JP2003323450A - Database retrieving device and method, computer program and computer-readable recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a database retrieving device capable of easily retrieving the solution suitable for a focused problem. <P>SOLUTION: The collection of records wherein a field assumed as a dependent variable relating to the focused problem is not null, and having the correlation among the fields, is retrieved by using an event database (S102). The event is expanded to parts (S104). The collection of the records configurated by the field (independent variable) of which an attribute value is not null with respect to all of the parts is created (S106). A trial function is determined by combining the plurality of fields selected from the independent variables to relate the dependent variable and the independent variable (S108). A solution space is identified by executing the exhaustive retrieval to classify the events by combining the systematical operation of the trial function and the sorting of a value of the dependent variable (S110). <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data mining technique for discovering a specific pattern from a database of complex phenomena such as scientific data and economic data, and more specifically, from a record group having many fields. In order to systematically obtain the causal relationships implicitly implied in the event database for the configured event database, the events are expanded into the constituent elements, and the operations between the attributes possessed by the constituent elements and the data The present invention relates to a database search device, a database search method, a computer program, and a computer-readable recording medium that perform reorganization.

[0002]

2. Description of the Related Art The behavior of a complicated system cannot be completely and explicitly expressed due to restrictions such as observation means, measurement methods, data description, etc. For complex events, the data stored in the event database will be approximate, unlike actual events. Therefore, the numerical data available from the literature, the description relating to the measurement method and the measurement conditions for obtaining the numeric data (hereinafter, this description is referred to as “related parameter”), and the content of these documents ( Bibliographic items related to (source) will be combined to build a fragmentary database. All attempts to build an event database using such a method have failed. It is extremely difficult to obtain a satisfactory evaluation from a user of an event database constructed based on an original data model composed of a combination of data and related parameters.

Generally, in using a database,
The user inputs a plurality of search conditions or selects a desired search condition from the search condition menu. The database creates a record set that matches each search condition, extracts a desired record group by logical operation of the record set, and displays or prints.

By such processing, in the case of the document database, the document information and bibliographic items suitable for the purpose can be obtained. In the case of the event database, attribute values such as numerical values, texts, figures and photographs can be obtained. Then, in either case of the literature database or the event database, a record group as a subset of the accumulated record group can be obtained together with the attribute value.

Further, in both the literature database and the event database, the retrieval result for the event that is not stored as a record in the database is an empty set. An event database constructed based on the original data model often returns an empty set in response to a search request due to the difficulty of data editing.

That is, in order to construct a database retrieval system that can satisfy users, it is assumed that the observed facts can be accurately described without excess or deficiency, and the lack of information will be reduced. It is important to consider how to complement.

[0007]

[Problems to be Solved by the Invention] Currently, in order to overcome "ambiguity of words" which is one of the attributes of natural language, keywords are prepared for each field, and searches utilizing such keywords are widely used. There is. However, it is not possible to enforce a strict definition and usage of keywords. For this reason, there is a limit to the support by the thesaurus and natural language processing, and the search result inevitably contains noise or misses the search. Therefore, the determination of the validity of the search result is left to the user.

Further, in the event database, when the data amount or the data quality is insufficient, the information amount of the search result is inevitably poor. In the former case, it is difficult to reach a valid conclusion, and in the latter case, it is easy to lead to a wrong conclusion. In other words, there is a problem that it is difficult to draw an effective conclusion because the construction of an event database that is uniform in quality and quantity requires a huge amount of work of extremely high quality.

Utilization of knowledge processing technology overcomes these limitations. However, knowledge processing technology is a technology related to the reuse of knowledge that is logically defined and established, and it is almost impossible to build a knowledge base for complex phenomena. In the case of complex events, it is necessary to acquire new knowledge from a data group whose logical relationship is only implicitly defined, and consider the use of the acquired knowledge on the spot. Therefore, there is a problem that the conventional knowledge processing technology is not effective for a solution search problem in which knowledge acquisition and knowledge utilization proceed simultaneously, such as a design solution, a solution for business strategy and preventive maintenance. .

