JP3519126B2 - Automatic layout system - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic layout system, and more specifically, a data design expert can appropriately predict how many record occurrences or link occurrences should be arranged on a screen at one time. It is possible, but it is difficult for non-specialists, so even non-specialists can use this GUI
(Graphical User Interface) The present invention relates to an automatic layout system capable of creating a GUI that an operator expects to see with the help of an automatic generator.

[0002]

2. Description of the Related Art Computer programs are becoming easier to use with the development of graphical user interfaces (GUIs). The proliferation of application programs with graphical user interfaces is due in particular to the advent of the Windows user environment (eg Microsoft Windows). Unfortunately, generating a graphical user interface typically requires the use of a user interface <toolkit> by highly skilled programmers. Although this user interface toolkit allows the programmer to specify how to display a particular data structure, such a toolkit is not involved in the overall interface design task.

Instead, in designing the entire interface, the application programmer usually has (1) knowledge of the toolkit and interface objects such as push buttons and scroll windows belonging to it, and (2) application input / output conditions. Select an appropriate interface object based on (3) customize the interface object for a specific application, (4) lay out the selected interface object on the interface display, (5) write code to start the interface, Is required to be stored with the application.

Thus, an interface for an application can be created at present, but if the format of this application is changed, the interface must be changed accordingly. Developing applications and interfaces in parallel is repetitive and expensive. Therefore, there is increasing interest in developing tools that automate the design of graphical user interfaces.

In database applications, there are several types of tools used to facilitate the design of user interfaces. It includes a stand-alone form manager, fourth generation language (4GL), graphical user interface builder, automatic graphical user interface generation system, automatic layout designer and the like.

[0006]

Various systems have been commercialized and are widely used as application development support tools for databases. These user interface screens are from traditional character-based to Windows
And X Window System (registered trademark of Massachusetts Institute of Technology, USA) and other multi-window based G
It is moving to UI (Graphical User Interface),
While the ease of use is increasing for application users, the problem is that it takes time and effort for the GUI to be created by developers.

Existing commercial tools do not allow developers to create screen configurations from scratch, but first create a GUI with a simple field arrangement based on the schema information, and then use it as a developer. We have adopted a method of customizing. Microsoft Access [Microsoft Corp .:
Microsoft Access Users Manual, Microsoft Corp., 199
3.], Borland Paradox [Boland Corporation:
Both Paradox for Windows Version 1.0J User's Guide, 1993.] are commercial tools on Windows, but they allow you to set fairly detailed settings for field alignment. However, the only data used to create the individual fields is the field name, data type and its length of the information in the data dictionary. Also, the correspondence between fields and GUI components is not abundant, and there is no variety in the form of fields themselves. Therefore,
The screen becomes more monotonous than when it is created by programming.

To solve this problem and allow the system to generate a wide variety of screens on behalf of humans, it makes use of statistical information in the data dictionary. As statistical information,
For example, the following values are sampled at certain intervals and managed by a DBMS (database management system). -Relational type, tuple length and number of tuples for each table.・ Types of possible field values and their degree of dispersion (eg:
The value of the field "Country" is "America", "Germany",
The degree of dispersion is 90%, 6%, or 4% in any of "Japan"). -When the number of elements is variable in the array type field, the maximum value and average value of the number of elements. -Average field value length of variable length character string type fields. The statistical information is used to optimize an inquiry such as selection of an access path, but there is no other example where the statistical information is used to generate a GUI screen.

The present invention has been made in view of such circumstances, and based on the statistical information, the number of array elements can be set so that the user can see the desired information at a glance.
Providing a screen structure in which the most frequently accessed field values are displayed by default, and in the initial system, depending on whether the data is multi-occurrence or single-occurrence, it is different for each record type. Since only the types of GUI components (hereinafter referred to as widgets) were associated, the GUI look &
Feel more diversified, improve GUI quality,
It is an object of the present invention to provide an automatic layout system that uses statistical information of a data dictionary for automatic generation of a GUI unit.

