Patel

Nature and Application of Databases for Spatial and Temporal data

By: Gargi Patel (MSc. Computing For Business) Student no-05903617 Module Leader: Andy Jones Subject: Database Administration and Management Subject Code: CE53701-3

TABLE OF CONTENTS

INTRODUCTION………………………………………………….…1 APPROACH……………………………………………………….….2 RESEARCH………………………………………………………......3 1. Spatial and temporal data…………………………………...………………3 1.1 Conventional Vs. Spatial and temporal data ………………………………3 2. Characteristics of spatial and temporal database system…………..….….4 2.1 Data modeling………………………………………………………………...4 • Different types of data modeling…………………………….4 o Spatial data modeling………………………………………...4 o Temporal data modeling……………………………………..5 o Spatio-temporal data modeling……………………………...5 2.2 Query processing and indexing………………………………………………6 •

Nature of different queries…………………………………..6 o Challenge related to query processing and indexing………6 o Query example………………………………………………..6 o Data mining…………………………………………………...7

2.3 Implementation & architecture of spatio-temporal DBMS…………….…9 3. Various applications of Spatial and temporal databases…………………....11

CONCLUSION……………………………………………………....12 REFERENCE………………………………………………………...13 APPENDIX…………………………………………………………..15

2

Introduction Space and time became a key factor to represent a real world. Many modern applications such as cadastralΨ systems that capture the histories of land parcels, weather recording system, Navigation systems, traffic management system, etc, consider space and time as vital information for their data analysis. These applications require the storage and manipulation of voluminous traditional (alphanumeric) and non-traditional data (image, text, geometric objects, time-series) sets. Efficient relational or object oriented database management systems can resolve the challenge for storing and manipulating these data significantly; however spatial and temporal data need more attention and different approaches to handle them. Hence, spatial and temporal database management systems have been developed to support these non-traditional data sets. Considering space and time features in the database systems emphasis additional complexity and arguments in various areas such as data modeling, query processing and implementation of new database systems, etc. This research aims to illustrate various issues related to nature and application of spatial and temporal database management systems.

Cadastral systems records various land information.

3

Approach Spatial and temporal database system is wide concept, as it needs more and effective research work. I have emphasis mainly on different books related to spatio-temporal concepts and related previous research work from different institutes. Precisely, I have tried to put more stress on trends in the world today related to spatial and temporal database system research. As well, future utilization of this database management system is clearly on-demand, consequently more understanding on various data management approaches, various query processing methods and technologies should be provided in order to accomplish most effective and real time database management systems. Many organizations have put their step forward to engage more research work on spatial and temporal database management systems, although some of the work is still only research based and requires significant amount of time and appropriate researching techniques. Also, various applications and advancements of spatial and temporal database management system are taken into consideration to provide the vision of their scope. Additionally, after reviewing appraisal report, essential comments and suggestions from appraiser are also taken in to consideration, which assisted to enhance the report even further. Certainly, it proved excellent aid for the whole research work

4

Research [1]

Spatial and Temporal Data In recent years, noticeably increased attention of GIS research, interactive mapping applications, marine research, and medical research, this demands specialized data such as spatial and temporal data. However, Spatial and temporal data are not simply the image or pictorial data, although they are containing sets of objects in space or they might be data related to moving objects. Efficient management of these data is extremely crucial task, which requires appropriate data management approach, various tools and efficient algorithms. [1.1]

Conventional Vs Spatial & Temporal data Most of the application needs standard data such as employee details, product details or account details etc. These data includes regular data attributes, for instance employee entity requires name, address, telephone no, email address of employee etc for different records. Typically, data management of these conventional data such as retrieval, querying, sorting etc. is relatively trouble-free. Whereas, there are some specific applications requires spatial data (data related to space) and/or temporal data (data changing over time). Main difference between traditional or conventional data and spatio-temporal data is that changes can be constant or continuously variable such as the movement of the object (i.e. moving car) is constant. Whereas, traditional database systems assumes that data changes due to explicit update. As a result, for avoiding these constant updating of database, the description of the changes has to be stored as a function of time.

In comparison with the conventional data, spatial-temporal data requires new data types and sophisticated data structures, different data modeling, efficient data mining techniques, various queries processing techniques and complex database management system. The common challenge in handling spatial- temporal data is collections of data sets are very large. Specifically, using various technologies, various algorithms, and

5

different querying techniques, data management of spatial and temporal data could be achievable.

[2]

Characteristics of Spatial and temporal database system Spatial and Temporal database system embodies spatial or temporal or spatio-temporal database concepts and captures spatial and temporal aspects of the data. When a spatial database is enhanced by temporal characteristics or temporal database is enhanced by spatial characteristics, we get a spatio-temporal database management system. New algorithms and tools are needed in order to manipulate these data.

