A Semantic Approach for the Modeling of Trajectories in Space and Time

1
A Semantic Approach for the Modeling of
Trajectories in Space and Time
Donia Zheni, Ali Frihida, Christophe Claramunt,
Henda Ben Ghézala

• SeCoGIS 2009

2
Contents

• Introduction
• Proposed Approach
• Query Examples
• Conclusions and Future Work

3
Introduction

• Spatio-Temporal Trajectory Recording movement
  in time
• New needs for representation
• Intelligent perception of the movement
• Efficient manipulation tools

t6
Time
t5
t4
t3
t2
t1
space
4
Introduction

• Existent DBMS does not satisfy these needs

• A semantically poor perception

Stop to do what?
How to move ? Why to move?

• No unified representation of the spatial,
  temporal and semantic components
• No efficient manipulation tools

5
Introduction

• The main contribution of this work is the design
  of the trajectory as
• Identifiable and semantic rich entity
• With a rich set of operations for the creation
  and manipulation at different levels
• Semantic
• Spatial
• Temporal
• Spatio-temporal

6
Proposed Approach Time Geography

• The Time Geography Conceptual framework for the
  study of the spatio-temporal behavior

Activity
Trip
7
Proposed Approach An Abstract Data Type

• Algebraic Abstract Data Type (ADT) design
• Formal description of a new type named STT

Point
ADT Spatio-Temporal Trajectory
Line
Region
Components
Operations
Time
Interval
8
Proposed Approach An Abstract Data Type

• The concept of Spatio-Temporal Trajectory (STT)
  models a trajectory history where a series of
  trajectory states is depicted by persistent
  observations
• Retrieval and analysis of STTs are expected to
  give insights concerning the manner phenomenon
  evolves (how) and the reason behind such an
  evolution (why)
• We hereafter model a trajectory as a
  spatio-temporal trajectory (STT)

9
Proposed Approach An Abstract Data Type

• A STT represent the spatio-temporal and semantic
  history of a moving objet during a period of time
• We propose an algebraic model to represent and
  encapsulate the syntactic and semantic
  specification of STTs into an ADT
• Description of two parts
• A Syntaxic Model
• A Semantic Model

10
Proposed Approach Syntaxic Model

• Represents the syntaxic interface of the type STT
• Its a signature ? (Sorts, ?)
• Sorts Set of types
• ? Set of operations
• OperationName Type1 Typen ? ResultType

11
Proposed Approach Syntaxic Model
12
Proposed Approach Syntaxic Model
13
Proposed Approach Syntaxic Model

• Operation Examples
• Semantic
• Temporal
• Spatial
• Set

14
Proposed Approach Semantic Model

• Represents a possible implementation of the
  proposed syntax
• A ?-Algebra
• A carrier set for each sort of the signature
• A formally defined semantic of the operations

15
(No Transcript)
16
Proposed Approach Semantic Model
Activity
Trip
a (l, ts, te, purpose)
Temporal Condition
Temporal Condition
a. ts lt a. te
a2
a
d2
Domain
Domain
te
a
a1
a
Temporal Condition
Temporal Condition
ts
a
ai.te lt ai1.ts
d1
TST Activities
TST trips
Spatio-Temporal Conditions
TST Domain
17
Proposed Approach Semantic Model

• Operations Semantics
• Formal description of the operation behavior
• Interpretation functions
• Offers a mathematic, rigorous, and DBMS
  independent description

fop(ag1,ag2,agn) interpretation
18
(No Transcript)
19
Query Examples

• Query examples to illustrate the capabilities of
  the query language
• An application that models activity time-paths of
  person activities
• Every person generates several activity
  time-paths corresponding to a period of time

TimePath(id String, idPer String, traj STT,
day Date)
spatio-temporal attirbute
ADT
20
Query Examples

• Let us consider the following self-explanatory
  tuples
• person(id String, name String)
• TimePath(id String, idPer String,
  traj STT, day Date)
• city(name String, zone Region)
• place(name String, location Point)
• mode(name String)
• We could formulate some query examples without
  loss of generality using an algebraic based
  semantic

21
Query Examples

• What is the transport mode that a person of
  identity p used just before its ith activity
  during the day d?
• Trip_Mode(TP.traj, Trip_Before_Activity(TP.traj,N
  th_Activity(TP.traj,i))) TP ? TimePath ?
  TP.idPersp ? TP.day d
• Which persons have the same spatio-temporal
  behavior as the person of id id during the day d?
• PR1.name PR1 ? person ? PR2 ? person ? TP1 ?
  TimePath ? TP2 ? TimePath ? PR1.id TP1.idPers ?
  PR2.id TP2.idPers ? PR2.id id ? TP1.day d
  ? TP2.day d ? Equal( TP1.traj, TP2.traj)

22
Conclusions and Future Works

• We introduced in this work an abstract data type,
  named STT, for a semantic-based representation of
  spatiotemporal trajectories
• The STT data type is able to reveal additional
  knowledge on the dynamics, changes and evolution
  of the represented trajectories
• It integrates several categories of setbased,
  thematic, spatial and temporal operations that
  together offer several query capabilities
• As future work, we would like to implement the
  model in an extensible database system

23
Thanks