JP2000231564A - Data mining auxiliary device, data converting method and recording medium with data format conversion program recorded therein - Google Patents

Abstract

PROBLEM TO BE SOLVED: To convert a table format where plural formats mix into an item column format about a data mining auxiliary device supplying data being suitable for a data mining engine performing a data analysis. SOLUTION: A user designates ID columns in which identifiers in a data unit are described, attribute columns in which attributes are described and value columns in which values are described about a processing object table 42. An attribute name is prepared in each combination of the kind of a value in the attribute columns and a name of the value column. The kind of a value described in the value columns in each attribute name is examined and an item is allocated to a part or all of those values. An item code is attached to each item on the basis of the number of appearance data units of each item in the processing object table. The item code is associated with respective data units where the same identifier is described in the ID columns to make them an item column format table.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a data mining auxiliary device, a data format conversion method, and a recording medium on which a data format conversion program is recorded. In particular, the present invention relates to a data mining auxiliary device, a data format conversion method, and a recording medium suitable for supplying appropriate data to a data mining engine that performs data analysis.

[0002]

2. Description of the Related Art Conventionally, a data mining engine that analyzes a correlation between data and finds a meaningful correlation is known. Conventional data mining engines are designed on the premise of binary or item string data. FIG. 13 shows an example of a binary data table. FIG. 14 shows an example of a data table in an item string format.

As shown in FIG. 13, a data unit in binary format (a unit that can be combined as having the same identifier (ID) in data processing) is a row,
In the column, attributes that take binary values (hereinafter, referred to as “items”) are arranged. Either of the two values is arranged in each row, and indicates the presence or absence of the item. In the table shown in FIG. 13, whether or not the items from Item 1 to Item 7 are present is represented by ○ in each of the data units ID1 to ID4.

[0004] As shown by reference numeral 10 in FIG. 14, the item string format is a format in which only existing characteristics are indicated by rows. In the item column format, the presence of an item in each row is often described by a code, and the item 1 in FIG.
The total recording amount can be reduced by separately providing a table 12 (a table for associating the correspondence between the code and the item) indicated by 2.

Some conventional data mining engines have a function of handling data in a spreadsheet format or transaction format. FIG. 15 shows an example of a data table in a spreadsheet format. FIG. 16 shows an example of a transaction format data table.

As shown in FIG. 15, the data unit in the spreadsheet format is a row as in the case of the binary format. In each column in the spreadsheet format, unlike the case of the binary format, attributes that do not always take binary values are arranged. In each line, a character string, a numerical value, or a null value representing a value (attribute value) corresponding to the attribute is arranged.

[0007] As shown in FIG. 16, in the transaction format, three columns of an ID column, an attribute column, and a value column (attribute value column) are used.
A column group consisting of types is used. The value column is one column, and the data unit extends over a plurality of rows. The value of the ID column of each row indicates the affiliation of the row to the data unit.

The applicant of the present invention has disclosed in Japanese Patent Application No. 10-40149 an auxiliary device for converting a spreadsheet format into a binary format. Also, Japanese Patent Laid-Open No. 9-1
Japanese Patent No. 34365 discloses an apparatus for converting a spreadsheet format into a binary format (named as 0-1 attribute in the publication). Further, according to the prior art, an auxiliary device as shown in FIG. 17 can be considered as a data mining auxiliary device for converting a spreadsheet format to an item string format.

[0009] The conventional auxiliary device shown in FIG. The data converter 14 converts a data table other than the spreadsheet format into the spreadsheet format. The spreadsheet format table 16 is generated by the data conversion unit 14 performing the above processing.

The information in the spreadsheet format table 16 is stored in an information search unit 18 for each attribute, an information search unit 20 for each attribute value,
And is supplied to the item sequence creation unit 22. The attribute-by-attribute information investigation unit 18 examines the type of value for each attribute of the spreadsheet format table 16, and creates a value item correspondence table based on the investigation result. The value item correspondence table is a table showing the relation between individual attribute values and the items assigned to them.

When the value item correspondence table is created, one item is basically assigned to one attribute value. However, when there are many types of values, for example, when the attribute values are continuous values, a range is set. And categorize the attribute values, and associate a plurality of values with one item. Also, a plurality of appropriately selected values may correspond to one item, or an item may not correspond to an appropriately selected value.

The attribute-by-attribute information research unit 20 is based on the value item correspondence table provided from the attribute-by-attribute information research unit 18 and, for example, a spreadsheet format table 16.
Find the number of occurrences of each item in. Then, the attribute value-based information investigating unit 20 creates an item code correspondence table by numbering each item with a positive integer as an item code in the order of the number of appearances.

