Deduction with Production Rules

0
0
1424 days ago, 489 views
PowerPoint PPT Presentation
Surmising with Production Rules. Forward and Backward anchoring through the tenets may be utilized The two frameworks each have their focal points and burdens and truth be told answer distinctive sorts of inquiry For instance in Mycin a forward affixing framework may answer the inquiry \"what do these manifestations suggest?\" a regressive tying framework may answer the inquiry \"does this patient experience the ill effects of amnes

Presentation Transcript

Slide 1

Surmising with Production Rules The path in which the information base is utilized is dictated by the derivation motor It is a fundamental guideline of generation frameworks that each govern ought to be an autonomous thing of learning and basically oblivious of different principles The induction motor could then just "fire" rules whenever when its premises are fulfilled.

Slide 2

Inference with Production Rules Forward and Backward tying through the principles might be utilized The two frameworks each have their focal points and weaknesses and in actuality answer diverse sorts of question For instance in Mycin a forward affixing framework may answer the question "what do these indications suggest?" a regressive fastening framework may answer the question "does this patient experience the ill effects of amnesia ?"

Slide 3

Inference with Production Rules Before taking a gander at Forward and Backward anchoring in more detail we have to clear up a few things There are two primary issues required in the usage of an administer based framework How is struggle determination actualized? Different systems How are certainties coordinated to rules? Design coordinating required And numerous more issues should be settled

Slide 4

Matching truths to principles First of all we have to characterize a grammar for the guidelines and the actualities In the accompanying slides we will talk about example coordinating expecting a forward affixing framework is utilized

Slide 5

Rule-based Representation Example R0: IF home loan is expected AND financial records has enough cash to pay contract THEN pay contract AND diminish checking parity by measure of home loan

Slide 6

Rule-based Representation Example The past govern utilizes characteristic dialect as a general rule we should confine the run base portrayal to a constrained machine-processable portrayal There is no standard manage punctuation For instance: We can think about the LHS of an administer just like a rundown that contains the name of a question took after by sets of properties and qualities related with that protest

Slide 7

Rule-based Representation Example R0: IF (charge  name contract  status due  sum 500) AND (account  name checking  adjust 500) THEN (attest (pay  thing contract)) AND (expel (charge  name contract  status due  sum 500)) AND (evacuate (account  name checking  adjust 500)) AND (attest (account  name checking  adjust 0))

Slide 8

Rule-based Representation Example Assume the working memory has the accompanying substance WM: (charge  name contract  status due  sum 500  account  name checking  adjust 500) The control will fire

Slide 9

Rule-based Representation Example bill is a question name , status , sum are traits trailed by their qualities evacuate and attest are primitives used to expel and add certainties to working memory Problem: A run would need to be made for each sum and each kind of bill To fathom it we can present factors and operations

Slide 10

Rule-based Representation Example R0: IF (charge  name <BILL>  status due  sum <AMOUNT>) AND (account  name checking  adjust <BALANCE>  (<BALANCE> ≥ <AMOUNT>)) THEN (attest (pay  thing <BILL>)) AND (evacuate (charge  name <BILL>  status due  sum <AMOUNT>)) AND (expel (account  name checking  adjust <AMOUNT>)) AND (declare (account  name checking  adjust (<BALANCE> - <AMOUNT>)))

Slide 11

Rule-based Representation Example Assume the working memory has the accompanying substance WM: (charge  name electric  status due  sum 100  account  name checking  adjust 400) The run will fire The reality adjust 300 will be added to the working memory

Slide 12

Rule-based Representation Example (disjunctions) R0: IF (charge  name <BILL>  status [due over_due]  sum <AMOUNT>) AND (account  name checking  adjust <BALANCE>  (<BALANCE> ≥ <AMOUNT>)) THEN (attest (pay  thing <BILL>)) AND (expel (charge  name <BILL>  status due  sum <AMOUNT>)) AND (expel (account  name checking  adjust <AMOUNT>)) AND (affirm (account  name checking  adjust (<BALANCE> - <AMOUNT>)))

Slide 13

Rule-based Representation Example (disjunctions) Assume the working memory has the accompanying substance (charge  name contract  status due  sum 700) (bill  name electric  status over_due  sum 200) (bill  name water  status not_due  sum 50) (account  name checking  adjust 900) what number circumstances will the lead fire?

