Project Topics
Get Complete Project Material File(s) Now! »
Event-based models
A significant amount of spatio-temporal modelling research has been conducted over the past few years towards event-based models (Allen, 1995; Worboys, 1995; Claramunt & Thériault, 2002). A survey published by Allen (1995) shows that events are useful abstractions, and should be modelled as entities in order to further explore the notion of causality. According to Galton (1995), a theoretical basis is necessary to describe events regarding particularly the representation of the conditions for their occurrence. It introduces a framework for formalizing our “common-sense” knowledge of the real world. Allen et al. (1995) also introduced the concept of causality as a modelling abstraction that can be integrated into a temporal GIS, and developed a conceptual data model using an entity-relation formalism. This modelling approach identifies the specific elements that can contribute to the search of a causal theory, and the relationships between them.
Story & Worboys (1995) introduced a spatio-temporal model that considers events as a main primitive object associated to spatial, temporal and thematic attributes defined as basic components of that object abstraction. Claramunt & Thériault (1995) proposed an event-based conceptual model in which time is regarded as a complementary dimension that favours the representation and qualification of processes, thus favouring extensibility and manipulation at the query level. Such event-based modelling approach has been in particular applied to the representation of geo-lifelines (Claramunt & Thériault, 2002). Another example is the spatiotemporal scheduling framework introduced by Stewart et al. (2013). This framework is based on an ontological approach that identifies key process-based and event-based classes as well as the relations that connect them. These classes are considered as either continuants modelling real-world entities or occurrents capturing dynamics and events. The framework links a domain ontology (e.g., campus events) with an application ontology (e.g., scheduling), and a task ontology which allows individuals to plan and manage their schedule on a daily basis.
Vector-based models
One of the logical data models most often applied in geographical space is based on the so-called vector model (Hallot, 2007; Noyon, 2007; Miller, 2009; Liu, 2014). The vector-based data model is particularly adapted to represent spatio-temporal data such as trajectories. A logical data model designed as a relative representation for trajectories has been introduced by Noyon et al. (2007), in which the trajectory of a given mobile entity is defined by its relative speed distance according to a reference entity. This method is well adapted to capture the semantic of the movement of an entity with respect to a second one considered as an entity of interest. Liu (2014) suggested a vector-based model based on Time Geography concepts and a preliminary contribution of Miller & Bridwell (2009). This model can be applied to estimate the density of similar movements, and in order to extract moving patterns. Overall, vector-based models provide efficient frameworks to capture the spatial semantics of geographical phenomena, but they should be closely associated to the conceptual level and additional descriptive data in order to provide a more complete representation framework.
Raster-based models
Another approach that can be applied to represent trajectories, although less intuitive, is the raster model roughly defined as a representation, by most often a regular grid, of a continuous phenomenon. For instance, Peuquet (1995) suggested an Event-based Spatio-Temporal Data Model whose objective is to represent change patterns over time. Changes identified at a given time are represented at the primitive levels and related to cell locations. In other words, locations associated to some common semantic values are stored together as members of a same component. Another example is the one of Cellular-Automata (CA) models that have been widely used to model and replicate land-use evolution using combination of stochastic evolution rules, probabilities and comparison to observed data (Wu, 2002; Menard, 2005). These methods when applied to possible evolution scenarios can be cross-compared with projected land demand. Raster-based approaches can be further extended towards three-dimensional representation frame where the third dimension represents the temporal dimension, and also combined with the vector-based model (Liu, 2015). The model is used to estimate a population demand vector field using a vector kernel density estimated from observed trajectories of a sample population (Figure II.13).
TRAVEL DISPLACEMENT MODEL
The conceptual database schema of human travel displacements is made of several schemas closely connected (so-called modules hereafter). Each one of the three modules identified plays a specific role regarding the analysis of travel displacements. The design of the model is inspired by the trajectory ontology proposed by Yan et al. (2008) and the construction of the model is designed according to the principles of the MADS model developed by Spaccapietra & Parent (2006).
The conceptual schema comprehends three complementary parts designated as modules:
• The Trajectory module is an important component of a human displacement. The Trajectory module abstracts the movements of human beings in the city. Since a large part of people outdoor movements generally happen along the transportation network, we first postulate that the concept of trajectory is closely connected to a network. Figure III.1 shows an example of trajectory decomposed in travel segments qualified by different transportation modes and an overall activity which is ‘going to work’.
• The Transportation Network module represents all kinds of network that facilitate human movements inside the city.
• Next, and as to explore human travel displacements, the different factors that generate moving patterns are classified and organized in a module denoted.