[0010]

In order to solve the above-mentioned problems, the present invention provides (1) a data structure used by an application in a relational database as E-
Enter an ER diagram that specifies the range of R and which entity to center in that ER, trace the entire ER entered, and entity block rooted at the centered entity Tree structure (called BPT)
A BPT creating means for creating a BPT and an entity expressed as a node of the BPT are blocks, and when the relationship between two entity blocks is a multi-occurrence type, one side is a fieldSet type and the other side is a table
Type or fieldList type, and if the entity block is a single type occurrence, it is a fieldSet type, and even if it is a single type occurrence, it is a fieldList type when it is an array type. Means and the schema information of the relational database, the data type of the field of the database included in the block is physically determined according to the mapping table for each logical screen object type of each block determined by the block generation means. Configuratio that maps to screen object and generates block definition file in which multiple configurations correspond to each block
n Referring to the generation means and the block definition file,
Randomly, (A) block morphology to another configurati
A plurality of layout candidates and their evaluation values are obtained by performing three types of work, that is, replacing with on, (B) exchanging the positions of two blocks, and (C) moving one block to another position by the simulated annealing method. And a layout candidate having an evaluation value of the layout candidate equal to or higher than a predetermined evaluation point, by adjusting the attributes of the physical screen object so as to fill the space between the blocks, A layout determining means for arranging an actual physical screen object and selecting a layout whose evaluation point exceeds a predetermined fixed level; and (2) the input E- in (1) above.
If the R diagram cannot be displayed on one screen, divide the BPT into multiple subtrees and divide each of the subtrees.
And a screen layout including a plurality of screens is generated, and (3) in (1) above,
For many-to-many or one-to-many relationships, the statistical information of the schema information of the relational database is referred to, and a plurality of link occurrences are arranged and displayed at one time according to the number of link occurrences. Further, (4) in (3), the number of link occurrences is determined from the minimum number of links, the maximum number of links, and the average number of links, which are statistical information of the schema information of the relational database, and (5) ) in (1), the configuration generating means, when mapping to a physical screen object, if the data type of the field definitive in the relational database is an array type, the schema information of the sequence type field
Determining the number of physical screen objects to be displayed on the screen by referring to the statistical information in (6).
In (5) above, the number of the physical screen objects
The number is determined from the average number of elements, the minimum number of elements, the maximum number of elements, and the standard deviation that are the statistical information in the schema information of the array type field , and further, (7) above (1)
In the configuration generating means, when mapping to a physical screen object, if the relational <br/> database to definitive field domain is a discrete set, in the discrete schema information of the aggregated field
Determining the data value to be incorporated as a default of the physical screen object by referring to the statistical information in the physical screen object, and (8) in (7) above ,
The data value to be included as the default of the project is
Determining from a field value that is statistical information in the schema information of the discrete set type field and an appearance probability, and (9) in (1) above, the configuration
Generating means, when mapping to a physical screen object, if the data type of the field definitive in the relational database is a variable type, the variable type fee
Variable by referring to the statistical information in the field schema information
Determining the field length of a physical screen object for displaying the length field value, and further, (10) in (9), of the physical screen object Fi
The field length is included in the variable length field schema information .
The average length is a definitive statistics, minimum length, maximum length, be determined from the standard deviation, further, in (11) wherein (1), the configuration generating means, one field
In order to diversify the field layouts in the Set type, it is characterized in that a plurality of configurations are generated by changing the number of columns, the arrangement with priority in the vertical or horizontal direction, and the like.

[0011]

The layout pattern of the conventional automatic layout function of 4GL (4th generation language) is fixed. In the layout automatic generator of the present invention, a simulated a
nnealing Ru can generate an arrangement of that in various blocks using (simulated annealing) algorithm. In the traditional system, it does not is the ability to create multiple candidates, they automatically mechanism without any to place, among the plurality of layout that can be a combination of them, which is also no mechanism to assess whether good layout. The present invention uses graphical objects (X win
Widget in the dow system: widget
(To call) is to use the statistical information in the data dictionary to determine the type and number.

