JPH0934760A - Device and method for optimizing retrieving instruction - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simply and easily decide the description of a retrieving instruction to be applied to a data base responding to the structure information of a data base system. SOLUTION: A transformation knowledge storage means 7 stores transformation knowledge for optimizing the retrieving instruction. A structure information possessing means 5 possesses the structure information of the data base system from a data base system 9 and outputs it to an optimizing means 3. The optimizing means 3 fits the applied retrieving instruction and structure information into the transformation knowledge stored in the transformation knowledge storage means 7. When it is judged that transformation is enabled, the applied retrieving instruction is transformed and outputted as an optimized retrieving instruction.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a search instruction optimizing apparatus used in a database system, and more particularly to conversion of search instructions.

[0002]

2. Description of the Related Art Today, a database system may be read on an application program to perform a search process or the like. For example, in an application program of accounting software, by accessing a database system, it is possible to obtain average age and average annual income data of employees and perform accounting processing.

When performing a search process on this application program, the search efficiency changes depending on how the SQL statement is described. This is because the processing characteristics of the database system are different for each database system, and the search efficiency is improved by reviewing the description of the SQL statement. This processing characteristic refers to a characteristic of a processing procedure such that a certain processing is first performed for SQL statements having a plurality of processing contents regardless of the description order. Even in the same database system, depending on the data structure (for example, whether or not an index has been created, the number of rows in each table, etc.), even if it is an SQL statement that searches for the same content, The speed may be different.

Therefore, generally, after the SQL statement to execute a search instruction is determined, the SQL statement is tuned.

However, such tuning is not easy. This is because it is necessary to understand the characteristics (including structural information) of the database system as to how the SQL statement can be used for high-speed retrieval. Therefore, the tuning is impossible for a beginner who does not understand the characteristics.

An object of the present invention is to easily and easily determine the description of a given search command according to the characteristics of a database system, and further to reduce the development process in application software design and development. In particular, it is an object to simply and easily determine the description of a search command given according to the structural information of the database system.

[0007]

[Means for Solving the Problems]

[0008]

In the search instruction optimizing apparatus according to the first aspect of the present invention, in order to convert the search instruction into a search instruction that can be processed at a higher speed, conversion knowledge for optimizing the search instruction is provided. The search command is stored and applied, and the given search command is applied to the conversion knowledge to convert the search command.

That is, as described in FIG. 1, which is a diagram showing the overall configuration of the invention, in the search instruction optimizing apparatus of claim 1, a conversion for storing conversion knowledge for optimizing the search instruction. A knowledge storage means; and an optimization means for applying a given search command to the conversion knowledge and converting the search command to output an optimized search command when it is determined that conversion is possible. And

According to a second aspect of the present invention, there is provided a search instruction optimizing apparatus comprising structure information obtaining means for obtaining structure information from a database system and outputting it to the optimizing means when a structure information obtaining instruction is given. When it is determined that the structure information of the database system is necessary, the converting unit outputs the structure information acquisition command, and uses the structure information acquired by the structure information acquisition unit in addition to the search command. It is characterized by determining whether or not the conversion is possible.

According to a third aspect of the present invention, in the retrieval instruction optimizing device, the conversion knowledge comprises a membership function and a rule, and the optimizing means transforms the retrieval instruction by fuzzy inference. To do.

In the search instruction optimizing apparatus according to the present invention, the optimizing means should search the table to be searched by the search instruction using the index based on the column selectivity and the number of rows. While judging whether or not
When it is determined that the index should be used for the search, the search command is converted so that the index search can be performed.

In the search instruction optimizing method of claim 5, the given search instruction is applied to conversion knowledge stored in advance, and if conversion is possible, the search instruction is converted and optimized. It is characterized by doing.

In the search command optimizing method of claim 6, when it is determined that the structure information of the database system is necessary, the structure information is acquired from the database system, and in addition to the search command, the structure information is further acquired. It is characterized by using the acquired structural information.