In view of such a problem, the present invention provides a database search device, a database search method, a computer program, and a computer-readable recording medium capable of easily searching for a solution suitable for a problem of interest. The purpose is to

[0011]

In order to achieve such an object, the invention as set forth in claim 1 is a database search apparatus, wherein a first record having a plurality of fields for describing an event is provided. , The second record having a plurality of attributes for describing the parts constituting the event, and the first record stored in the storage means, the field of interest is empty. Extraction means for extracting a record that is not a value, setting means for setting at least two trial functions whose variables are the attributes selected from the second record stored in the storage means, and a trial set by the setting means It is characterized by comprising a function and a classifying unit for classifying the event based on the value of the field of interest in the record extracted by the extracting unit.

The invention described in claim 2 is the same as claim 1.
The database search device described in (4) is characterized by further comprising: creating means for creating a new attribute based on the attribute stored in the storage means.

The invention described in claim 3 is the same as claim 1
Alternatively, in the database search device according to the second aspect, the setting unit receives by the first receiving unit that receives selection of the attribute, second receiving unit that receives input of mathematical operation, and the first receiving unit. Assigned attributes,
Means for creating the trial function based on the mathematical operation accepted by the second accepting means.

The invention according to claim 4 is the same as claim 1.
In the database search device according to any one of 1 to 3, the categorizing means performs calculation of the trial function set by the setting means, and the calculating means on a space defined by the trial function. And a value of the field of interest corresponding to a reference point selected from the calculation results placed by the placing means, and other values existing around the reference point. Comparing means for comparing the value of the field of interest corresponding to the calculation result of 1.

The invention described in claim 5 is the same as claim 4
In the database search device according to [1], a display means for displaying a space in which the calculation result is arranged by the arrangement means, and a display means for changing the display form of the calculation result according to the value of the field of interest. It is further characterized by being provided.

The invention according to claim 6 is characterized in that a first record having a plurality of fields for describing an event and a second record having a plurality of attributes for describing a part constituting the event. Of the first record, a record in which the field of interest is not a null value is extracted from the first record, and the attribute selected from the second record is changed to a variable. Is set, and the events are classified based on the set trial function and the value of the field of interest in the extracted record.

The invention according to claim 7 is a computer program, comprising a first record having a plurality of fields for describing an event, and a plurality of parts for describing a part constituting the event. A second record having a second attribute, and a computer that stores a second record having a second attribute is extracted from the first record, in which the field of interest is not a null value, and the attribute selected from the second record is used as a variable. It is characterized in that at least two trial functions to be set are set, and the events are classified based on the set trial functions and the value of the field of interest in the extracted record.

An eighth aspect of the present invention is a computer-readable recording medium having the computer program according to the seventh aspect recorded therein.

In the present invention, a solution search problem such as a compound design solution is solved by performing an exhaustive search of a solution space that combines a trial function and a solution sorting function using an event database. Design is the act of deriving a set of facts satisfying any of the conditions from requirement specifications and available resources, axioms and theorems for synthesizing resources and predicting attributes. The solution space of the design solution is the space derived from the records corresponding to the facts about this act.

The event database is a functional system of trial functions for approximating a space in which a solution exists by performing an exhaustive search of a solution space in which execution of a trial function composed of a plurality of field operations and a sorting function are combined. To solve the problem.

The history of the solution space search can be recorded as a history of trial and error of the solution search, and can be stored as a reusable knowledge base after protocol analysis.

[0022]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. In the following description, an event refers to a recognized individual composed of one or more attributes (fields). For example, considering an event of human health management, this event is composed of attributes such as name, blood pressure, blood sugar level, and the like. In addition, the data model is the process of abstracting information in the abstracted world. A model for making rules independently of the algorithm of each application program.