The statistical information includes the number of record occurrences included in one record type, what field value exists for one field, and the number of times that field value appears. Statistics information for snapshots of various database states. Therefore, we will explain how to use statistical information 4
It was classified into two patterns. In each pattern,
I showed how to express statistical information. There is no example of applying statistical information to automatic generation of user interface.

The system "GUI automatic generator" of the present invention
Is an expert system that determines the type and number of appropriate graphical objects for humans by using the statistical information in the data dictionary. The system of the present invention is characterized in that a plurality of layout candidates are generated for each block (this is called configuration) and the layout candidates are arranged by simulated annealing. Layout of layout candidates
To perform simulated annealing
For details, see Japanese Patent Application No. 5-274675 proposed earlier.
It is listed. In this way, a data design expert can properly predict how many record occurrences or link occurrences should be lined up on the screen at one time, but it is difficult for non-specialists, so Also, with the help of this GUI automatic generator, it is predicted that the operator wants to see G
You will be able to create a UI.

[0014]

Embodiments will be described below with reference to the drawings. First, a method of creating a GUI unit corresponding to each record type will be described, then a method of specifying a field arrangement will be described, and an example of a screen created by the automatic layout system will be described.

Generation of GUI for Each Block In the automatic layout system of the present invention, a set of database fields, which is a unit of meaning, is associated with a rectangular area called a block on the screen. The block is surrounded by a frame on the screen. There are various expressions on the GUI for each block. The system prepares multiple GUI forms for the block in advance, and the block definition file (block d
efinition file). Each GUI form belonging to a block will be called a configuration.

FIG. 1 is a diagram showing a correspondence relationship between a block and a plurality of configurations in one block definition file. In the figure, 1 to 5 are blocks, 11a to 11e,
12a to 12c, 13a, 13b, 14a to 14e, 1
5a and 15b are configurations. Block 1
The number of configurations included in each 5 is different. 1
One configuration is randomly selected for the block. In the figure, the shaded configuration 11b,
12c, 13a, 14d, and 15a represent the selected ones. One GUI screen is configured by integrating the selected configurations.

As described above, the shaded area is selected c
onfiguration. In this way, one of the configurations is selected, and, for example, the screen as shown in FIG. 2 is configured. Although there are five blocks on the screen of FIG. 2, each corresponds to the selected configuration. In this system, the schema information for the set of database fields is regarded as an object, GUI is generated as its method, and it is used as the return value.
Prepare GUI (). To create the configuration, the method makeGUI () that creates a GUI may be executed for the schema information object. make G
UI () can be thought of as a mapping from the database space onto the GUI.

As shown in FIG. 3, the schema information includes record name, field name, type, width and height. Information from other data dictionaries will be described later. For width and height information
Since it is assumed that the text_field widget is assigned as the configuration, the unit is the number of characters, not the number of pixels.

Method make GUI () In the automatic layout system, ER (Entity-Relatio
nship) Map the chained many-to-many relationships on the GUI part based on a certain rule. The many-to-many relation is first decomposed into one-to-many relations, and one side of the one-to-many is a single-occurrence type block and the many-side is a multi-occurrence type block. In conventional system specifications, there were only two types, single occurrence and multi-occurrence, but in makeGUI (), the types are table (field), fieldSet (fieldset), and fieldList.
Prepare three types of (field list). The type of widget generated depends on the type. Figure 2
The block type of is t in the block definition file in Figure 1.
It is described in the ype: field. Below, table, fieldSet, fi
The eldList will be explained.

Table: A tabular widget with scroll bars is created. The field arrangement of each record occurrence is arranged as a list item. Used for multi-occurrence data manipulation. fieldSet: Used for single-occurrence data manipulation.
A set of different types of widgets is generated. In this system, depending on the field type, the database field is associated with the widget with the title label according to the mapping shown in FIG.

