21 Hierarchical, network and relational data
Dr Dinesh Kumar
1. Learning Objective
The objective of this module is to understand the concept of Database Management System (DBMS) and database Models. The present module explains in details about the relevant data models used in GIS platform along with their pros and cons.
2. Introduction
GIS is a very powerful tool for a range of applications across the disciplines. The capacity of GIS tool to store, retrieve, analyze and model the information comes from its database management system (DBMS). The DBMS of GIS system stores and retrieves the spatial and non-spatial information and optimizes the process for users. It can be an embedded package within the system which allows the user to set up, use and maintain the database. It offers a useful set of functions and capacity to store terabyte information.
DBMS enables the GIS system to extract the information based on certain query and also use data model which defines the data structures. Many useful properties of DBMS can be summarized as below:
- Storage and retrieval
- Data integrity
- Conditional based query
- Data backup and recovery
- Data redundancy
- Tabular
- Relational
- Network
- Hierarchal
- Object Oriented
Among these models, Relational, Network and Hierarchal database model will be discussed in detail.
3.1 Relational Data Model
The term relational database was defined and coined by Edgar Codd at IBM Almaden Research Center in 1970. Relational database model relies upon common attributes within the dataset. It utilizes the common attribute to clusters the dataset. Most of the GIS tables are stored using relational data model. The relation is expressed in terms of tuples based on common attribute and a table composed of rows and column. Relational DBMS is based on a set of mathematical principles also called as relational algebra, which provides specific set rules which designs the database models. The relational algebra lays down rules such as
- Table can’t duplicate the other row data in any tuple.
- Column(s) can be used for primary search key.
- Single column or even multiple columns.
- Primary key cannot be left blank to avoid duplication.
The users can match the data from one table to another table by relational join. The primary key (the first column) is used to match the data with foreign key (second column). Going further, second column can act as first column and next column as foreign key to match the data which may continue further based on requirement. In the process of joining or matching the data, a key (column) must be common. The process of joining makes the process simple for complex searches maintaining a very simple, well-defined and easily developed set of tables.
3.1.1. Relationship
RDBMS can relate among the tables based on below mentioned relationship (Fig-2).
One to One relationship: One record from destination table is related to only one record of source table.
One to many Relationship: one record from destination table is related to many records from source table.
Many to one relationship: Many records from destination table is related to only one record of source table.
Fig-1: Relational database Model structure and Table. Source: Textbook of Remote Sensing and Geographical
Information Systems by M. Anji Reddy
3.2. Network database model
The network model’s was defined and invented by Charles Bachman, and was developed into a standard specification published in 1969 by the CODASYL Consortium. More flexible database model to relate multiple linkages and a little upper hand over hierarchal by allowing rapid linkage among adjacent features. It defines its property by its name i.e. network which means it is a network (many to many) of linkage making query process more user friendly (Fig-3). In another point of view, it represents itself as upside down tree with branches as member information. In addition, the relationship that the information has in the network database model is defined as many-to-many relationship because one owner file can be linked to many member files and vice versa.
This model supports the many parents theory where child can have many parents. It is taken as up gradation of hierarchal data model as the later one lacks many to many relationships.
Fig-2: Data relationship representation diagrammatically. Source:AGS72
Network database model lacks simple database in terms of larger number of pointers. Large number of pointers uses substantial storage space. Larger numbers of pointers are needed to manage by additional database models.
3.2.1. Terminology
Network DBMS maintains an owner record (parent record) and a member record (child record) which is used as one to one, one to many and many to many linkages.
Fig-3: Diagrammatic representation of Network database Model 3.3. Hierarchical data Model
This database model maintains the data in a tree like structure in hierarchical way. It consists of records that are connected to each other through links. Each record represents a collection of fields (attributes), each of which have one data value. Association between two records is called as link.
The model grows with single root into many branches following the many children, single parent logic Fig-4).
Data model is simple database like real-world relationship like book index. In this model, data is structured into tree-like structure with one- one-to-many relationship between two different types of data.
This data model is more common to environmental sciences as they offer simple and easier way to understand. Being a simple data model, it provides quick and convenient means of data access easier for data retrieval. To have information of query, one need go from root to target. There are some limitations of this model as it follows hierarchy.
For the purpose of editing or addition of new records at down as it has to be linked with root.
- It has limitation to branching network itself.
- Query for near or far feature data
- Series of pointers for accessing data
- Redundancy of data in case of many-to-many relationships
Fig-4: Hierarchal Data structure diagram
Above discussion can be compiled in way that database model has improved significantly over the year. The initial database being simple in nature, were capable of handling efficiently smaller number of datasets with limited queries. Further, database model evolves into networking to relational enhancing its capacity to handle more and more data with more complex queries. With the exponential growth of information technology, information volume has exponentially increased which needs more complex database models to handle. The core objectives of data models are to fulfill the requirements arising out in new data scenario.
4. Summary
The DBMS is primary requirement to process any query, analysis and updation. It becomes very important for retrieving and backing up the data. With the exponential growth of information and data, DBMS have evolved from simpler to more complex. While hierarchical data model provides simple and easy way to handle the queries and analysis, Network and relational data model provides more complex analysis and queries.
Suggested Readings
- https://courses.washington.edu/gis250/lessons/introduction_gis/relational_database_model.html
- https://www.databasejournal.com/sqletc/article.php/26861_1428721_3/Introduction-to-Databases-for-the-Web-Pt-1.htm
- http://wiki.gis.com/wiki/index.php/Relational_database
- https://courses.washington.edu/gis250/lessons/introduction_gis/relational_database_model.html
- https://www.databasejournal.com/sqletc/article.php/26861_1428721_3/Introduction-to-Databases-for-the-Web-Pt-1.htm
- https://www.mapsofindia.com/gis/attribute-data.html
- Principles of Geographic Information system, An Introductory text book, Otto Huisman and Rolf A. De
- Textbook of Remote Sensing and Geographical Information Systems, M. ANJI REDDY, Professor, Institute of Science and Technology, Jawaharlal Nehru Technological University, Hyderabad