Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management...

39
Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona

Transcript of Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management...

Page 1: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Semantic Model Support for Geographic Information Systems

Sudha RamDepartment of Management Information Systems

The University of Arizona

Page 2: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Agenda

• Research Motivation– GIS Data Management Problems

– Need for a Semantic Model

• USM* Constructs and Formal Definitions• Architecture of the Data Management System• Advantages of Using the USM*• Lessons Learned• Future Research Directions

Page 3: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Natural Resource Management

• Requires large amount of spatial & temporal data. • Example: Risk management assessment

concerning fire:– Lack of understanding about how fire behaves.

– Typical questions that are investigated: “How fast will the fire move under specified conditions?” “How hot will the fire be at certain points?”

– Understanding action of fire can be done using models. Simulation Models + GIS Databases Requires experts from both areas

Page 4: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Risk Management Assessment Process - Fire Model

• A fire simulation requires:– Weather model

temperature, wind speed, cloud cover, and other climatological factors

– Fire model weather model, fuels, time of the

day, terrain, etc.

– Data about the area of interest terrain, slope, aspect, fuels,

firebreaks, etc.

• Two experts are required:– Modeler

– GIS technician

• A fire simulation model

Two Experts User

+ +

Weather Model Fire Model GIS Data

Page 5: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Objective of Research

• To define a formal model (USM*) to capture semantics of spatiotemporal data.

• To facilitate the integration of simulation models with GIS databases.

Page 6: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Why do we need a Semantic Model

• To narrow the gap between users and data/models.

• Currently existing data models do NOT:– combine both spatial and temporal data.

– link simulation models and GIS data.

– capture dynamic behavior of spatiotemporal objects (e.g., fire).

• Why not object-oriented data model?– Relationships between/among spatial objects are explicit in a

semantic model.

– Constraints are explicit in a semantic model.

– Dynamic behavior is explicitly captured in a semantic model.

– All of the above is hidden in the methods in OO models.

Page 7: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Existing Data Models Proposed in the Literature

• Semantic ModelsChen (1976), Hammer and McLeod (1981), Hull and King (1987), Abiteboul and Hull (1987), Peckham and Maryanski (1998), Ram (1995)

• Object-Oriented ModelsAlves, D. (1990), Herring (1992), Lipeck and Neumann (1986), Wiegand and Adams (1994), Worboys et al. (1990)

• Spatial ModelsGambosi et al. (1988), Goodchild (1992), Guptill (1990), Gutting (1988), Orenstein and Manola (1988), Raper and Bundock (1991), Worboys (1992)

• Temporal ModelsElmasri and Wuu (1990), Newell et al. (1992), Langran and Chrisman (1988), Wuu and Dayal (1992)

Page 8: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Constructs for Modeling Spatiotemporal Data (USM*)

• Spatiotemporal entity class (e.g., riparian zone)

• Dynamic entity class (e.g., fire, erosion)

• Spatiotemporal aggregates– Spatial aggregate (e.g., states - cities, counties, rivers…)– Temporal aggregate (e.g., climate - aggregate of weather)– Simple spatiotemporal aggregate (e.g., weather)

• Spatiotemporal interaction relationship– Topological relationship (e.g., overlap, disjoint, touching)– Spatial order relationship (e.g., above, below, left_of, right_of)– Metric relationship (e.g., within)– Fuzzy relationship (e.g., near, far)– Motion relationship (e.g., spreads, enters)

• Cause-effect relationship (e.g., changes, erodes)

Page 9: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Example USM* - Part 1

FIRE

Spatial Agg

GEOLOGY

S

SOIL_EROSIONenum

FUEL

describesdescribed_by

coverscovered_by

spreadschanged_by impacts

induces

erodeseroded_by

WATER

SOIL_LAYER

SOIL

RIPARIAN_ZONE

VEGETATION

Spatial Agg

ASPECTSLOPEELEVATION

TERRAIN

WEATHER

COUNTIES

found_in

contains

Page 10: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Example USM* - Part 2

P

S

adjacent

PRECIPITATION

INFILTRATION

determined_bydetermines

falls

infiltrates dischargedischarged_by

WATER

SUBSURFACE_WATER

SURFACE_WATER

GROUND_WATER

POND STREAM

SUR_SUBSURFACE_WATER

SOIL

RIPARIAN_ZONE

Simple ST Agg

Tem

pora

l Agg

WIND

RAINFALL

HUMIDITY

TEMPERATURE

SUNSHINE

WEATHERCLIMATE

RUNOFF

OVERLAND_FLOW

CHANNEL

SS

SSP

U

S

S

P

Page 11: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Graphical Representation of USM* Constructs