This system has the following specifications for mapping to fieldSet. That is, as shown in FIG. 4, one field type has multiple confi
It may be mapped to guration. For example, if the number of possible field values is limited, database fields are mapped to the option_menu (open menu) or radio_button (radio button) widgets. fieldList: A list of multiple fields of the same form. Used for multi-occurrence record operations.

The types of widgets will be described below. A variable character (variable length character string) is a type that stores it as a variable length when the field length changes greatly for each data. A variable character field is a text_field with a scrollbar
lled_text) or text_field widget. The scale widget is a slider with knobs and is used when you want to continuously change the analog value.

When the widget labeled "Latin America" in FIG. 2 presses the label with option_menu, a menu pops up. Radio_button widgets are grouped by multiple, and one is exclusively selected. radio_button
Is used to select evaluations No. 1 to No. 8 in 4 of FIG. 8 described later.

How to use the statistical information in the data dictionary will be described. preprocessor st before makeGUI ()
The atistical data collector is run and fetches the statistical information in the format shown below from the snapshot at that point. From that statistic, the system will properly determine the number of widgets to line up in the GUI. There are the following four types of usage of the statistical information inside makeGUI ().

(1) Number of Elements of Array Type An array type database field can take a plurality of values. The number of elements in the array may vary from field to field.
In this system, the following 5-tuple information regarding the number of array elements is obtained and used to determine the number of widgets arranged in the fieldList. The five items of the average number of elements, the minimum number of elements, the maximum number of elements, m−3σ, and m + 3σ are held as statistical information. The average value is m, and in the case of normal distribution, m ± 3
During σ, the probability is 99.7% (σ is standard deviation).

(2) Proportion of different field values When the domain of the field is a discrete value set, the number of appearances for each field value is counted, and (value,%) pair information is held as statistical information for each field value. To do. It is useful to create a user interface that is easy for the operator to use by incorporating a value with a high appearance frequency as a default of a widget, and by incorporating a value as a menu item in order of frequency. There is also an advantage that the developer can save the trouble of specifying a set of field values that may appear.

(3) Variable-length (variable) data length Even when a variable-length field is defined, the average length, minimum length, maximum length, m-3σ, and m + 3σ are the same as the number of array-type elements.
Keeps statistics on. This value is used by the system to determine how large a widget should be to display a variable length field value.

(4) Number of link occurrences For many-to-many associations or one-to-many associations, it is often the case that it is better to display a plurality of link occurrences at once so that the overall situation can be grasped at once. The system counts the number of link occurrences for all record occurrences of the origin record type R [i] for one link type L [i] (R [i], R [j]). Calculate 3-tuple statistics.

That is, the 3-tuple statistical information is the minimum number of links, the maximum number of links, and the average number of links. For example, if the relationship is a one-to-one correspondence, (the minimum number of links,
The maximum number of links and the average number of links) are “(1,1,1)”. If there are a few records without links, for example, "(0,1,0.99)". On the contrary, if most of them are not linked, statistical information such as "(0,1,0.1)" will be taken. When the maximum number of links is 8, it becomes "(0,8,0.1)", for example. Of these three values, the maximum value is the most important. The heuristic rules in the present invention are shown below.

If the maximum number of links is 3 to 5, the link occurrences are arranged on the screen by the maximum number. As shown in FIG. 6, the link occurrence number 3 has the highest appearance frequency and is often appropriate in terms of layout. -If the maximum number of links is 6 or more, use the table widgets 7a-7c with scroll bars instead of the text_field widget. If there is no space for the table, it will be displayed as a popup window.

Designation of Field Arrangement In a fieldSet type block, the configuration can be made more diverse by giving fields and changes. Allow the following options to be added to the makeGUI () method. As a unit of layout, two widgets that represent fields (label and t
(Ext etc.) as a unit.

Stand_in Describes horizontal priority or vertical priority. Will you pack it again? If unpacked, the label will be sized to the largest of the rows. label Whether the label string is right justified, centered, or left justified. fold How many folds should be placed in the specified direction before wrapping? item_grid Whether to add an extra margin so that it is aligned with the horizontally or vertically adjacent widgets.