According to a seventh aspect of the present invention, in the search instruction optimizing method, the conversion knowledge is composed of membership functions and rules, and whether or not the conversion is possible is determined by using fuzzy inference. And

In the search instruction optimizing method according to claim 8, it is determined whether or not the table to be searched by the search instruction should be searched using the index based on the column selectivity and the number of rows. When it is determined that a search should be performed using the index, the search command is converted so that the index search can be performed.

According to a ninth aspect of the present invention, there is provided a storage medium storing the program, wherein the apparatus or method according to any one of the first to eighth aspects is realized by using a computer.

[0018]

[Definition of terms] The concept of terms used in the claims used to express the technical idea devised to solve the problems is defined as follows, and the relationship between the terms and the embodiments will be described. .

"Structural information of database system": Structural information means information about a processing target specified by a search command of a conversion target. In the embodiment, it corresponds to the table information about the table referred to in the conversion target SQL, the column information about the column, and the index information about the column. This structural information includes individual required structural information. The individually necessary structure information is data necessary for determining whether or not the structure information meets the condition of the conversion knowledge.

"Column selectivity": A value obtained by the number of rows in a column / the number of unique data. The larger the value, the smaller the column selectivity, and the smaller the value, the greater the column selectivity.

[0021]

In the search instruction optimizing apparatus or method according to claims 1 and 5, when the given search instruction is applied to the conversion knowledge stored in advance, conversion is possible. Converts and optimizes the search command. Therefore, it is possible to obtain a search command that enables higher speed processing. As a result, even a person who has a poor understanding of database processing can create a search command similar to that of a skilled person. That is, it is possible to reduce the skill of adjusting SQL statements regarding the processing characteristics of the database, semi-automating, and further reduce the design development process.

In the search instruction optimizing apparatus or the method thereof according to claims 2 and 6, when it is determined that the structure information of the database system is necessary, the structure information is acquired from the database system and the search is performed. In addition to the command, the acquired structural information is further used. Therefore, it is possible to obtain a search command suitable for the structure of the database system. This enables higher speed processing.

In the search instruction optimizing apparatus or the method thereof according to claims 3 and 7, the conversion knowledge is composed of membership functions and rules, and the fuzzy inference is used to judge whether the conversion is possible or not. Using. Therefore, since inference that satisfies a plurality of conditions can be described with knowledge that matches the human sense, knowledge can be easily constructed and modified.

According to another aspect of the present invention, there is provided a search instruction optimizing apparatus or method thereof, wherein a table to be searched by the search instruction is searched using an index based on column selectivity and the number of rows. If it is determined that the search should be performed using the index, it is converted so that the index search can be performed. Therefore, the search instruction can be optimized in consideration of column selectivity and the number of rows.
As a result, it is possible to reduce the need for de-skilling and semi-automating the adjustment of SQL statements regarding the processing characteristics of the database, and further reduce the design development process.

[0025]

An embodiment of the present invention will be described with reference to the drawings. The search instruction optimizing device 1 shown in FIG. 1 includes a conversion knowledge storing means 7, an optimizing means 3, and a structure information acquiring means 5.

The conversion knowledge storage means 7 stores conversion knowledge for optimizing the search command. Structural information acquisition means 5
Is connected to the database system 9. The database system 9 is an ordinary relational database system, and includes a database unit that stores a plurality of data and a database management system that manages this database unit. When the structural information acquisition command is given, the structural information acquisition unit 5 acquires the structural information of the database system from the database system 9 and outputs it to the optimization unit 3.

When the optimizing means 3 determines that the structural information of the database system is necessary when determining whether the conversion is possible, the optimizing means 3 outputs the structural information acquisition command and outputs the structural information to the search command. In addition, it is determined whether the conversion is possible by using the structure information acquired by the structure information acquisition means. That is, the given search command and the structural information given from the structural information acquisition means 5 are converted into the conversion knowledge storage means 7.
Applied to the conversion knowledge stored in. When it is determined that conversion is possible, the given search command is converted and output as an optimized search command.