The item string creating unit 22 determines an item corresponding to each attribute value for each row of the spreadsheet format table 16 based on the value item correspondence table. Further, the item sequence creating unit 22 determines the code of the determined item based on the item code correspondence table, and creates an item sequence in which the item codes are arranged.

The auxiliary device shown in FIG. 17 further includes an engine parameter determining unit 24. The engine parameter determining unit 24 determines parameters when the data mining engine operates. For example, when an upper limit and a lower limit of the item appearance frequency are set to operate the data mining engine, those parameters are determined by the engine parameter determining unit 24.

[0015]

As described above, according to the apparatus shown in FIG. 17, conversion from a spreadsheet format to an item string format can be performed. Similarly, according to the conventional technique, the conversion from the transaction format to the item string format can also be performed. However, according to the conventional technology, when handling a table format in which a spreadsheet format and a transaction format are mixed, it is necessary to convert data to one of the formats in advance,
In each case, it was necessary to design and manufacture the data converter.

According to the conventional technique, when a data mining engine is applied by determining a common data unit across a plurality of tables, the plurality of tables are converted into one table in advance, or , They needed to provide a view that users could use to combine them into a single table.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and a wide range can be obtained by enabling conversion from a table format in which a spreadsheet format and a transaction format are mixed to an item column format. It is a first object of the present invention to provide a data mining assisting device that makes it easy to apply a data mining engine to such data. Further, the present invention, when integrating a plurality of tables into one, by performing the integration immediately before converting the data format to the item string format, it is possible to perform analysis for each table in parallel, It is a second object of the present invention to provide a data mining auxiliary device which can reduce an area used for table integration.

[0018]

According to the first aspect of the present invention,
A data mining auxiliary device that supplies a data mining engine that discovers a correlation between data to a data mining engine in a table format that can be processed by the engine. A table view generation unit for a user to specify an ID column for describing an identifier, an attribute column for describing an attribute, and a value column for describing a value; and a value type of the attribute column and a name of the value column. An attribute name determining unit that creates an attribute name for each combination of the attribute names and creates an attribute name table representing the correspondence between the combination and the attribute name, and describes in the value column for each attribute name defined in the attribute name table Investigate the types of values that have been set, determine items for some or all of those values, and indicate the value-item correspondence that corresponds to the item. An attribute-based information research unit that creates an item code correspondence table by attaching an item code to each item based on the number of appearance data units of each item in the processing target table. In the processing target table, an item code corresponding to each data unit in which the same identifier is described in the ID column is checked based on the value item correspondence table and the item code correspondence table. And an item sequence creation unit that creates an item sequence representing a correspondence with the item code.

According to a second aspect of the present invention, there is provided the data mining auxiliary device according to the first aspect, wherein the attribute-based information investigating section determines, for a predetermined type of the values described in the value sequence, The present invention is characterized in that categorization is allowed and that it is possible to allocate items for each category.

The invention according to claim 3 is the invention according to claim 1 or 2
The data mining assisting device according to claim 1, further comprising an engine parameter determining unit that determines a parameter for controlling an operation state of the data mining engine.

According to a fourth aspect of the present invention, there is provided the data mining auxiliary device according to any one of the first to third aspects,
The attribute name determination unit has a row set for each type of value of the attribute column specified by the table view generation unit, and a same column as the value column specified by the table view generation unit In addition, an attribute name table is automatically generated in which the attribute name is defined by combining a value of the attribute column and a name of the value column according to a predetermined rule.

According to a fifth aspect of the present invention, there is provided the data mining assisting apparatus according to the fourth aspect, wherein the attribute name determining unit receives a correction of the attribute name for each attribute name, and And at least one of means for collectively accepting the corrections described above.

According to a sixth aspect of the present invention, there is provided the data mining auxiliary device according to any one of the first to fifth aspects, wherein:
The attribute name determining unit automatically creates a one-line attribute name table having the name of the value column as an attribute name when only the ID column and the attribute column are specified in the table view generating unit and the value column is not specified. It is characterized by generating.

According to a seventh aspect of the present invention, there is provided the data mining assisting apparatus according to the third aspect, wherein the engine parameter determining unit determines which item is determined by using a value item correspondence table obtained from the attribute-based information research unit. Is included in which attribute and which attribute column is included in the value of which attribute column is presented to the data mining engine, and the data mining engine accepts a parameter setting in units of the value of the attribute column. It is a feature.