FIGS. 7A to 7F are diagrams showing examples of the option, and show six configurations.
In order to create the fieldSet array shown in FIG. 2, what options should be specified is shown. Note that one rectangle No. 1 to 10 represents a pair of a label widget and another widget.・ Reviewed Paper make GUI (fieldSet): stand_in column: fold 2: it
em_grid on (same for off): label left ・ Evaluation make GUI (fieldSet): stand_in column: fold 3: it
em_grid on (off is the same in this case): label left

The options that give the variety of field arrangements are also prepared by other database application development support tools, and it is hard to conclude that the specifications of the present invention are much better. However, in other systems, it is used only for the purpose of interactively specifying the template of the initial layout by the user, whereas in this system, these options are used in order to increase the variety of configurations. There is an advantage over the conventional system in that the number of configurations that are candidates is increased by changing the combination of and the ones that get a good evaluation score are selected from them. The steps to increase the variety of configurations are summarized below.

Diversity of representation of many-to-many relation How to represent many-to-many relation of a block is specified by a type (table, fieldSet, fieldList) of a logical screen object. Mapping to physical screen object Depending on which widget set is actually used, the widget to be mapped from the logical object will be different. The mapping of FIG. 4 corresponds to the OSF / Motif [Open Software Foundation: OSF / Motif developed in the present invention.
Programmer's Reference Revision 1.2, Prentice Hal
L, 1993, Englewood Cliffs, NJ] based widget set. Furthermore, even with the same physical widget, multiple configurations can be created by changing the combination of the attribute values.

Variety of field arrangement By providing a plurality of fieldSet arrangements,
Increase the number.

Assuming that there are n, m, and 1 (el) pieces of the above-mentioned three types of diversity, the order of the total number of created configurations is (n × m × l). Automatic layout
In the simulated annealing process, since the blocks are combined, the number of combinations is considerably large.

Example of automatic layout screen In the automatic layout system, when the arrangement of blocks is decided, configura which is an actual widget in the block is displayed.
Try to place tion. A block may be allocated a space larger than the size of its configuration depending on the positional relationship with surrounding blocks. At that time, (1) the widget is arranged at the center, and (2) the following is performed to expand the space in the block.・ The widget with scroll bar expands the size. -FieldList and table type widgets increase the number of occurrences. These last adjustments make the widget layout look good. FIG. 5 and FIG. 8 are diagrams showing an example of a GUI generated when a different screen size is designated with the same operation content as in FIG. From these configuration
It can be seen that various GUI forms can be generated by the combination of.

System Overview FIGS. 9 and 10 are flowcharts showing the overall procedure of how an application developer creates a database operating GUI by the automatic layout system of the present invention. (1) Specification of Database Structure Using Specif
ication Tool) The developer can check the ER range of the application on the ER diagram and what enti in the ER.
Select whether to center ty. App the created file
It is called a license Specification Diagram (ASD). We call this central entity the master entity.

(2) Creation of BPT (ASD Tracing) Screen constituent elements are grouped into blocks for each data operation unit. In the current system specifications, each entity and relat
Blocked for each ionship. Also, menu buttons for routine data operations are also blocks. relati
Onship can also be a block like entity, but it is not blocked if it is a relationship that does not include fields. Inside the system, the block follows the relationship and stores it as a tree structure rooted at the master entity. The system traces the entire target ER and creates a tree structure of entity blocks rooted at the centered entity. This is called a Block Precedence Tree (BPT), and an example thereof is shown in FIG.

[0041] Node B P T represents a block, BPT
The branch between the nodes of represents the relationship. BPT
Among the node blocks of, the ones surrounded by thick lines indicate the multi-occurrence type blocks. Also, those enclosed by a thin line indicate a single-occurrence type block. If the system determines that the target ER is large and cannot be expressed in one screen, the tree structure is divided into several subtrees to form a plurality of screens. According to the current specifications, blocks from the root to three levels are stored on one screen, and the other blocks are individually pop-up displayed as dialog screens.

