JPH07262018A - Structuralized knowledge data base preparing method - Google Patents

Abstract

PURPOSE:To provide a general-purpose method for preparing a structuralized knowledge data base from a data base in a table format with a ruled transformatmon procedure by giving irreducibly minimum definition for transformation. CONSTITUTION:A data structuralizing definition statement defining the structure of the structuralized knowledge data base and the items of data required for preparing the structuralized knowledge data base is prepared, the required data are extracted from the table format data base of a transformation source according to the defined contents of the data structuralizing definition statement, and the hierarchy information of the structuralized knowledge data base to be prepared is prepared from the data extracted from the transformation source table format data base. Then, inter-hierarchy data are connected according to the hierarchy information, and the data extracted from the transformation source table format data base are structuralized.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of creating a structured knowledge database from a tabular database, and more particularly, to the structure of a structured knowledge database to be created and the structured knowledge database. The present invention relates to a method for creating a data structured definition document that defines necessary data items and mechanically creating a structured knowledge database from a conversion source table format database according to the definition content of the data structured definition document.

[0002]

2. Description of the Related Art In a knowledge engineering application system, in order to utilize information accumulated in the form of a conventional tabular database, a structured knowledge database represented by an object-oriented database can be constructed from the tabular database. Was often there.

Conventionally, in the work of creating the structured knowledge database, the contents are analyzed for each conversion source table format database, a conversion procedure is determined, and a dedicated conversion program or conversion device is set in accordance with the conversion procedure. It was done individually by creating.

[0004]

However, the above-mentioned conventional structured knowledge database creation method is extremely efficient because the contents must be analyzed and a conversion program created for each conversion source tabular database. Was bad. Therefore, it is difficult to utilize the previously accumulated information, which has been an obstacle to the use of the structured knowledge database.

Therefore, an object of the present invention is to provide a general method for creating a structured knowledge database from a tabular database by providing a minimum required definition for conversion and a regularized conversion procedure. Especially.

[0006]

In order to achieve the above object, the structured knowledge database creating method according to the present invention is necessary for creating the structure of the structured knowledge database and the structured knowledge database. Create a data structured definition document that defines items of various data, extract necessary data from the conversion source table format database according to the definition content of the data structured definition document, and extract from the conversion source table format database Creating hierarchical information of a structured knowledge database to be created from the created data, connecting data between the layers according to the hierarchical information, and structuring the data extracted from the conversion source table format database. To do.

[0007]

The data structured definition document created by the structured knowledge database creating method of the present invention defines the structure of the structured knowledge database to be created and the data items necessary to create the knowledge database. Therefore, the structure of this knowledge database and the data items to be extracted are common minimum necessary information for creating a structured knowledge database from a tabular database.

On the other hand, a series of processes of data extraction, hierarchical information creation, and data structuring performed by the structured knowledge database creation method of the present invention converts various tabular databases into structured knowledge databases. This is a common conversion procedure.

In the structured knowledge database creating method of the present invention, the information necessary for converting the tabular database into the structured knowledge database is obtained from the data structured definition document, and regularized based on this. The structured knowledge database is created by the common conversion procedure.

Therefore, according to the structured database creating method of the present invention, the structured knowledge database can be created by the conversion procedure which is regularized only by defining the predetermined items of the structured definition document. it can. By storing a plurality of the above-mentioned data structured definition documents in a single conversion device and selecting and applying the data structured definition documents, a desired structured knowledge database from a number of tabular databases by a single conversion device. Can be created.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A structured knowledge database creating method of the present invention will be described below with reference to the accompanying drawings.

FIG. 1 schematically shows a conversion device for executing the structured knowledge database creating method according to the present invention and the flow of the processing thereof. In FIG. 1, the conversion device 1 is
A data structured definition document storage unit 2 storing a plurality of data structured definition documents, a conversion processing unit 3, a table format database storage device 4 storing a conversion source table format database, and a created structured knowledge database. And a structured knowledge database storage device 5 for storing

In the conversion device 1 described above, when the structured knowledge database is created, the conversion processing unit 3 uses the data structured definition document storage unit 2 and the tabular data storage device 4 to generate a predetermined data structured definition document, respectively. The conversion source table format database is read, a predetermined conversion process is performed, and the created structured knowledge database is output to the structured knowledge database storage device 5.

