Inference processor using data consistency holder

Claim 2

An inference processor comprising: a compiler-type rule processor comprising: a rule file for storing user-defined rules; a rule compiler for expanding said rules into a rule network; a production memory for storing compiled rules; a working memory for storing intermediate inference processing results; and an inference engine which makes an inference by using said production memory and said working memory; an interpreter-type rule processor comprising: a rule file for storing user-defined rules; a rule interpreter for successively interpreting said rules; a production memory for storing interpreted rules; an inference engine which uses said production memory for inference processing; and a working memory for storing intermediate inference processing results; and a data consistency holder for maintaining, based on the rule files of both said compiler-type and interpreter-type rule processors, the consistency of data used in common by said compiler-type rule processor and said interpreter-type rule processor by matching the data when said data consistency holder is started by said compiler-type ruler processor and said interpreter-type rule processor after inference processing is completed by both said compiler-type and interpreter-type rule processors.

Claim 3

The inference processor according to claim 1, wherein said inference processor comprises a rule manager to refresh the contents of said production memory in said interpreter-type rule processor from outside said intterpreter-type rule processor without making any logical contradictions in the contents of said production memory.

Claim 4

The inference processor according to claim 1, wherein said inference engine of said interpreter-type rule processor uses a time series characteristic between conditions of said rules being implemented.

Claim 5

An inference processing system comprising: first and second rule files for storing respective user-defined rules; a rule compiler for generating conversion functions using said user-defined rules stored in said first rule file and for expanding said user-defined rules stores in said first rule file into a rule network; and a rule processor which receives said conversion functions and rule network from said rule compiler and which receives said user-defined rules stored in said second rule file and which includes a compiler-type rule processor including (a) a first production memory for storing the rule network received from said rule compiler, (b) a first working memory for storing first intermediate inference processing results and (c) a first inference engine for executing a first inference processing using said first production memory and said first working memory, an interpreter-type rule processor including (a) a rule interpreter for successively interpreting said user-defined rules received from said second rule file, (b) a second production memory for storing interpreted rules obtained by said rule interpreter, (c) a second working memory for storing second intermediate inference processing results, and (d) a second inference engine for executing a second inference processing using said second production memory and said second working memory, and a data consistency holder, formed of said conversion functions received from said rule compiler, for linking said first and second working memories using said conversion functions; wherein said rule compiler includes means for generating said conversion functions when a user-defined rule to be processed from said first rule file is directed to data of said first working memory of said compiler-type rule processor and shared with said second working memory of said interpreter-type rule processor, and wherein the thus generated conversion functions are for converting the data used by said respective interpreter-type rule processor and said compiler type rule processor to maintain the consistency of the data shared by said compiler-type rule processor and said interpreter-type rule processor.

Claim 6

An inference processing system comprising: first and second rule files for storing respective user-defined rules; a rule compiler for generating conversion functions using said user-defined rules stored in said first rule file and for expanding said user-defined rules stores in said first rule file into a rule network; and a rule processor which receives said conversion functions and rule network from said rule compiler and which receives said user-defined rules stored in said second rule file and which includes a compiler-type rule processor including (a) a first production memory for storing the rule network received from said rule compiler, (b) a first working memory for storing first intermediate inference processing results and (c) a first inference engine for executing a first inference processing using said first production memory and said first working memory, an interpreter-type rule processor including (a) a rule interpreter for successively interpreting said user-defined rules received from said second rule file, (b) a second production memory for storing interpreted rules obtained by said rule interpreter, (c) a second working memory for storing second intermediate inference processing results, and (d) a second inference engine for executing a second inference processing using said second production memory and said second working memory, and a data consistency holder, formed of said conversion functions received from said rule compiler, for linking said first and second working memories using said conversion functions; wherein said rule compiler includes means for generating said conversion functions when a user-defined rule to be processed from said first rule file is directed to data of said first working memory of said compiler-type rule processor and shared with said second working memory of said interpreter-type rule processor, and wherein the thus generated conversion functions are for converting the data used by said respective interpreter-type rule processor and said compiler type rule processor to maintain the consistency of the data shared by said compiler-type rule processor and said interpreter-type rule processor.

Claim 7

The inference processor system according to claim 4, wherein said rule interpreter comprises a rule manager to refresh contents of said second production memory from outside said interpreter-type rule processor without making any logical contradictions in the contents of said second production memory.

Claim 8

The inference processor according to claim 4, wherein said second inference engine uses a time series characteristic between conditions of said rules being implemented.