(3) Object Type Decision and Block Definition
(Definition) When the relationship between two entity blocks is one-to-many, the one side is fieldSet type. Since multiple side wants to display multiple data at the same time, table and fieldList
And Multi-occurrence type fieldList or table Single-occurrence type fieldSet Furthermore, in the following cases, another block is newly created.

In the case of a single type occurrence, of the target database fields, only the array type field is a separate fieldList independent of the original fieldSet.
The block of is set up. -It may be better to display multiple link occurrences at once for a many-to-many or one-to-many relationship. Therefore, many-to-many or one-to-many relationsh
The number of ip link occurrences is determined using statistical information.

In FIG. 6, the relationship between the block "Reviewer Name" and the block "Reviewed paper" is 1
There are many, and this GUI displays 3 link occurrences at the same time. When displaying a plurality of link occurrences, as shown in FIG. 6, it is necessary to arrange the blocks corresponding to the records at both ends side by side so that the relationship can be understood. Display the names of up to three peer reviewers assigned to a single “Revewed Paper”, this time tracing the associations in reverse order and assigning the article names (generally multiple) assigned to the reviewers using the table widget. it's shown.
If the number of link occurrences is 2 or more as in this example, the system treats the two records at both ends of the link as one block.

Instead of leaving it to the system, an application developer who wants to display a plurality of long path search results such as ER-E 'or ER-E'-R'-E''... on the same screen. It also needs to be able to describe the request explicitly. Therefore, this system allows specification of (<path data>, <number of data>) as user preference.

(4) Mapping from Database Fields to Physical Graph
ical Objects) The fieldSet, fieldList and table mentioned above are logical screen object types, and then the system needs to perform a mapping from logical to physical screen objects. This is done according to the mapping table as shown in FIG. Each database field in the fieldSet can have multiple physical representations, as described above.

(5) Determination of Physical Graphical Object Resource Value Dec
ision) Further, even in each physical object, a plurality of forms can be considered depending on the attribute value. Conventionally, the attribute value determination has only given a plurality of candidate default values as templates, but in the present invention, the statistical information obtained from the data dictionary is used for attribute value determination. The number of elements of the array type database field and the length of the variable length field are determined from these statistical information. If it is found from the statistical information that the number of field values is finite, the statistical information is used to arrange the selection candidate values in the order of appearance frequency. Statistical information is also used as a criterion when the system decides how many data should be displayed simultaneously for the table type block.

(6) Variety of field arrangement (Laying O
ut Physical Graphical Objects in afieldSet) Furthermore, in order to diversify the field layout within one fieldSet, by changing the number of columns, priority in the vertical / horizontal direction, etc., many configura
Create tion. Multiple configurations per block
As a file. This is a Block Definition File, as shown in Figure 1.
Call.

(7) Block Placement and Layout Evaluation (Block Placement Generator) Next, the system of the present invention arranges a plurality of blocks constituting the screen by a simulated annealing method to create different block layouts one after another. This simulated annealing is described, for example, by S. Kirkpatrik, CD Gelatt and MP Vecchi in “Optimization by Simulated Annealin.
g (optimization by simulated annealing) ”(Science, 2nd
0, No. 4598, May 1983, pp. 671-680). The simulated annealing procedure is based on the use of random functions that move randomly through the solution space. The solution space corresponds to any potential arrangement within the database interface of object block sets that corresponds to a particular set of block description files 130. Following each iteration, the solution "quality" is compared to the solution "quality" obtained during the previous iteration. In general, new solutions are only taken if they show an improvement over the previous solution. However,
To avoid being constrained to the solution space approximated by a local minimum,
The simulated annealing function accepts certain lesser solutions.