The conversion processing in the conversion processing unit is roughly divided into various processes such as data extraction, layer information creation, and data structuring for connecting data between layers. Hereinafter, this conversion process will be described.

FIG. 2 shows an example of the structured knowledge database created. A database of this type is
Generally object-oriented database (OODB: Obje)
ct Oriented Data Base) and is said to be suitable for creating a database with a complex structure such as CAD data or a database with a different data format.
An OODB is an object (object) having each element of data as an entity managed by its name, and a data element including the object (object) is an upper layer of the object, and each layer is an attribute defined in the upper layer. It has the characteristic of inheriting.

In the example of FIG. 2, an element (instance) of data is managed by the name of the part PART-A-P6. The element name of the upper hierarchy (in this case, PART-NAME-PRAT-A 7) is defined in this instance. This upper layer PART-NAME-PART
-A7 defines attributes specific to the hierarchy. PA
For RT-A-P 6, PART-NAME-PRA
Since the T-A7 attribute is inherited, it is not necessary to define the attribute belonging to the upper layer, and the attribute defined here is only the unique information of the instance.

Such an object-oriented database is very compatible when used in a knowledge engineering application system, and it is very easy to adopt a search method in which specialized knowledge is added to data search. .

FIG. 3 shows an example of a tabular database which is a conversion source. In the tabular database of FIG. 3, each row in the horizontal direction is a data element (record).
Each example in the vertical direction is an item of data, and each item of data is distinguished by the number of digits. The data items of the database in FIG. 3 are arranged from the left side of the table in the order of plant name, system name, component name, size, pressure, data, and type.

In this tabular database, as is apparent from FIG. 3, it is necessary to define all the contents of the data items which are the search keys of each record. For example, the data item "SYSTEM-1" must be defined redundantly in each of the first to fourth records containing it,
The data item "PART-B" must be defined redundantly in the second and third records containing it.

According to the structured knowledge database creating method of the present invention, the structured knowledge database shown in FIG. 2 is created from the tabular database shown in FIG. 3 and is unique to each conversion case as shown in FIG. This information is defined in advance in the data structured definition document.

FIG. 4 shows the specific contents of the above data structured definition document. This data structured definition document includes information for extracting data in the tabular database and the extracted structured knowledge database. And information about the structure of the.

The information for extracting the data includes the variable name 8 of the data, the data type 9, the data size 10, and the number of items 11 of the data to be extracted.

On the other hand, in the information on the structure of the structured knowledge database, the number of structured hierarchies is 12, the number 13 of data items forming the hierarchy, the number 14 of data items forming the object, and the object forming the object. The data item number 15 to be executed is included.

In addition, the data structured definition document includes information 16 such as a database name to be created, a title, and an inheritance source, which is unique to each data structured definition document.

In the present invention, the information necessary for structuring is extracted from the database itself. In other words, a certain data item is the target of structured data extraction (data that should be the name of an object)
In that case, the data of the entire database is rearranged and unnecessary parts are removed for the item, the element name created in the hierarchy is extracted, and the information necessary for structuring the hierarchy is obtained.

In this embodiment, a case will be described in which the structured knowledge database shown in FIG. 2 is created from the data structured definition document shown in FIG. 4 and the tabular database shown in FIG. In this case, the structured knowledge database created based on the definition contents of the data structured definition document in FIG. 4 is configured such that the upper hierarchy (class) is composed of each data of the plant name, the system name, and the part name, and the instance ID Is a composite name and type.

