Excerpt
CONTENT
I. INTRODUCTION
II. STATE OF THE ART
III. THE LAYING DATA MODEL
IV. REASONING OVER GIS APPLICATIONS
V. CONCLUSIONS
REFERENCES
Abstract — Geographical Information Systems (GIS) research area have been evolving with time. Those systems have become useful beyond spatial and geographic information representation and computer aided analysis using maps. Some of the most important fields of application for GIS are fleet control, tourism analysis and Meteorological analysis. For the last mentioned field, advisory and prediction models should be enhanced aiming the avoidance of critical damage associated to Hurricanes. Intelligent systems have become the optimal solution when decision-making situations are extreme and advanced reasoning is expected. Existing models involve high mathematical analysis, which is complex for humans and also sometimes computational costs have to be considered. There are some artificial intelligence fields that are exploring the possibility of making inference about existing data, enriching the information and enhancing the obtained results. In the present work we propose a hybrid GIS which includes some behavioral aspects of intelligent systems for Hurricane analysis.
Keywords – GIS, Hurricane Analysis, Intelligent Systems, Ontology, Movement Semantics
I. INTRODUCTION
The analysis of hurricanes and their impact in determinate areas is mandatory if we have to preserve human life and economic resources. To precisely predict how will evolve and behave one tropical storm in time is the key for meteorologists on this venue. Even we have precise mathematic and geo-graphic models for understanding this phenomena, the diverse nature and the high number of variables that interact in the life cycle of one hurricane makes extremely difficult to say we are exact in prediction, even more with a minimum probability of failure. The answer lies in the ”understanding of hurricanes as they mean”, taking into account what origin them, which factors have influence in their evolution and behavior and what features exactly characterize all them in general or what they have in common. Then we are considering semantic analysis. Moreover, we have to pay special attention to the analysis of the motion semantics, especially to the representation of movement of spatial entities and the associated reasoning. All these spatial objects that change their position or shape over time are called moving objects [14].
The analysis of motion semantics has become in an impor-tant issue for the correct representation and subsequent rea-soning about moving objects in the world (trajectory analysis, movement patterns recognition, movement prediction). When we represent planes or vehicles for tracking purpose, we can talk about moving points. For meteorologists hurricanes (that move or change their extension) can be represented by moving regions. Finally, when we represent or track something like a moving fleet of ships or convoys, which can increase or diminish their extension, we have then a moving line. Most of these application fields are extremely important and some of them need accurate information in order to avoid unwanted disasters, for example aircraft tracking.
Some important proposals are: [16], [17], [10], [3], [13], [12], [19], [1], which we classify as foundational or visionary papers. Many of these papers deal with temporal and spatial semantics separately, even when describing the present prob-lem. On the other hand, there is another point of view for the representation and reasoning of spatiotemporal objects seman-tics: using logical and mathematical analysis (e. g. description logics, fuzzy, math algorithms, qualitative reasoning).
In this work we present an adaptation of a GIS application that have the capability of reasoning and inference using an ontological model. The proposed model helps on enhancing the obtained results in a heavily critical field related to Hurricane tracking and analysis.
This document is structured as follows: After the introduc-tion, section II describes the state of the art and related works. Then, section III explains the data model characteristics and importance. In section IV, the current proposal is presented, describing the reasoning process ans the system main com-ponents and, finally, section V shows conclusions and some aspects for been considered in the future.
II. STATE OF THE ART
One common direction followed by researchers and scien-tists today is to use Qualitative Reasoning (QR) and Qualitative Spatial Reasoning (QSR). In this last group is the Region Connection Calculus (RCC). The original idea of this work is found in [11], but we will focus on [4]. In RCC Cohn et al. take regions as primary spatial elements. RCC contains axioms that allow us to deal with spatial and physical objects, analyzing their relationships (e.g. containment relationships) and shape. Concerning moving objects, there is an approach called reasoning about continuous change, which tries to explore the semantic inside the relationship between regions (e.g. inside, outside).
