Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
expert system (cH)
Transcript of expert system (cH)
advantages & disadvantages
allow a user to access a
EXpSys to act as a human expert
how it is made up
what it is used for
good points, bad points and other consequences
expert system shell
domain specific facts and rules
created by knowledge engineer from human experts
built up during each consultation session
session information entered by user
session information generated by inference engine from using rules
process rules in knowledge base
request information from user
give reasons for asking the user a question
clarify user's understanding
leads to better quality answers
give reasons for advice given
states line of "thinking" leading to conclusion
may convince the user to trust the advice
(cc) photo by medhead on Flickr
suppose an expert system was used to diagnose a patient's condition
suppose a doctor treated the patient based on the advice
suppose in fact the diagnosis was wrong
suppose the patient's health was affected
is it moral in fact to use expert systems in medical situations
(some research shows that human doctors may incorrctly diagnose a condition 50%)
suppose an oil company used an expert system to choose a drilling site
suppose they spent millions of pounds setting up the operation
suppose there was in fact very little oil available and it was abandoned
who is to blame?
the knowledge engineer?
the programmer who coded the knowledge base?
the human consultants whose expertise was used?
the company for acting on the advice?
(cc) image by rocketboom on Flickr
once the expert system is set up the software can be saved
many copies can be made, unlike a human the system will not grow old, forget or die!
although an expert system may cost a lot to set up, once made it does not need upkeep
a human would expect fees every consultation
the expert system can be copied and is available 24/7/365
an expert system will always reach the same conclusions based on the same evidence
humans can change their mind for many reasons and sometimes due to error
an expert system on the web is available everywhere, humans need transport!
an expert shell contains a ready-made user interface and inference engine
by adding a custom knowledge base a new expert system can be made
using a shell may not provide a solution that is as finely tuned to the client's needs
a traditional expert system requires a complete new set of components
they have specialist knowledge, gained after lots of study and years of experience
a consultation is a meeting where a client and an expert can have a conversation, the client may pose a request.
the expert may in turn ask the client questions to fill in some details that are needed to come to a conclusion
during the conversation the client may not be clear on why the expert wants to know a particuar fact
by explaining why the facts are needed the expert may help the client to answer the questions more accurately thus making the final advice more reliable
after the conclusion the client may want to know how the expert came to a conclusion.
the expert will give reasons to justify ideas and reassure the client
an expert may confer with other experts to check their ideas
an expert system is software that can be used to give advice
ususally it would be used to assist a human expert
sometimes it may offer public advice
created by knowledge engineers following software development process
provide useful advice
There are many areas where expert systems can be used.
Solicitors can check that they are giving the correct advice as it is difficult to keep up to date in every aspect of all the different laws.
A human legal expert should always give the final advice to a client.
A GP or junior doctor might use an expert system to help diagnose a patient's symptoms.
However this would always be followed up by referring the patient to a human consultant.
Bank or financial
When applying for a loan an expert system help decide if it should be granted.
The expert system can be used by the applicant to gather all the necessary facts and figures.
This avoids the need for human staff to spend a lot of time interviewing applicants.
The recommendations of the expert system will be checked before a human actually approves the loan.
Customers are happy because the application process takes place more quickly and there are no human errors to slow down the process.
Knowing where to drill is obviously important.
Traditionally lots of test drilling took place to find the right spot.
Now data such as geological surveys and sample from test drilling can be used by an expert system to more quickly and reliably pinpoint the ideal location.
software development is not cheap, consulting with the best human experts will be expensive
keeping the expert system up to date with latest research means involving human experts again
expert systems do not learn by them selves
the knowledge engineer has to code new knowledge with the help of human experts
human experts although specialists still have wider experiences
they may sometimes use unrelated knowledge to help solve a problem
experts systems can only use the restricted domain
the knowledge engineer will gather expertise from human specialists
the domain has to be quite narrow
for example a medical expert system would typically be used for one specific type of illness
a human expert would be likely to have a specialist area but not quite so narrow
another form of inflexible
a human expert will still use everyday sense which of course is very wide-ranging
this is completely beyond any ai system!