JPH02259873A - Method for synthesizing table processing program - Google Patents

Abstract

PURPOSE:To synthesize a complete program code, which is coincident with request information, by extracting suitable program parts from a meaning network knowledge base, in which the existent table processing program parts are accumulated, and matching an interface between the parts. CONSTITUTION:An inference processor 2 and a meaning network knowledge base 3 are provided and the table processing program, which is already prepared, is accumulated as the program parts in the expression of meaning network knowledge corresponding to a meaning network. By using the data type of a table writing item extracted from the request information, the suitable program parts are selected from the meaning network, in which the existent table processing program parts are accumulated, by pattern matching, and the interface state of a selected program parts group is decided. When there is mismatching, a mismatched part is corrected by inference and as the program parts group to be matched to the request information, the whole program code is generated from the group of these program parts. Thus, the table processing program can be efficiently generated.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a software development support system, and is particularly suitable for when a program designer creates a new program using the definition of an existing program. Regarding technology for storing and using program information.

[Conventional technology]

Conventionally, a table creation method as described in Japanese Patent Application No. 126052/1982 has been known.

[Problem to be solved by the invention]

The above conventional technology focuses on hierarchical relationships between table items.
Describe the hierarchical relationships between table items and the attributes of the items themselves,
This is a method of generating a table processing program that the user wants to develop, and the user can understand only the logical relationships of the table without knowing the hardware characteristics of the device that creates the table (screen memory address, screen display control code, etc.). Although it is possible to automatically generate a table processing program by writing a table, there is a problem in that the user must provide all the relationships between items and the attributes of the items.

An object of the present invention is to create a table processing program that matches the user's request information for table processing by using the table processing structure and semantic content of the table processing program that has already been developed. The goal is to generate it efficiently.

[Means for solving the problem] The above purpose is to store a table processing program for table creation as a program component in a semantic network type knowledge representation by making it correspond to a semantic network, and to store table description items extracted from request information. Using the data type, an appropriate program component is selected by pattern matching from a semantic network in which existing table processing program components are accumulated, and the interface state of the selected program component group is determined.

If there are any inconsistencies, the inconsistencies are corrected by inference and a group of program parts that are consistent with the required information are created.
This is achieved by generating the entire program code from a group of these program parts.

[Effect]

The method of the present invention expresses an existing table processing program using a semantic network of table processing structures, and extracts necessary items for table processing from the request specifications by analyzing the request specifications from the user. Components corresponding to the created table processing items are extracted from the semantic network.

A semantic network is a format in which a table processing program gives a main structure in which the root part has an order relationship between leaves and corresponds to the table processing creation item, and the table processing item corresponding to the root and the items included in that item are Expresses the relationship with lower-level table processing items.

An appropriate tree structure is selected by matching the data type of the table processing item in the request information with the data type of the root node of the tree structure, which is a partial structure in the semantic network. Each tree structure becomes a program component of a table processing program.

We aim to harmonize the interfaces of appropriate program component groups for the cross-linked request information. The status of the interface between program parts is: ■ Normal connection, ■ If there is a conflict,
(2) When there is a shortage of lower-level program parts, (2) The data type of a necessary F-rank program part may be missing from the upper-level program parts.

In the cases of ■, ■, and ■, the program parts are made consistent by inference.

A method for generating a program code that meets the required specifications is to create a single program code by combining codes corresponding to each of the harmonized program parts of the interface between parts into a single program code along the data flow.

〔Example〕

Hereinafter, one embodiment of the present invention will be explained with reference to FIGS. 1 to 11.

First, the configuration of this embodiment will be explained. FIG. 2 shows an overall configuration diagram of this embodiment.

Reference numeral 1 denotes an input/output terminal device, which is used to input requests that the user wants to create, and to display and respond to inquiries from the inference section. 2 is an inference processing device, and an input/output terminal device 1
This is a device that uses inference to generate a table processing program that meets the required specifications based on the input from the computer. 3 is a semantic network type knowledge base, in which existing table processing programs are expressed and stored as program parts and used as an inference processing bag.

Station 2 uses it as a knowledge base.

The inference processing bag [2 includes a request information analysis section 4, a program component selection inference section 5, a program component matching inference section 6, and a program code generation section 7. The request information analysis unit 4 analyzes the request information from the input/output end floor bag f11 to the table processing program of usage wealth, and extracts the data type of the table processing item required by the table processing program. The program component selection inference section 5 is.

Data type of table processing item extracted from request information.

