JPH02230330A - Inference device - Google Patents

Abstract

PURPOSE:To facilitate an input/output procedure by automatically generating a procedure giving and receiving data between data of a data file and a knowledge base by means of a procedure type language in accordance with a relation stored at the time of converting an inference rule into the procedure type language by a language conversion means. CONSTITUTION:A relation description means 2 describes a corresponding relation between an inference variable given from the input means 1 of the inference variable and the field of a file in which data that is to be read into the inference variable is stored, and a storage means 3 stores the corresponding relation. A rule description means 4 describes the inference rule constituting a knowledge base by the knowledge type language, and the described inference rule is stored in the storage means 3. The language conversion means 5 converts the corresponding relation of the stored inference variable and external data, and the inference rule into the procedure type language. A program described by the procedure type language is translated in the machine language by a translation means 6 and the knowledge base 7 is constituted by the translated inference rule. Thus, the input/output processing of data can be executed by the description of the corresponding relation between external data and the variable in a device.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an inference device, and more specifically, to an inference device that executes inference while exchanging data with an external database or program. [Prior Art] When an inference device uses data in a database external to the device or outputs results obtained by executing inference processing to an external database, etc.
Previously, input/output procedures written in traditional procedural languages to call external programs were executed each time within the knowledge base. The actual exchange of data was executed by an external program as an input/output procedure to which control was transferred from the inference device. An example of such a method is disclosed in Japanese Patent Application Laid-Open No. 62-2318.
It is disclosed in the invention of the publication number. In this invention, as shown in FIG. 5, the external data calling function 10 is written in a conventional procedural language separately from the knowledge base 7, and after this is translated into machine language by the translation means 6,
It is configured to connect with a knowledge base. On the other hand, the inference rules are written in a knowledge type language by the rule description means 4, converted into a procedural language by the language conversion means 6, and then translated into machine language by the translation means to form the knowledge language 7.

[Problem to be solved by the invention]

With conventional inference devices such as those mentioned above, each time a situation arises that requires human output of data, it is necessary to program procedures for inputting and outputting data using a procedural programming language, which requires a large amount of man-hours. ．． Furthermore, calling an input/output procedure program inside the inference device is generally executed by specifying the name of an external program as an input/output procedure, so it is not possible to actually use only the description on the knowledge base. There is also a problem in terms of readability, as it is difficult to determine the contents of the data exchange that is executed, and it is difficult to change and maintain the inference device. The present invention was made in view of these circumstances, and
When an inference device inputs and outputs data to and from an external database, etc., the input/output procedure can be performed in a simple manner, eliminating the need for programming, and the contents of the data input/output can be determined from the contents of the knowledge base of the inference device. The purpose is to provide an easily readable reasoning device. [Means for Solving the Problems] The inference device of the present invention creates a knowledge base by writing inference rules, and also determines the correspondence between the inference variables used therein and the data structure of a file that executes data input/output. Create a relationship and store it in a storage device. Specifically, this correspondence is the correspondence between the data item of the file from which data should be read and the inference variable of the knowledge base in which it should be stored, or the inference variable of the knowledge base and its contents are output and stored. This is a description that shows the correspondence with the data item of the file to be used. This correspondence is interpreted by a language conversion means, and a program for executing data input/output processing is created using a procedural language. At the same time, the knowledge base is also interpreted by the language conversion means and converted into a procedural language, but at this time, input/output instructions included in the knowledge base are replaced with instructions for calling a program for data input/output processing.

[Effect]

When the procedural language program obtained through the above processing is input to a machine language translation device, an executable machine language knowledge base is finally created as the output. [Embodiments of the Invention] Hereinafter, the present invention will be described in detail based on drawings showing embodiments thereof. FIG. 1 is a block diagram showing a schematic configuration of an inference device according to the present invention. 1 in the figure is an inference variable input means, which inputs inference variables and outputs them to the relationship description means 2. The relationship description means 2 describes the correspondence between the inference variable given from the inference variable input means 1 and the field of the file in which the data to be read into the inference variable is stored.

The storage means 3 is, for example, a magnetic disk device or the like, and stores the correspondence described by the relation description means 2. For example, 3 rfieldlfi as external files
When using rfileJ consisting of eld2 and field3J and reading IlieldN in that rfileJ into the inference variable rvariable l J, rf
For example, the structure of ilelJ is described as shown in Figure 2,
Also, the inference variables rvariable I J and rfile
The correspondence relationship with the field of lJ is described as shown in FIG. 3 and stored in the storage means 3. 4 is a rule description means, which describes inference rules forming the knowledge base in a knowledge-based language. That is, as an inference device, there is a general “if “then” type rule execution unit,
That is, write an input operation command that specifies the file name after "then". For example, as shown in FIG.
The command “READ (file
)" is written. Note that a general-purpose editor can be used as the rule description means 4, and the written inference rule is stored in the aforementioned memory 1000 stages 3. 5 is a language conversion means, As described above, the correspondence between the inference variables and external data and the inference rules stored in the memory 3 are converted into a procedural language.The processing by the language conversion means 5 is performed using the file shown in FIG. According to the structure description and the correspondence between quasi-variables and fields shown in Figure 3, read the data, that is, the field contents, from the file and execute the input procedure to assign the necessary field values to the corresponding inference variables. This program is written in a procedural language. Also, when converting an inference rule, an input command specifying a file name in the inference rule execution part shown in is converted into an instruction to call the above program.

The output of the language conversion means 5, which is a program written in a procedural language, is given to the translation means 6. The translation means 6 translates the program given from the language conversion stage 5 into machine language using a program language compiler.

A knowledge base 7 is constructed from the inference rules translated into machine language by the translation means 6.

〔Effect of the invention〕

As detailed above, according to the inference device of the present invention, there is no need to program input/output procedures in conventional procedural languages, and instead of writing a description of the correspondence between external data and internal variables of the device. Data input/output processing can be executed using high-level descriptions, reducing program creation time and improving the readability of the knowledge base.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of the inference device according to the present invention, FIG. 2 is an explanatory diagram showing the structure of an external data file to be read into the inference device of the present invention, and FIG. 3 is the inference device of the present invention. An explanatory diagram showing the correspondence between inference variables and external data files to be read into them, Figure 4 is a schematic diagram showing an example of writing an inference rule for performing data reading processing, and Figure 5 is a conventional diagram. FIG. 2 is a block diagram showing the configuration of an inference device. ■... Inference variable input means 2... Relationship description means 3... Storage means 4... Rule description means 5.
...Language conversion means 6...Translation means

Claims (1)

[Claims] 1. Rule description means for writing inference rules constituting a knowledge base in a knowledge-based language, and converting the inference rules written in the knowledge-based language into a procedural language by the rule description means A system comprising: a language conversion means; and a translation means for translating inference rules converted into a procedural language by the language conversion means into machine language; and between an external data file written in a procedural language and the knowledge base. an inference device that sends and receives data, comprising a storage means for storing a predetermined relationship between the data in the data file and the knowledge base, and the language conversion means is configured to perform processing when converting inference rules into a procedural language. The inference device is characterized in that it automatically generates in a procedural language a procedure for transmitting and receiving data between the data in the data file and the knowledge base according to the relationship stored in the storage means. .