Database Project Ideas

Right, beer free (relatively!) and ready to go...

I really have to agree with ala_frosty - your professor / tutor / whatever should be able to give you some good ideas. Typically there will be a list of possible titles floating around somewhere and past projects the tutor has worked on could form the basis for future study. I would also suggest two other sources of inspiration:

1. Choose a subject are you have particular interest in for reasons of your own.
When I wrote my first thesis (potential use of a ducted fan in 2 seat light aircraft - I studied aeronautical engineering and know far more of the theory behind aerodynamics than software design!) I chose it to form the starting block of a long term ambition. I still have the drawings somewhere and WILL build that 'plane!

2. Take a look at industry; they may have good ideas and be willing to pay you to research them.
A university friend of mine was given a Masters research project by Lola (the racing car people) to investigate the airflow around different types of tyre on racing cars. As far as I remember, it wasn't a massively interesting project but the pay kept him in beer for the year. Again, your tutor should be able to help you out if you go down this route. It may also be worth getting in touch with whoever maintains the SQL ANSI standard and see if they're heading off in any exciting new ways(!).


One idea I had - this isn't original so please don't laugh too loud if it's pants! - is to take a standard database and investigate ways of 'auto-building' an object model from it. The input and output format are pretty much irrelevant (i.e., it should be able to read any given RDBMS database and output to any given language - this is pure symantics anyway) but there should be no reason not to develop the idea in, say, Oracle & Java and explain how you'd take it from there. The end result would have to be able to query the database, make sense of the table relationships, etc. and build a skeleton object model from it. This skelton should have at least the basic data structure, property definitions, standard update / retrieve / delete methods and so on. Some things you'd need to consider are...
* optional 1-1 joins. If a base table has a variety of child tables hanging of it on 1-1 joins, the base table can probably become a base class and the child tables inherited classes.
* 1 - many joins. Implies a collection.
* any standardised naming conventions that would help your system work
* possibly a way of inspecting sprocs to see if they can bcome standard methods in your objects
* impact on the devlopment community (is it worth-while?). A system that generates a useful OO framework would save a lot of development time handling basic property generation and so on. A system that generates a poor OO framework would be worse than useless. IMHO, a well-rounded thesis should assess the socio-economic impact of the subject matter. In this case, would it be a benefit to system builders?, could it form a first-step to true self-writing code?, would a product based on your ideas be marketable? and so on.

Hope this helps (with a little more sincerity than last time!),

s46.