SpatiotemporalEntity Class

SpatiotemporalAggregate

Entity Class

DynamicEntity Class

SpatiotemporalAggregate

Relationship

SpatiotemporalInteraction

Relationships

Cause-EffectRelationships

Entity NameEntity Name Entity Name

STAGG

Page 12: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Formal Definition:Spatiotemporal Entity Class

• Definition - A set of spatiotemporal entities which have spatial (i.e., location) and temporal information, denoted as S = { ei },

– where ei = n-tuple { a1, a2, …, an, si, t }, and i = 1, 2, …, n.

– Attributes ai (i = 1, 2, …, n) are non-spatiotemporal attributes,

– si is a spatial index that describes locational information for ei and is defined as a subset of the Cartesian product of coordinate sets, si, and

– t is a temporal index (e.g., time stamp).

• If the value of either one of ai or si changes, a new ei is added with a new value t.

• The underlying data can be represented in either a raster-based or a vector-based system.

• Examples - riparian zone, vegetation, surface water, runoff, etc.

Entity Name

Page 13: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Formal Definition:Dynamic Entity Class

• Definition - A roster-defined subclass of the spatiotemporal entity

class, denoted as D = { di },– where di = n-tuple defined the same way as ei in the spatiotemporal entity class.

• Dynamic (process-oriented) behavior.

• A definite birth-to-death state.

• The granularity of time interval for the birth-to-death state may be different for different dynamic entity classes.

• One of the most crucial components in the simulation environment.

• Examples - fire, wind, soil erosion, etc.

Entity Name

Page 14: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Formal Definition: Spatiotemporal Aggregates

• Definition - The members of the spatiotemporal aggregate entity class are physically or logically made up of members or sets of members from some other spatiotemporal entity class(es), denoted as STA = n-tuple { stai },

– where each stai is defined as an n-tuple { Ei }, where Ei is a spatiotemporal entity class.

– Each stai has

a set of spatial indexes si (defined earlier) and

a set of temporal intervals (ti, tj) associated with it, where i, j { 1, 2, …, n } and i j.

• Examples - states, counties, climate, weather, etc.

STAGG

Entity Name

Page 15: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Formal Definition: Spatiotemporal Aggregates

• Spatial Aggregate– A constant time index

– For example, entity class STATES is a spatial aggregate whose members are sets of cities, counties, rivers from entity classes CITY, COUNTY, and RIVER.

• Temporal aggregate– A constant spatial index.

– For example, Arizona CLIMATE is an aggregation of WEATHER in Arizona over time.

• Simple spatiotemporal aggregate– A constant spatial and temporal index

– For example, Tucson summer WEATHER

Page 16: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Formal Definition: Spatiotemporal Interaction Relationships

• Definition - A spatiotemporal interaction relationship among n spatiotemporal entity classes, E1, E2, ..., En, defines a set of associations among entities from each of these classes, and denoted as , STI = {stii}

– where stii associates n entities (e1, e2, …, en) and each entity ei belongs to Ei, 1 i n, where Ei is a spatiotemporal entity class.

• Associations between and among only spatiotemporal entity classes.

• Examples - touching, overlap, inside, above, within, near, far, etc

Page 17: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Formal Definition: Spatiotemporal Interaction Relationships

• There are five subclasses, each depicted by type t where t = { {topological relationships}, {spatial order relationships}, {metric relationships}, {fuzzy relationships}, {motion relationships}}.

– Topological: covers, disjoint, equal, inside, touching, overlap, etc.– Spatial order: above, below, left, right, north, south, in_front_of, behind, etc.– Metric: describe distances and directions, e.g., “list all public buildings within 3

miles from White House.”– Fuzzy: near, far, etc.– Motion: spreads, enters, etc.

• These subclasses are NOT mutually exclusive.

• The interaction between (instances of) entity classes may change over time as a result of an event.

Page 18: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Formal Definition: Cause-Effect Relationship

• Definition - An association that relates a set of affectors (or affectees) to a set of affectees (or affectors), and defined as CR = ordered tuple { {Ei}, {Ej} },

– where (Ei D Ej D ) (Ei D Ej D ), and D is the set of dynamic entity classes.

