IPT home IPT A Virtual Approach by Peter Whitehouse
Quick Links:
 
 
SQL home
Information and Intelligent Systems Social and Ethical Implications Human Computer Interaction Software and Systems Engineering eXercise Files Course Outline and Assessment A-Z of Geeky Acronyms Terrace Work Program 2004 Sillybus FAQ = Frequently Asked Questions Help
 
 

The Relational Model of Data

Relational Model Operations

Relational Databases are managed by Relational Database Management Systems. Access, MySQL, Oracle, Postgress, Interbase and so on are RDBMSs and all use SQL in some form or another to access data. To be considered RELATIONAL, a database language (or development platform) must allow a number of basic operations. These operations have been categorised into DDL, DML and SECURITY below:

Data Definition (DDL)

  1. Users with appropriate permissions should be able to creating table shaped containers. They should be able to define special relationships like mandatory/optional fields and primary/foreign keys
  2. .

  3. It should be possible to add columns and alter table definitions
  4. .

  5. Users with approprite authority should be able to populate their tables (ie. add row instances), modify those rows, view them and remove them. Note that in some flavours of SQL, explicit permissions are set for EACH of these operations.


  6. Users with approprite authority should also able to destroy tables.

In MySQL, table creation can be achieved either through the use of command-line like DDL commands, or graphically (using many different graphical user interfaces) or dynamically via webpage calls. Additionally, MySQL Data definition can be achieved 'live' via scripts embedded in server-side web pages.

In Access, we use table builder grids and relationship builders to achieve the same (although it is possible to create tables and relationships using commands in Access, it is not a common way to do this).

Data Manipulation (DML)

Once your data containers are defined and filled with data, then the next step is doing something with that data. There is little point in entering data if nothing can be doneto it - SQL provides many complex and powerful data manipulation commands which have been logically grouped below. Further explanation of these groups occurs later in this section.

  1. SELECTION - we should be able to see only the rows that are relevant to a particular problem = row subsetting
  2. (show us rows corresponding to some condition)

  3. PROJECTION - we should be able to see only those columns that are relevant to our problem = column subsetting (display certain columns only)
  4. INTERSECTION - viewing data that is common between compatible tables (list data that is in both places)


  5. UNION - merging data from compatible tables (merge this list with this list to produce a single, distinct list)


  6. DIFFERENCE - listing data that is present in one table but absent in another - like XOR (show me values that occure here but do not occur there)


  7. JOIN - combine rows of 2 or more related tables to re-join facts (connect rows here with corresponding rows from there)


  8. ORDERING - cosmetic rearrangement of rows based onon specific criteria (arrange these rows based on this criteria)


  9. GROUPING - agglomerating rows according to some criteria (combine all these values and do someting with the combined data)


  10. DERIVATION - using pre-defined functions to work out certain facts from stored info (using this information, work something out and display the answer)

Data manipulation is invariably achieved using QUERIES. There are many types of queries, some merely requesting data, others forcing calculation, still others used to alter existing table data.

In this section of the course you will be given a better than working knowledge of querying - the power of relational databases lies in the ability to use the stored data in really flexible ways. You will also be given a visual lexicon (language/method of representing query ideas so as to better find simplicity in complex situations).

ACCESS PRIVILEGE

SQL traditionally allows the designer of the database to determine who is allowed to access (either see/update) particular fields in a database. The designer can customise a VIEW of the database, protecting the base tables from unorthorised access.

MySQL databases and tables are STRICTLY controlled via username and password. Database designers assign permission levels to various users - read, add, modify, delete etc. Unless you have specific permissions to perform actions, you are prohibited from doing so. The public MUSIC database available online allows you to merely read the data- you have no add, modify or delete priveliges. The PUBLIC database is open to any user - all users have complete permissions over the tables stored there.

In Access, if you can open the database, you have complete control over it unless your access to the database is governed by some programmed INTERFACE.

We will not concentrate on security in this unit, but be aware that any form of commercial database needs careful security measures to prevent unauthorised or accidental damage to the data. In the ideal world, security measures are unnecessary, nothing goes wrong and the information wants to be free. In the real world shit happens and if 'accidents' can be prevented they should be.


Relational Diagrams - 'Query Shorthand'

The following diagramming method will be used throughout this section to explain (in a physical way) the action of a query. Often it is easier to design queries, particularly complex ones, symbolically before attempting to encode them.

This method of query representation is similar in many ways to QBE (Query By Example) which is a popular method of querying in modern RDBMS's.

Please note that, although the following diagrams are drawn very neatly, a scribble sketch is sufficient for student use (ie. don't get out your ruler to do these) - they are merely visual aids, hope they help.

A detailed explanation of SQL Query diagrams is presented in a paper given by Mr Peter Whitehouse at the 1999 QSITE State Conference with many of the examples presented being recycled here as well

projection

list serial numbers and album names from the albums table

selection with projection

list the album name and publishing date of all albums by the artist "Pink Floyd" arranging the data in reverse chronological order

a simple derivation

list album name and artist along with their age that I have on CD

conditional OR

list albums and the media they are on from either Edgar Froese or Tangarine Dreat, arranged alphabetically

no repeats = distinct

list all group names without repeats

group by having

list artists that I have more than 2 albums of

sub query inclusion, projection and selections

list artists and album name of all CDs that feature the song called 'blister in the sun'

 
subquery exclusion with distinct projection

list artists that do not play bass

 
union

list all artists recorded in alphabetical order

 

natural inner join projection with multiple selections

list the album name and songs of any album released by Barry Manilow

The actual SQL commands for the diagrams above will be explored in class, along with many more complex problems. The diagrams are presented to help you 'see' the solution.

Union T1 U T2 (so long as the tables are type compatible)

     eg.       TableA  name     age     TableB  name     age
                       fred     14              cyril    17
                       marg     12              blue     13
                       glen     4               fred     14
                                                marg     12

     TableA U TableB =

               name     age
               fred     14
               marg     12
               glen     4
               cyril    17
               blue     13
Columns involved in union must be type compatible

Combinations of the above operations are allowed: (X where b = q [a]) U (Y where c = y [a]) where column a is compatible

Intersection T1 intersection T2 (ie. common elements)

     TableA intersection TableB = 
                                      name      age
                                      fred      14
                                      marg      12

Difference T1 - T2 (ie. those in T1 that are not in T2)

     TableA - TableB =                name      age
                                      glen      4

Joining (cartesian product) - without constraint

        Grades             Classes       Result - unconstrained
          year               form          year     form
          5                  blue          5        blue
          6        X         gold          5        gold
          7                                6        blue
                                           6        gold
                                           7        blue
                                           7        gold

Joining - another unconstrained join: Produce a 'play list' for GPS

     school1       school2       result (unconstrained)
     BSHS          BSHS          school1       school2
     NC        X   NC            BSHS          BSHS     'a self game'
     GT            GT            BSHS          NC
                                 BSHS          GT
                                 NC            BSHS     'repeat'
                                 NC            NC       'a self game'
                                 NC            GT
                                 GT            BSHS     'repeat'
                                 GT            NC       'repeat'
                                 GT            GT       'a self game'

a constrained join:

a partially constrained join
     school1 X school2 where school1 < school2

          result
          school1       school2
          BSHS          NC
          BSHS          GT
          GT            NC
 

wonko@wonko.info
©Copyright t 1992..2018+. Edition 26.150117
wonkosite
Creative Commons License
This work is licensed under a
Creative Commons Attribution-NonCommercial-ShareAlike 2.1 Australia License
.