In this embodiment, the conversion knowledge storage means 7
The transformation knowledge stored in is composed of membership functions and rules. Then, the optimizing means 3 converts the search instruction by fuzzy inference.

FIG. 2 shows the retrieval instruction optimizing apparatus 1 shown in FIG.
1 shows an example of a hardware configuration realized by using a CPU.

The search instruction optimizing device 1 includes a CPU 23, a memory 27, a hard disk 26, a CRT 28, a keyboard 28, an FDD 25, an input / output interface 32 and a bus line 29.

The CPU 23 controls each unit via the bus line 29 according to a control program stored in the hard disk 26.

This control program is read from the flexible disk storing the program via the FDD 25 and stored (installed) in the hard disk 26. The hard disk 26 stores correction knowledge (see FIGS. 6 and 7) described later. The memory 27 stores the calculation result and the like. CRT28 has
The correction plan processed by the CPU 23 is displayed.

A database management system (DBMS) 34 is connected to the input / output interface 32. The database unit 35 stores various types of data, as well as information about the logical structure and physical structure of the database system, for example, addresses at which tables are stored, how indexes are added to tables, and the like.

In this embodiment, an Oracle database system (manufactured by Oracle) is used as the database system and DG / U is used as the operating system.
The X operating system (manufactured by Data General) was adopted.

Next, referring to FIG. 3, the hard disk 2
The program stored in 6 will be described.

First, the CPU 23 determines the processing target knowledge number i
Are initialized (step ST1 in FIG. 3). Next, CPU
23 reads the target SQL sentence and stores it in the memory 27 (step ST3). Here, SQ of FIG. 5A
It is assumed that the L sentence has been read.

The CPU 23 reads the i-th correction knowledge into the memory 27 (step ST5). As shown in FIG. 6, the correction knowledge is composed of a knowledge ID, necessity of database structure information, and SQL sentence correction knowledge. In this case, since i = 1, the CPU 23 reads the corrected knowledge of the knowledge ID 0001 into the memory 27.

Next, the CPU 23 executes the SQ shown in FIG. 5A.
It is determined whether the L sentence corresponds to the corrected knowledge of the knowledge ID 0001 shown in FIG. 6 (step ST10 in FIG. 3).

The process of step ST10 will be described in detail with reference to FIG. First, the CPU 23 determines whether the correction knowledge of the knowledge ID 0001 is the correction knowledge that requires the database structure information (step ST2 in FIG. 4).
1). In this case, as shown in FIG. 6, the knowledge ID 0001
For the correction knowledge of, the database structure information is "necessary"
Since it is stored, the process proceeds to step ST23. In step ST23, the CPU 23 determines whether or not the database structure information has already been extracted. In this case, since the database structure information has not yet been extracted, step ST2
Go to 5.

In step ST25, the CPU 23 uses the input / output interface 32 shown in FIG.
4 to extract the database structure information from the database dictionary (not shown) of the database unit 35.

The database structure information is the S to be converted.
It refers to the data related to the processing target specified in the QL statement. Specifically, it is table information regarding a table referred to in the conversion target SQL, column information regarding a column, and index information regarding the column.

In this embodiment, the table name, the table owner name, and the row number sequence of the table are adopted as the table information. In addition, the column name, the table name, the table owner name, the data type, and the number of unique data are adopted as the column information. Also, as index information, index name, index owner name,
A column name, a table owner name, and a column order are adopted. That is, for the SQL statement in FIG. 5A,
The table information, the column information, and the index information as shown in C are obtained. Regarding the index information shown in FIG. 9C, the column “NAME” does not exist because no index is added to the column.

In FIG. 9B, the data type "NUMBE
“R” means numeric type data, and the data type “VARCHER
"2" refers to character string type data. The number of unique data is a value indicating how many rows have different values in the data of the column.