Slide 14

Matching truths to rules Once a control linguistic structure has been characterized we should determine how the coordinating of certainties to states of principles will be performed I.e. attempting to figure out whether the LHS of a manage is fulfilled by the certainties in working memory It has been assessed that 90% of a control based framework's run time is spent on performing redundant example coordinating amongst principles and truths in the working memory

Slide 15

The Rete Matching Algorithm What is an answer for this example coordinating issue? Attempt comprehensively to match tenets to actualities one by one Use ordering methods The Rete calculation was the principal effective answer for the truths rules design coordinating issue It stores data about matches in a system structure

Slide 16

The Rete Matching Algorithm Nodes of the system compare to individual condition components Conditions and conjunctions of conditions Each hub has two sets related with it The primary set contains all the working memory components that the condition hub coordinates The second set contains blends of working memory components and the ties which create a steady match of the conditions that tie up to the hub condition

Slide 17

The Rete Matching Algorithm With this arrangement tedious testing of all lead conditions in each cycle is stayed away from Only the hubs influenced by a recently embedded or adjusted reality are checked For instance, consider the principles IF a(X,1) and b(X,Z) THEN g1(X,Z) IF a(X,2) and b(X,Z) THEN g2(X,Z)

Slide 18

The Rete Matching Algorithm Initially the working memory is void begin b(X,Z) a(X,Y) - There is a beginning hub and a hub for each of the manage conditions and conjunctions of conditions. - Arcs are marked with variable ties Y=1 Y=2 a(X,2),b(X,Z) a(X,1),b(X,Z)

Slide 19

The Rete Matching Algorithm Fact a(3,1) is added to the working memory begin a(3,1) b(X,Z) a(X,Y) a(3,1) is stored in the hub named a(X,Y) and will spread through the curve named Y=1 Y=2 a(3,1) a(X,2),b(X,Z) a(X,1),b(X,Z) Rule doesn't coordinate

Slide 20

The Rete Matching Algorithm Fact b(3,4) is added to the working memory begin a(3,1) a(X,Y) b(X,Z) b(3,4) b(3,4) is saved in the hub named b(Y,Z) and will engender through the circular segments named Y=1 and Y=2 Y=1 Y=2 a(3,1),b(3,4) b(3,4) a(X,2),b(X,Z) a(X,1),b(X,Z) Rule matches Rule doesn't coordinate

Slide 21

The Rete Matching Algorithm Fact a(3,2) is added to the working memory begin a(3,1),a(3,2) a(X,Y) b(X,Z) b(3,4) a(3,2) is kept in the hub named a(X,Y) and will proliferate through the bend named Y=2 Y=1 Y=2 a(3,1),b(3,4) a(3,2),b(3,4) a(X,2),b(X,Z) a(X,1),b(X,Z) Rule matches Rule matches

Slide 22

The Rete Matching Algorithm The Rete calculation (and expansions) are broadly utilized as a part of govern based frameworks It considers a proficient coordinating procedure (by and large) An innocent calculation that tries all mixes of principles and realities has exponential many-sided quality

Slide 23

Inference with Production Rules There are two primary issues required in the usage of a manage based framework How is strife determination executed? Different systems How are certainties coordinated to rules? Design coordinating required And numerous more issues should be settled

Slide 24

Inference with Production Rules There are many issues to be chosen while executing the surmising component In what arrange do we check rules? In what arrange do we check realities? Calculations may create certainties that are superfluous to objectives. How would we abstain from creating such certainties? In reverse affixing does not experience the ill effects of this issue

SPONSORS