Relationships between Modules
Relationships are essential abstractions used to design the Travel Displacement schema. Most often connections between the different modules are materialized by specific relations that associate the different modelling abstractions identified. For instance, the Follow relationship between the object type Route and the object type Trajectory is a key abstraction of the TravelPattern module (Figure III.6), such as the temporal relationship Compose between Route and TransportationSegment. Therefore, Route is the object type that implicitly connects the three modules. The cardinality between Trajectory and the line Follow is (0, n). This is due to the fact that a Trajectory might not follow a Route, while on the contrary a Route should at least follow one Trajectory. As a Route is made of a list of TransportationSegments, the cardinality from Route to TransportationSegment is (1, n), while from TransportationSegment to Route is (0, n). A TravelPattern is characterized by different factors in the TravelPattern module. As for the relationship between the Travel Pattern Module and the Trajectory Module, multiple relationships have been modelled. The relationship between the Trajectory Module and the Travel Pattern Module denotes the indirect relation between TravelPattern and Individual, and finally represents the fact that every Trajectory belongs to an Individual. Indeed, the Individual abstraction plays a crucial role when modelling travel displacements, and it can be seen as a dominant abstraction when modelling a travel pattern. One Individual can own one-to-many Trajectories to reflect some travel patterns, his/her travel decisions being the result of different constraints. A District is in spatial relationship with the Transportation Network. Considering the possible topological relationship for instance between TransportationSegments and Districts, it appears that the possible relationship include Touch, Within and Intersect. These relations are summarized in the database schema by the topological relation TopoGeneric (Figure III.7). POIs also play an important role, as Trajectories, POIs reflect source, intermediate or destination nodes of a given trajectory. A POI is connected by an Equal topological relation with a Node, but with the respective POIs and Node sets being disjoint. Finally, POIs are likely to bring many opportunities for further analysis when the whole model is logically implemented and experimented with real data.
Table of contents :
CHAPTER I: Introduction.
I.1 Problem Statement
I.2 Aim and Objectives
I.3 Approach Overview
I.4 Contributions
I.5 Thesis Structure
CHAPTER II: Related Work
II.1 Introduction
II.2 Human Travel Patterns
II.3 The Concept of Trajectory
II.4 Trajectory Data Model
II.4.1 Conceptual Models
II.4.2 Event-Based Models
II.4.3 Process-Based Models
II.4.4 Logical Models
II.4.4.1 Vector-Based Models
II.4.4.2 Raster-Based Models
II.4.5 Novel Approaches
II.5 Summary and Discussion
CHAPTER III: Travel Displacement Modelling
III.1 Modelling Research Objective
III.2 Modelling Principles
III.3 Travel Displacement Model
III.3.1 Trajectory Module
III.3.2 Transportation Network Module
III.3.3 Travel Pattern Module
III.3.4 Relationships between Modules
III.4 Discussion
CHAPTER IV: Travel Displacement Database and Data Analysis
IV.1 Database Implementation Objectives
IV.2 Travel Displacement Database Principles
IV.3 Travel Displacement Analysis
IV.3.1 Data Analysis Categories
IV.3.1.1 Semantic Level
IV.3.1.2 Temporal Level
IV.3.1.3 Spatial Level
IV.3.1.4 Spatio-Temporal Levels
IV.3.2 Data Analysis Functions and Queries
IV.4 Discussion
CHAPTER V: Implementation & Experimental Evaluation
V.1 Implementation Objective
V.2 Experimental Data Set
V.3 Computing and Software Environments
V.3.1 Development Environment
V.3.2 Database Environment
V.4 Framework Construction
V.4.1 Data Pre-Processing
V.4.1.1 Construction of the Transportation Network
V.4.1.2 Map Matching
V.4.2 Database Construction and Analysis
V.4.2.1 Data Structure Generation
V.4.2.2 Route Extraction Algorithm
V.5 Results and Discussion
CHAPTER VI: Conclusion & Perspectives
VI.1 Outline
VI.2 Main Contribution
VI.2.1 Conceptual Modelling
VI.2.2 Logical Design
VI.2.3 Data Manipulations & Functions
VI.2.4 Implementation
VI.2.5 Travel Behaviour Exploration
VI.3 Perspectives
VI.3.1 Modelling Extensions
VI.3.2 Data Analysis Extensions
VI.3.3 Data Manipulation Extensions
VI.3.4 Decision Making & Travel Prediction
APPENDIX A: Résumé étendu de la thèse
A.1 Introduction
A.2 Approche de Modélisation
A.2.1 Modèle de Mobilité
A.2.2 Module Trajectoire
A.2.3 Module Transportation Network
A.2.4 Module Travel Pattern
A.2.5 Relations entre Différents Modules
A.3 Analyse de Patrons de Déplacements
A.3.1 Interrogat ions de comportements de déplacements (Travel Displacement Queries)
A.4 Implementation et Validation
A.4.1 Conception de Fonctions
A.4.2 Données Expérimentales de Trajectoires
A.4.3 Extraction des Trajectoires
A.4.4 Résultats
A.4.5 Evaluation des Performances
A.5 Conclusion
APPENDIX B: List of Publications
APPENDIX C: Short Curriculum Vitae
REFERENCES
