• Proceedings of the Postgraduate Annual Research Seminar 2006 195

Sponsored Links

  • FileName: Object-relational_Features_for_modeling_and_analysis_of_Spatio-temporal.pdf [preview-online]
    • Abstract: Proceedings of the Postgraduate Annual Research Seminar 2006 195Object-relational Features for modeling and analysis of Spatio-temporalDataA.Z.M Kamruzzaman, Mohd Taib Wahid

Download the ebook

Proceedings of the Postgraduate Annual Research Seminar 2006 195
Object-relational Features for modeling and analysis of Spatio-temporal
A.Z.M Kamruzzaman, Mohd Taib Wahid
Department of Information Systems, Faculty of Computer Science & Information Systems
University Technology of Malaysia, K.B. 791
81310 Skudai, Johor, Malaysia
Tel: (607)-5532422, Fax: (607) 5565044
[email protected], [email protected]
Abstract fauna and peat swamp. Malaysia has been
loosing much of its natural resources including
The age of discovery, description, and plants and animal species through ecosystem and
classification of biodiversity is entering a new degeneration.
phase. In responding to the importance and uses
of the large number of biodiversity data of Malaysia has been endowed with vast amount of
Malaysia make it essential for their biodiversity natural resources including luxuriant tropical
to be conserved. In order to manage large forest which is one of the most diverse and
number of biodiversity data, it is essential to complex ecosystems of the world. Malaysia is
develop a data model which is a new integrated rated as one of the world's 12 "mega diversity"
and support spatio-temporal data. Recently, we countries. It boasts over 150,000 species of
propose a conceptual spatio-temporal object- invertebrates, 286 mammal species, 736 bird
relational biodiversity data model and which is species and 15,000 types of flowering plant.
being tested with Malaysian biodiversity data Forest resources have been one of the major
and has been loaded to oracle database. The main sources of revenue in the Malaysian economy.
objectives of this paper are new aspects of using However, it is decreasing every year. The change
object-relational feature of the underlying of land covered by forests has been reduced from
database management system Oracle to improve 65.9% in 1990 to 59.5% in 2002 [1].
the extensibility and flexibility of data models,
enhance interoperability and analyzing potential Different types of forests can be found in the
as well as to ensure consistency by defining Peninsular, Sabah and Sarawak. From the
standards based on abstract data types. The management perspective, forests of Peninsular
object-relational extension of data models can be Malaysia can be classified into dipterocarp, peat
combined with data pyramid framework swamp and mangrove forests. Among which
techniques to analyze present, past and future approximately 95% is covered by dipterocarp
data. In addition to this, by using this model forests, 3.34% by peat swamps and 1.84% by
forest department, researchers, as well as natural mangrove forests (UNEP, 2002). In Sabah, there
recourses will be helpful. is a gradual succession of forest vegetation from
the coastal beach forests and mangrove forests to
Keywords: Biodiversity, Spatio-temporal, lowland dipterocarp forest and eventually
Object-relational, Oracle 9i montane forests. In Sarawak, five types of
natural forest types are abundant namely: Hill
Mixed Dipterocarp Forest, Peat Swamp Forest,
1. Introduction Mangrove Forest, Kerangas Forest (Heat Forest)
and Montane Forest (UNEP 2002). Currently
Biodiversity, or biological diversity, refers to the there are total 610,606 hectares of mangrove
variety of life on earth and the most important forest estimated in Malaysia. The largest
factors Influencing the stability and health of concentration are found in Sabah (56%) followed
ecosystems. Information about biodiversity and by Sarawak (27.5%) and the remaining in
ecology is vital to a wide range of scientific, Peninsular Malaysia (16.5%) [2]. However, this
educational, industrial, governmental activities. estimation is still on process and areas in Sabah
Malaysia is undoubtedly on the richest rain and Sarawak have not been actively surveyed
forests in the world with diverse floristic botanically. Based conservation, management,
composition and complex ecosystem. It is not data analysis, data accessing, monitoring and
only rich for plant biodiversity data also rich in
Proceedings of the Postgraduate Annual Research Seminar 2006 196
data complexity of biodiversity data, data model develop a new approach to develop biodiversity
is required to protect biodiversity data. data model. In the following a brief introduction
of object-relational modeling concepts is given.
After almost two decades research, As the implementation is based on an ORACLE
representation of space and time in databases and 9i database, the corresponding SQL-dialect is
functional applications are still problematic [8]. followed. Object relational approach with its
Problem with previous data models is that spatial characteristic of inheritance and aggregation is
and temporal aspects of databases are modeled capable of capturing the various notions of space
separately [9], [10]. Spatial database focuses on and time and reflecting them into a single
supporting geometries [11], while temporal framework extensible to different applications.
databases focus on the past state [12].