FIG. 1 is a block diagram showing a hardware configuration of a database search device according to an embodiment of the present invention.

The database search device 20 shown in FIG.
0 is a system memory 204 and a CPU (Central Proc
essing unit) 206, an event database (DB) 208 including a set of a plurality of event data, and a ROM (Rand
om Access Memory) 210 is connected to the system bus 221.

The system memory 204 stores the DB control program 212 and the data 213 referred to by this program.

The DB control program 212 is a program for performing processing such as data registration, update, and deletion on the event database 208. Event database 2
08 is a set of data registered by the execution of the DB control program 212. A set of these data is registered in an internal or external storage device such as a hard disk, a magnetic disk, an optical disk, a magneto-optical disk, or the like.

A display 230, a speaker 232, a keyboard 234, and a mouse 236 are connected to the system bus 221 as peripheral devices.

The display 230 is for displaying an image edited by the CPU 206, and is an LCD (Li
quid Crystal Display) and CRT (Cathode-Ray Tube)
Etc. can be used. The speaker 232 converts the electric signal into sound and outputs the sound. The keyboard 234 and the mouse 236 are input devices used for controlling the cursor on the display 230 and inputting commands to the CPU 206. Display 230, speaker 232, keyboard 234, and mouse 236
Are all connected to the system bus 221 via an input / output (I / O) interface (not shown).

The various functions related to the present invention include the CPU 206.
Can be achieved by reading and executing the program stored in the ROM 210 or the system memory 204, or by supplying a computer-readable recording medium storing the program according to the present invention to the database search device 200. In the latter case, the CPU 206 of the database search device 200 reads out and executes the program stored in the recording medium.

As a recording medium for supplying the program to the database search device 200, specifically, a flexible disk, a hard disk, a magneto-optical disk, an optical disk, a CD-ROM, a CD-R, a magnetic tape, a non-volatile memory card and a ROM. In addition, a flash memory such as CompactFlash (registered trademark) can be used, but the present invention is not limited to these media.

The term "program" used in the present embodiment means, for example, a program written in a programming language such as BASIC, C language, C ++ or other high level programming language, assembly language or machine language. The invention is not limited to these languages.

Next, the contents of the event database installed in the database search device 200 according to the present embodiment will be described with reference to FIG.

The event database 208 includes event data 601 of a compound composed of two kinds of elements and part data 60.
I remember 2 and 2. The event data 601 is a set of records having a plurality of fields (attributes) for describing an event. The records that make up event data are melting point, density, hardness, lattice constant, precipitate size, dislocation density, crystal grain size, neutron cross section, tensile strength (tens
ile strength) and yield strength (yield strength). The part data 602 is a set of records having a plurality of fields (attributes) for describing the parts constituting the event, and includes a Mendereev number, an elastic modulus,
It consists of fields such as Young's modulus, elastic constant, and electronegativity. In the following description, in order to avoid confusion between the fields of event data and parts data, the fields of event data are described as “fields” and the fields of parts data as “attributes”.

Next, the database search method according to this embodiment will be described with reference to the flowchart of FIG.

In this embodiment, for a compound consisting of two elements, the characteristics of some compounds such as crystal structure or melting point and the attribute of the constituent element (element) such as Mendeleev number or electronegativity are described. Search for existing hidden correlations. The attributes of the two elements are combined by some mathematical operation (eg maximum atomic number),
This combination is called a trial function. The attribute and a plurality of mathematical operations on the attribute are selected by the user. The database search device 200 combines the selected attributes by mathematical operation to set a trial function, and executes the trial function for all elements to search for a solution. In the following explanation,
The purpose of the search is to find solution spaces that, like crystal structures, allow the properties of different compounds to be well separated by boundaries.

First, in step S102, the database search device 200 searches the event database 208 and sets the field of interest as a dependent variable (Dependent Variant).
able: DV). Then, a record that is a field of interest and in which a correlation exists between the fields is a non-null record, and a set of the extracted records is created.

