial Objects • intrinsic orientation, dimensionality, size, shape • Path Objects • routes, lines, turns, arcs • Links • Topological relations • Dimension and Orientation • Metrics • Motions • functions from regions to regions

Previous Work • Egenhofer and Franzosa (1991) • Randell et al. (1992) • Cohn and Hazarika (2001) • Nguyen and Worring (2003) • Hollink, Nguyen, Schreiber, Wielemaker, Wielenga, and Worring (2004) • Morarescu (2006) • Mani et al (2008)

Qualitative Spatial Reasoning Thanks to Anthony Cohn for slides

Qualitative Spatial Reasoning • Many aspects: – ontology, topology, orientation, distance, shape. . . – spatial change – uncertainty – reasoning mechanisms – pure space v. domain dependent

Qualitative Spatial Reasoning • Traditional QR spatially very inexpressive • Applications in: – – – – – Natural Language Understanding GIS Visual Languages Biological systems Robotics Multi Modal interfaces Event recognition from video input Spatial analogies. . .

Reasoning about Geographic change Consider the change in the topology of Europe’s political boundaries and the topological relationships between countries disconnected countries surrounding others Did France ever enclose Switzerland? (Yes, in 1809. 5) continuous and discontinuous change …

Ontology of Space • • extended entities (regions)? points, lines, boundaries? mixed dimension entities? What is the embedding space? – connected? discrete? dense? dimension? Euclidean? . . . • What entities and relations do we take as primitive, and what are defined from these primitives?

Why regions? • encodes indefiniteness naturally • space occupied by physical bodies – a sharp pencil point still draws a line of finite thickness! • points can be reconstructed from regions if desired as infinite nests of regions • unintuitive that extended regions can be composed entirely of dimensionless points occupying no space! • However: lines/points may still be useful abstractions

Topology Fundamental aspect of space “rubber sheet geometry” connectivity, holes, dimension … interior: i(X) union of all open sets contained in X i(X) X i(i(X)) = i(X) i(U) = U i(X Y) = i(X) i(Y) Universe, U is an open set

Boundary, closure, exterior • Closure of X: intersection of all closed sets containing X • Complement of X: all points not in X • Exterior of X: interior of complement of X • Boundary of X: closure of X closure of exterior of X

History of QSR (1) • Little on QSR in AI until late 80 s – some work in QR – E. g. FROB (Forbus) • bouncing balls (point masses) can they collide? • place vocabulary: direction + topology

History of QSR (2) • Work in philosophical logic – Whitehead(20): “Concept of Nature” • defining points from regions (extensive abstraction) – Nicod(24): intrinsic/extrinsic complexity • Analysis of temporal relations (cf. Allen(83)!) – de Laguna(22): ‘x can connect y and z’ – Whitehead(29): revised theory • binary “connection relation” between regions

History of QSR (3) • Mereology: formal theory of part-whole relation – – Lesniewski(27 -31) Tarski (35) Leonard & Goodman(40) Simons(87)

History of QSR (4) • Tarski’s Geometry of Solids (29) – mereology + sphere(x) – made “categorical” indirectly: • points defined as nested spheres • defined equidistance and betweeness obeying axioms of Euclidean geometry – reasoning ultimately depends on reasoning in elementary geometry • decidable but not tractable

History of QSR (5) • Clarke(81, 85): attempt to construct system – more expressive than mereology – simpler than Tarski’s • based on binary connection relation (Whitehead 29) – C(x, y) x, y [C(x, y) C(y, x)] z C(z, z) – spatial or spatio-temporal interpretation – intended interpretation of C(x, y) : x & y share a point

History of QSR (6) • topological functions: interior(x), closure(x) • quasi-Boolean functions: – sum(x, y), diff(x, y), prod(x, y), compl(x, y) – “quasi” because no null region • Defines many relations and proves properties of theory

RCC Theory • Randell & Cohn (89) based closely on Clarke • Randell et al (92) reinterprets C(x, y): – don’t distinguish open/closed regions • same area • physical objects naturally interpreted as closed regions • break stick in half: where does dividing surface end up? – closures of x and y share a point – distance between x and y is 0

Defining relations using C(x, y) (1) DC(x, y) df ¬C(x, y) x and y are disconnected P(x, y) df z [C(x, z) C(y, z)] x is a part of y PP(x, y) df P(x, y) ¬P(y, x) x is a proper part of y EQ(x, y) df P(x, y) P(y, x) x and y are equal alternatively, an axiom if equality built in