In the present embodiment, since the Oracle database is adopted, the SQL statement shown in FIG.
The table information shown in A, the SQL statement shown in FIG. 8B can obtain the column information shown in FIG. 9B, and the SQL statement shown in FIG. 8C can obtain the index information shown in FIG. 9C.

Next, the CPU 23 executes the SQ shown in FIG. 5A.
It is determined whether or not the L sentence satisfies the condition of the corrected knowledge of the knowledge ID 0001 shown in FIG. 6 (step ST2 in FIG. 4).
7).

The conditions for the corrected knowledge of the knowledge ID 0001 shown in FIG. 6 are the following three.

Condition (1) The expression in the WHERE clause of the SELECT command is
Condition (2) Connected by OR logical operator, condition (2) Index is added to each expression sequence connected by OR logical operator, Condition (3) Sub-knowledge S1 is true About SQL statement shown in FIG. 5A , It is determined whether these three conditions are satisfied. First, regarding condition (1), W
HERE NAME_ID = 100 OR SALARY = 250,000, which is satisfied. Regarding condition 2, as shown in Fig. 9C, "NAME_ID"
And, "SALARY" is indexed and satisfied. Regarding condition 3, it is determined whether or not the sub-knowledge 1 shown in FIG. 7 is true. In this embodiment,
As shown below, such a judgment is made by making a fuzzy inference using a rule and a membership function.

Specifically, it is determined as follows. Using the rules shown in FIG. 10, the membership function of the antecedent part shown in FIGS. 11A and 11B, and the membership function of the consequent part shown in FIG. 11C, the suitability of the antecedent part is obtained for each rule. The suitability of the consequent part is calculated based on the suitability of the antecedent part. After processing the consequent part for all rules (two in this case), combine the consequent part membership functions of each rule obtained by the consequent part process to generate a combined membership function, and generate The synthesized membership function is subjected to defuzzification by the center of gravity method or the like to determine whether it is true or not.

Here, the column selectivity can be obtained by the following formula.

Column Selectivity = Number of Column Rows / Number of Unique Data (Equation 1) Specifically, in this case, the selectivity of the column “NAME_ID” is 80,000 (table 9A) and the number of unique data is 80,000 (see FIG. 9B), the selectivity of the column “NAME_ID” = 1. In this case, when applied to the membership function shown in FIG. 11A, it can be seen that the degree of suitability of “high column selectivity” is 1 for the column “NAME_ID”. Also, since the number of rows in the table is 80,000, when the membership function shown in FIG. 9B is applied, the goodness of fit of “large number of rows of table” is 0.6 and the goodness of fit of “small row of table” Is 0.3.

The selectivity of the column “SALARY” is 80,000 table rows (see FIG. 9A) and the unique data number is 10,000 (see FIG. 9B). Selectivity = 8. In this case, applying the membership function shown in FIG. 11A, the column "SALARY"
It can be seen that the conformity of “high column selectivity” is 1.

Since the number of rows in the table is 80,000, when the membership function shown in FIG. 11B is applied, the degree of conformity for "large number of rows in table" is 0.6 and "small number of rows in table". The degree of conformity of is 0.3. Next, FIG.
For each rule shown in 0, the conformance of the consequent part is obtained.
In this case, the rule 1 has a goodness of fit of 0.6, and the rule 2 has a goodness of fit of 0.3. The generated composite membership function is defuzzified by the center of gravity method, and is determined to be true.

Therefore, since the SQL sentence shown in FIG. 5A satisfies the above three conditions, it is determined that the condition is satisfied in step ST27 in FIG. 4 and step ST1 in FIG.
Go to 5.

At step ST15, the CPU 23 creates a correction plan. In this case, the knowledge ID 000 shown in FIG.
If the condition 1 is satisfied, "S using the UNION set operator
Since it is the knowledge of "convert to QL sentence", the SQL sentence shown in FIG. 5A is converted into a correction plan as shown in FIG. 10A by using an SQL sentence using a "UNION" set operator. The CPU 23 stores the correction plan in the memory 27 and displays it on the CRT 28 (step ST1 in FIG. 3).
7). In this way, the SQL sentence is converted.