Appropriate program parts are selected and extracted by matching the data types of table processing items stored in the semantic network type knowledge base 3. Program component consistency inference section 6
The program harmonizes the interfaces between the selected and extracted program parts by inference. The data value input component creation unit 30 creates a program component that transfers data values from another system (such as a database) to the value area of a table item. The program code generation unit 7 generates an entire program code that meets required specifications from a group of harmonized program parts.

As shown in FIG. 11, the table processing program is a data value input program 61 from another system.

It consists of a display dialog driver program 62. Among these, creating the program 62 is the central technical issue. Therefore, first, a method for the program 62 will be described.

The structure of the program parts in the semantic network type knowledge base 3 is shown in FIG. 3. 8, 9, etc. indicate the program parts in the semantic network type knowledge base 3. 10.
13 represents table processing items of program component 8゜9;
The nodes 11, 12, 14, 15, etc. below 10.13 are the data types of items used to define the data type of table processing item 10.13. The program parts are 8
9, it corresponds to water that constitutes a part of the entire network structure. In the example shown in Figure 3, the subtree structures showing program parts 8 and 9 are both multi-layered, but 1
A single-layer subtree structure consisting of only 0, 11, and 12 can also be treated as a program component.

The program component 9 is expressed as shown in FIGS. 4 and 5. In Figure 4, the item 16 is listed as 17.18.
The relationship between 17° and 18 is 19
Table constructor shown in (nesting and order relationship of table structure)
It shows that it is expressed as . FIG. 5 shows the definition of the table processing program component shown in FIG. 4 as a data type definition. Here, the left side of the definition expression represents the data type name of the parent (root) item of the tree structure, and the right side represents the relationship that defines that data type (the data types that define the left side and their order relationship). Attributes for each item, such as the total item being processed in the 0 table shown, are stored in the attribute information storage area for each item.

Next, the operation of this embodiment will be explained using the operation flow shown in FIG.

In this embodiment, first, request information from the user is analyzed in step 32. Request information can be entered using keywords or Japanese text that can extract the data type of table processing items, or by displaying a table created by an existing table processing program component on the input/output terminal 1 and specifying the necessary items therein. The 0th order obtained by doing this, 33, the 1st order that extracts the data type of the table processing item from the analysis result of 32, and the 1st order, 34.

Components corresponding to the data type obtained in step 33 are extracted from the semantic network. Extraction from the semantic network is performed by pattern matching between the data type of the table processing item in the request information and the data type of the definition of the program component in the semantic network.

Next, the interface between the program parts corresponding to the data type extracted in step 35 is checked. the result,
At step 36, it is determined whether or not there is a mismatch between program parts interfaces.If there is a mismatch, the type of the mismatch is determined at step 37.

If the type of inconsistency is missing, an attempt is made in step 38 to additionally extract parts of the missing data type from the semantic network. As a result, in step 41, it is determined whether or not a program component corresponding to the missing data type exists in the semantic network.

If it exists in the semantic network, it is extracted from the semantic network and added by 42. If it does not exist in the semantic network, 43 creates a program part corresponding to the missing data type through interaction.

37 types of inconsistency include when unnecessary parts exist,
39, unnecessary parts are specified and deleted.

As a type of inconsistency in 37, if there is a mismatch in the interface between program parts corresponding to the extracted data type, the inconsistent part is specified by 40. Next, the specified inconsistent part is specified by 44. Determine whether or not it can be corrected by inference. As a result, if it can be modified, it is modified by reasoning using knowledge in step 45. If it cannot be corrected, the conflicting portion is resolved through dialogue in step 46.

Next, after processing steps 42, 43, 45, and 46, it is determined whether or not there is any inconsistency in the interface between the entire program components, and if there is a match, a program component 47 for inputting data values from other systems is created. .

To create a program component for inputting data values from other systems, create a program that inputs files from other systems such as databases and outputs pairs of item names and item values used in the table display program. Program components that input data values from other systems use a module for accessing other systems such as a database system (access components, which are often attached to databases), search for the necessary data for table items, and Next, in step 48, the program codes of the program parts are synthesized to output a completed program code.

The method will now be explained using a simple example.

FIG. 6 shows an outline of the example problem. Reference numerals 48, 49 and 50 are program parts in the semantic network. 51 is request information. 52 is a table generated by this method. 7 to 9 show tables for program parts 48 and 49.50, respectively. Next, data type definitions of table processing items for program parts 48 and 49.50 are shown.