The field of interest is a field related to the problem that the user is trying to analyze. For example, when attention is paid to the strength of a material (chemical substance) as a property of a compound that is a design target, that is, when the strength of a material is a dependent variable, the fields of tensile strength, yield strength, and hardness are Can be a field. Similarly, a field (lattice constant, size of precipitates, dislocation density, size of crystal grain) corresponding to a parameter describing the compatibility of the two-phase interface can be set as the “field of interest”. On the other hand, like the cross section of neutrons,
Fields unrelated to the strength of the material are not "fields of interest".

Correlation between fields means that, for example, when the environment of the compound is changed, such as when the temperature of the compound is changed, the attribute value of the field assumed to be the dependent variable changes with a certain relationship. It includes a relation to a physical change as well as a relation to a temporal change.

Whether or not there is a correlation can be determined as follows. Now, the problem that the user is trying to analyze is based on the characteristics of the two elements, between the crystal structure of those compounds (whether it is the NaCl or CsCl structure) and the attribute value of the elements constituting the compound. Suppose to find a correlation. When determining the correlation for temperature, a set of records for event data at room temperature (Dataset 1) is created for all compounds that crystallize to the NaCl or CsCl structure.
Then, for all of those compounds, a set of records (data set 2) for event data when the temperature is changed is created. Finally, data set 2 is subtracted from data set 1. Then, the difference between the data sets 1 and 2 is evaluated.

The process of assuming the field of interest as the dependent variable can be set by displaying a list of all the fields of the event data on the display 230 and accepting the selection of the field from this list. In this case, the dependent variable is determined according to the selection operation by the user using the input device.

Further, a problem to be noticed (for example, material strength, material stability, etc.) is displayed as a selection menu on the display 230, and the problem is selected according to the problem selected by the user using the input device. The field corresponding to may be set as the dependent variable.

Next, in step S104, the database search device 200 develops the event into parts.

Since the event database 208 stores the events of the compound consisting of two kinds of elements, the events in this case are developed into basic constituent elements, that is, parts called elements. For example, Ni used as a shape memory alloy
Ti develops into Ni and Ti. The database search device 200 extracts the component data of the developed component from the event database 208.

At the same time, the database search device 200 calculates a new attribute value that can be an independent variable for the dependent variable of interest, if necessary, based on the attribute value stored in the part data in advance. For example, when the strength of a material is the dependent variable, the total energy of atoms, the first derivative of the total energy, the second derivative of the total energy, etc. can be considered as the independent variables. If stored in advance, the first derivative and the second derivative can be calculated based on the stored data.

Then, in step S106, the database search device 200 records, for all the expanded parts, attribute values that can be independent variables with respect to the above dependent variables and that are not null values. Create a set of.

The database retrieval device 200 combines fields that are assumed to have independent variables with respect to the above dependent variables and must have values for all parts into one record, and collects a set of parts records (part database). create.

Next, in step S108, the database search device 200 accepts the selection of a specific attribute from the attributes constituting the parts database and the input of a mathematical operation for the selected attribute. Then, the trial function is set based on the accepted attribute (independent variable) and the mathematical operation.

In this case, the user selects at least one desired attribute value from the parts database in order to associate the dependent variable (field of interest) with the independent variable (attribute of each element). Select at least two mathematical operations on the attribute. The database search device 200 sets and stores the trial function by combining the attribute value selected by the user and the mathematical operation.

The attribute value assumed to be the dependent variable of interest can be set by displaying a list of attribute value candidates that can be independent variables on the display 230 and accepting the selection of the attribute value from this list. it can. In this case, it is set according to the selection operation from all attribute values by the user using the input device. Similarly, a mathematical operation can be set by displaying a list of candidates on the display 230 and accepting a selection from this list.

