JPH11316766A - Multidimensional analytical construction system and database for analytical processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To construct a multidimensional analytical system in a short time and efficiently by analyzing a multidimensional database (DB) based on a format that is produced by an analytical format producing mechanism. SOLUTION: When an operator of a computer 3 for a client requests the download of a JAVA applet 10 through a WWW browser 14, a WWW server 9 downloads the applet 10 to the computer 3. A construction server 4 produces an analytical format 5 and a multidimensional DB 6 while conversing with a definition tool 12 and a management tool 13. An extracting part 7 extracts data from an extraction source DB 8 and writes it to the DB 6. An analytical server 11 receives a request from a JAVA applet 10' downloaded to the computer 3, produces image information based on an analytical format corresponding to the request and data in the DB 6 corresponding to the analytical format and transfers it to the applet 10'.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a multidimensional analysis construction system and an analysis processing database.

[0002]

2. Description of the Related Art Multi-dimensional analysis processing is a method of multiplying data that is currently performed to solve the upper limit of data analysis by a relational database composed of a two-dimensional array of rows and columns. It is extended as a dynamic analysis method using dimensionality. For example, products,
If there are three dimensions such as time and branch, the person in charge of the product analyzes the sales of the corresponding product in each region in chronological order, and the person in charge of the branch chronologically analyzes the sales of all the products in his branch. analyse. Also, executives take a variety of perspectives at various angles. It is a feature of multidimensional analysis that such multilateral analysis is possible.

[0003] Techniques for advancing multidimensional analysis include slice, dice, drill-down, and roll-up techniques. For example, in FIG. 17, assuming that there is sales data by product for each business office in each fiscal year, when performing multidimensional analysis, first, it is classified into an axis representing a fiscal year, an axis representing a business office, and an axis representing a product. The mapping of each axis in the x, y, and z directions corresponds to the upper left cube. The display data in this case is the sales of the product A by sales office in quarterly units. Switching this to sales data of another product such as the product C, that is, designating which dimension of the multidimensional data to extract is called a slice operation. On the other hand, switching the sales data by sales office in quarters to sales data by product in sales offices, that is, changing the combination of the selected dimensions is called a dice operation. Drill-down or roll-up refers to dividing an item in more detail or summarizing it into a large classification. For example,
Classifying in order of year, half year, quarter, and month is called drill-down, and conversely, summarizing from month to year is called roll-up.

[0004]

To create a multidimensional analysis system, it is necessary to design a multidimensional database, create an extraction program for writing extracted data from an existing database to a multidimensional database, and create a multidimensional analysis application. It is. In performing multidimensional analysis, there are generally the following problems. (A1) It is difficult to design a screen. (A2) It is difficult to design a database.

[0005] Regarding item (a1), in order to perform multidimensional analysis, it is necessary to easily change the setting of dimensions and to easily design a standard analysis. However, in reality, the purpose of the analysis is not clear, so we use methods such as prototyping to match the user and the designer,
The method of introducing is taken. According to this method, there is a problem that it takes a long time to introduce the device and the number of steps increases.

For item (a2), it is necessary to design a database to be analyzed together with the setting of dimensions. However, it is assumed that the information stored in a company has different meanings between systems. For example, if a gender code is designed for one male and two females in one system, and another male in M and female in W, it is necessary to match them in order to analyze by gender. There is. In addition, various types of information may be distributed to each system. In other words, a database design and extraction application corresponding to unification and integration of distributed data is required. There is a problem that this design requires specialized know-how and is not easy.

In a conventional multi-dimensional analysis system based on an RDBMS (relational database management system), only the detailed data to be analyzed is managed. It makes it difficult to do it at high speed. (B1) Aggregation calculation at the time of analysis processing (b2) Joining of analysis target data tables at the time of analysis processing (b3) Management of correlation between hierarchies in dimensions

The term (b1) generally requires a plurality of tabulation operations in the vertical / horizontal tabulation calculation used in multidimensional analysis. However, in a multidimensional analysis system based on RDBMS, this tabulation is generally performed. The calculation is performed during the analysis processing, which makes it difficult to perform the analysis processing at high speed.

Regarding the item (b2), the detailed data to be analyzed may span a plurality of tables, and it is necessary to combine the detailed data tables. The joining of the tables is generally performed at the time of the tabulation operation, and it is difficult to perform the analysis processing at high speed by performing it at the time of the analysis processing as in the case of (b1).

