Knowledge base systems

A knowledge base system or expert system looks to mimic human knowledge and experience a particular field , The knowledge base system finds solution to problems with given set of conditions.

Human experts all act in a similar way

1. someone asks the expert for advice about a particular problem
2. the expert asks them a series of questions to find the best solution 
3. keep the dialogue open until more than one solution is found 
4. prioritise soultions based on budget, time etc
5. A knowledge base system seeks to replicate this

The knowledge base has four components

• the human computer interface - this is the system that allows a non-expert user to query (question) the expert system, and to receive advice. The user-interface is designed to be a simple to use as possible.
• the knowledge base -This is a collection of facts and rules. The knowledge base is created from information provided by human experts
• the rule base
• the inference engine - This acts rather like a search engine, examining the knowledge base for information that matches the user's query

Expert system




inference engine is the part of knowledge base system that works out a reasonable solution using the results of the rule base.

Human computer interface is

The NHS choices knowledge base interface asks a series of questions to attempt to find out what kind of illness or pain you are suffering from and what the cause of it is and the solutions etc. It is very good  to find out a general. basis of the injury or illness , the symptoms ,the causes , the treatment and the complications .However if you don't know exactly what you have then it can prove very difficult to find out what it is when using  it for example you can answer the series of questions to get to a certain point when it gives a list of possible injury's or illnesses but there are often so many possibility's  that it can be very difficult to actually find out what it is. Therefore  making the knowledge base only useful up to a certain point before it becomes very difficult to use.

Some more examples of knowledge base systems are

• diagnostic tool for fixing machinery and vehicles
• on-line medical systems for diagnosing a problem
• telephone based help desk
• finance firm making credit decisions
• government services such as working out tax benefits

Advantages

• expert advice all of the time
• knowledge of an expert can be captured before they move on
• can be used for staff training to increase expertise of all staff
• does not get tired or over worked
• efficient way of getting answers as it is an automated system
• makes rational decisions without emotional overhead 

Disadvantages

• only covers a narrow range of knowledge 
• a lot of effort and cost required to make a good knowledge base system
• not as good as having human expert hands
• most systems are menu driven which may not deal well with ambiguous problems
• advanced interfaces still have some way to go before they can be truly effective
• can only learn from mistakes if user feedback and human maintenance is part of the on going development.

How to set up an expert system?

To set up an expert computer system you will need a database with information about the different types of issues and the solution to each problem. This will need to be linked to a rule base that will contain the probabilities of each problem happening so the most likely problem will be chosen, an unresponsive computer will most likely be fixed with a reboot and not new hardware. Finally a Human Computer Interface will be needed for the users to interact with the system and the inference engine can query the database / knowledge base. 

Examples of expert systems?

• Medical diagnosis
• strategy games (e.g chess against the computer)
• Providing financial advice - whether to invest in a business, etc.
• Helping to identify items such as plants / animals / rocks / etc
• Helping to discover locations to drill for water / oil 
• Helping to diagnose car engine problems
• Government tax credits
• Tax claims
• Auto pilots 

Drawbacks of expert systems?

• Can't easily adapt to new circumstances (e.g. if they are presented with totally unexpected data, they are unable to process it)
• Can be difficult to use (if the non-expert user makes mistakes when using the system, the resulting advice could be very wrong)
• They have no 'common sense' (a human user tends to notice obvious errors, whereas a computer wouldn't)
• Can make mistakes, just as humans do – even a low error rate e.g. in the diagnosis of a disease, may cause people to mistrust it
•  Expert systems do not learn from their mistake – new knowledge has to be entered into the knowledge base as it becomes available
•  Difficult to acquire all the required knowledge from the human experts in order to build the expert system