The following pseudo code describes a simulated annealing procedure. Simulated Annealing Initialize; placement = GeneratelnltialPlacement; cost = Evaluate (placement); while (loop_count <MAX_ITERATIONS and temperature> MIN_TEMPERATURE) new_placement = Perturb (placement) new_cost = Evaluate (new_placement); if (Accept (new_placement, new_cost, temperature)) placement = new_cost; loop_count ++; temparature = NewTemparature (loop_count); return (placement, cost);

In the simulated annealing, the operation of randomly changing the arrangement, size and shape and performing evaluation is repeated. During this loop, three types of work are performed: randomly replacing the block form with another configuration, exchanging the positions of two blocks, and moving one block to another position. As the loops are piled up, the amount of change is gradually reduced and converged. For each loop, the system gives a rating to the layout.

(8) Arrangement of screen objects in the block layout (Physical Graphical Object Assignmen)
t) For block layouts that exceed a certain evaluation point, try placing actual physical objects. At that time, the resizeable portion of the screen object is adjusted by changing the attributes of the object so as to fill the space between the blocks. Each generated layout is scored by the rating system and only those above a certain level are selected.

(9) Generation of UIL code (Display Code
Generator and Compiler) The completed GUI screen is visible in OSF / Motif
It is converted into a code in UIL (User Interface Language) and displayed through the UIL compiler.

[0054]

As is apparent from the above description, the present invention has the following effects. That is, the present invention increases the variety of database operating GUIs by using the statistical information of the data dictionary. In the automatic layout system of the present invention, first, the physical screens that are actually arranged for each block are displayed. Prepare multiple elements (widgets) and call them configuration. One block can represent multiple physical screen element expressions, which means that multiple configurations are associated with each block. By using the statistical information of the data dictionary in associating with this physical screen element, the conventional configu
It is now possible to prepare more configurations than ration. The statistical information in the data dictionary is useful not only for visual diversity, but also for improving the operability of the GUI by making the default display value frequently accessed. For developers who are knowledgeable in database design, a GUI can be created by arranging an appropriate number of widgets of an appropriate type.
It was possible to realize a system that creates a GUI at the same level as a GUI created by an expert, even if the developer is unfamiliar. In this sense, a system that creates a GUI layout that is easy to use based on statistical information is an expert system. In addition, the statistical information that has been conventionally used only for optimizing a query can also be used for improving the operability of GUI. Moreover, the use of statistical information has made it possible to automatically generate more diverse GUIs.

[Brief description of drawings]

FIG. 1 is a diagram showing a correspondence relationship between a block and a plurality of configurations in a block definition file according to the present invention.

FIG. 2 is a diagram showing an example (No. 1) of a GUI screen generated according to the present invention.

FIG. 3 is a diagram showing an example of a schema information object according to the present invention.

FIG. 4 is a diagram showing mapping permitted by fieldSet in the present invention.

FIG. 5 is a diagram showing an example (No. 2) of a GUI screen generated according to the present invention.

FIG. 6 is a diagram showing an example of a screen in which three link occurrences are displayed at once in the many-to-many association of the present invention.

FIG. 7 is a diagram showing an example of designation of a field arrangement in a fieldSet according to the present invention.

FIG. 8 is a diagram showing an example (part 3) of a GUI screen generated according to the present invention.

FIG. 9 is a flowchart (No. 1) for explaining the automatic layout system according to the present invention.

FIG. 10 is a flowchart (part 2) for explaining the automatic layout system according to the present invention.

FIG. 11: Block Precedence Tree in the present invention
It is a figure which shows the example of (BPT).

[Explanation of symbols]

1-5 ... Block, 6a-6c ... Link occurrence, 7
a to 7c ... table widget, 11a to 11e, 12
a-12c, 13a, 13b, 14a-14e, 15
a, 15b ... configuration.