According to an eighth aspect of the present invention, information of a plurality of tables to be processed having different table formats is converted into a format that can be processed by a data mining engine.
4. The data mining assisting device according to claim 1, wherein the table view generation unit allows specification of the ID column, the attribute column, and the value column for each table format, and the attribute name determination unit. Creates the attribute name table for each table format, the attribute information investigation unit creates the value item correspondence table for each table format, and the attribute value information investigation unit The item code correspondence table, and the item string creation unit creates a single item string based on the value item correspondence table, the item code correspondence table, and the plurality of processing target tables. It is characterized by the following.

According to a ninth aspect of the present invention, there is provided a data format conversion method for supplying data in a table format that can be processed by the engine to a data mining engine that discovers a correlation between data. A table selection step of selecting a table having a data structure of, as an object table, an ID column describing an identifier of a data unit, an attribute column describing an attribute, and a value column describing a value for the object table. I to specify
D column / attribute column / value column selecting step, an attribute name table for creating an attribute name for each combination of the value type of the attribute column and the name of the value column, and representing a correspondence between the combination and the attribute name Creating an attribute name table, and examining the type of value described in the value column for each attribute name defined in the attribute name table, and selecting an item for a part or all of those values. Determining and creating a value item correspondence table representing the correspondence between values and items, and attaching an item code to each item based on the number of appearance data units of each item in the processing target table An item code correspondence table creating step for creating an item code correspondence table, and the same as the ID column in the processing target table. An item code corresponding to each data unit in which an identifier is described is examined based on the value item correspondence table and the item code correspondence table, and an item string representing the correspondence between each data unit and the item code is determined. And a step of creating an item string to be created.

According to a tenth aspect of the present invention, in the data format conversion method according to the ninth aspect, only the ID column and the attribute column are specified in the ID column / attribute column / value column selecting step,
When the value column is not specified, the one-line attribute name table having the name of the value column as the attribute name is automatically generated in the attribute name table creating step.

According to an eleventh aspect of the present invention, there is provided the data mining assistance device according to the ninth or tenth aspect, wherein information of a plurality of tables to be processed having different table formats is converted into a format which can be processed by the data mining engine. A data format conversion method, wherein the ID column / attribute column / value column selecting step includes a step of specifying the ID column, the attribute column, and the value column for each table format, The table creation step has a step of creating the attribute name table for each table format, and the value item correspondence table creation step has a step of creating the value item correspondence table for each table format, and the item code In the correspondence table creation step, the item code correspondence table is generated for the entire processing target table. And wherein in the item string creation step, a single item string is created based on the value item correspondence table, the item code correspondence table, and the plurality of processing target tables. .

According to a twelfth aspect of the present invention, there is provided a recording medium recording a data format conversion program for supplying a table format data processable by the engine to a data mining engine for finding a correlation between data. The program causes a computer to select a table to be processed according to a predetermined input, and describes an ID column and an attribute describing an identifier of a data unit in the table to be processed according to a predetermined input. Recognize an attribute column and a value column describing a value, create an attribute name for each combination of the value type of the attribute column and the name of the value column, and represent the correspondence between the combination and the attribute name An attribute name table is created, and for each attribute name defined in the attribute name table, the type of the value described in the value column is checked. Recognize items for some or all of the values, create a value item correspondence table representing the correspondence between values and items, and assign an item to each item based on the number of appearance data units of each item in the processing target table. By attaching a code, an item code correspondence table is created, and in the processing target table, an item code corresponding to each of the data units in which the same identifier is described in the ID column is stored in the value item correspondence table and the item code. The present invention is characterized in that an investigation is performed based on a correspondence table, and an item string representing the correspondence between each data unit and the item code is created.

According to a thirteenth aspect of the present invention, in the recording medium according to the twelfth aspect, the program specifies only an ID column and an attribute column and does not specify a value column for the processing target table. In the event that
It is characterized in that a one-line attribute name table having the name of the value column of the processing target table as an attribute name is automatically generated.

According to a fourteenth aspect of the present invention, in the recording medium according to the twelfth or thirteenth aspect, there is provided a program for converting information of a plurality of tables to be processed having different table formats into a format that can be processed by a data mining engine. A recorded recording medium, wherein the program causes a computer to execute an ID column, an attribute column, and
A value string is recognized for each table format, the attribute name table is created for each table format, the value item correspondence table is created for each format, and the item code correspondence table is created for the entire processing target table. And generating a single item string based on the value item correspondence table, the item code correspondence table, and the plurality of processing target tables.

[0032]

Embodiments of the present invention will be described below with reference to the drawings. Elements common to the drawings are denoted by the same reference numerals, and redundant description will be omitted.