FIG. 5 shows a flow of processing for creating the structured knowledge database. Prior to the processing of FIG. 5, the data structured definition document of FIG. 4 is created in advance. In the process of FIG. 5, when the process is started, first, a predetermined conversion source table format database and a data structured definition file are opened (step 100). Next, from the title name of the opened data structured definition document, the structured knowledge database (output file) with the title name is opened (steps 110 and 120). Next, the number 11 of data items to be extracted from the data structured definition and the number of each of these data items are read (step 130). Then, the data conversion name 8 of each data item, the data type 9 and the data size 10 are read. The number of digits is read (step 140). Next, the data of one record (one line of the table) is read from the tabular database, and each variable is assigned according to the number of digits (step 150). Here, when the tabular database of FIG. 3 is read, the first data record has a form as shown in FIG.

At this time, since the item numbers 13 constituting the hierarchy specified in the data structure definition document are 1, 2 and 3,
Data used for structuring are plant name, system name,
It becomes a component name and the intermediate file shown in FIG. 7 is output (step 160). This intermediate file becomes the data used to create the class, that is, the hierarchical information.

Next, the data item number (instance ID item number) 15 forming the object is 3 (P
Since it is ART-NAME) and 7 (TYPE), the instance ID (instance name) is "PART-A-
P ”(step 170).

Next, the instance, that is, the substantial part of the object, is output in a predetermined format like the first part of FIG. 8 (step 180). The output format of the instance in this example is commonly called an S-expression, and first the instance ID, then the class ID, in this case PART-N.
The upper layer of AME-PART-A and SAMPLE-DATA is described, and then the substance part of the data is described. That is, the output of this instance includes connection information between layers as well as the substance of the instance.

Steps 150 to 180 described above
Process is performed for each data record of the conversion source tabular database, it is determined whether the file of the conversion source tabular database is finished after processing each record (step 190), and when it is determined that the class is finished, Move to creating parts.

To create the class portion, the intermediate file of FIG. 7 output when creating the above instance is read, and the data of the intermediate file is read (step 200). In this data, the component name is the lowest class and is directly connected to the instance. Here, a class is created from the lowest part (step 210). When creating a class, if the element already exists, move to higher class creation processing. If the class does not exist, a class ID is created from the data and the class is output (step 220).

The above process is repeated up to the highest class, the next data is input when there is no higher class data, and the same operation is repeated until the intermediate file is completed (step 240). Specifically, the created class part becomes like the second and subsequent parts in FIG.

The set of data in FIG. 8 is shown in FIG. 2 when illustrated. That is, by performing the above processing for each object, a class element and an instance part are created, and each element is an upper class I.
The structured knowledge database of FIG. 2 containing D information is completed.

In the processing shown above, various tabular databases can be converted into structured knowledge databases by the same processing method and the same processing device if various data structured definitions are defined. As a result, it is possible to eliminate the inefficiency of the work of creating the conversion program or device each time the structured knowledge database is created.

Next, a method of applying the present invention to a relational database (relational model) consisting of a plurality of tabular databases related to each other will be described. In a relational database, a tabular database file is represented by a set of tuples (sets) that is a set of data.

When a structured knowledge database is created for the above-mentioned database, the main database file (hereinafter referred to as Master File) and the data structured definition document are read so that the associated database file ( (Slave File), the related file name and common attributes. Next, SlaveF
Read the data structured definition for ile. The common attribute, the number of required data items, the variable name of each attribute, the variable type, the number of characters, etc. are obtained from each data structured definition document. In addition, the size of one record is obtained from the number of data items, the variable type of each attribute, and the number of characters.

FIG. 9 shows an example of the data structured definition document of the relational database. In the data structured definition document in FIG. 9 in which the same parts as those in FIG. 4 are denoted by the same reference numerals and the description thereof is omitted, as the information for reading the related database, the number of related files 17, the relationship file name 18, the common attribute 19 is added.

When a structured knowledge database is created from the relational databases associated with each other as described above, first, a database connection process is performed. Figure 1
0 shows the flow of processing to combine two database files.