Finally, in step S110, the database search device 200 combines the systematic calculation of the above trial functions with the sorting of the values of the dependent variables in order to perform classification by the values of the dependent variables of the set P. And perform an exhaustive search for classification. By this operation,
A solution search is attempted in order to find an optimal trial function for distinguishing the values of the dependent variables based on some threshold.

FIG. 4 is a flowchart showing the processing in step S110 in more detail. The database search device 200 executes the plurality of trial functions set in step S108 for a combination of certain elements (S801). Specifically, when the set trial functions are "ratio of Mendeleev number of two elements" and "maximum value of Mendeleev number of two elements", the database search device 200 picks up two kinds of elements and Perform trial function calculations.

The database search device 200 determines, for each calculation result, whether or not the value of the dependent variable in the compound composed of the two elements used in the calculation exceeds a certain threshold (category of event). According to this judgment, the calculation result shows that the value of the dependent variable is less than the threshold value,
It can be divided into a characteristic in which the value of the dependent variable is a threshold value or more (S802).

Then, the calculation result is placed in the solution space defined by each trial function (S803). That is,
The calculation result is placed on the solution space with the x-axis as “ratio of Mendeleev number of 2 elements” and the y-axis as “maximum value of Mendeleev number of 2 elements”. Such calculations and arrangements are performed brute force for all elements (S8
04), finally all the calculation results are arranged in the solution space (sorting).

Next, the database search device 200
One point is set as a reference point from the points arranged on the solution space (S805), and the characteristic of the reference point is determined. Next, one or more points closest to the point are found, the characteristic of the found point is compared with the characteristic of the reference point, and it is determined whether they are the same characteristic (S806). Then, the probability that the characteristic of the point closest to the reference point is the same as the characteristic of the reference point is calculated.

Next, one or more points closest to the reference point are found, and the probability that the characteristic of the found point is the characteristic of the reference point is obtained in the same manner as above. Such processing is repeated for all the points arranged in the solution space (S807).

Next, the database search device 200
Based on the obtained probability, at a certain distance from the reference point,
Judge whether the characteristics of the points are well separated. Specifically, this determination is performed by obtaining a boundary line in the solution space in which the probability obtained above is most rapidly reduced. Then, the solution space with the best separation is identified as the solution space suitable for the problem in question.

In this step S110, the calculation result (that is, the solution) of the trial function classified by the value of the dependent variable
Is provided to the user by displaying each point arranged in the solution space on the display by color-coding according to the determination result.

In this way, the systematic operation of the trial function and the sorting of the values of the independent variables are combined to perform an exhaustive search for the solutions, whereby the identification of the solution space that better classifies the events can be performed. Will be realized.

In the present embodiment, there is no record that matches the search condition, and only the systematic sorting performed based on the attribute values of the elements that make up the substance makes it possible to use the macroscopic characteristics of the substance and the basic structure. It is particularly effective when the solution space for the correlation with the element (element) cannot be identified.

Next, an example of a screen displayed on the display in the database search method according to this embodiment will be described with reference to FIGS. In the following description, the problem for the user is the stability of the compound,
The analysis of this question is based on the compound formation (com
pound former), and compound non-formation (c
It is carried out by determining an open former (when constituent elements are randomly arranged at lattice positions or become amorphous). Note that this example is used as a sample because the data has been thoroughly examined, but the selection as a dependent variable is not limited to this.

FIG. 5 shows an example of the input box displayed on the display in step S108. In the figure, the input box 300 includes a menu 301 of independent variables and a menu 302 of mathematical operations. In the independent variable menu 301, selectable independent variables include bulk modulus, Young's Modulus, and modulus of rigidi.
ty), Mendeleev number, and other attribute values are displayed. In addition, the addition (S
um), subtraction (Difference), etc. are displayed. The database search device 200 uses the trial function set based on the result of this menu selection to perform the search in the solution space search.