Definition of Part A (Profit and Loss Statement): : = (Sales) <Cost of Sales><Selling, General and Administrative Expenses><Cost of Sales> : := <Product Inventory Amount><Product Manufacturing Cost for the Current Period><Total><Ending product inventory><Grossprofit> Definition of part B <Product cost report> : :=<Material cost°><Laborcost>
Indirect costs〉〈Product manufacturing cost for the current period〉 Definition of part C〈Cost of sales〉 ::= (Starting inventory of merchandise〉〈Purchases of simple items for the current period〉〈Ending inventory of merchandise〉 Gross profit on sales〉〉〈Cost of goods sold for the current period〉 >::=<Total purchase amount of g products><Purchase discount and return amount><Total> It is assumed that the following information is obtained as required information that is input to this method.

1. Type of table to be created Income statement 2. Table processing items Sales, cost of goods sold, selling, general and administrative expenses, etc. 3. Target industry Distribution industry (not manufacturing, therefore not product items) ,
(requires product item) In this method, the following inferences are made regarding the interface between parts.

1. Between part A and part B Part B is a subordinate item of part A, and if all the items of part A connected by the semantic network are required, part ■ is necessary from the logical relationship. It is. However, in the example problem, the product item for part A is unnecessary, so part B is also unnecessary.

2. The data type definition of cost of sales between parts A and C is inconsistent between parts A and C (mismatching) 0 Part A is defined as a product, but part C is defined as a product. In example 0, which is defined as a product, it is a distribution industry, so it is necessary to define it as a product. Conflicts between table processing items are resolved by reasoning using knowledge (if the industry that prepares the income statement is a distribution industry, the cost of goods sold is calculated based on products). As a result, a table shown in FIG. 10 is created.

According to this embodiment, by regarding the program parts that make up one table processing program as the data type of table processing item, its definition, and the program code, the selection of program parts for the requested information is performed based on the data type, and the final The program generated can be a program code that is convenient for computer execution, and is effective in that a high-performance program can be efficiently obtained.

〔Effect of the invention〕

According to the present invention, the accumulated table processing program.

Even users who are not familiar with the storage method can select and extract appropriate program parts from the semantic network knowledge base that has accumulated existing table processing program parts by inputting the request information for the table processing program. Since it is possible to harmonize the interfaces between programs and synthesize complete program code that meets the requirements, it has the effect of significantly reducing the number of man-hours required by people involved in program development. Furthermore, since existing high-quality programs can be used, the quality of the software being developed can be improved.

In recent years, programmers have been increasingly required to improve their productivity, and one person is forced to handle multiple business software systems or take on new tasks one after another in a short period of time. Therefore, there is an increasing need for a method that not only improves program productivity but also allows end users of computer systems to develop their own programs by effectively utilizing existing high-quality programs.

[Brief explanation of drawings]

FIG. 1 is an operational flow diagram of an embodiment of the present invention, FIG. 2 is a block diagram showing the functional configuration of an embodiment of the present invention, and FIG. 3 is a diagram showing the configuration of a semantic network type knowledge base of an embodiment of the present invention. Explanatory diagrams, Figures 4 and 5 are explanatory diagrams showing the data type definition structure of table processing items, Figure 6 is an explanatory diagram showing an overview of this embodiment, and Figures 7, 8, and 91fi are programs. FIG. 10 is an explanatory diagram showing an example of an output table. FIG. 11 is a block diagram showing the relationship of program configurations in an embodiment of the present invention. 1... Input/output terminal, 2... Reasoning processing device, 3...
Semantic network knowledge base for program parts, 4.
...Requirement information analysis section, 5. Program component selection/inference section, 6. Program component matching inference section, 7. Program code generation section. ■ Figure 1. Figure 1. Figure 1. Figure 1. Figure 1.

Claims (1)

[Claims] 1. In a table processing program creation method comprising a request information analysis section, a program component search utilization section, and a program code output section, table processing is performed using a tree structure in which leaves have an order relationship with each other and a tree structure. Since it can be expressed by the attributes of each contact point,
Existing table processing programs are represented in a tree structure and stored as program parts in a semantic network type knowledge base, and request information for the table processing program is analyzed and appropriate program parts are created by matching with the stored program parts. and inferring the consistency state of the interface between the extracted program parts, and inferring the inconsistency of the interface between the parts so that missing, mismatched, and unnecessary parts of the interface between the parts meet the requirements. 1. A method for synthesizing a table processing program, the method comprising: a step of resolving the problem, and a step of generating a program code from program parts that have harmonized interfaces between program parts. 2. Input of the requested information can also be specified by displaying the table format of an existing table processing program on an input/output terminal and allowing the user to select necessary table items, as described in claim 1. A table processing program synthesis method. 3. Input the values of table processing items by correlating them with item names from other system devices such as databases,
A table processing program synthesis method according to claim 1, which is capable of displaying a table that meets a request from a user and setting values for items.