Regarding (b3), in a multidimensional analysis operation, in order to perform an operation such as drilling down or rolling up, it is necessary to have a correlation between hierarchies in a dimension. For example, it is necessary to perform management such that the total data of the first half and the second half is included in the fiscal year, and the total data of April to September is included in the first half. In order to realize this, it is necessary to have a mutual relationship between hierarchies in a tree structure, but it is difficult to manage this with a general RDBMS.

The present invention has been made in view of this point, and an object of the present invention is to enable a multidimensional analysis system to be efficiently constructed in a short time. Another object of the present invention is to provide an analysis processing database suitable for a multidimensional analysis system.

[0012]

In order to solve the above-mentioned problems, the inventor considered that instead of creating an analysis format from scratch, switching ideas and tracing back from the method of using analysis results. In other words, general business forms and output formats of analysis tools are classified and systematized according to the form type and purpose of use. By incorporating it into the system as a sample, it is possible to easily match the purpose of use and the awareness of the results of realization with the user. Thus, the operation of selecting complicated dimensions and designing a screen can be replaced with the operation of simply selecting a form, and the time and cost required for introduction can be greatly reduced.

[0013] In the design of the database, since the screen configuration is determined, it is only necessary to define the relationship between the items on the screen and the database on the server stored in the company. The system of the present invention automatically generates an analysis processing database and data extraction definition based on these. The user does not need to be aware of the multidimensional database, and only needs to be aware of the location of the server data from which the data is to be extracted.

The configuration and usage of the sample format will be described. In the sample format, information indicating the type of axis such as the time axis, information on the hierarchy of drilldown and rollup, information on items such as the definition of the aggregation method, display information on items such as digit separator and scale, column width , Display information of the entire screen, such as the display order of the data on the screen and the title, are pre-installed. Whether the slice operation can be performed based on the axis information, the level of the sample format used can be drilled down or rolled up based on the information on the hierarchy, and the calculation formula of the embedding operation such as a profit table is defined in advance based on the information on the items. After that, by defining how the item information corresponds to the columns of the RDB (relational database) on the server, an analysis format for multidimensional analysis is created. Of course, it is also possible to customize the configuration of the screen or change the method of summing the data of the extraction source according to the user's business.

The implementation of the multidimensional analysis will be described. In the system of the present invention, a database for analysis processing, which is a multidimensional database, is generated based on the components of the screen defined in the analysis format and the columns of the extraction source. The analysis processing database includes an RDB that manages total data and a multidimensional management file that manages a hierarchical matrix.

In general, cross tabulation from detailed data requires a plurality of tabulation processes. In the present invention, the result of the cross tabulation is stored in the RDB. Further, in order to obtain an optimal structure for realizing the multidimensional analysis processing, all combinations between layers of each axis such as a vertical axis, a horizontal axis, and a slice axis are incorporated in the RDB. This makes it possible to omit the tallying process at the time of analysis, thereby improving the performance.

The multidimensional management file will be described. In order to realize analysis processing such as drill-down and roll-up, it is necessary to have a correlation between hierarchies of each axis. For example, it manages the relationship that the total data of the first half and the second half is included in the fiscal year, and the total data of April to September is included in the first half. To achieve this, it is necessary to have a correlation between hierarchies in a tree structure, but managing it on a database is inefficient. In the present invention, it is realized as a multidimensional management file.

General New Creation Wizard and Sample
The difference between the format definitions will be described. The New Creation Wizard is a way to create a format from scratch,
Unlike the present invention, it is impossible to incorporate know-how that can be actually used in business or to extract customized items suitable for a user.

The difference between reference creation and sample-based format definition will be described. Reference creation only uses existing ones, and unnecessary or unnecessary information is included in the information.
There is information that cannot be operated as it is, but it is not possible for the user to determine which information is useful and which information is unnecessary or needs to be modified simply by creating a reference. In the case of the sample format base, there is no need to have unnecessary information, and information that can be corrected (or needs to be corrected) for each format is extracted in advance and can be customized by a third party.