In FIG. 10, one of the database files to be combined is a master file (Master File).
e), the other database file is defined as a slave file, and the data structured definition of the Master File is opened (step 300). Next, above Master Fi
The common attribute information is read from the data structured definition of le (step 310), and similarly the data structured definition of Slave File is opened (step 320).

Next, it is checked whether or not a common attribute exists between the two databases based on the common attribute (step 330), and if there is no common attribute, it is considered as an error (step 340). If there are common attributes,
Open each database (step 350). Then Mast
Read each record of Slave File while collating each record of er File with common attribute value (Step 36
0 to step 400), attribute root of Master File and Sl
The attribute roots of the ave File are combined in the enumerated form (step 410). The above process is performed for each data record of the Master File for all data records of the Slave File (step 420), and is repeated for all data records of the Master File (step 430). Since the combined result has duplicate attributes with exactly the same contents,
One of the duplicate attributes is deleted, and the process moves to instance creation using the data associated with the information. Since the combined database files are in the form of one database file in the instance creation, the structured knowledge database can be created by the same method as described in FIG.

[0042]

As is apparent from the above description, according to the structured knowledge database creating method of the present invention, there is a peculiarity to each conversion case which differs depending on the tabular database to be converted and the structured knowledge database to be created. Information is defined in the data structured definition document, and based on this data structured definition document, a structured knowledge database is created by data extraction common to all conversion cases, hierarchical information creation, and data structuring processing. In knowledge engineering application systems,
When using existing external data, it is possible to save a great deal of work such as data analysis involved in data conversion processing and formulation of conversion procedures, making it possible to import data very efficiently. .

[Brief description of drawings]

FIG. 1 is a diagram for explaining a conversion device that executes a structured knowledge database creation method of the present invention and a schematic process flow thereof.

FIG. 2 is a diagram exemplifying a structured knowledge database created by a structured knowledge database creation method of the present invention.

FIG. 3 is a diagram showing an example of a conversion source table database.

FIG. 4 is a diagram specifically showing the contents of a structured definition document.

FIG. 5 is a flowchart showing a processing flow of a structured knowledge database creation method according to the present invention.

FIG. 6 is a diagram specifically showing the contents of one extracted record.

FIG. 7 is a diagram specifically showing the contents of hierarchical information.

FIG. 8 is a diagram specifically showing the contents of structured data.

FIG. 9 is a diagram specifically showing the contents of a structured definition document of a plurality of relational databases having common attributes.

FIG. 10 is a flowchart showing a flow of processing for creating a structured knowledge database from a plurality of relational databases having common attributes.

[Explanation of symbols]

 1 conversion device 2 data structured definition document storage unit 3 conversion processing unit 4 tabular database storage device 5 structured knowledge database storage device 8 data variable name 9 data type 10 data size 11 number of data items to be extracted 12 number of layers 13 layers Data item number that composes 14 Number of data items that compose the number of objects 15 Data item number that composes the number of objects 16 Unique information of data structured definition document

Claims (3)

[Claims] 1. A data structured definition document that defines the structure of a structured knowledge database and the items of data required to create the structured knowledge database, According to the definition contents, necessary data is extracted from the conversion source tabular database, the hierarchical information of the structured knowledge database to be created from the data extracted from the conversion source tabular database is created, and the hierarchy information is created according to the hierarchy information. A method of creating a structured knowledge database, characterized in that the data extracted from the conversion source table format database is structured by connecting data between them. 2. The conversion source table format database is composed of a plurality of relational databases having common attributes, the data structured definition document includes information about the common attributes, 2. The structured knowledge database creation method according to claim 1, wherein the data extraction, the creation of hierarchical information, and the structuring process are performed by associating a data aggregate of a relational database having the common attribute. 3. A hierarchy of a structured knowledge database in which data is extracted from the conversion source table format database by sequentially reading all data items of the conversion source table format database and creating each read data item. The structured knowledge database creation method according to claim 1 or 2, wherein