─────────────────────────────────────────────────── ─── Continuation of the front page (56) References Japanese Patent Laid-Open No. 7-334656 (JP, A) Yukari Shirata, Atsushi Iizawa, Development system of application software for database operation of wind database, Research Report of Information Processing Society of Japan, Japan , IPSJ, March 22, 1993, Vol. 93, No. 29, 93-D BS-92, pages 87-95 Yukari Shirata, Auto Piza no, Atsushi Iizawa, GUI for database operation with automatic layout function Generation System, Research Report of Information Processing Society of Japan, Japan, Information Processing Society of Japan, July 21, 1993, Vol. 93, No. 95, 93-DBS-94, pages 215-222 (58) Fields investigated (Int .Cl. ⁷ , DB name) G06F 17/30-17/30 419 G06F 3/00 650-658

Claims (11)

(57) [Claims] 1. A data structure used by an application in a relational database is entered by entering an ER diagram that specifies the range of the ER and which entity is centered within that ER. E-
BPT creating means for creating a tree structure (called BPT) of an entity block whose root is an entity to be traced over the entire R, and an entity expressed as a node of the BPT as a block between two entity blocks If the relationship is a multi-occurrence type, one side is a fieldSet type, the multi-side is a table type or fieldList type, and if the entity block is a single type occurrence, it is a fieldSet type, even if it is a single type occurrence. When it is an array type, as a fieldList type, block generation means for determining the logical screen object type of the block,
By referring to the schema information of the relational database, the data type of the field of the database included in the block for each logical screen object type of each block determined by the block generation means is converted into a physical screen object according to a mapping table. A configuration generating unit that performs mapping to generate a block definition file in which a plurality of configurations are associated with each block, and randomly refers to the block definition file to replace the (A) block form with another configuration. (B) Swap the positions of two blocks,
(C) Move one block to another position 3
Do some kind of work with simulated annealing method,
Block layout means for generating a plurality of layout candidates and their evaluation values, and for a layout candidate having an evaluation value of the layout candidate equal to or higher than a predetermined evaluation point, a physical screen object An automatic layout system comprising: layout determining means for changing an attribute to adjust and arranging an actual physical screen object, and selecting a layout whose evaluation point exceeds a predetermined fixed level. 2. The input ER diagram is 1
One of the case can not be represented on the screen, the automatic layout system according to claim 1, by dividing the BPT multiple subtrees, characterized in that so as to generate a screen for each said sub-tree. 3. A many-to-many or one-to-many relationship
On the other hand, with reference to the statistical information of the schema information of the relational database, a plurality of link occurrences are arranged and displayed at one time according to the number of link occurrences. Automatic layout system. 4. The automatic layout system according to claim 3, wherein the number of link occurrences is determined from a minimum number of links, a maximum number of links, and an average number of links, which are statistical information of schema information of the relational database. . Wherein said configuration generating means, when mapping to a physical screen object, if the data type of the field definitive in the relational database is an array type, contact the schema information of the sequence type field
The number of physical screen objects to be displayed on the screen is determined by referring to the statistical information on the screen.
Automatic layout system described in. 6. The number of physical screen objects is
The automatic layout system according to claim 5, wherein the automatic layout system is determined from the average number of elements, the minimum number of elements, the maximum number of elements, and the standard deviation, which are statistical information in the schema information of the array type field . Wherein said configuration generating means, when mapping to a physical screen object, if the domain of the field definitive relational database is discrete set, the schema information of the aggregated fields the discrete
The automatic layout system according to claim 1, wherein a data value to be incorporated as a default of a physical screen object is determined with reference to statistical information in the information . 8. The data value incorporated as the default of the physical screen object is the discrete set type field.
8. The automatic layout system according to claim 7, wherein the automatic layout system is determined from the field value and the appearance probability, which are statistical information in the schema information of the field. Wherein said configuration generating means, when mapping to a physical screen object, if the data type of the field definitive in the relational database is a variable type, the schema information of the variable length type field
Display variable length field values, see statistics in
The automatic layout system according to claim 1, wherein a field length of a physical screen object for indicating is determined. 10. The fee of the physical screen object
The field length is included in the variable length field schema information .
10. The automatic layout system according to claim 9, wherein the automatic layout system determines the average length, the minimum length, the maximum length, and the standard deviation that are statistical information. 11. The configuration generating means generates a plurality of configurations by changing the number of columns, vertical or horizontal priority arrangement, etc. in order to diversify the field layout in one fieldSet type. The automatic layout system according to claim 1, wherein: