Learning Center
Plans & pricing Sign in
Sign Out

Rule-Based Expert Systems

VIEWS: 203 PAGES: 28

  • pg 1
									Rule-Based Expert
           Expert Systems
 Acknowledge that computers do not
  posses general knowledge (common
 Attempt to train computer in a “limited
 Experts have deep, often complex
  knowledge, but generally for a limited
    Can A Computer Do What an
           Expert Does?
 Limited domain
   The system has a finite, and relatively small number
    of things it needs to “know” about.
   The processing might be complex, but computers are
    good at that.
 Can the expert knowledge be extracted?
   Knowledge Engineers do the extracting
   Experts provide their knowledge
   Programmers encoded the knowledge into an “expert
             Some History
 In the early to mid 1980’s the desire to
 build “expert systems” was very high in the
 US and elsewhere.
   Japanese 5th generation computing project
   AI researchers aggressively recruited by
   Expert systems were considered by many to
    be “the next big thing.”
 Overstatement of Capabilities
 The results of several expert systems
 were oversold.
   Prospector didn’t really discover millions of
    dollars worth of Molybdenum.
      The originators of the system never made this
      claim, but despite their efforts to stop it, the story is
      repeated to this day.
   Frequently Cited Examples of
         Expert Systems
   Infectious blood disease diagnosis
 Dendral
   Analyzed mass-spectra (chemistry)
   Geological analysis
 R1
   Configure a VAX computer
Expert knowledge can be difficult to
 Experts often do not really know how they do
  things themselves. Although the expert can
  perform the tasks, he/she does not necessarily
  have access to the mechanism used.
 Experts have reasons to be uncooperative in the
  process of disseminating their expertise.
 Experts often disagree on both processes and
 The process might require judgment that is not
  easily codified.
    Rule-based Expert Systems

 Sets of IF-THEN rules are established to
 codify expert knowledge.
   If <antecedent> then <consequent>
       Antecedents can be combined using logical
         If <a> and <b> or <c> then <consequent>
   IF “3 enemy stones in a row” AND
      NOT “3 friendly stones in a row”
  Then “place a stone in the row with 3 enemy
       Knowledge Engineers
 Tasked with working with the expert to
  extract expertise and codify in a set of
 Has training in the development of expert
  systems, but not necessarily in the
  application domain.
 Know the capabilities of the technology
  and knows how to apply it.
       Expert System Shells
 Separate the mechanisms for making
  inference from the rule base
 Facilitate the entry of rules by non-
 Provide reuse for what would otherwise be
  redundant code across expert systems
   Expert System Components
 Inference Engine
   Forward or Backward-chaining
   Conflict resolution algorithms
 Rule-base
   IF-THEN rules
 Database
   Current state on which IF-THEN rules are applied.
 Explanation Facilities
   An important advantage rule-based expert systems
    hold over other types of AI
          Inference Engines
 Forward-chaining
   Submit current data to all rules
   Rules make conclusions, which in turn,
    generate new data
   “Inference Chains” result from initial data and
    the data generated in conclusions.
 Backward-chaining
   Try to prove a conclusion by working
    backwards from ways to prove it.
Forward-chaining Example (A,B,E,
        and D are given)
   If Y and D then Z
   If X and B and E then Y
   If A then X
   If C then L
   If L and M then N
    A           X

    B                    Y


                              Example from Negnevitsky
         Backward Chaining
 Prolog uses backward chaining
   Work backward from the goal.
   Check rules that can provided the desired
    Backward Chaining Example
   If Y and D then Z
   If X and B and E then Y
   If A then X
   If C then L
   If L and M then N
    A            X

    B                    Y    Z
    D                    D

Forward or Backward Chaining?
 What do experts use?
 Are we trying to prove a particular
   Backward chaining
 Are we trying to find all possible
   Forward chaining
 What does the rule set look like?
   Could be either one or a combination of both.
          Conflict Resolution
 What happens when two rules provide
 conflicting conclusions?
   If it has feathers then it is a bird
   If it can’t fly then it is not a bird
   What if has feathers, but can’t fly?
   Conflict Resolution Methods
 Use rule-order as an implied priority
    The first rule to provide an answer is used.
 Assign a priority to each rule, the rule with the
  higher priority is sustained.
 Longest Matching Strategy uses the rule with
  the most specific information.
    If it cannot fly and has feathers then it is a bird.
 Certainty-based conflict resolution
    Measures of certainty are provided for data and rules.
     Most certain rule is sustained.
 Frame-based Expert Systems
 Frame
   Marvin Minsky (1975)
 Frame-based Expert Systems utilize
  frames to encapsulate data and methods
  about an entity.
 Frames are similar to objects, but the data
  types and processing methods are quite
 A frame is a data structure with typical
  knowledge about a particular object or concept
    Frame is a collection of attributes called “slots”
       Example slots for a truck
             Engine size
             Number of wheels
         Slots consists of attribute/value pairs called facets
         Value/18
         Default Value/4
         Range/[3-18]
         User Query/”Enter the number of wheels:”
 Slots or facets can contain procedures that are
  executed with the data is accessed or changed
   When_Changed demon is executed when new
    information is placed in a slot.
       Might include forward chaining or backward chaining rules
   When_Needed demon is executed when information
    is read from a slot
       Might include code to read sensors or try to prove a goal
 Frames can inherit from other frames.
   Frame implicitly contains all the slots
    contained in the inherited frame unless the
    frame overrides the slot with its own definition.
   Inheritance is established with the “IS-A”
   In frames, inheritance is principally used to
    provide default values, rather than structure
    and methods.
   Other Frame Relationships
 Aggregation (a-part-of)
   An engine is a-part-of a car
   A spark plug is a-part-of an engine
 Association (Other semantic relationships)
   Examples
      Ownership (computer has-owner Joe)
      Uses (dentist uses drill)
      Location (Joe is-near theDesk)
    No Limits on Relationships
 Frame can employ multiple inheritance
 Frame can have any number of
 Relationships can be of any type that is
 Interactions of Frames and Rules
 Different frame-based systems use
  different mechanisms.
 Rules are often invoked by demons.
  Some systems allow different rule sets to
  be applied to different frames
         Why Use Frames?
 In large systems, frames can provide the
  system the capability to find relevant
  information quickly.
 Making inferences from the most relevant
  information can provide greater efficiency
  and allow searches to be constrained.
 Relationships between frames can be
  provided at a relatively low cost.
 Advantages of Expert Systems
 Provide an explanation capability
   What rules fired to provide the conclusion?
   Why other conclusions were not made.
 For simple domains, the rule-base might
  be simple and easy to verify and validate.
 The system might use a method similar to
  what the expert uses.
 Expert system shells provide a means to
  build simple systems without programming
    Disadvantages of Expert Systems
 When the number of rules is large, the effect of
    adding new rules can be difficult to assess.
   Expert knowledge is not usually easily codified
    into rules.
   Expert often lack access to their own analysis
   Validation/Verification of large systems is very
   Track record does not seem to contain many
    successes. Relatively high-risk to implement.

To top