Skip to main content

5. Expert Systems

All Past Paper Questions: https://docs.google.com/document/d/14POvsigutX83XWY9ESokphmqd_9j19Bf3Z7973Wkm_A/edit?usp=sharing

Expert Systems​

alt text

alt text

  • Components of an Expert System
    • knowledge base
      • has a database of facts and rules base
      • facts is built up by the engineer
        • on information and knowledge of the subject specialists
        • also from database that may exist for topic
      • rules base if a set of rules in the form: 'IF...THEN'
    • shell includes
      • user interface
      • explanation system
        • explains the user
          • the chain of reasoning
          • used to arrive at a particular order
      • inference engine
      • knowledge base editor
        • enables knowledge engineer to edit rules and facts
          • of the knowledge base
    • inference engine does the reasoning
      • uses forward chaining
      • backward chaining
      • or both
    • user interface is how the computer interacts with the user
      • displaying questions
      • information on screens
      • enables user to type in answers

Uses​

  • general answer

    • UI asks questions (about illness)
    • patient types in symptoms
    • rules base is a set of rules
    • the inference engine
      • uses data/facts from knowledge base
      • is able to find possible diagnoses
        • by using a form of reasoning
        • uses forward reasoning or backwards chaining or both
      • compares symptoms to whats in knowledge base
        • uses rules base of 'IF...THEN'
    • knowledge base editor
      • enables knowledge enginner to edit
    • possible diagnoses output to user-interface
    • explanation system
      • produces reasons for suggestions
      • and outputs to user-interface

  • scheduling system to route vehicles

    • store locations of each distribution point
    • store type of vehicles being used
    • store working hours
    • scheduler would type in destinations
    • system would match those against available types of vehicle
      • match against locations
    • suggest how many drivers would be needed
    • suggest vehicle needed for each driver
    • suggests allocation of orders to each vehicle
    • ?? suggest list of goods in reverse order ??
      • ?? so that, each vehicle had the first order loaded on to the vehicle last ??
    • system would take into account the fuel and time
    • outputs the most efficient rule

User Interface​

  • used for both input and output
  • how computer interacts with the user
  • description
    • allows user to enter problems
    • questions output by system
    • allows user to input answers to questions
    • further questions are output to the UI
      • based on previous answers
    • outputs suggestions of possible solutions
      • and explanation of findings

Techinician​

  • Enters data using an interactive user interface
  • questions based would be asked from techinician
  • answers are entered to questions
  • techinician would see probabilities of diagnoses output from the system

Knowledge base​

  • comprises of database of facts
    • and a rules base
  • built up by the engineer on information
  • knowledge of subject specialists
  • description
    • holds database of facts
    • that the interface engine searches
    • contains rules base
    • which consists of 'IF...THEN' constructs

Knowledge base editing software​

  • enabled the knowledge engineer to edit rules and facts
    • of the knowledge base

Knowledge engineer​

  • employeed to create an expert system
  • responsible for maintaining the system
  • edits facts in knowledge base
  • edits rules in rules base part of knowledge base
  • deleted unused rules from the rules base

Rules base​

  • set of rules
  • in the form: 'IF...THEN'

Inference Engine​

  • reasoning part of the system
  • description
    • compares data to whats held in knowledge base
    • uses rules base, in the form of 'IF...THEN'
    • produces suggestions
    • uses explanation system to produce explanations
      • as how to solutions were arrived at

Forward Chaining​

alt text

alt text

alt text

  • starts with available data
    • uses inferenece rules to extract more data
    • until a goal is reached
  • use is to search the inference rules
    • until it finds a rule
    • then, uses the 'THEN' part
    • to cause addition of new information
  • rejects any rule that is not true
  • this is iterated repeatedly
    • until a goal is reached
  • this is 'data-driven'

Backward Chaining​

  • starts with a list of goals
    • and works backwards
  • would search inference rules
    • until it finds one
    • which has a 'THEN' part that matches a desired goal
  • if the IF part of the rule is not false,
    • then it is added to the list of goals
  • this is 'goal-driven'
    • because list of goals determines which rules are selected and used

Explanation system​

  • explains to a user the chain of reasoning
  • used to arrive at a particular conclusion