Home Page

Lecture 2 - Rule Based Expert Systems

Workshop

Assignment 2

Course Lecturers

Christian Gibson, Tiberiu Lupascu

E-mail addresses

c.h.s.gibson@hva.nl  t.d.lupascu@hva.nl

Lecture 2

content

Rule Base Expert Systems

In this lecture we dealt with Rule Base Expert Systems. Basically a rule based expert system is an intelligent systems which provides the sort of information that a user could expect from a human expert - for example a doctor, a car mechanic or a mining engineer. The system works with IF THEN rules and with input data provided by the user. The rules and the expert knowledge are contained in a database and rules of inference in the form of IF THEN AND OR can be used to provide useful information. IF THEN AND OR statements (rules) can represent relations, recommendations, directives, strategies and heuristics (text book p. 27).

Relation  IF   the fuel tank is empty THEN the car will not move   Recommendation IF      the season is autumn AND the sky is cloudy AND the forecast is drizzle THEN the advice is 'take an umbrella'   Directive IF       the car will not move AND  the fuel tank is empty THEN the action is refuel the car   Strategy IF    the car will not move THEN the action is check the fuel tank; step 1 is complete IF       step 1 is complete AND  fuel tank is full THEN check the batter; step 2 is complete   Heuristic IF      the spill is liquid AND the acidity or alkalinity is < 6 pH AND spill smells of vinegar THEN spill material is acetic acid

The inference engine of an expert system can reach its conclusions through 'forward chaining' and 'backward chaining' (p. 37 to 40).

For further information about rule based expert systems we advise students to study chapter 2 of the text book and to take a look at the power point presentations. In particular, the example of an expert system and the chapter summary in pages 41 to 53 are important.

Workshop

Following the lecture, the students formed 3 groups of 7 students per group to discuss criteria for grading the assignments submitted after lecture one. The purpose of this exercise was to focus student's attention on the ideas submitted by their colleague. It is basically a Wikipedia approach - an attempt to achieve the best coverage of a subject by submitting the text to review by other parties. Of course in the Wikipedia approach articles are open to review by the world at large, which allows for a potential 6 billion reviewers! Unfortunately it is asking slightly too much to allow for 6 billion assessors to review the course assignments, but despite this we are hopeful that the results of around 6 reviews per assignment might also be worthwhile!

Grading

Most of the students took their task seriously. They first of all agreed on criteria to be used to grade the assignments. They then examined the exercises critically, one by one. The resulting grading turned out to be fairly strict. The grades the students awarded themselves was on average less than they would normally have earned from their lecturers. The average grade was only 6.2. What the students didn't know was that the course lecturers had also submitted their own exercises for the grading: their result was 7.0 and 4.6.....

Official results

A number of students were concerned that the grading would actually count for the official grade on completion of the course. This is NOT so. Of course the grades the students have awarded will help the lecturers to form their own judgements, but in the end the results will be graded according to criteria and insight provided by the lecturers.

Assignment 2

Build a knowledge database with around 10 rules and 15 facts. You can, for example, choose one of the following domains (but you are not limited to these - you are free to use your imagination)

Information security

Project management

Knowledge management

Reduce the database to 5 rules and 8 facts and

Perform a forward chaining

Perform a backward chaining

Additional information (using the text book! If you don’t yet have the book, copy the pages from a colleague, make sure that you have the information in good time. Don’t leave it till the last moment!)

Create an inference chain as described on page 37 of the text book

Draw the chaining in a diagram as shown in  pag.38-39 of the text book

Don’t just copy the inference chain given in the text book. Use your own creativity and expert knowledge in your chosen domain.

Submitting your completed assignment

Upload the document to iknow, with the file name: <studentnr_naam>.doc

Return