A fast method for a bidirectional inference

Claim 2

A bidirectional inference method comprising the steps of: (a) transforming a rule condition part into a first network comprising a node representative of each condition constituting said rule condition part, an arc representative of each relation between said conditions, and a root node; (b) transforming a rule consequence part into a second network comprising a node representative of each statement constituting said rule consequence part, an arc representative of each relation between said statements, and a root node; (c) obtaining a third network by coupling said first and second networks by terminal nodes thereof; (d) applying data representative of the status of an object to said third network from the root node of said first network toward the root node of said second network, and determining if the condition represented by each node is satisfied or not; and (e) applying said data representative of the status of said object further to said second network to obtain the result for a rule having said satisfied condition part.

Claim 3

A bidirectional inference method according to claim 1, wherein a goal to be instantiated is applied to said third network from the root node of said second network toward the root node of said first network, identifying a rule by which said goal is guided, applying said goal to said first network coupled to said second network, and obtaining a new sub-goal.

Claim 4

A bidirectional inference method according to claim 1, wherein as an arc constituting said third network having two root nodes representative of said rule condition and consequence parts, added to said third network are an arc (181, 182) on which data flow bidirectionally and an arc (183, 184) on which data flow only unidirectionally, to thereby form a partial bi­directional network.

Claim 5

A bidirectional inference method according to claim 1 or 2, wherein in said third network having two root nodes representative of said rule condition and consequence parts, an arc (64) is set which couples said first and second networks passing over the portion where the common statement/condition is represented, to thereby cause the data representative of the status of said object and said goal to flow via said arc.

Claim 6

A bidirectional inference method according to claim 1 or 2, wherein in said third network having two root nodes representative of said condition and consequ­ence parts, the data representative of the status of said object and said goal are stored at an arc in said third network to control the flow of said data representative of the status of said object, whereby a forward inference is realized through a backward inference by controlling the flow of said goal.

Claim 7

A bidirectional inference method according to claim 5, wherein the data representative of the status of said object and said goal are supplied at the same time to said third network to realize an inference in cooperative association with both forward and backward inferences.

Claim 8

A bidirectional inference method according to claim 1, wherein the relation among patterns described by said rule condition and consequence parts is expressed by a graph which uses as its structural ele­ment two types of patterns including condition and consequence part patterns, and wherein said third network is generated by using said graph.

Claim 9

A bidirectional inference method according to claim 7, wherein as said graph for generating said third network, a hierarchical graph is used which used as its structural element two types of patterns including condition and consequence part patterns, and which uses an arc representative of the inclusion relation between patterns.

Claim 10

A bidirectional inference method according to claim 7, wherein as said graph for generating said third network, a hierarchical graph is used which uses as its structural element two types of patterns including condition and consequence part patterns, and which uses an arc representative of the inclusion relation between patterns and a matchable arc between patterns having the same upper pattern.

Claim 11

A bidirectional inference method according to claim 8, wherein out of a hierarchical graph which uses two types of patterns including condition and consequence part patterns and an arc representative of the inclusion relation between patterns, only patterns constituting said condition part are derived, a difference of condi­tions between patterns coupled by said arc is expressed as a node for said arc, and a terminal node representative of a satisfied pattern is set, to thereby generate said first network.

Claim 12

A bidirectional inference method according to claim 8, wherein out of a hierarchical graph which uses two types of patterns including condition and consequence part patterns and an arc representative of the inclusion relation between patterns, only patterns constituting said consequence part are derived, a difference of conditions between patterns coupled by said arc is expressed as a node for said arc, and a terminal node representative of a satisfied pattern is set, to thereby generate said second network.

Claim 13

A bidirectional inference method according to claim 7, wherein out of said graph, a pattern constituting said condition part and an arc representative of the inclusion relation are derived, a difference of condi­tions between patterns coupled by said arc is expressed as a node for said arc, and a terminal node representative of a satisfied pattern is set, to thereby generate said first network.

Claim 14

A bidirectional inference method according to claim 7, wherein out of said graph, a pattern constitut­ing said consequence part and an arc representative of the inclusion relation are derived, a difference of conditions between patterns coupled by said arc is expressed as a node for said arc, and a terminal node representative of a satisfied pattern is set, to thereby generate said second network.

Claim 15

A bidirectional inference method according to claim 9, wherein in case where arcs from upper patterns to patterns coupled by a matchable arc are present respectively within said condition part network and said consequence part network or vice versa, and in case where an arc to a terminal pattern representative of a satisfied pattern and an arc to the node for determin­ing if a lower pattern than said satisfied pattern is satisfied are present respectively within said condition part network and said consequence part network or vice versa, a shortcut arc is set for coupling therebetween.

Claim 16

A bidirectional inference method according to claim 14, wherein the arcs from the root node of said condition part network to a first occurrence short­cut arc are inhibited to make data flow toward said root node.

Claim 17

A bidirectional inference method according to claim 14, wherein the arcs from the root node of said consequence part network to a first occurrence shortcut arc are inhibited to make data flow toward said root node.

Claim 18

A bidirectional inference method according to claim 14, wherein a plurality of arcs are joined together, said arcs except one arc are inhibited to make data flow toward the junction.

Claim 19

A bidirectional inference method which uses a rule relation graph generated by the steps of: (a) generating a graph representative of a hierarchical relation between patterns, based on the inclusion relation between patterns described by a rule condition part and rule consequence part; (b) detecting patterns having the same upper pattern in said graph, and setting an arc between said detected patterns; (c) selecting a pattern constituting said condition part and an arc representative of the inclusion relation between patterns from said graph, and using a difference between patterns coupled by said arc as a node for said arc, to thereby generate a condition part network; (d) selecting a pattern constituting said consequence part and an arc representative of the inclusion relation between patterns from said graph, and using a difference between patterns coupled by said arc as a node for said arc, to thereby generate a consequence part network; (e) detecting a pattern whose first arc to the terminal node of said condition and consequence part networks and second arc to the node of a pattern lower than said pattern are present within both said condition and consequence part networks; (f) if said pattern is detected, setting an arc coupling said first and second arc; and (g) detecting that third arcs to patterns coupled by a matchable arc are present within both said condition and consequence part networks, and setting an arc coupling said third arcs.

Claim 20

A bidirectional inference method which uses a rule relation graph generated by the steps of: (a) generating a graph representative of a hierarchical relation between patterns, based on the inclusion relation between patterns described by a rule condition part and rule consequence part; (b) detecting patterns having the same upper pattern in said graph, and setting an arc between said detected patterns; (c) selecting a pattern constituting said condition part and an arc representative of the inclusion relation between patterns from said graph, and using a difference between patterns coupled by said arc as a node for said arc, to thereby generate a condition part network; (d) selecting a pattern constituting said consequence part and an arc representative of the inclusion relation between patterns from said graph, and using a difference between patterns coupled by said arc as a node for said arc, to thereby generate a consequence part network; (e) detecting a pattern whose first arc to the terminal node of said condition and consequence part networks and second arc to the node of a pattern lower than said pattern are present within both said condition and consequence part networks; (f) if said pattern is detected, setting an arc coupling said first and second arc; and (g) detecting that third arcs to patterns coupled by a matchable arc are present within both said condition and consequence part networks, and setting an arc coupling said third arcs.