[2.1]

Data modeling In general, data modeling is an approach or method to structure and organize the data in way that it can be used easily by the databases. So, data modeling is systematization of data. Thus, data can be analyzed and integrated appropriately by database systems.

6

A spatio-temporal database system manages data whose geometry changes over time. These spatio-temporal datasets have some unique characteristics, as modeling these dataset is obviously different than the traditional relational and transactional datasets. Different types of data modeling: Spatial data modeling: By assuming 2-D GIS application two basic things need to be represented, first is objects in space, which have their own geometric description such as cities, mountains or forests etc., those can be modeled as single objects. To represent single objects point, line and region can be used. Other one is space that explains every point in the space itself, which can be modeled by spatially related collections of objects. To represent collection of objects partition or networks can be used. Commonly, system called ROSE1 Algebra is used to classify different spatial data types and operations. Temporal data modeling: Typically, Time based model records the timestamp (time series) for any change and associated details describing each specific change in temporal order. Temporal locations will be recorded with certain time domain. For successive or continuous change, recording timestamp can be determined by certain rule or when the change accumulates to a meaningful status. There certain numbers of data models proposed for the temporal data modeling such as Bitemporal Conceptual Data Model (BCDM) 2. Spatio-temporal data modeling: Data models, which combine both the capabilities of spatial and temporal data, are not straightforwardly available. There is not agreed-upon conceptual approach to model spatial and temporal data. Although, there are some ways in which we can model spatial and temporal data together. Two main ways to model Spatio- temporal data on which we can build databases tools. First is, to add spatial properties and operations in temporal databases and second is to add temporal properties and operations in spatial databases. It is something like cross product of spatial and temporal databases. Generally available data models are either spatial or temporal models but very few for Spatio-temporal model. Characteristics of spatial and temporal database systems are combined without any specific integration effort. There some efforts have been put towards combining these both facilities which a conceptual model is called MADS 3. MADS currently supports spatio-temporal features needed by the applications such as land management or utility network. Specifically, these data models can be used to design spatial and temporal database systems.

1

1Explanation of ROSE Algebra is provided in Appendix. Explanation of BCDM is provided in Appendix. 3 MADS represents, Modeling of Application Data with Spatio-temporal feature. 2

7

[2.2]

Query processing and indexing Query processing for spatial and temporal data are much more complicated, complex data types and volume of the data is very large. As, to develop query processors requires complex algorithms and indexing techniques to deal with spatial and temporal data. Query language should consider both space and time factor. 8

•

Nature of different queries:

In general, there are different classifications of the queries such as spatial, temporal or spatio-temporal queries. Spatial queries deal with spatial relationships between objects. They return spatial attributes or attributes calculated from them and are formulated in terms of spatial operators. It might be image of space or any graph or any other metric value. Temporal queries contain minimum one temporal operator and no spatial operator. Also, they deal with temporal relationships between objects at the valid time. Whereas, spatio-temporal queries deal with spatial relationships along with the time and they return both the spatial components at specific time. •

Challenges related to query processing and indexing:

Furthermore, we are facing some challenges to handle these queries. Most of spatiotemporal queries are continuous in nature; as well they are available in wide variety and some queries are concurrent. Also, indexing for moving object is quite complicated. Binary tree (B* tree) data structure is largely used for indexing. Even though, emerging database applications require new indexing structures beyond binary tree. Data can be available in any for except text data such as image, video or multidimensional data. There are some other indexing techniques and query technologies available for spatial and temporal database system such as quad tree, trie data structures used by Oracle spatial product, bitmap representations of record number pointers that match queries, called collections and Boolean combinations of inter list and actual bitmaps. •

Query examples:

Querying spatial, temporal or spatio-temporal data is quite crucial. Following provides some highlights of spatial queries other than database is often queried for specific temporal data at that point some database provides period table in a star schema or other similar schemas. Spatial Selection: All the post offices no more than 40kms from London SELECT p.name FROM post p Where dist (p.center, London.center) < 40 and p.pop>50k Spatial join: For each river pass through UK, find all cities within less than 50 kms SELECT c.cname FROM rivers r, cities c WHERE r.route intersects UK.area and dist (r.route, c.area) < 50KM Also, spatial and temporal data need to be combined together often in real world, e.g “moving object or objects” (e.g. satellite movement along with the earth movement) or “find all the pubs in 10 mile distance on Friday and Saturday night and rank by distance from my location ”. In these cases space and time are involved together. It is even possible to track moving objects in 3D by introducing a fourth dimension in addition to X, Y and time: Z.