The difference from the package will be described. The package is a system engineer who builds a system based on the existing application package by programming, but the system of the present invention eliminates the need for such programming skills and specialized skills in analysis tables. The difference is that it is provided as a system.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIGS. 1 to 6 show examples of analysis operations. In the embodiment of the present invention, a format list is displayed in a tree view format so that an analysis format suitable for the purpose can be easily found. The display of the analysis result is performed by selecting an analysis format on the tree view. The analysis result is displayed on the analysis result window. The analysis operation can be performed on the analysis result window. As an analysis result, drill down, roll up, and slicing operations are possible.

Drill-down and roll-up operations can be performed by selecting a hierarchy button representing the hierarchy of each axis displayed on the analysis result window. Further, when it is desired to display only a part of the data in detail, it is possible to perform a partial drill-down by selecting the data. As for the slice operation, a slice operation screen is displayed by selecting a slice operation item on a menu bar or a pop-up menu on the window. Items of each layer are displayed in a tree view format on the slice operation screen, and when the user selects an item to be analyzed, the result is reflected on the analysis result window.

FIG. 1 is a diagram showing the transition from the format selection to the analysis screen. In the illustrated example, “profit and loss analysis of the entire business unit” is selected as the analysis format.
When "Profit and loss analysis of entire business unit" is selected, an initial screen as shown in FIG. 2 is displayed. This initial screen shows the profit (loss) of the entire division every year. FIG. 3 is a diagram illustrating the drill-down (overall). When the “category” button is selected in FIG. 2 and the semi-annual period is selected, a profit-loss analysis by category for each semi-annual period as shown in FIG. 3 is displayed.

FIG. 4 is a view for explaining the drill-down (part). In FIG. 3, for example, double-clicking the “freezer-freezer 2 door” cell displays details of the selected portion. FIG. 5 is a diagram illustrating a slice operation. In the illustrated example, “Nagoya” is selected on the slice operation screen. When “Nagoya” is selected, an analysis result screen as shown in FIG. 6 is displayed.

FIGS. 7 to 10 are diagrams for explaining the creation of an analysis format. FIG. 7 shows an initial screen for creating an analysis format. On the initial screen, a list of sample formats is displayed in a tree view format. For example, when the operator selects the sample format of “sales composition analysis (by product)”, a sample of the analysis result of “sales composition analysis (by product)” as shown in the figure is displayed.

FIG. 8 is a diagram for explaining the association between the analysis format and the extraction source. The information list of the existing DB (database) is displayed at the upper left of the display screen of FIG.
The database is a relational database. When the sales master of the sales department is selected in the information list of the existing DB, the item names of the sales master are displayed at the upper right of the display screen. In the example shown, "date",
"Product", "quantity", "unit price", etc. are displayed.

In the example shown in the figure, an analysis format item list is displayed on the lower side of the display screen, and an entry frame for extraction source items is provided below the list. For example, after selecting the item "quantity" in the item list of the analysis format, and bringing the "quantity" of the existing DB to the entry frame of the item to be extracted by dragging and dropping, the column called "quantity of the analysis format" Is associated with a column called the quantity of the sales master.

FIG. 9 is a view for explaining the schedule setting for each analysis format. When set as shown in the figure, data is extracted from the extraction source DB every Monday,
The extracted data is written to a multidimensional DB corresponding to the analysis format of the sales transition by product.

FIG. 10 is a diagram showing a configuration example of the multidimensional analysis system of the present invention. In the figure, 1 is a server computer, 2 is a client computer, 3 is a client computer, 4 is a construction server, 5 is an analysis format, 6 is a multidimensional DB, 7 is an extraction unit, 8 is an extraction source DB, 9 Is a WWW server, 10 is a JAVA applet, 10 'is a JAVA applet downloaded to the client computer 3, 12 is a definition tool, 13 is a management tool, 14 is WW
Each W browser is shown.

The server computer 1 includes a construction server 4, an analysis format 5, a multidimensional DB 6, an extraction unit 7, and an extraction source DB.
8, a WWW server 9, and a JAVA applet 10. The client computer 2 has a definition tool 12 and a management tool 13. The client computer 3 has a WWW browser 3 and a JAVA ablet 10 '.

The definition tool 12 is for creating an analysis format, customizing the format, and performing extraction definition. Creating an analysis format means generating an analysis format based on a sample format when a new analysis format is created. Format customization means designating a hierarchy definition and a screen display format at the time of analysis for the analysis format. What is an extraction definition?
It means data association with the core DB or data warehouse, designation of data conversion, and the like.