Embodiment 1 In the present embodiment, a description will be given of a case where the results of inspections performed on a plurality of inspection points in a certain manufacturing process are recorded in one table, and a data mining engine is applied to the table. More specifically, unique inspection items are set for each of the plurality of inspection points, and the inspection results are recorded in one table in a format in which a spreadsheet format and a transaction format are mixed. Will be described.

FIG. 1 is a block diagram showing the configuration of a data mining auxiliary device according to this embodiment. FIG. 2 is a flowchart for explaining the operation. In the table view generation unit 30 shown in FIG. 1, first, the table selection step (S101) shown in FIG. 2 is executed. S101
Then, one table is selected by the user as a processing target of the data mining engine. FIG.
Shows an example of the table selected in the table selection step S101. The table 42 shown in FIG. 3 (hereinafter, referred to as a “processing target table 42”) is a table in which inspection results at a plurality of inspection points are recorded in a format in which a spreadsheet format and a transaction format are mixed, as described above. is there.

Next, the table view generation unit 30
A D column / attribute column / value column selection step (S102) is executed. In S102, the user gives a distinction between an ID column, an attribute column, and a value column to the format of the processing target table 42. FIG. 3 shows “Product I” in the processing target table.
“D” and “input ID” are in the ID column, “inspection point” and “inspection method” are in the attribute column, and “result 1” and “result 2”
And "Result 3" are specified in the value column.

The attribute name determining section 32 shown in FIG. 1 executes an attribute name table creation step (S103) shown in FIG. In S103, the processing target table 42 shown in FIG.
An attribute name table is created which has the same number of rows as the types of values present in the attribute column and the same number of columns as the number of columns in the value column.

FIG. 4 shows the attribute name table 44 created by the process of the attribute name table creation step S103.
And 46 are shown. In the present embodiment, in S103, first, a process of creating the attribute name table 44 is performed. That is, the value of the attribute column (ie, “p” and “1”) of the processing target table 42 and the name of the value column (ie, “result 1”) are separated by the delimiter “_” and “ The process of automatically creating the attribute name table 44 is performed by generating the attribute name “p1_1” or the like by omitting the characters of “result” in common and combining them, and arranging the attribute name in each row and each column. Will be By automatically creating the attribute name table 44 in this way, it is possible to reliably prevent the process from proceeding to the subsequent procedure without specifying the attribute name.

In the present embodiment, in S103 described above, the user is allowed to modify the value of the attribute name table, that is, the attribute name. The device according to the present embodiment displays the automatically generated attribute name table 44 on a user interface displayed in, for example, a spreadsheet format, and corrects a cell on the spreadsheet by correcting the attribute name. There is provided a means for receiving a correction for each attribute name such as interpreting. Further, the apparatus according to the present embodiment further includes, for example, preparing an attribute name in a CSV format file,
There is provided a means for receiving a batch correction of the entire attribute name table, such as correcting the attribute name via a file interface for reading the file and a user interface for selecting a file.

In the above-described step S103, the attribute name table 44 is converted into, for example, an attribute name table 46 by performing the above-mentioned correction by the user. The attribute names used in the attribute name table 46 are:
Compared to the attribute names used in the attribute name table 44, it is easier for the user to intuitively understand. Therefore,
By allowing the attribute name table to be modified as described above, the operability for the user can be improved.

When the attribute name table 44 or 46 is generated, the table view generating unit 30 shown in FIG. 1 converts the table view having the same information as the processing target table 42 (FIG. 3) into the format shown in FIG. Can be provided to the user. The apparatus of the present embodiment has a function of answering a question about the table view 48 in addition to the functions described below.

After the table view 48 shown in FIG. 5 is created, the attribute-by-attribute information examination unit 34 shown in FIG.
Step of creating a value item correspondence table shown in FIG.
The processing of 4) is executed. In S104, the type of value (m, n,...) For each attribute (p type, p current, etc.) of the table view 48 provided by the table view generating unit 30.
3, 1.4, etc.), and the user creates a value item correspondence table 50 as shown in FIG. 6 based on the result.

In the process of S104, an item is defined by combining an attribute name and an attribute value corresponding to each attribute value. In this processing, one item is basically assigned to one attribute value. However, when there are many types of attribute values, for example, the attribute values take a continuous value like “p current” in FIG. In such a case, the range is determined, the attribute values are categorized, and a plurality of attribute values correspond to one item.