9

•

Data mining:

Data mining (DM) is the process of extracting knowledge, from large amount of data from different perspective and summarizing it into useful information, which often called as knowledge-Discovery in database (KDD). Spatial and temporal data available in gigantic volume and they changed continually. As, digesting large volume of data and discovering useful knowledge from them, is again greatly challenging task. Generalpurpose tools used for data mining are not very effective for spatio-temporal data mining. Data mining is mainly application oriented research domain; data are collected for particular analysis or survey. However, this research is based on generic issues rather than, on the specific case study. Spatial data mining differs from general data mining, as, in case of spatial data mining attributes of the neighbors of some objects of interest may have an influence on the object as they have to be considered as well. Extracting interesting patterns and rules from spatial datasets is complex technique, the reason behind the image data sizes are too large to be mined in reasonable amount of time using exiting data mining algorithms. As, for efficient and effective data mining on spatial data, one widely used approach is Peano Count Tree (P- Tree), which facilitates significant pruning techniques and proves to be a promising approach for spatial data mining. Also, other new algorithms for spatial characterization and spatial trends analysis were developed. Temporal data mining can be defined as the activity of looking for interesting correlations or patterns in large sets of temporal data accumulated for other purposes. Temporal data mining process encompasses collection of various time related data using appropriate algorithms and techniques to get reliable conclusions. It comprises the analysis of events ordered by one or more dimensions of time, which illustrates two main broad issues, one concerns, discovery of causal relationships among temporally-oriented events and other concerns the discovery of similar patterns within the same time sequence or among different time sequence, which is also known as Time Series Analysis. Temporal data mining can be useful to investigate and analyze various important data such as measuring patient temperature or treatments effectiveness or any time based scientific information.

Moreover, mining spatial and temporal both relations at the same time complicates the task even further. Most recent techniques applied to spatial and temporal data sets are generally use very simple representation of geographic objects. As, data mining of spatial and temporal both datasets, are still demands more research work.

10

[2.3]

Implementation and architecture of spatial and temporal database Implementing spatio-temporal database is complex task that involves representing objects with spatial extensions and attributes values that change over time. To deal with spatiotemporal data, specialized database management system is required. To build spatiotemporal database management system, the traditional DBMS architecture and functionality have to be extended. In order to manage spatio-temporal data, requires spatio-temporal data types and operations, extensions to the query and data manipulation language, and index support for spatio-temporal data. Principally, there is not much difference between conventional DBMS and spatio-temporal DBMS. In the case of conventional DBMS data types such as STRING is used where as spatio-temporal DBMS use data types such as REGION or POINT etc. 11

There are three main approach to build the spatio-temporal DBMS, following diagrams illustrates different spatio-temporal DBMS.

Layered architecture is the traditional method to design an information system for advanced data types and operations; it uses off-the-self DBMS and implements a layer on top providing data types and services for the specific application domain requirements. Most of these architectures only support standard data types; layered architecture is also known as dual architecture. Monolithic architecture extends application systems with DBMS functionality instead of porting it to a standard DBMS, which aid query functionality, transaction management, and multi-user capabilities that layered architecture does not facilitate. However, the layered and monolithic architecture do not support an easy adaptation of the DBMS to new requirements of advanced applications, which can be resolved by the use of extensible database that provides a generic system capable to deal with new data types and functionality required for specialized applications.

12

[3]

Various applications of spatial and temporal database systems The applications of Spatial and Temporal DBMS are listed below to point out the accomplishments of these DBMS in current world. •

Defense unit is very important for any country, as it is responsible to take essential military decisions. The use of spatial data often provides a geographic framework for evolution of tactical situations. Also, Time series information, including weather effects, movement of battlefield obscurants and countermeasure techniques must be included in the visualization of a battlefield or more regulated test environment. These sorts of critical decision-making and risk analysis intricacy can be solved by use of spatial and temporal databases.

•

Medical sector is another extremely crucial area, where almost all the decisions are safety critical and important. Spatial and temporal data are often required to understand different movements of the body parts or other related medical images such as x-ray, MR-images, etc. Also, applications such as arcView GIS can be used for epidemiology and public health monitoring and to map public health indicators, identify disease clusters, and explore the spatial distribution of the disease and health care facilities. As, spatial and temporal database systems could serve the purpose of research and development in medical field.

13

•

Spatial and temporal information are valuable to various business processes in many industries to analysis their business needs and efficiently manage their businesses. As, some spatial database technologies provided by company such as IBM, has developed an infrastructure, which includes spatial-temporal database feature, which can help businesses to improve strategic business problem solving environment, and it could be competitive advantage for the businesses.

Conclusion This research on spatial and temporal database systems describes various concepts; arguments and applications related to spatial and temporal database management system. It principles to materialize several methods, techniques and certain technologies associated with spatial and temporal relationships. The report demonstrates the rapid growth of IT technologies to handle various database management systems. Spatial and temporal database management systems can resolve many future problems and enhance the DBMS pitfalls. Positively, it attests various features of spatio-temporal activity in the current world and future work allied to spatial and temporal DBMS.