In the SQL statement shown in FIG. 5A, the column "NA
As for ME_ID ”and“ SALARY ”, indexes are added as shown in FIG. 9C. Therefore, as shown in FIG. 12A, an index search (search using an index) is performed by separating both WHERE clauses. You can This enables high-speed processing.

In this embodiment, step ST
At 23, the CPU 23 determines whether or not the database structure information has already been extracted. As a result, once the database structure information has been extracted, the time for re-reading is unnecessary, and higher speed processing is possible.

Next, the processing when the SQL sentence shown in FIG. 5B is given will be described. The CPU 23 initializes the processing target knowledge number i (step ST1 in FIG. 3). Next, the CPU 23 reads the target SQL statement shown in FIG. 5B and stores it in the memory 27 (step ST
3). Then, the i-th correction knowledge is read (step ST5). In this case, i = 1, so CPU2
3 reads the correction knowledge of the knowledge ID 0001 into the memory 27.

Next, the CPU 23 executes the SQ shown in FIG. 5B.
It is determined whether the L sentence corresponds to the corrected knowledge of the knowledge ID 0001 shown in FIG. 6 (step ST10 in FIG. 3). In the judgment of step ST10, as already described,
The CPU 23 determines whether the correction knowledge of the knowledge ID 0001 is the correction knowledge that requires the database structure information (step ST21 in FIG. 4).

In this case, as shown in FIG. 6, the knowledge ID 0
Regarding the correction knowledge of 001, since the database structure information is stored as "necessary", the process proceeds to step ST23. In step ST23, the CPU 23 determines whether or not the database structure information has already been extracted. In this case, since the database structure information has not yet been extracted, the process proceeds to step ST25 and the database structure information is extracted.

Next, the CPU 23 executes the SQ shown in FIG. 5B.
It is determined whether or not the L sentence satisfies the condition of the corrected knowledge of the knowledge ID 0001 shown in FIG. 6 (step ST2 in FIG. 4).
7). In this case, the SQL statement shown in FIG. 5B does not meet all the above three conditions, and therefore the SQL statement shown in FIG.
If the condition is not satisfied in step ST27, the process proceeds to step ST11 in FIG. In step ST11, the CPU
23 determines whether i is the last. In this case, since it is not the final one, the process proceeds to step ST12, i is incremented, the process proceeds to step ST5, i-th, that is, ID0.
The correction knowledge of 002 is read.

The CPU 23 executes the SQL statement shown in FIG.
It is determined whether or not it corresponds to the correction knowledge of the knowledge ID 0002 shown in FIG. 6 (step ST10 in FIG. 3). Step ST
In the judgment of 10, as described above, the CPU 2
3 determines whether the correction knowledge of the knowledge ID 0002 is the correction knowledge that requires the database structure information (step ST21 in FIG. 4).

In this case, as shown in FIG. 6, the knowledge ID 0
Regarding the correction knowledge of 002, since the database structure information is stored as "unnecessary", the process proceeds to step ST27. In step ST27, the CPU 23 determines whether or not the SQL statement shown in FIG. 5B satisfies the condition of the correction knowledge of the knowledge ID 0002 shown in FIG. 6 (step ST in FIG. 4).
27). In this case, in the SQL sentence shown in FIG. 5B,
No group function is specified in the HAVING clause of the SELECT command. Therefore, since the condition of the correction knowledge of the knowledge ID 0002 is satisfied, the process proceeds from step ST27 of FIG. 4 to step ST15 of FIG. 3 to create and display the correction plan (step ST17). In this way, S shown in FIG.
The QL sentence is converted into the SQL sentence shown in FIG. 12B.