For example, as shown in the “p number” column of FIG. 6, one item may be associated with a plurality of appropriately selected attribute values. Further, in FIG.
As shown in the attribute value m of the “p type” and the attribute value 1 of the “p number”, the item may not be associated with an appropriately selected attribute value. In FIG. 6, attribute values not corresponding to the items are set at four places.

The attribute value-based information research unit 36 shown in FIG.
An item code correspondence table creation step (S105) shown in FIG. 2 is executed. In S105, based on the value item correspondence table 50 (FIG. 6), for example, the order in which the items appear in the processing target table 42 (FIG. 3) is checked. In S105, a positive integer is assigned to each item as an item code in the order of the number of appearances, so that the item code correspondence table 52 shown in FIG.
Is created.

The item sequence creating section 38 shown in FIG. 1 executes an item sequence creating step (S106) shown in FIG. In S106, first, for each of the IDs of the table view 48 shown in FIG. 5 (IDs in which the combination of the product ID and the input ID is one unit), the items corresponding to the respective attribute values are stored in the value item correspondence table shown in FIG. 50. Next, in S106, the code of the item corresponding to each ID is determined based on the item code correspondence table 52 shown in FIG. And each ID
And an item code corresponding to the ID, an item string format 54 in which the item codes are arranged as shown in FIG. 8 is created.

The engine parameter determining section 40 shown in FIG.
Then, the process of the engine parameter determination step (S107) is executed. In S107, parameters used when the data mining engine operates are determined.
More specifically, when it is desired to operate the data mining engine with, for example, an upper limit and a lower limit of the item appearance frequency, parameters for realizing a desired setting are determined in S107.

As described above, according to the data mining assisting apparatus of the present embodiment, a table in which the spreadsheet format and the transaction format are mixed, that is, a table in which the meaning of the column differs depending on the row, can be easily set in the item column. Can be converted to a format. For this reason, according to the use of the auxiliary device of the present embodiment, the data mining engine can be applied to a wide range of data formats without designing and manufacturing individual data conversion units.

Embodiment 2 Next, a data mining assistance device according to a second embodiment of the present invention will be described with reference to FIG. FIG. 9 shows a table 56 to be processed by the data mining assistance device (hereinafter, a second processing target table 56) in the present embodiment. As shown in FIG. 9, the second processing target table 56 has a data structure in a spreadsheet format.

The data mining assisting apparatus according to the present embodiment operates in the same manner as the apparatus according to the first embodiment when handling the processing target table 42 as shown in FIG. The data mining assistance device of the present embodiment operates as follows when a spreadsheet format table 56 as shown in FIG. 9 is to be processed.

That is, when the auxiliary device of this embodiment processes the second processing target table 56, the table view generation unit 30 (see FIG. 1) selects an ID column, an attribute column, and a value column (S102). Is executed, only the ID column and the value column are designated by the user as indicated by reference numeral 58 in the figure. When recognizing that the user has specified only the ID column and the value column in S102, the attribute information research unit 34 (see FIG. 1) performs the second process target table 56 in the attribute name master table creation step (S103). The attribute name table is created by using the name 60 of the value column of “1” as the attribute name.

According to the above processing, the processing target table 4
The attribute name table can be easily created without performing the automatic generation of attribute names and the correction of attribute names (see FIG. 4) required in the processing of Embodiment 1 for the second embodiment.
Therefore, according to the auxiliary device of the present embodiment, data in a spreadsheet format can be easily converted to an item string format. As described above, according to the auxiliary device of the present embodiment, a table in a spreadsheet format can be easily converted into a table that can be handled by a data mining engine.

Embodiment 3 FIG. Next, referring to FIG.
Embodiment 3 of the present invention will be described. Some data mining engines can be processed more efficiently by being given a hierarchical structure of items. When the data mining engine is such an engine, the auxiliary device of the present embodiment performs the processing shown in FIG. 6 during the processing of the engine parameter determination step S107 (see FIG. 2) in the engine parameter determination unit 40 (see FIG. 1). Item and attribute name ("p
A type such as “type”) and an attribute name and a value of an attribute column (such as “p1”) are extracted as shown in FIG. 10, and the hierarchical structure is presented to the data mining engine.

According to the data mining assisting device giving the attribute hierarchy described above to the data mining engine as a part of the engine parameters, the information provided by the assisting device of the first embodiment includes the attribute column value, the attribute name, , It is possible to give the engine a deeper hierarchical structure. Therefore, according to the auxiliary device of the present embodiment, the data mining engine can perform the processing more efficiently.

Embodiment 4 Next, FIG. 11 and FIG.
A data mining assisting device according to a fourth embodiment of the present invention will be described with reference to FIG. In the present embodiment, the results of inspections performed at a plurality of inspection points each having a unique inspection item set in a certain manufacturing process are recorded in a plurality of tables, and a case where a data mining engine is applied to the tables is described. explain.