The user can operate the mouse 236 to select a desired independent variable from the menu 301 on the input box 300 and a desired mathematical operation from the menu 302. In the figure, all the lines of the menu 302 are selected for the “M1 Mendereev number” of the menu 301. The user can set the trial function by pressing the “Next” button 303. In this way, the processing of accepting the selection of the element attribute, accepting the mathematical operation, and combining the accepted attribute and the mathematical operation to create the trial function is realized.

FIG. 6 is a diagram showing an example of display of the calculation result in step S110.

Once the calculation of the trial function is completed, the screen shown in the figure is displayed on the display. As shown in the figure, in the window 401 on the left side, the existence probabilities of points having the same characteristics as the reference points are displayed as distances (Neig) from the reference points.
hbours). For example, “20” on the horizontal axis indicates a distance of 20 units from the reference point in the solution space.

A window 403 on the lower right side displays a list of calculation results for each trial function sorted according to the integral of the curve at the closest distance from the reference point (that is, one unit from the reference point).

Finally, in the upper right window 402, a list of trial functions is displayed. The window 402 is composed of three items, “Profile”, “1”, and “50” in order from the left. Of these, the item "P
The name of the trial function is displayed in "file". The number in parentheses in the item "1" indicates the probability that the characteristic of the point closest to the reference point is the same as the characteristic of the reference point. Further, the number in parentheses in the item “50” indicates the probability that the characteristic of the point at the position closest to the fiftieth from the reference point is the same as the characteristic of the reference point.

According to this window 402, "38-
"03 / 38-06 / 00-00" in the solution space defined by the trial function, 100% of the points at a distance of 1 unit from the reference point and about 92 of the points at a distance of 50 units.
% Have the same crystal structure as the compound corresponding to the reference point. The curve 404 in the left window 401 is
The separation state for the trial function "38-03 / 38-06 / 00-00" is shown. These results show that the compound former and the compound nonform.
er) shows a very good "separation" for the two crystal structure patterns. Therefore, it can be said that the probability of correctly predicting the structure of a new compound in which the events are best classified and the database does not include the value of the dependent variable is high when this trial function is used.

FIG. 7 shows a graphical representation of the solution space for this trial function. Window 50 on the left side of FIG.
In 1, the calculation result of the trial function is arranged with the x-axis as the ratio of Mendeleev numbers and the vertical axis as the maximum value of Mendeleev numbers. In addition, among the points indicating the calculation results, the black dots represent that the substance composed of the parts used in the calculation forms a compound, and the white dots represent that the compound does not form. As can be seen from the figure, good separation between the crystal structure patterns (white dots and black dots) is obtained.

By identifying the solution space in this way,
Designers can predict crystal structure patterns for materials for which there is no data. In addition, the application of this method,
It is not limited to the crystal structure pattern.

Further, since each point corresponds to the technical literature, it can be predicted based on the points in the solution space that, for example, a place where the points are densely studied is actively researched. It is possible to visually grasp the research and development trend of compounds.

In the example shown in the figure, the solutions are categorized by the color of the dots, but they may be displayed by using other representational forms such as shapes.

Although the embodiments of the present invention have been described in detail above, it is needless to say that the present invention is not limited to the above-mentioned embodiments and can be implemented in various other forms.

For example, the conventional orthodox method for elucidating the mechanism of a complex event requires careful consideration in setting and executing the trial function, and the event database system to which the conventional search method is applied requires a huge amount of calculation time. , Computational resources are needed. Therefore, the calculation process, which is the rate-determining process in the above database search process, may be performed by a calculation server that is configured as a separate body, and a method of registering the calculated value in the event database may be adopted.

The space represented by the trial function is 2
It is not limited to dimensions. FIG. 8 shows an example of the solution space when the space represented by the trial function is three-dimensional.

Further, the mathematical operation of the trial function is calculated by four arithmetic operations,
By using a simple functional system with only comparison operations, the time required to verify the work hypothesis can be greatly reduced to a few minutes, and the utility as an information processing device that accelerates the user's idea in the solution process is acquired. be able to.