2. Spatio-temporal Data Modeling
Another problem is representation of data should Approaches
be natural to human. The structures of space and
time are identified as essential for the realization
of cognitive systems [13]. According to Donna J.
2.1 Pyramid framework and event
Peuquet and her group [14], models of spatial based approach
temporal data in geographical database In the design of object-relational biodiversity
representations must incorporate human data model to support spatio-temporal data,
cognitive principles. Human knowledge of the pyramid frame can make enhancement to support
dynamic geographical world comprises of three a biodiversity data model design. A conceptual
different (and interrelated) subsystems that framework (also known as pyramid framework)
handle what, where and when aspects of object was designed which guide the implementation of
properties [15]. the semantic geographical information system
(GIS) data model [23]. Conceptual frame work
Recent years have seen an explosion of digitally (pyramid) is interrelated with two separate parts
available information about biological diversity one is data component and another is knowledge
[3]. Data management tools and software need component. Data component can be divided into
more sophisticated facilities to face new three categories: location (position in the spatial
requirements from emerging application areas three dimensions), time (position alone a time
and non-traditional user interactions. In line) and theme (a set of observations,
particular, better concepts and tools for measurements, or attribute values associated
manipulating spatio-temporal data are needed. with a particular location and time). Data
Major DBMS tools are incorporating facilities components stores spatio-temporally referenced
for spatial or temporal data management (e.g., observational data such as spectral, climate,
Oracle’s Spatial Cartridges and Informix’s vegetation and other environmental attributes
Databases). Temporal systems are still somehow that maybe queried and visualized to reveal
behind, with no generic products on the embedded spatio-temporal pattern and
marketplace, just a few ad hoc systems or relationships. Data component can be
application-specific developments (e.g., for time computerized as a multi-dimensional, spatio-
series management). However, current tools do temporal referenced ‘hyper-cube’ of
not match the user perception of and reasoning observational data that is similar to the ‘feature
about the application data space’ concept commonly cited in analysis of
remote sensing imagery.
So far no data model for analysis of biodiversity
data has been performed using GIS and database Knowledge component stores information about
technology, particularly in regard to new higher-level semantic ‘objects’ the geographic
possibilities provided through object-relational entities or process that are describe by the data.
features. Since database models get closer to real Information concerning on object’s location,
world objects and events by allowing the setup time and composition. All the objects are also
of an integrated and adjusted representation placed within two hierarchical relationship
based on user-defined types. A complete structures central to cognitive knowledge
description of spatial objects in database systems representational and object-oriented modeling,
can be provided for the first time. Therefore it is plant taxonomy (generalization) and partonomy
obvious to make use of object-relational (aggregation). The taxonomy structures groups
structures provided by the underlying DBMS to similar objects within a category and stores a
Proceedings of the Postgraduate Annual Research Seminar 2006 197
rule-base that describes how those objects may Location attributes record space information of
be identified within the data space. These rules the object, such as whether it is a point, line, or
may be derived from expert knowledge or from polygon, and its location. Location attributes can
inductive analysis of the observation data. In an be inherited into subclasses such as the point
object-relational data model consists of a set of class, line class and polygon class. Time
object classes (of different types or schemas). attributes are represented through timestamps
Each object class has an associated set of associated with theme and location information.
objects; each object has a number of attributes When properties or forest information of an
with values drawn from certain domains or object changed, a new timestamp will be
atomic data types. Of course, there are additional recorded for that object. That means only initial
features, such as object valued attributes (Oid), information and changed information will be
methods, object class hierarchies, etc. Besides recorded. Each timestamp is an instant. Interval
objects, attributes describing geometries is the time between two timestamps
including time are of particular interest. Hence
we would like to define collections of abstract In this model, theme, spatial, time attributes are
data types, or in fact many-sorted algebras recorded as 3 tables. Each attributes use one list.
containing several related types and their For example, a flora object has theme attributes
operations, for spatial values changing over time. as “Name”, “FloraID”, “FloraType” and
This section presents a simple and expressive spatial attributes as “Area”. During different
system of abstract data types, comprising data season and time of a year, the volume of the lake
types and encapsulating operations, which may will change, and the border of the lake expands
be integrated into a query language, to yield a or contracts. Whenever the volume or the border
powerful language for querying spatio-temporal of the lake changes, a timestamp will be
data. recorded. Thus, as shown in Figure 1, each
object has a name list, a value (1 to N+1), a
Location has area code (1, 2, 3 and n+1) and a
timestamp or events (Day, Month, Year, Second,
minute, hour). Data can be collected same day
same time from different area by different data
collectors. Sometime data collects from different
forests for several days, months and years. Time
stores as an events records as EventID,
StartTime, Duration and Attribues (types of
data). Diagram bellow indicates fields that can
be used to associate objects between tables.
3. Object-relational Features
As we know that within the classic relational
database model there are only scalar but no
complex data types. With the introduction of
Figure 1: Relationship between theme, space and time object types the definition and composition of
abstract data types is possible. An abstract data
Theme attributes record the information of “what type can be comprised of a multitude of scalar
is this object” and other related property types and again of user-defined complex types
information. For example, if the theme of this and enhances consistency when creating
object is “Forest”, a Forest must have attributes database models.
such as Area or Province it belongs to, Forest
Size, and Forest Type, etc. Theme can be create or replace type Address_TY as object (
inherited according to certain classification City VARCHAR2 (30),
hierarchy. For example, theme can be classified Sate VARCHAR2 (30),
as natural resources (forest, earth quake, and Area_code NUMBER (8),
rainfall), transportation (rail and highway), Country VARCHAR2 (30));
settlement (country, city), water system (river,
lake, and pool), and cadastre etc.
Proceedings of the Postgraduate Annual Research Seminar 2006 198
Here create or replace the Address_TY data type In the classical Entity-Relationship-Model
after that create Org_Address_TY by using aggregations and compositions are modelled
Address_TY data type. through master-detail-relationships. Object-
relational dbms provide collection types that
create or replace type Org_Address_TY as contain multiple elements and thus are suitable
object ( to express 1: N relationships directly. Each
Org_name VARCHAR2 (30), element or value for a collection has the same
Address Address_TY, substitutable data type. The most popular
Org_url VARCHAR2 (50), collection types are varrays and nested tables.
Phone NUMBER (12),
Email VARCHAR2 (30)); A varying array contains a variable number of
ordered elements. Varying array data types can
Data types only represent descriptions of data be used as a column of a table or as an attribute
structures. To ensure persistency a table must be of an abstract type. Named table types can be
bound to the data type. A relation with column created in an Oracle database using SQL. These
objects can be created that is a set of table types can be used as nested tables to
Observer_ID with Observer_name, provide the semantics of an unordered collection.
Observer_org and Observer_Address as: As with varray a nested table type can be used as
create or replace type Observer as object ( a column of a table or as an attribute of an object
Observer_ID VARCHAR2 (15), type.
Observer_name VARCHAR2 (30),
Observer_org VARCHAR2 (50), Named table types can be created in an Oracle
Observer_Address Org_Address_TY ); database using SQL. These table types can be
used as nested tables to provide the semantics of
Object views allow user to treat relationally data an unordered collection. As with varray a nested
as they allow synthesizing object from data that table type can be used as a column of a table or
continues to be stored in relational tables. Object as an attribute of an object type.
views have similar functionality like object
tables. They can have methods, belongs to Create type Address_TY as table of
collections, reference one another, have object Org_Address_TY;
identity and can be accessed from SQL. In
addition tables are assigned to object views can Multi-Level-Collections that lead to multiple
be uploaded by using special instead of triggers. nested tables can be realized if useful for
Any instance (row) of an object class contains a applications but it’s up to the user to balance – a
unique ID called Object_ID (ID). Generally more intuitive representation of data vs. higher
OIDs are system-generated but can also be complexity of accessing the data.
derived from a primary key column or can be
user-defined. An object type declaration can also include
methods that are defined on values of that type.
Relationships between objects can be defined When using these objects types in tables their
using reference types. A reference column stores methods are also applied to the data of these
OIDs of associated (row) objects since column tables. The method is declared in the create type
objects do not have inherent OIDs and therefore command and the code for the function itself (the
cannot be referenced. Row objects that belong to definition of the method) is in a separate create
a reference can be selected and dereferenced type body command.
using the DEREF VALUE operator. Modeling
object relationships with OIDs and REFs is often create or replace type Taxon_TY as object (
compared with foreign key relationships inside Taxon_Lname VARCHAR2 (20),
the relational model but implicates some benefits Tanon_Ename VARCHAR2 (20),
like the ability to distinguish between equal and Taxon_Species Species_TY,
identical objects. Objects are identical if they Taxon_family Family_TY,
have one common OID. They are equal if they Taxon_area Area_TY,
have different OIDs but coincide with their Taxon_description VARCHAR2 (50),
attributes and values. Taxon_start_date DATE,
Taxon_end_date DATE,
Proceedings of the Postgraduate Annual Research Seminar 2006 199
Member function DURATION (start_date, in xml-documents, all stored natively within the
end_date in Date) return Number); database.