This is an interesting way of tracking the approximate movement of regions using a set of snapshots of the relation-ships between them and analyzing the continuity. The main problem here is the lack of temporal semantics consideration, but for the mentioned Interval Temporal Logic, given that the authors consider that it closely mirrors the region-based spatial reasoning presented in RCC. Moreover, the complete abandonment of the point-based reasoning leaves out one critical element for the analysis of moving objects: moving points. On the other hand RCC provides many useful axioms, mainly associated with relationships between regions, which should help in the development of a future improved frame-work for the analysis of evolving entities trajectory, location management and other geographical reasoning, all this given the region-like shaped representation of spatial entities on a map (e.g. forest, buildings, cities).
The Qualitative Constraint Calculi (QCC) is another kind of qualitative reasoning. It has been applied in some areas for qualitative spatial reasoning, representing binary relationships between spatial elements like constraints and analyzing consis-tency or satisfiability of a given fact, and also for discovering information implicit in constraints. One clear example of the usage of QCC for our purpose is [7], where Gantner introduces a useful tool based in the Qualitative Constraint Calculi.
Finally, the most interesting and recent paper concerning the area of representation and reasoning about trajectories and tracking objects is The Hybrid Model [18]. The main goal that Yan et al. planned to achieve was to have a generic model that allows enriching semantically the trajectories obtained from GPS or any other mobility feeds. For this purpose, they provide a computing platform that enable the semantic enrichment of mobility data. Three structural logic components (Data Model, Conceptual Model and Semantic Model) are in charge of analyzing GPS feed sequences associating them by time intervals, establishing correlations between some features and finally adding knowledge that could be, for example, socio-political or geographical information. The obtained semantic concepts of mobility data could be used for reasoning in an ontology-based approach. The concept of ontology became stronger in the 90’s, when Gruber defined ontology as an “explicit specification of a conceptualization” [9].
From that moment, ontologies became an important ele-ment inside the Artificial Intelligence field and still are today. Ontologies were first proposed in order to provide a common vocabulary for communication between systems and systems with persons. Over the years, they have become useful in many other application scenarios, including some approaches in the spatial and temporal field. The surging of ontology reasoners and rule based approaches have enabled the possibility of creating new information based on inferences using existing one. This heavily helps on enhancing the further analysis.
In this vein, some works for been considered are: [8], where Grenon presents the basic elements of a framework for spatio-temporal knowledge representation and reasoning. This framework helps on facing endurants and perdurants diverse modes of existence. Separated modules consider the semantics of both classifications. Some given axioms consider mereological and mereotopological features, intra-ontological and trans-ontological relationships. The establishment of these relationships between ontologies is an interesting goal for this proposal, given that we too consider that the integration of existent ontologies is a viable solution to the existing problem; [6], where Frank establishes a discussion about spatio-temporal proposed ontologies and methods for constructing them and their limitation and problems, aiming to find better proposals, suggest improvements and find a consistent ontology for spatio-temporal GIS. The proposed ontology deeply analyzes metric and topological properties, projections, snapshots and other topological and mereological elements and [15], where a clear idea of an ontological representation for moving objects is given. The main purpose of Tryfona and Pfoser is to allow interoperability between moving objects applications and, even so, important goals are achieved. They also propose architectural issues for mobile ontology-based applications. This can be considered a starting point for further development. Based on this approach, some improvement is needed, mainly in those aspects considering regions mobility and moving lines cases.
Related to GIS area is the rising of the Ontology Driven Geographical Information Systems (ODGIS), as stated in [5], with a tremendous importance. But most of existing systems only consider spatial and geographical semantics (with all as-pects it involves), nothing included from the needed movement semantics, totally focused in the spatio-temporal domain. The solution could be to think in an enhanced kind of ODGIS, starting from new spatio-temporal ontologies for the moving objects domain, not only the existing spatial or geographical ones.On the other hand, no intelligent-driven approach is assumed, which have a lack on accuracy and applicability.
[...]
Comments