The management tool 13 is for performing extraction management and tree view maintenance. The extraction management means designating the scheduling and the extraction method at the time of executing the extraction. The maintenance of the tree view means categorization, format movement, deletion, addition (only categorization), rename, and the like.

When the operator of the client computer 3 has WWW
When the download of the JAVA applet 10 is requested via the browser 14, the WWW server 9 downloads the JAVA applet 10 to the client computer 3. This JAVA applet is for displaying a screen for an analysis operation as shown in FIGS.

The construction server 4 creates the analysis format 5 and the multidimensional DB 6 while talking with the definition tool 12 and the management tool 13. The extraction unit 7 is an extraction source DB
Data is extracted from the data 8 and written into the multidimensional DB 6. The analysis server 11 receives a request from the JAVA applet 10 ′ downloaded to the client computer 3, and displays screen information based on an analysis format corresponding to the request and data in the multidimensional DB 6 corresponding to the analysis format. The created screen information is passed to the JAVA applet 10 '.

FIG. 11 is a diagram showing another example of the configuration of the multidimensional analysis system of the present invention. In the figure, reference numeral 15 denotes another computer, and reference numeral 16 denotes a construction server. In addition,
The same reference numerals as those in FIG. 10 indicate the same components. In FIG.
The server computer 1 and the other computers 15 are connected via a network. The extraction source DB 8 and the construction server 16 exist in the other computers 15. Construction server 16
Extracts data from the extraction source DB according to the data transfer request from the extraction unit 7 and passes the extracted data to the extraction unit 7. The extraction unit 7 writes the received data into the multi-dimensional DB 6. Other points are the same as those in FIG.

FIG. 12 is a diagram for explaining the creation of an analysis format according to the present invention. In the figure, 5 is an analysis format storage unit, 17 is an analysis sample storage unit, 18 is an analysis result selection unit, 19 is an analysis result presentation unit, 20 is a format definition generation unit, 21 is a screen format change unit, and 22 is a multi-dimensional change unit. A structure change unit, 23 is an extraction source data location display unit, 24 is an extraction definition generation unit, 25 is a screen format definition, 26 is an extraction definition, 27 is an analysis application, and 28 is an extraction application.

Parts 17 to 24 in FIG.
It exists in the definition tool 12. The analysis application 27 is the same as the analysis server 11, and the extraction application 28 is the same as the extraction unit 7. The analysis format storage unit 5 stores a screen format definition and an extraction definition.

The analysis sample storage unit 17 stores a sample format focused on tasks such as information-related task processing. The analysis result selection unit 18 has a function of designating which of the samples stored in the analysis sample storage unit 17 is to be used. In the analysis sample, a set of data that is deemed necessary according to the purpose is defined. The analysis result presentation unit 19 has a function of displaying the analysis result according to the definition.

The screen format changing unit 21 customizes items that do not affect the DB (or multidimensional configuration). Examples of the customizing means include intuitive specification by layout and specification of a selection method by a wizard. Examples of screen format changes include display column width, background color, title, analysis operation control, and the like.

The multidimensional structure changing unit 22 is responsible for changing data relating to the components of the multidimensional DB. In terms of processing,
Generate data by copying items that are present in the sample. As a result, the information on the screen format for the change in the configuration of the multidimensional data can be largely automatically generated. Only items that cannot be copied (for example, display literals on the screen) are specified in the wizard or layout display. Although not shown, the multidimensional structure change unit 22 is divided into an axis management unit, a hierarchy management unit, an item management unit, and an item definition unit.

The axis management unit manages addition and deletion of axes. When adding an axis is specified, specify the axis name for the copy source axis. For example, there is a product, time, store axis,
Copy the store axis to create a regional axis. Store axis information is copied to the regional axis, and of the copied information,
The axis name and hierarchy name (copy additional items of the following information) can be specified in order by the wizard.

The hierarchy management unit manages addition and deletion of a hierarchy. When a layer is added, the characteristics of the layer are obtained from the layers before and after the addition position, and the pattern of the layer is determined (copied). For example, to add a hierarchy between year and month,
The interim is half a year or a quarter. Let the user specify which one to insert and determine the type of hierarchy. Thereby, the association between the year, the insertion hierarchy and the month can be determined automatically. For example, there is a store that retrieves items from a DB like a store. In this case, there is no direct connection between the layers on the screen, and only the designation of the layer name is performed.