Names the function that is a “member” of the For modeling highly dynamic changes of real
Taxon_TY datatype. Since Duration will return a world objects the SQL type TIMESTAMP can be
value, it is a fuction. To define DURATION used which is a high-precision time and date type
function, use the create body command, whose that allows storing fractions of a second [5]. The
full syntax is shown in the Alphabetic Reference. timestamp values are basically points on a linear
time axis. For slower changes in state we can use
Let’s create the Duration function as a member DATE or create own adapted abstract data types
function within the Taxon_TY datatype. The that can express discrete dates or times intervals.
DURATION function will return duration, in The data types can be flexibly extended to meet
days, of the taxon. an application’s accuracy requirements and
provide attributes to express the fuzziness of a
create or replace type body Taxon_TY as time record, for instance if a time interval cannot
member function DURATION (start_date, be determined as accurate as the data type could
end_date DATE) store [6].
return NUMBER is
begin create or replace type TIME_TY as object (
return totaldays (start_date + end_date ); year NUMBER (4,0),
end; month NUMBER (2,0),
day NUMBER (2,0),
The following query executes the duration hour NUMBER (2,0));
method in the data type Taxon_TY and returns
the duration of all collection objects in relation create or replace type TIME as object (
Observation. Values of components of an object start_date TIME_TY,
(attributes and methods) are accessed with the end_date TIME_TY,
dot notation. start_date_per NUMBER (1,0),
end_date_per NUMBER (1,0));
select Taxon_ID, O.Taxon_Area, Duration
(O.start_date, O.end_date) 4. Object-relational Application
from Observation O ; Development
Object types can be mapped to the prevalent OO In terms of Biodiversity data of Malaysia, object-
languages like C++ and Java. Thus object type relational database technology can be used to
instances in the database can be accessed and model different applications, e.g. the spatio-
modified to and from C++, Java applications [4]. temporal process of flora or fauna based on its
influencing factors or real time deer tracking.
In the current version Oracle database 10g native
data types for storing raster and persistent For the development biodiversity data model
topology data were added to further extend some major influencing factors have been proved
possibilities to model real world objects based on to be crucial. In the model we have judged like
object-relational database structures. The data should be natural to human. The structures
completeness and therefore the quality of spatial of space and time are identified as essential for
features’ descriptions can be further enhanced the realization of cognitive systems. Models of
with the introduction of a new system-defined spatial temporal data in geographical database
datatype XMLType that can be used as the representations must incorporate human
datatype for columns in tables and views to cognitive principles. Finally model support
create, extract, and index XML data. A feature’s human knowledge of the dynamic geographical
position and spatial representation is then world comprises of three different (and
modeled by using vector resp. raster types, its interrelated) subsystems that handle what, where
characteristics are captured by (complex) and when aspects of object properties.
attributes and any further information on the
spatial object, that cannot be structured to be create or replace type Position_TY as object (
kept in the data types mentioned, is maintained Map VARCHAR2 (30),
Proceedings of the Postgraduate Annual Research Seminar 2006 200
Grid25 NUMBER (4), contained in a relation column and computes a
Grid100 NUMBER (4). new evolving region. Internally, this operator is
based on a binary function MAXST applied to two
Create or replace type SpatialData as object ( evolving regions R1 and R2 'and yielding a new
Forest_ID NUMBER (6), evolving region in the following way:
Forest_name VARCHAR2 (30),
Forest_area AREA_TY, MAXST (R1, R2):= {(t, r) |t Є time r MAXgeo (R1
Forest_position Position_TY); (t), (R2 (t))}
This definition uses a function MAXgeo which is
Based on this type an object view is created and applied to two regions R1 and R2 and which
OIDs are assigned to the spatial datasets of the returns larger of both regions.
corresponding forest area: ⎧ r1 If area (r1) > area (r2)
MAXgeo(R1, R2)= ⎨
create or replace view Forestdetails_OV ⎩ r2 Otherwise
of Position_TY Altogether this means that for 2 evolving regions
with object identifier (ForestID) as R1,…, Rn we first compute the evolving region R
select Forest_name, position, = MAXst (R1,….MAXst(Rn-1, Rn )..). Afterwards,
from Forest_area we apply the raise area of R, which computes the
where AreaName = Johor; area of R at all times as a temporal real number.
