Saint Louis University |
Computer Science 371
|
Dept. of Math & Computer Science |
Topic: Database Revision
Related Reading:
Due:
8:00pm, Monday 7 November 2005
Please make sure you adhere to the policies on academic integrity.
This assignment is a followup to the earlier assignment in which you developed your own database schema and implementation. Presumably, you will be working in the same group for this assignment as you did for that earlier assignment. However if there is some reason why the previous group dynamics no longer works, please speak to me as soon as possible so that we can devise a plan for this assignment.
For this assignment, you are to take the feedback which you have received from the original work, and to perfect your design, your implementation and your demonstration of the database. Problems which were clearly addressed in the feedback to the original assignment should be remedied in your revision. Furthermore, we are asking everyone for additional work which was not in the original assignment. The precise efforts for each group will vary somewhat depending on the state of the original work, but the following are the major themes for the new assignment
Accurate ER model
You must submit a new version of your ER model, which is
accurate in the following ways:
If there were major design problems with your original model, those should be corrected.
You should use the precise diagramming conventions as used in our class and text, including annotations for the cardinality constraints on the participating entities for each relationship.
The ER model on paper should match as closely as possible with the actual SQL implementation you are giving. This means that the names of the entities, relationships and attributes should be the same, when appropriate.
However please note that the ER model is not the same as the SQL implementation. There may be times where an entity and relationship is the ER diagram are combined into a single table in the relational schema. Similarly, the foreign keys in the relational schema which represent participating entities for a relationship are not explicitly diagramed as attributes in the ER model (but rather are implicit in the semantics).
Accurate SQL scripts
We will be testing your revised SQL scripts on turing. You are
to ensure their correct operation. Furthermore,
we ask that you explicitly break your previous SQL
script(s) into distinct pieces as follows:
An SQL script titled schema.sql, which defines all aspects of your database schema (i.e., the tables, constraints, and any associated domains). This script should be written so that we can recreate your database schema from scratch.
An SQL script titled populate.sql, which populates your database instance (this will be called once, after schema.sql has been executed)
An SQL script titled cleanup.sql, which can be used to drop all parts of your database, getting us back to a clean slate.
New SQL Queries
To see your database in action, you are to write queries to
answer five natural questions for your database domain. In
particular, five specific questions have been suggested for your
project, as part of the feedback to the original assignment.
You must write queries for those five questions based on your
(redesigned) SQL schema. You should create a separate script
query.sql with those queries, and you should save the
output of those queries on your populated database in a separate
text file results.
Populate your database with a richer instance
In generally, the original assignments were quite weak in the
area of populating the database. Though it is not necessary to
have thousands of entries, it is important to have enough data
entered to demonstrate the aspects of the database. In
particular
Demonstrate the integrity constraints
Presumably, all of the insertions in the populate.sql
script should succeed. However it should be possible to
demonstrate the enforcement of the integrity constraints
(e.g. primary key, unique, and foreign key constraints) by
attempting other data manipulations which violated such
constraints. To do this, provide a separate script,
failure.sql, which gives a series of commands
which trigger the failure of the various constraints in your
schema.
Based on the above description, each group must submit the following documents:
A revised version of your ER diagram.
Five SQL scripts, schema.sql, populate.sql, query.sql, failure.sql, cleanup.sql.
A text file, results, giving the output of the executed queries on your populated database.
We will attempt to base our evaluation of this assignment as fairly as possible upon the work which has been done during the revision (as opposed to the quality of the original assignment).