When the HAVING clause is converted to the WHERE clause,
The fact that the search processing time is shortened depends on the characteristics of the Oracle database system. That is, in the Oracle database system, when the SQL statement shown in FIG. 5B is given, the column “LO” of the table “ADDRESS” is given.
"CATION", the grouping process, that is,
Group data with the same place name. And, "TOK
The average age is calculated for the data of the row having the place name which is neither “YO” nor “YOKOHAMA”. On the other hand, in the case of the SQL statement shown in FIG. 12B, the table “ADDRE
For the column "LOCATION" of "SS", click "TO
Data of a line having a place name that is neither "KYO" nor "YOKOHAMA" is extracted. Then, these are grouped. Therefore, unnecessary grouping processing is unnecessary, and higher speed processing is possible.

Similarly, the SQL sentence shown in FIG. 5C is converted as shown in FIG. 12C using the knowledge ID 0003. High speed processing is possible by such conversion because when an index exists, columns always exist, and therefore counting an indexed column enables higher speed processing.

The SQL statement shown in FIG. 5D is a knowledge ID.
0004 is used to convert as shown in FIG. 12D.
The reason why high speed processing is possible by such conversion is as follows. Given the SQL statement shown in FIG. 5D, for the column “account_name” of the table “transaction”, the “account_name” and “trans” of the columns in which the first to seventh characters are “CAPITAL” are given.
"date" and "amount" are calculated. On the other hand, in the case of the SQL statement shown in FIG. 12D, the prefix match search including “CAPITAL” is performed using the index. Therefore, high-speed processing can be performed as much as the index can be used.

As described above, it is determined whether or not each given SQL sentence corresponds to the correction knowledge stored in advance, and the convertible SQL sentence is converted.

[Other Embodiments] In the above embodiment, a flag indicating whether or not database structure information is necessary is added to each correction knowledge as shown in FIG. 6, and the correction knowledge is referred to in the SQL statement to be converted. The related table information, column information, and index information are extracted for each table and column. However, the present invention is not limited to this, and the data may be extracted only when necessary for the condition determination as described below. Such data is called individual required structure information.

That is, the individual required structure information is the S
Of the structural information of the database necessary for performing the processing specified by the QL sentence, it means the data necessary for judging whether or not the condition of the conversion knowledge is met. Specifically, regarding the SQL statement of FIG. 5A, the individual required structure information in the SQL statement 1 is “none” for the condition (1),
The condition (2) is index information regarding the columns “NAME_ID” and “SALARY”, and the condition (3) is shown in FIG.
It is the number of rows of the table name "EMPLOY" shown in A and B, and the unique number of columns "NAME_ID", "NAME", and "SALARY".

The determination using the individual required structure information (step ST10 in FIG. 3) will be described with reference to FIG. The CPU 23 determines whether or not the individual required structure information is required (step ST41). For example, the SQ shown in FIG. 5A
In the case of the L sentence, it can be seen from the condition (2) that the individual required structure information is required.

Next, the CPU 23 determines the SQL to be converted.
The individual required structure information is determined with reference to the sentence and the correction knowledge (step ST43). In this case, referring to the SQL statement shown in FIG. 5A and the correction knowledge 0001 shown in FIG. 6, the index information regarding the columns “NAME_ID”, “SALARY”, the number of rows of the table name “EMPLOY”, and the column “NA” are displayed.
The unique numbers of “ME_ID”, “NAME”, and “SALARY” are determined to be the individual required structure information. Next, the CPU 23
Accesses the DBMS 34, and the database unit 35
The individual required structure information is extracted from the database dictionary (step ST45 in FIG. 13).

As described above, of the structure information, the SQL
By extracting only the individual required structural information necessary for processing the sentence, the speed can be increased by the extraction time of the unnecessary structural information. In this case, the necessity of the database structure information shown in FIG. 6 need not be stored.

In each of the above embodiments, the fuzzy reasoning is used for the judgment of the sub-knowledge S1, but the judgment may be made by the normal reasoning.

In this embodiment, the CPU 23 is used to realize the above-mentioned functions, and this is realized by software. However, some or all of them may be realized by hardware such as a logic circuit.