– Affector : a set of entities that affects the status of other entities, attributes, and relationships

– Affectee : a set of entities, attributes and relationships being affected by the affectors.

• Examples - A fire changes the composition of vegetation.

Page 19: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Formal Definition: USM* Schema (Extended Part Only)

• A 2-tuple S = (S, R), where– S is a set of all spatiotemporal entity classes (defined earlier).

– R is a set of all spatiotemporal relationships (defined earlier).

Page 20: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

USM* Metamodel (Extended Part Only)

SpatiotemporalEntity Classes

SpatiotemporalAggregate

Entity Classes

DynamicEntity Classes

SimpleSpatiotemporalEntity Classes

SpatialAggregate

Entity Classes

TemporalAggregate

Entity Classes

SimpleSpatiotemporal

Agg Entity Classes

SS S

S SS

P

P

USM*Models

Agg

ObjectClasses

S

SpatiotemporalRelationships

SpatiotemporalAggregate

Relationships

SpatiotemporalInteraction

Relationships

SpatialAggregate

Relationships

TemporalAggregate

Relationships

SimpleSpatiotemporal

Agg Relationships

S S

S SS

P

P

S SS

TopologicalRelationships

S

FuzzyRelationships

MetricRelationships

Spatial OrderRelationships

MotionRelationships

S

X

Cause-EffectRelationships

S

Related By

RelatesCreated By

Spatial Index

Temporal Index

Attributes

S SS

SS

Time Stam

pTim

e Interval

StructuralAttributes

SimpleAttributes

DerivedAttributes

Entity Class

Interaction Entity Class

Composite/Grouping Entity Class

Weak Entity Class

S

Agg

Generalization/Specialization

Aggregate

Partition (Coverwith No Overlap)

Mutually Exclusive(No Overlap)

X

P

Legends

Inteaction Relationship

Page 21: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Architecture of the Data Management System

Non-GISData

DATAACCESSOR

DATAMODELER

Integrity Checker

SIMULATOR USM* REPOSITORY

Meta

data

Dicti

onar

y

Map

ping

Dict

iona

ry

Mod

el De

scrip

tor

GISDatabase 1

GISDatabase 2

SimulationOutput

Database

Data

Hand

ler

Mod

el Ba

ses

GRAPHICAL USER INTERFACE

USERS

Page 22: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Accessing Spatiotemporal Data Through the USM*

• Metadata Access– Example query: “Display all different types of fuels about which data is available.”

– The browser uses the metadata dictionary and mapping dictionary.

• Metadata and Data Access (Spatial Queries)– Example query: “Display all counties where each type of vegetation is found.”

– The data accessor uses the metadata dictionary, mapping dictionary and metadata directory, and calls proper data access object wrappers.

• Data Access– Example query: “Show a particular county about which vegetation data is available.”

– The data accessor uses the metadata dictionary, mapping dictionary and metadata directory, and calls proper data access object wrappers. Mediators will perform necessary conversions and resolve (or inform) the semantic conflicts.

• Access to Output from Simulation– Example query: “How fast is the fire spreading and which area will be affected by it?”

Page 23: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 24: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 25: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 26: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 27: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 28: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 29: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 30: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 31: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 32: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 33: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 34: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 35: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.
Page 36: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Implementation

• Common repository resides in Oracle Server 8.• Domain knowledge is encoded using Java.• Applications are written in Java (& Swing).

– stand-alone application

– accessed via Java enabled web browser.

• Cross platform (NT Pentium Workstation & SUN Ultra 1 Workstation).

• Three-tier architecture.

Page 37: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Advantages of using the USM*

• Users can easily understand the contents of databases and models available in underlying geographic databases.

• Users do not need to learn to program using the GIS commands.

• Users are spared from having to learn the details of the storage formats and naming conventions.

• Data can be accessed in terms of what the user understands.

• Allows access to both spatial and temporal data.

Page 38: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Lessons Learned

• Initial evaluation of the system has been done by a group of ecosystem analysts at the University of Arizona.

• The metadata repository is a very important.

• USM* constructs evolved over time (e.g., dynamic entity classes and cause-effect relationships).

• Access to simulation output through the semantic model.

• Access to derived data and models.

Page 39: Semantic Model Support for Geographic Information Systems Sudha Ram Department of Management Information Systems The University of Arizona.

Future Directions

• Detecting & resolving semantic heterogeneity in spatiotemporal data.

• Semantic mediators.

• Ontology for semantic mediators.

• Support for schema evolution.