Defining relations using C(x, y) (2) • O(x, y) df z[P(z, x) P(z, y)] – x and y overlap • DR(x, y) df ¬O(x, y) – x and y are discrete • PO(x, y) df O(x, y) ¬P(x, y) ¬P(y, x) – x and y partially overlap

Defining relations using C(x, y) (3) EC(x, y) df C(x, y) ¬O(x, y) x and y externally connect TPP(x, y) df PP(x, y) z[EC(z, y) EC(z, x)] x is a tangential proper part of y NTPP(x, y) df PP(x, y) ¬TPP(x, y) x is a non tangential proper part of y

RCC-8 8 provably jointly exhaustive pairwise disjoint relations (JEPD) DC EC PO TPP NTPP EQ TPPi NTPPi

Expressivenesss of C(x, y) • Can construct formulae to distinguish many different situations – connectedness – holes – dimension

Notions of connectedness • One piece • Interior connected • Well connected

Gotts(94, 96): “How far can we C? ” • defining a doughnut

Other relationships definable from C(x, y) • E. g. FTPP(x, y) – x is a firm tangential part of y • Intrinsic TPP: ITPP(x) – TPP(x, y) definition requires externally connecting z – universe can have an ITPP but not a TPP

4 -intersection (4 IM) Egenhofer & Franzosa (91) • 24 = 16 combinations • 8 relations assuming planar regular point sets disjoint overlap in coveredby touch cover equal contains

RCC 8 Relations

Extension to cover regions with holes • Egenhofer(94) • Describe relationship using 4 -intersection between: – – x and y x and each hole of y y and each hole of x and each hole of y

9 -intersection model (9 IM) 29 = 512 combinations 8 relations assuming planar regular point sets potentially more expressive considers relationship between region and embedding space

9 -Intersection Model for Line-Region Relations Egenhofer and Herring (1991)

LR Intersection

LR Intersection

LR Intersection

LR 9 Intersection

Mereology and Topology • Which is primal? (Varzi 96) • Mereology is insufficient by itself – can’t define connection or 1 -pieceness from parthood 1. generalise mereology by adding topological primitive 2. topology is primal and mereology is sub theory 3. topology is specialised domain specific sub theory

Direction-Relation Matrix (Goyal & Sharma 97) • cardinal directions for extended spatial objects also fine granularity version with decimal fractions giving percentage of target object in partition

Distance/Size • Scalar qualitative spatial measurements – area, volume, distance, . . . – coordinates often not available – Standard QR may be used • named landmark values • relative values • comparing v. naming distances – linear; logarithmic – order of magnitude calculi from QR • (Raiman, Mavrovouniotis )

How to measure distance between regions? • nearest points, centroid, …? • Problem of maintaining triangle inequality law for region based theories.

Distance distortions due to domain (2) • Human perception of distance varies with distance – Psychological experiment: • Students in centre of USA ask to imagine they were on either East or West coast and then to locate a various cities wrt their longitude • cities closer to imagined viewpoint further apart than when viewed from opposite coast • and vice versa

Distance distortions due to domain (3) • Shortest distance not always straight line in many domains

Distance distortions due to domain (4) • kind of scale – – figural vista environmental geographic • Montello (93)

Shape • topology. . . . . fully metric – what are useful intermediate descriptions? • metric same shape: – transformable by rotation, translation, scaling, reflection(? ) • What do we mean by qualitative shape? – in general very hard – small shape changes may give dramatic functional changes – still relatively little researched

Qualitative Shape Descriptions • • boundary representations axial representations shape abstractions synthetic: set of primitive shapes – Boolean algebra to generate complex shapes

boundary representations (1) • Hoffman & Richards (82): label boundary segments: – curving out > – curving in – straight > < | – – angle outward > angle inward < cusp outward cusp inward > > >

boundary representations (2) • constraints: – consecutive terms different – no 2 consecutive labels from {<, >, , } – < or > must be next to or • 14 shapes with 3 or fewer labels • { convex figures • { polygons

boundary representations (3) + • maximal/minimal points of curvature (Leyton 88) – – – Builds on work of Hoffman & Richards (82) M+: Maximal positive curvature M-: Maximal negative curvature m+: Minimal positive curvature m-: Minimal negative curvature 0: Zero curvature

Previous Spatial Annotation Efforts • Space. ML (Cristani and Cohn) • Describes geographic objects associated with web pages • Space. ML (Morarescu) • Annotates spatial information in text. • Spatial. ML (MITRE) • Maps PLACE information in text to data from gazetteers and other databases • GML •