FIG. 11 is a block diagram showing a configuration in the case where the data mining assisting apparatus of the present embodiment processes two tables. FIG. 12 is a flowchart for explaining the operation.

In FIG. 11, the table view generator 6
2 and 64, the attribute name determining units 66 and 68, and the attribute-based information checking units 70 and 72 have the same functions as the table view generating unit 30, the attribute name determining unit 32, and the attribute-based information checking unit 74 shown in FIG. are doing. Also, the table selection steps (S201a and S201a) shown in FIG.
b), ID column / attribute column / value column selection step (S202a)
And S202b), attribute name master table creation step (S203a and S203b), value item correspondence table creation step (S204a and S204b)
Then, respectively, the table selection step (S
101), ID column / attribute column / value column selection step (S10)
2), attribute name master table creation step (S10)
3), value item correspondence table creation step (S10)
The same processing as in 4) is performed.

In the present embodiment, the attribute value-based information investigating unit 74 shown in FIG. 11 executes the process of the item code correspondence table creating step (S205) shown in FIG. In S205, first, based on the value item correspondence tables as shown in FIG. 6 provided by the attribute-based information research units 70 and 72, for example, the tables as shown in FIG. The number of occurrences of the item in the view is determined. Then, in S205, a positive integer (item code) is assigned to all items in the order of the number of appearances of each item, thereby creating one item code correspondence table as shown in FIG.

In this embodiment, the item string creating unit 3
8 and the engine parameter determination unit 40 have the same functions as those of the first embodiment (see FIG. 1). That is, in the present embodiment, the item sequence creation step (S205) and the engine parameter determination step (S206) shown in FIG. 12 are the same as the item sequence creation step (S106) and the engine parameter determination step (S107) shown in FIG. Is performed.

As described above, according to the data mining assistance device of the present embodiment, one item string format can be created for a plurality of tables to be processed.
Therefore, according to the auxiliary device of the present embodiment, the first embodiment
The data mining engine can be applied to a wider data format than the device of the above.

Further, in the data mining assisting apparatus of this embodiment, when a plurality of tables are to be processed, the processing before the value item correspondence table creating step (S204a and S204b) is executed for each table. You. When the value item correspondence table is created, an item may not be assigned to some attribute values, as described in the first embodiment.

The attribute value to which no item is assigned is
When creating an item code correspondence table, it can be handled by excluding it from processing. Therefore, by executing the processing up to the previous stage of the item code correspondence table creation step (S205) for each table, after specifying attribute values that can be excluded from the processing target for each table,
An item code correspondence table can be created. In this case, it is smaller than the case where a plurality of tables are mixed at the stage of the ID column / attribute column / value column selection step (S202a, S202b) or the stage of the attribute name table creation step (S203a, S203b). A desired operation can be performed in the processing area.

[0062]

Since the present invention is configured as described above, it has the following effects. According to the first, ninth, or twelfth aspect of the present invention, a table in which the meaning of a column differs depending on a row can be converted into an item string format that can be handled by a data mining engine. Therefore, according to the present invention, the data mining engine can be applied to a wide range of data formats via the item sequence format without individually designing and manufacturing the data conversion unit.

According to the second aspect of the present invention, items can be assigned by categorizing attribute values, so that it is possible to prevent an unnecessary increase in the number of items. Therefore, according to the present invention, it is possible to reduce the calculation load in the data conversion process.

According to the third aspect of the present invention, by determining the parameters of the data mining engine in the auxiliary device, the engine can be operated efficiently.

According to the fourth aspect of the present invention, since the attribute name table is automatically created, it is possible to prevent the processing from proceeding to the subsequent procedure without specifying the attribute name.

According to the invention described in claim 5, the attribute names defined in the attribute name table can be corrected to names that are easy for the user to understand. For this reason, according to the present invention, the operability of the device can be improved.

According to the sixth, tenth, or thirteenth aspect of the present invention, when a table in a spreadsheet format is set as a processing target table, it is easier to automatically generate an attribute name and to correct the attribute name. An attribute name table can be created.

According to the present invention, an attribute hierarchy can be given to the data mining engine as a part of the engine parameter. Thus, according to the present invention,
It is possible to efficiently operate an engine that can increase the operation efficiency based on the attribute hierarchy.