The item management unit manages addition / deletion of items to be counted. A function associated with the item to be totaled can be defined. When adding an item, the user specifies whether or not this function can be used. The item definition section defines an item with an expression that specifies the item. This makes it possible to incorporate the equation in the sample in advance.

The format definition generation unit 20 embeds the customization information of the screen format change unit and the multidimensional structure change unit in a sample, and generates a screen format definition. The extraction definition changing unit 24 embeds the location of the data to be extracted in the information defined by the multidimensional structure changing unit as a frame, and holds the data as the extraction definition. Although not shown, the extraction definition generation unit is divided into a tally definition generation unit and an axis definition generation unit.
The tally definition generation unit associates the tally item on the screen with the tally item of the extraction source data. Based on the display format on the screen and the format of the item to be extracted, the type of the total item of the multidimensional DB is determined, and the data conversion method is determined. The axis definition generation unit associates the item data displayed in the axis with the extraction source data. Depending on the type of axis, it is divided into time and DB items. In the case of time, associate it with the source time column as an axis,
By specifying the time format of the extraction source, the association of each layer is automatically generated. In the case of a DB item, the user is asked whether to use the item as it is, to switch to another column for display, or to specify and display in another format by the user.

The analysis application 27 is an interactive application for performing analysis processing based on a screen format. The extraction application 28 is an application that extracts / converts data from the extraction source server based on the extraction definition.

FIG. 13 is a diagram showing a processing flow for constructing a multidimensional analysis system. FIG. 13A shows a conventional processing flow, and FIG. 13B shows a processing flow of the present invention.

In the conventional processing flow, the following processing is performed. Listening to the request of the end user, the design of the multidimensional DB and the selection of the data to be extracted are performed. Create an application that extracts data from the extraction source DB to the multidimensional DB. Create an analysis application for analyzing multi-dimensional DB data. Test your suite of applications to see if they meet your end-user requirements. If not, or go back to fix the application. Have the end user use it as a test operation and evaluate it. Modify based on or back to the evaluation result.

In the processing flow of the present invention, the following processing is performed. Select the sample closest to your end-user requirements.
If necessary, ask the end user to evaluate based on the sample. Perform customization definition for sample data according to end user definition. According to the screen configuration, the location of the data of the extraction source DB is defined. The extraction definition, multidimensional DB definition, and analysis format definition can be made through the above-mentioned, and the end user performs a test operation based on the definition file. Modify the definition based on or back to the evaluation results. FIG.
By comparing FIG. 13A with FIG. 13B, it can be seen that the system of the present invention can be constructed more easily.

FIG. 14 is a diagram showing the structure of a multidimensional database. In the figure, 29 is a data mart, 30 is a vertical axis literal management table, 31 is a horizontal axis literal management table, and 32 is a slice axis literal management table.

As shown in FIG. 14, the multidimensional database has a data mart 29, a vertical axis literal management table 3
0, a horizontal axis literal management table 31, and a slice axis literal management table 32. The data mart 29 is an RDB, and the vertical axis literal management table 30,
Each of the horizontal axis literal management table 31 and the slice axis literal management table 32 is a TEXT file.
Each record (row) of the data mart has a field indicating a vertical axis code, a field indicating a horizontal axis code, a field indicating a slice axis code, and a field indicating values of n numeric items. The vertical axis code of each record of the data mart 29 points to a row of the vertical axis literal management table 30, and the horizontal axis code is the horizontal axis literal management table 31.
, And the slice axis code points to the row of the slice axis literal management table 32. In the illustrated example, four lines (excluding the line in which the item name of the first line is written)
Although only shown, there are actually many more lines.

Since the vertical axis literal management table, horizontal axis literal management table, and slice axis literal management table have the same structure, only the vertical axis literal management table will be described. First of vertical axis literal management table
In the line, literal information “T001, whole, 1, 0, 4” is written. “All, 1, 0, 4” indicates that the display literal is “all”, that its own identifier is 1, that there is no parent, and that the number of children is 4. In the second row of the vertical axis literal management table, “T
002, network, 2, 1, 5 "is written. “Network, 2, 1, 5” indicates that the display literal is “network”, its own identifier is 2, the parent identifier is 1, and the number of children is 5.

On the vertical axis, since the network exists in the hierarchy immediately below the whole, the literal information of the network is written in the line next to the whole literal information. Similarly, on the vertical axis, since the COBOL system exists in the hierarchy immediately below the DP, the COBOL system literal information is written in the line next to the DP literal information.