Spatial. ML The interpretation of spatial language has been hampered by the lack of a markup scheme As a result, lack of resources such as corpora and evaluation methods for systems that process spatial language Spatial. ML is a markup scheme for representing places mentioned in text and their relationships The main focus has been on geo-coding of natural language, i. e. , the mapping of geographic references in text to data in gazetteers and other databases. sourceforge. net/projects/spatialml

Some Challenges in Annotation-Based Methods 1. 2. 3. Annotation Scheme: Expressiveness versus Usability Maturity of Guidelines System Adaptation Cost • Languages • Domains

Objectives of Spatial. ML • Develop a research program for spatial information extraction for natural language – Develop Spatial. ML annotation scheme – Integrate spatial and temporal reasoning and machine learning • Transition into existing GIS applications • Improve system by leveraging feedback from analysts

Related Work Spatial. ML borrows ideas from Schilder et al. (2004) , Garbin and Mani (2005), and Toponym Resolution Markup Language of Leidner (2006). Spatial. ML is compatible with Automatic Content Extraction (ACE) English Annotation Guidelines for Entities (Version 5. 6. 6 2006. 08. 01), specifically their GPE, Location, and Facility entity tags and the Physical relation tags. Unlike ACE, Spatial. ML: – – Grounds mentions with geo-coordinates where possible Handles relative locations involving distances and orientation relations Doesn’t group mentions into coreference classes Doesn’t address metonymy Spatial. ML can be integrated with the Geography Markup Language (GML) defined by the Open Geospatial Consortium (OGC). Spatial. ML leverages ISO (ISO-3166 -1 for countries and ISO-3166 -2 for provinces). Mappings: Spatial. ML to KML , and from Meta. Carta output to Spatial. ML.

Spatial. ML Example a

Spatial. ML • Recently-developed markup language for representing spatial expressions in natural language documents • Allows for potentially better integration of text collections with resources providing spatial information about a domain, including gazetteers, physical feature databases, mapping services, etc. • Available at sourcefourge. net and the LDC

Spatial. ML Tags • PLACE tags have attributes including type of place, gazetteer reference (e. g. , USGS GNIS and NGA Geonames gazetteers) and geo-coordinates, among other features • LINK tags are used to express containment, connection, or other topological relations between a pair of locations. • RLINK (RELATIVE-LOCATION-LINK) tags express one location relative to another.

Types of Spatial and Temporal Descriptions of Interest • Temporal – Absolute • Fully specified March 14 2005 (141930 CMAR 05) • Partial Tuesday; this week – Relative • Precise offsets two weeks ago today • Vague offsets three days after the incident; later • Spatial – Absolute • Fully specified southern Kerala district of Cudallah • Partial El Qahira – Relative • • Precise offsets five miles from Bedford; behind National Cathedral Vague offsets in the vicinity of Georgetown University; east on a service road; in his residence

PLACE TYPES Coarse-grained, to make it easier for humans and machines to annotate Drawn opportunistically from Alexandria Digital Library Feature Types Thesaurus NGA Geonames USGS GNIS

TYPE Codes

LINKS • a

Region Connection Calculus • • These 8 relations describe static spatial relationships. Spatial. ML uses a modified version of RCC 8: – TPP, NTPP = IN – Near = NR – Extended External Connection = EEC

Topological Relations MOD Code. Example Direction Code Example B the bottom of the [well] B [behind] the house BR [Burmese] border C central [district] A [above] the roof E eastern [province] BL [below] the tree-line L left on [Bourbon Street] N [North India] E [E] of NEAR near [Harvard] R turn right at the [Mc. Donald’s] S southern [India] T W ESE, WSW, etc. F [in front of] theater the top of the [mountain] N [north] of west [Tikrit] S [south] of W [W] of

Identifying Spatial Entities • Geo-referencing of text can allow information in unstructured data to be integrated with Geographical Information Systems • Today’s tools can ground proper names in a coordinate system • However, since they don’t handle general spatial language, users can miss a lot of relevant information

Spatial. ML Example • a

Limitations • Spatial. ML (and other theories based on systems like RCC) are good for representing static information about space • Ideally, we want to know: – What happened – When did it happen – Where did it happen • Anchored Spatial. ML: Use event and time information from Time. ML and space information from Spatial. ML to locate events

Desired ISO-Space Elements • Regions • Geographic, Geopolitical Places, Functional Locations • Arbitrary Locations • Entities as Spatial Objects • intrinsic orientation, dimensionality, size, shape • Path Objects • routes, lines, turns, arcs • Links • Topological relations • Dimension and Orientation • Metrics • Spatial Functions • behind the building, twenty miles from Boulder • Movements and Spatial Processes • functions from regions to regions