According to the eighth, eleventh, or fourteenth aspect, one item string format can be created for a plurality of tables. Therefore, according to the present invention, the data mining engine can be applied to a wider data format as compared with the first aspect. Further, according to the present invention, since information obtained from a plurality of tables is not mixed until a step of creating an item code correspondence table, data conversion processing can be performed in a relatively small work area.

[Brief description of the drawings]

FIG. 1 is a block diagram of a data mining auxiliary device according to a first embodiment of the present invention.

FIG. 2 is a flowchart for explaining the contents of processing executed by the auxiliary device shown in FIG. 1;

FIG. 3 is an example of a processing target table handled by the auxiliary device shown in FIG. 1;

FIG. 4 is an example of an attribute name table created by the auxiliary device shown in FIG. 1;

FIG. 5 is an example of a table view created by the auxiliary device shown in FIG. 1;

FIG. 6 is an example of a value item correspondence table created by the auxiliary device shown in FIG. 1;

FIG. 7 is an example of a value item correspondence table created by the auxiliary device shown in FIG. 1;

FIG. 8 is an example of an item string format table created by the auxiliary device shown in FIG. 1;

FIG. 9 is a table in a spreadsheet format that is used as a processing target table in the data mining assistance device according to the second embodiment of the present invention.

FIG. 10 is an example of information on an attribute hierarchy provided to the data mining engine from the data mining auxiliary device according to the third embodiment of the present invention.

FIG. 11 is a block diagram of a data mining assistance device according to a fourth embodiment of the present invention.

FIG. 12 is a flowchart for explaining the contents of processing executed by the auxiliary device shown in FIG. 11;

FIG. 13 is an example of a table in a binary format.

FIG. 14 is an example of an item sequence table and an item code correspondence table.

FIG. 15 is an example of a table in a spreadsheet format.

FIG. 16 is an example of a transaction format table.

FIG. 17 is a block diagram of a conventional data mining auxiliary device.

[Explanation of symbols]

30; 62, 64 table view generation unit; 32;
66, 68 Attribute name determination unit, 34; 70, 72 Attribute information investigation unit, 36; 74 Attribute value information investigation unit, 38 Item column creation unit, 40 Engine parameter determination unit, 42 Processing target table,
44, 46 attribute name table, 48 table view, 50 value item correspondence table, 52 item code correspondence table, 54 item column format table, 56 second processing target table.

Claims (14)

