Week 12 - other intelligent systems

This is our last session before the Christmas vacation, and we will be following on from the sessions on expertise and expert systems but looking at some other systems that demonstrate intelligent characteristics, such as genetic algorithms and artificial neural networks.  The slides are in the ES tab.

Testing, testing...

It is useful to have a table to describe the test plan that you create and implement for your software.  Please see the following handout from Dawson's book:

Link to testing from Dawson's book

Dawson's book at Amazon

In the seminar after today's lecture (week 11, Thursday 13 December) we looked at examples of tests, and showed how any problems were addressed. Not all tests are successful first time, so we need to identify and errors and explain what had to be done to correct them.
We also need to consider how the systems deal with error data as well as valid data:

if you are calculating wages:

• 40 hours worked per week = OK
• 60 hours worked per week = looks like overtime was involved
• 80 hours worked per week = check if overtime was authorised
• 200 hours worked per week = whoa!  There are only 168 hours in a week!

You do not want your system to calculate payment for 200 hours worked, as this is impossible.

You should also check whether your system can cope with invalid data types, such as what happens when a character is typed instead of a number - does the system crash, wait for more data, or display an error message?

Note that testing refers to the testing of your queries (including passing parameters, and the switchboard).

Checking the data that makes up the database will be in the organisation chart section if it refers to staff; in the case of all other data, any errors (and associated actions and corrections) will be included in the section of data cleansing.

  

Weeks 10 and 11 - Expertise and expert systems

In these two weeks we will consider what is meant by expertise, and how we can capture expertise in order to transfer the skills of an expert into a system that can be used by non-expert users in order that they can have the benefit of expert advice.