FIG. 15 is a view for explaining the structure of the data mart. Each record of the data mart 29 has a code on the vertical axis, a code on the horizontal axis, a code on the slice axis, and total values 1 to n. The code on the vertical axis is for searching for a row in the vertical axis literal management table, the code on the horizontal axis is for searching for a row in the horizontal axis literal management table, and the code for the slice axis is a slice axis literal. This is for retrieving a row in the management table. Note that c and p on the left indicate the data type.

FIG. 16 is a view for explaining the structure of the horizontal axis literal management table. The horizontal axis literal management table has codes, literals, hierarchy IDs, parent codes, number of children, original data, and extraction execution flags. A description of each item is provided on the right side. Other literal management tables have a similar structure. In addition,
c5, c6, etc. indicate the data length.

[0055]

As is apparent from the above description, according to the analysis format creation mechanism of the present invention, an analysis format and an analysis processing database required for a multidimensional analysis system can be easily created. Further, according to the database for analysis processing of the present invention, multidimensional analysis processing such as drill-down, roll-up, and slice can be efficiently performed, and the memory and the CPU can be used efficiently.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating an analysis operation.

FIG. 2 is a diagram illustrating an analysis operation.

FIG. 3 is a diagram illustrating an analysis operation.

FIG. 4 is a diagram illustrating an analysis operation.

FIG. 5 is a diagram illustrating an analysis operation.

FIG. 6 is a diagram illustrating an analysis operation.

FIG. 7 is a diagram showing creation of an analysis format (No. 1).

FIG. 8 is a diagram showing creation of an analysis format (part 2).

FIG. 9 is a diagram showing creation of an analysis format (No. 3).

FIG. 10 is a diagram illustrating a configuration example of a multidimensional analysis system.

FIG. 11 is a diagram illustrating another configuration example of the multidimensional analysis system.

FIG. 12 is a diagram illustrating creation of an analysis format according to the present invention.

FIG. 13 is a diagram showing a processing flow for constructing a multidimensional folding system.

FIG. 14 is a diagram showing a structure of a multidimensional database.

FIG. 15 is a diagram illustrating the structure of a data mart.

FIG. 16 is a diagram illustrating the structure of a literal management table.

FIG. 17 is a diagram illustrating slice, drill-down, and dice.

[Explanation of symbols]

 Reference Signs List 1 server computer 2 client computer 3 client computer 4 construction server 5 analysis format 6 multidimensional DB 7 extraction unit 8 extraction source DB 9 WWW server 10 JAVA applet 10 'JAVA applet 12 definition tool 13 management tool 14 WWW browser 15 and others Computer 16 Construction server 17 Analysis sample storage unit 18 Analysis result selection unit 19 Analysis result presentation unit 20 Format definition generation unit 21 Screen format change unit 22 Multidimensional structure change unit 23 Extraction source data location display unit 24 Extraction definition generation unit 25 Screen Format definition 26 Extraction definition 27 Analysis application 28 Extraction application 29 Data mart 30 Vertical axis literal management table 31 Horizontal axis literal management table 32 Slice axis literal management table

Claims (5)

[Claims] 1. A mechanism for creating a multidimensional database based on a format created by an analysis format creation mechanism, a function of extracting data from the database to the multidimensional database, and creation by an analysis format creation mechanism A multidimensional analysis system construction system characterized in that a multidimensional database can be analyzed based on a specified format. 2. The analysis format creation mechanism according to claim 1, wherein the analysis format creation mechanism has a plurality of sample formats in which a screen format definition is pre-installed, and further includes a customizing means for extracting only user-specific correction information. 1. Multidimensional analysis system construction system. 3. The analysis format creation mechanism according to claim 1, wherein when the operator instructs to modify the screen format definition, the format definition on the screen is freely modified according to the instruction. 2. Multidimensional analysis system construction system. 4. In the analysis format creation mechanism,
Only the operator specifies the location of the data to be extracted based on the screen component items of the defined screen format,
4. The multidimensional analysis system construction system according to claim 1, wherein a mechanism for extracting data in the multidimensional database is automatically created. 5. A multi-dimensional database, comprising: a database having total data and having all combinations of items for each layer of an axis; and a multi-dimensional management file having a correlation between the layers in a tree structure. Analysis processing database of the analysis system.