[Claims] 1. A data mining assisting device for supplying a data mining engine for finding a correlation between data to a data mining engine in a table format that can be processed by the engine, the processing target table having a data structure in a table format A table view generating unit for the user to specify an ID column describing an identifier of a data unit, an attribute column describing an attribute, and a value column describing a value; and An attribute name determining unit that creates an attribute name for each combination with the name of the value column, and creates an attribute name table representing the correspondence between the combination and the attribute name; and for each attribute name defined in the attribute name table, Investigate the types of values described in the value column, determine items for some or all of the values, and express the correspondence between the values and the items. An attribute information investigating unit for creating a value item correspondence table; and an attribute value information for creating an item code correspondence table by attaching an item code to each item based on the number of appearance data units of each item in the processing target table. An investigating unit, and an item code corresponding to each data unit in which the same identifier is described in the ID column in the processing target table, based on the value item correspondence table and the item code correspondence table, A data mining assistance device, comprising: an item sequence creating unit that creates an item sequence representing a correspondence between a data unit and an item code. 2. The attribute-by-attribute information investigating unit allows categorization of a value of a predetermined type among values described in the value string, and allows allocating an item for each category. The data mining assisting device according to claim 1, wherein: 3. The data mining assistance device according to claim 1, further comprising an engine parameter determining unit that determines a parameter for controlling an operation state of the data mining engine. 4. The attribute name deciding unit is the same as the row set for each type of value of the attribute column specified by the table view generating unit and the value column specified by the table view generating unit. And automatically generating an attribute name table in which the attribute name is defined by combining a value of the attribute column and a name of the value column according to a predetermined rule. The data mining assistance device according to claim 1. 5. The method according to claim 1, wherein the attribute name determining unit includes at least one of a unit that receives the correction of the attribute name for each attribute name and a unit that receives the correction of all the attribute names collectively. Item 5. The data mining auxiliary device according to Item 4. 6. The attribute name determining unit, wherein only an ID column and an attribute column are specified in the table view generating unit,
6. The data mining assisting device according to claim 1, wherein when a value column is not specified, a one-line attribute name table having a name of the value column as an attribute name is automatically generated. . 7. The engine parameter determining unit uses a value item correspondence table obtained from the attribute-by-attribute information research unit to determine which item is included in which attribute and which attribute is included in the value of which attribute column. 4. The data mining assistance device according to claim 3, wherein the data mining engine is presented with the data, and the data mining engine receives a parameter setting in units of the value of the attribute column. 5. 8. A data mining assisting device for converting information of a plurality of tables to be processed having different table formats into a format that can be processed by a data mining engine, wherein the table view generating unit includes the ID column, the attribute column ,
And, the specification of the value column is permitted for each table format, the attribute name determination unit creates the attribute name table for each table format, and the attribute information investigation unit supports the value item for each table format. Creating a table, the attribute value-based information investigating unit creates the item code correspondence table for the entire processing target table, the item string creating unit includes the value item correspondence table, the item code correspondence table The data mining assistance device according to any one of claims 1 to 7, wherein a single item sequence is created based on the plurality of processing target tables. 9. A data format conversion method for supplying a data mining engine for finding a correlation between data to a data mining engine capable of processing the data in a table format, wherein the table has a data structure in a table format. A table selection step of selecting a table as a processing target table; an ID column and an attribute specifying an ID column describing an identifier of a data unit, an attribute column describing an attribute, and a value column describing a value for the processing target table; A column / value column selecting step, an attribute for creating an attribute name for each combination of a value type of the attribute column and a name of the value column, and creating an attribute name table representing a correspondence between the combination and the attribute name Creating a name table, and examining the type of value described in the value column for each attribute name defined in the attribute name table; A value item correspondence table creating step of determining an item for a part or all of these values and creating a value item correspondence table representing the correspondence between the value and the item; and occurrence data of each item in the processing target table. An item code correspondence table creation step of creating an item code correspondence table by attaching an item code to each item based on the number of units; and a data unit in which the same identifier is described in the ID column in the processing target table. And an item column creation step of investigating an item code corresponding to the item code based on the value item correspondence table and the item code correspondence table, and creating an item sequence representing a correspondence between each data unit and the item code. Data format conversion characterized by Method. 10. When only an ID column and an attribute column are specified in the ID column / attribute column / value column selecting step and no value column is specified, the attribute name table creating step sets the name of the value column to 10. The data format conversion method according to claim 9, wherein a one-line attribute name table as an attribute name is automatically generated. 11. A data format conversion method for converting information of a plurality of processing target tables having different table formats into a format that can be processed by a data mining engine, wherein the ID column / attribute column / value column selecting step comprises: The ID
Specifying a column, the attribute column, and the value column for each table format, and the attribute name table creating step includes creating the attribute name table for each table format, The item correspondence table creation step includes a step of creating the value item correspondence table for each table format. In the item code correspondence table creation step, the item code correspondence table is created for the entire processing target table. 11. The method according to claim 9, wherein in the item string creation step, a single item string is created based on the value item correspondence table, the item code correspondence table, and the plurality of processing target tables. The described data mining auxiliary device. 12. A recording medium recording a data format conversion program for supplying a table format data processable by the engine to a data mining engine for finding a correlation between data, the program comprising: And causing the computer to select a processing target table in accordance with a predetermined input, and in response to the predetermined input, in the processing target table,
Recognize an ID column that describes an identifier of a data unit, an attribute column that describes an attribute, and a value column that describes a value, and assigns an attribute name to each combination of a value type of the attribute column and a name of the value column. At the same time, an attribute name table showing the correspondence between the combination and the attribute name is created, and the type of the value described in the value column is checked for each attribute name defined in the attribute name table. Recognize items for some or all of the values, create a value item correspondence table representing the correspondence between the values and the items, and generate an item for each item based on the number of appearance data units of each item in the processing target table. By attaching a code, an item code correspondence table is created, and in the processing target table, a data unit in which the same identifier is described in the ID column. A storage medium that causes an item code corresponding to each item to be examined based on the value item correspondence table and the item code correspondence table, and to create an item sequence representing a correspondence between each data unit and the item code. . 13. The program, when only an ID column and an attribute column are specified for the processing target table and no value column is specified, the computer stores the name of the value column of the processing target table in the computer. 2. A one-line attribute name table as an attribute name is automatically generated.
2. The recording medium according to 2. 14. A recording medium recording a program for converting information of a plurality of tables to be processed having different table formats into a format that can be processed by a data mining engine, wherein the program stores the program in a computer. ID column, attribute column, and value column for each table format, the attribute name table is created for each table format, the value item correspondence table is created for each format, and the item code correspondence table is created. It is created for the entire processing target table, and further, a single item column is created based on the value item correspondence table, the item code correspondence table, and the plurality of processing target tables. 14. The recording medium according to claim 12 or 13.