In addition to the above-described design solution, the present invention is also applicable to a business strategy in which a required specification in design is a business goal, a required specification to search for a solution related to preventive maintenance for maintaining the soundness of an artifact, and the like. Can also be applied.

[0077]

As described above, according to the present invention,
A solution space suitable for the problem of interest can be easily identified.

Therefore, an information system platform is constructed to support the process of acquiring an idea by making full use of the event database, which is a collection of facts, and the process of adding the acquired idea as a fact.

[Brief description of drawings]

FIG. 1 is a block diagram showing a hardware configuration of a database search device according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of contents of event data and component data stored in an event database.

FIG. 3 is a flowchart showing a procedure of a database search method according to the embodiment of the present invention.

FIG. 4 is a flowchart showing the process in step S110 in more detail.

FIG. 5 is a diagram showing an example of an input box displayed on the display in step S108.

FIG. 6 is a diagram showing an example of a display of calculation results in step S110.

FIG. 7 is a diagram showing an example of a graphic representation of a solution space displayed on the display in step S110.

FIG. 8 is a diagram showing an example of a solution space in the case where the space represented by the trial function is three-dimensional.

[Explanation of symbols]

200 Database search device 204 system memory 206 CPU 208 Event database (DB) 210 ROM 212 DB control program 213 data 221 system bus 232 speaker 234 keyboard 236 mouse 300 input box 301, 302 menu 303 "Next" button 401, 402, 403, 501 windows 404 curve

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Pierre Villas             Swiss Sea H-6354 Fitznow               Schwanden 400 F term (reference) 5B075 NK46 PP02 PP03 PQ02 PQ13                       QM07

Claims (8)

[Claims] 1. Storage means for storing a first record having a plurality of fields for describing an event, and a second record having a plurality of attributes for describing a component constituting the event. An extracting unit that extracts a record in which the field of interest is not a null value from the first records stored in the storage unit, and an attribute selected from the second record stored in the storage unit as a variable Grouping the events based on the setting means for setting at least two trial functions, the trial function set by the setting means, and the value of the field of interest in the record extracted by the extracting means. A database search device, comprising: 2. The database search device according to claim 1, further comprising a creation unit that creates a new attribute based on the attribute stored in the storage unit. 3. The setting means, first acceptance means for accepting a selection of the attribute, second acceptance means for accepting an input of a mathematical operation, and attributes accepted by the first acceptance means. 3. The database search device according to claim 1, further comprising: a unit that creates the trial function based on the mathematical operation received by the second receiving unit. 4. The classification unit arranges the calculation result executed by the calculation unit in a space defined by the calculation unit, the calculation unit executing the calculation of the trial function set by the setting unit. And a value of the field of interest corresponding to the reference point selected from the calculation results arranged by the arranging means,
4. The database search apparatus according to claim 1, further comprising: a comparison unit that compares the value of the field of interest corresponding to another calculation result existing around the reference point. 5. A display means for displaying a space in which the calculation result is arranged by the arrangement means, further comprising a display means for changing a display form of the calculation result according to a value of the field of interest. The database search device according to claim 4, wherein 6. A database search device which stores a first record having a plurality of fields for describing an event and a second record having a plurality of attributes for describing a component constituting the event. The method for searching a database according to claim 1, wherein a record in which a field of interest is not a null value is extracted from the first record, and at least two trial functions having the attribute selected from the second record as a variable are set. Then, the database search method is characterized in that the events are classified based on the set trial function and the value of the field of interest in the extracted record. 7. A computer for storing a first record having a plurality of fields for describing an event and a second record having a plurality of attributes for describing a component constituting the event. From the first record, a record in which the field of interest is not a null value is extracted, and at least two trial functions having the attribute selected from the second record as a variable are set, and the set trial is set. A computer program for classifying the events based on a function and a value of the field of interest in the extracted record. 8. A computer-readable recording medium on which the computer program according to claim 7 is recorded.