14

15

Reference

16

1. Dr. Joerg Sander. (). Topics in Databases: Spatial Data Management and Data Mining. Available: http://www.cs.ualberta.ca/~joerg/courses/cmput697/fall2006/IntroSDBMS&KDD .pdf. Last accessed 13 October 2006. 2. Bonan Li, Guoray Cai. (). A general object-oriented spatial temporal data model. Available: http://www.isprs.org/commission4/proceedings02/pdfpapers/403.pdf#search='obj ect%20oriented%20spatial%20data%20model'. Last accessed 13 October 2006. 3. Norman W. Paton, Alvaro A.A. Fernandes, Tony Griffiths. (). Spatio-Temporal Database: Contentions, Components, and Consolidation. Available: http://www.cs.man.ac.uk/~norm/papers/asdm.pdf#search='spatiotemporal%20data base'. Last accessed 13 October 2006. 4. Cyrus Shahabi. (). Introduction to Spatial Database Systems. Available: http://infolab.usc.edu/csci599/Fall2001/presentations/session11.pdf. Last accessed 13 October 2006. 5. Robert P., Haining (2003). Spatial Data Analysis: Theory and Practice. Cambridge: Cambridge University Press. P54-111. 6. Manolopoulos, Yannis (2005). Spatial Database: Technologies, Techniques and Trends. USA: Idea Group Inc. p16-315 7. Wachowicz, John M. (1999). Object-Oriented Design for Temporal GIS. UK: CRC Press. p25-120. 8. David Beddoe. (2005). Spatial Information as a service. Available: http://www.directionsmag.com/article.php?article_id=2058&trv=1. Last accessed 11 Nov 2006. 9. Dharmajyoti Bhaumik, William A. Lanich, William Ribarsky, Joseph E. Swistak. (1997). Visualization of Temporal and Spatial Data for Defense Applications. Available: http://www.spie.org/web/meetings/calls/or97/or06.html. Last accessed 11 Nov 2006. 10. Wesley W. Chu, Alfonso Cardenas, Ricky Taira. (). KMeD: Knowledge-based Image Retrieval with Spatial and Temporal Constructs. Available: http://www.cs.ucla.edu/csd/IDM99/Proceedings/Summaries/Cardenas.htm. Last accessed 11 Nov 2006. 11. Qin Ding. (2003). Data Mining on Spatial Data. Available: http://math.nist.gov/mcsd/Seminars/2003/2003-02-25-ding.html. Last accessed 16 November 2006.

17

12. Dr. Hans-Peter Kriegel. (). Spatial Data Mining. Available: http://www.dbs.informatik.uni-muenchen.de/Forschung/KDD/SpatialKDD/. Last accessed 16 November 2006. 13. Martin Breunig, Can Turker, Michael H. Bohlen, Stefan Dieker. (). Architectures and Implementations of Spatio-temporal Database Management systems. Available: http://www.spatial.cs.umn.edu/CS8715/STD17_chrono.pdf. Last accessed 17 November 2006. 14. Ralf Hartmut Güting, Markus Schneider. (). Realm-Based Spatial Data Types: The ROSE Algebra. Available: http://www.informatik.fernunihagen.de/pi4/papers/ROSEAlgebra.pdf. Last accessed 17 November 2006. 15. John F Roddick. (). A Model for Schema Versioning in Temporal Database Systems. Available: http://www.hgo.se/~gustaf/vers/A%20Model%20for%20Schema%20Versioning% 20in%20Temporal%20Database%20Systems.pdf. Last accessed 17 November 2006. 16. Christian S. Jensen. (). Introduction to Temporal Database Research. Available: http://infolab.usc.edu/csci599/Fall2001/paper/chapter1.pdf. Last accessed 17 November 2006.

Appendix 18

•

ROSE Algebra

ROSE represents the algebra or sort of data types of the spatial data which defined on realms (A realm as a general database concept is a finite, dynamic, user-defined structure underlying one or more system data types.) and offers general types to represent point, line, and regional features together with a comprehensive set of operations. It is described within a polymorphic type system and interacts with a DBMS data model and query language through an abstract object model interface. •

BCDM (Bitemporal Conceptual Data Model)

BCDM is the timestamp data model employed by the TSQL2 language design; however it can be applied to temporal models. Under the BCDM, time is conceptually modeled as consisting of bitemporal chronons (Elementary rectangles) existing in the twodimensional space between transaction and valid-time. The idea behind the BCDM is to retain the simplicity of the relational model while also capturing the temporal aspects of the facts stored in a database.

19