The columns of the relational base tables are
5. Conclusion
now accessible as row objects through their
corresponding object views. Any detail tables of
Object-relational technology in DBMS helps to
the underlying relational model could now be
model and analyze spatial-temporal events as it
simulated by further object views with references
provides highly flexible means for storing,
that could point to the row objects of the
manipulating and validating diverse and complex
particular upper object view, to allow accessing
data structures. As for applications in Malaysian
the data in either way.
biodiversity data, the real challenge is to
understand the relationship of the most complex
Projection operations on moving objects map
processes and find definite and describable
either to their spatial or to their temporal aspect.
entities that can be part of a database model.
Assume that we are interested in the geometric
Thus it is often useful to extend the concept of
locations where the data was changed at the year
objects from a conventional point of view to a
January 20, 1990. These can be obtained by:
more general sense, so to speak to describe real
SELECT Flora_ID, AreaName
world scenes (instead of objects) that show on
FROM Flora, Area
their part some correlation with a set of
WHERE (((Forest_area =#Johor# AND
commonly used objects. To achieve a smoth
interplay between the embedded language and an
embedded system of abstract data type, a few
This operation computes the spatial projection of
interface facilities and notations are needed,
a spatio-temporal object for the Johor forest area.
expressible in one form or another in most
For an evolving region the trajectory operation
object-oriented or object-relational query
returns an object of the spatial type region which
results from projecting the union of the region
values for the Date 20th January, 1990.
6. References
The following query inquires about the largest
collection of flora areas. [1] Thang, H.C. (2004). Forest Management and
SELECT Area (max (Extent)) FROM Development in Malaysia. Paper presented at the
ForestArea Malaysian Timber Council’s Familiarisation
WHERE Type = "Flora" Programme for European Trade Representatives,
The query demonstrates an example of a spatio- 12 July 2004, Kuala Lumpur, Malaysia.
temporal aggregation operation max which is an Unpublished. UNEP (2002) Malaysia-Study
extension of the well known aggregation Area. http://www.rrcap. unep.org/lc/cd/
operator in SQL of the same name. It is here
applied to a collection of evolving regions
Proceedings of the Postgraduate Annual Research Seminar 2006 201
[2] Jeremy L, Mernnis, Donna J. Peuquet, [15] Mennis, J. L., D. J. Peuquet, et al. (2000).
Diansheng Guo, (2000), “A Semantic GIS “A Conceptual Framework for Incorporating
Database Model for the Exploration of Spatio- Cognitive principles into geographical database
temporal Environmental Data”, GIS/EM4, representation.” International Journal of
Alberta, Canada, September 2-8, 2000 Geographical Information System 14(6): 501-
[3] Bisby, F.A. (2000), “The Quiet Revolution:
Biodiversity Informatics and the Internet.” [16] M. Bouzeghoub, “Morse: a functional query
Science 289, 2309-2312 language and its semantic data model”, proc. of
int. Conf. on Trends an applications of databases,
[4] Lee, G., (2003), “Simple Strategic for IEEE-NBS, Gaithersburg, USA, 1994
complex Data: Oracle 9i Object-relational
Technology: An Oracle Technical White Paper”,
October 2003.
[5] Qian, L., (2004), “Oracle database 10g-
Development Spatial Application Using Oracle
Spatial and Map Viewer: An Oracle technical
White Paper, February 2004
[6] Plabst. S., (2001), “Object-relational Data
models for Geo information System”, Technical
University Munchene
[7] Kevin Loney, George Koch, (2003),
“Oracle9i: The complete reference”, McGraw-
Hill Publication, USA.
[8] Langran, G. (1992), “Time in Geographic
Information Syetems:, Taylor & Francis, London
[9] Sellis, T. (1999). Research issues in
Spatiotemporal Database Systems, Springer-
Verlag Berlin Heidelberg.
[10] Martin Erwig, R. H. G., Markus Schneider,
Michalis Vazirgiannis (1999). “ Spatio-Temporal
Data Types: An Approach to Modeling and
Querying Moving Objects in Databases.”
Geoinformatica 3(3): 269-296.
[11] Guting, R. H. (1994). “An Introduction to
spatial Database Systems.” VLDB Journal 4:
[12] Tansel, A. U. and J. Clifford (1993).
Temporal Databases: Theory, Design, and
Implementation, Benjamin/Cummings
Publishing Company
[13]Freksa, C. (1998). Spatial and Temporal
Structures in Cognitive Processes, Springer
Verlag Heidelberg.
[14] Peuquet, D. J. (2001). “Making Space for
Time: Issues in Space-Time Data
Representation.” GeoInformatica 5(1): 11- 32.

Use: 0.381