In the above embodiment, the program is stored in the flexible disk and installed in the hard disk 26. However, other storage media such as ROM, IC card, optical disk and CD are used.
-The program may be installed from a ROM or the like, and the program may be directly read from these storage media and executed.

Note that the same can be applied to a database system that performs fuzzy search.

[Brief description of drawings]

FIG. 1 is a functional block diagram of a search instruction optimizing apparatus 1 according to the present invention.

FIG. 2 is a diagram showing a hardware configuration in which the search instruction optimizing device 1 shown in FIG. 1 is realized by a CPU.

FIG. 3 is a processing flowchart of the search instruction optimizing device 1.

FIG. 4 is a detailed flowchart of step ST10 in FIG.

FIG. 5 is a diagram showing an input SQL sentence (before conversion).

FIG. 6 is a diagram showing an example of correction knowledge for converting an SQL sentence.

FIG. 7 is a diagram showing an example of modified sub-knowledge.

FIG. 8 is a diagram showing an SQL statement for acquiring database structure information.

FIG. 9 is a diagram showing an example of acquired database structure information.

FIG. 10 is a diagram showing a rule for fuzzy inference as to whether or not it matches the sub-knowledge S1.

FIG. 11 is a diagram showing a membership function for fuzzy inference as to whether or not it matches the sub-knowledge S1.

FIG. 12 is a diagram showing a modified SQL sentence (after conversion).

FIG. 13 is a detailed flowchart of another determination method in step ST10 of FIG.

[Explanation of symbols]

 3 ... Optimization means 5 ... Structural information acquisition means 7 ... Conversion knowledge storage means 40 ... Database system

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hideki Yatake 10 No.10 Hanazono Dodocho, Ukyo-ku, Kyoto Prefecture Kyoto Prefecture Omron Co., Ltd. Within Muron Co., Ltd.

Claims (9)

[Claims] 1. A conversion knowledge storage means for storing conversion knowledge for optimizing a search command, applying a given search command to the conversion knowledge, and converting the search command when it is determined that conversion is possible. And an optimization means for outputting an optimized search instruction. 2. The search command optimizing device according to claim 1, wherein when a structure information acquisition command is given, structure information acquisition means for acquiring the structure information of the database system from the database system and outputting the structure information to the optimization means. When the optimization unit determines that the structure information of the database system is necessary, the optimization unit outputs the structure information acquisition command, and the structure information acquisition unit acquires the structure information acquisition command in addition to the search command. A search instruction optimizing device, characterized in that it is determined whether or not the conversion is possible using structure information. 3. The search instruction optimizing apparatus according to claim 1 or 2, wherein the conversion knowledge comprises a membership function and a rule, and the optimizing means converts the search instruction by fuzzy inference. A search instruction optimizing device characterized by: 4. The search instruction optimizing device according to claim 2 or 3, wherein the optimizing means: 1) based on column selectivity and the number of rows for a table to be searched by the search instruction. In addition to determining whether to search using the index, 2) if it is determined to search using the index, convert the search command to enable index search. And a search instruction optimizing device. 5. A search instruction optimizing method for optimizing a search instruction given to a database system, which can be converted by applying the given search instruction to conversion knowledge stored in advance. In this case, the search command is optimized by converting the search command. 6. The search instruction optimizing method according to claim 5, wherein when it is determined that the structural information of the database system is necessary, the structural information is acquired from the database system, and in addition to the search instruction, A method for optimizing a search command, comprising using the acquired structure information. 7. The search instruction optimizing method according to claim 5 or 6, wherein the conversion knowledge includes a membership function and a rule, and fuzzy inference is used to determine whether the conversion is possible. The method for optimizing search instructions is characterized by: 8. The search instruction optimizing method according to claim 6 or 7, wherein a table to be searched by the search instruction should be searched using an index based on column selectivity and the number of rows. A search instruction optimizing method comprising: determining whether or not to search by using an index, and converting the search instruction so that the index can be searched. 9. A storage medium storing a program for realizing the apparatus or method according to any one of claims 1 to 8 by using a computer.