4g/i Expert systems

4g/i Expert systems

1. Expert system: a computer software replicating the role of a human expert using knowledge and rules of a specific subject area to diagnose / solve a problem.

   • e.g. NHS choices, government tax calculation tool, vehicle diagnostic tool, printer trouble shooting
   • Fault diagnosis for car engine: Connect engine management system to complete tests and find faults
   • Social services to determine benefits due from the claimant's form and why the benefits apply.
   • Suggests what solution is.
   • Can ask hypothetical questions and model what if scenarios

2. Knowledge base

   • Contains factual knowledge and rules
     • Collected from experts by knowledge engineers

3. Interference Engine

   • Uses decision tree based on rules in knowledge base and acts like a search engine to examine and match relevant knowledge to the user's answers.
   • Asks more questions to eliminate possible diagnosis and narrow down to the correct diagnosis

4. User Interface

   • User interacts with it
   • Asks questions and delivers diagnosis

5. Disadvantages

   • X Cost to set up: A lot of time needed to interview experts and gather all knowledge as well as input rules for diagnosis
     • X- Takes a lot of time and effort to update- knowledge base and change rules
   • X Lack of human touch: May be needed to see things like temperature or size of tumour on medical expert systems
   • X GIGO: Faulty knowledge or rules or answers will give incorrect diagnosis.
     • User may have a more serious illness, but will not know.
     • May lead to nonsensical diagnosis, e.g. the bubonic plague.
   • X- Most require internet access

6. Advantages

   • O- Always available
   • O- Can be used as training aid
   • O- No emotions => Rational decisions
   • O- Doesn't get tired, consistent performance
   • Can use a natural languages interface to seem friendlier.