JPH07262196A - Data base retrieval device using knowledge base - Google Patents

Abstract

PURPOSE:To easily improve a hit ratio even by a person other than an expert by automatically checking the combination of a retrieval condition by which the number of hits can be reduced by using a knowledge base in advance and correcting the retrieval condition to a correct one in such a way in a data base retrieval device. CONSTITUTION:This data base retrieval device equipped with a data base 6, an input part 1 to input the retrieval condition and a retrieval part 4 which retrieves the data base according to an inputted retrieval condition is provided with the knowledge base 3 on which the combination of a condition to set an event as true and a condition to set it as false is described and a check part 2 which judges whether or not the inputted retrieval condition satisfies the description of the knowledge base.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a database searching device, and more particularly to a database searching device with an improved hit rate.

[0002]

2. Description of the Related Art In a database search device, when a plurality of search conditions are specified, if the search conditions are specified without knowing the relationship between the search conditions, the number of hits of the search becomes extremely low, and sometimes the hits are not made at all. Sometimes I don't. For example, in the material database of glass, if physical properties are used as search conditions, those with high density (> 3.5 g / c)
m ³ ) has a high refractive index (> 2.0) and a person who does not have knowledge has a search condition out of this relation, for example, density <
When the database is searched by setting search conditions such as 3.5 g / cm ³ and refractive index> 2.0, the number of hits is 0.
It can happen.

Therefore, in the conventional database search apparatus, a person having specialized knowledge makes good use of empirical or theoretical specialized knowledge to create appropriate search conditions, thereby improving the hit rate of the search. It was

[0004]

However, since the number of users who can make appropriate search conditions by making full use of their specialized knowledge is small, the database search device cannot be fully utilized. In addition, when a person who does not have sufficient specialized knowledge tries to perform a search, it is difficult to obtain a satisfactory search result.

According to the present invention, in a database search device, a combination of search conditions that reduces the number of hits is automatically checked in advance using a knowledge base, and the search conditions are corrected to correct ones. The purpose is to make it easier for non-specialists to improve the hit rate.

[0006]

In order to achieve the above-mentioned object, the present invention provides a database search apparatus comprising a database, an input section for inputting search conditions, and a search section for searching the database according to the input search conditions. In (1), a knowledge base in which a combination of conditions for making an event true or a condition for making an event false is described, and a check unit for determining whether or not an input search condition satisfies the description of the knowledge base is provided.

[0007]

When a plurality of search conditions are input from the input unit, the check unit refers to the knowledge base created based on empirical or theoretical knowledge, and the relationship between the search conditions is described in the knowledge base. Judge whether or not satisfy. When it is determined that there is a relationship between the search conditions that does not satisfy the description of the knowledge base, the check unit displays correct knowledge about the corresponding search conditions on the display device, for example.

By referring to the display, the searcher can know the error in the search condition entered by himself, even if he does not have any special expertise. Then, the database search can be started after correcting the incorrect search condition to the correct search condition. When the database search is started, the search unit can perform a normal search process on the database and obtain an appropriate number of search results.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS As an embodiment of the present invention, an example in which a database search apparatus using the knowledge base of the present invention is applied to a glass material database search apparatus will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a database search device using the knowledge base of this embodiment. In the figure, an input unit 1 for inputting a search condition and correcting the input search condition is provided, and the input search condition is output to a check unit 2. The check unit 2 checks the input search condition by referring to the knowledge base 3, outputs the check result to the display unit 4, and outputs the input search condition or the corrected search condition to the search unit 5. Output. The search unit 5 searches the material database 6 according to the search conditions and outputs the search result to the display unit 4.

An outline of the operation of the apparatus shown in FIG. 1 will be described with reference to the flowchart of FIG. Regarding the material database of glass, there are two search methods: a method of obtaining the composition of glass by using the physical properties of glass as a search condition and a method of obtaining the physical property of glass by using the composition of glass as the search condition. An example of creating a search condition for the physical properties of glass and searching the composition of glass that matches the search condition from the material database 6 will be described below.

The searcher specifies search conditions for physical properties such as density, refractive index, thermal expansion coefficient, Young's modulus, dielectric constant, and inputs these search conditions from the input section 1 to the check section 2 (step S1). . When the search conditions for a plurality of physical properties are input, the check unit 2 refers to the physical property mutual relation knowledge base as the knowledge base 3 to check the search conditions (step S2). The detailed contents of this search condition check will be described later. As shown below, the physical property mutual relationship knowledge base stores empirical or theoretical knowledge about the physical properties of glass for each combination of two physical properties.

1. High density (> 3.5g / c)
m ³ ) also has a high refractive index (> 2.0). 2. High density (> 3.5 g / cm ³ ) has a high coefficient of thermal expansion (> 100 × 10 ⁻⁷ / K). 3. Higher density (> 3.5 g / cm ³ ) has lower Young's modulus (<3000 kg / mm ² ).

4. High density (> 3.5g / c)
m ³ ) also has a high dielectric constant (> 20). 5. Those with a high refractive index (> 2.0) also have a high dielectric constant (>
20). The check unit 2 creates a combination of two physical properties for the input search conditions, sequentially searches the physical property mutual relation knowledge base according to the combination, and compares the condition value on the knowledge base with the condition value of the search condition. .

As a result of checking the search condition, the check unit 2 displays the relevant knowledge or message on the display device 4 when the condition value of the search condition does not satisfy the physical property value condition on the knowledge base ( Step S3). From the message displayed on the display device 4, the searcher knows that there is a contradiction between the condition values of the input search conditions. When trying to correct this error, the process returns from step S4 to step S1 and the search condition is corrected based on the message. Therefore, even if the searcher does not have sufficient knowledge, it is possible to set search conditions consistent with each other. In step S4, a special system data search can be performed by proceeding to the search without modifying the search conditions.

When the search condition in which the contradiction is corrected is input in step S1, steps S2 and S3 described above are repeated. As a result of the correction of the search condition, if there is no contradiction in the condition value, a message about the contradiction of the search condition is not displayed on the display device 4, so the searcher proceeds from the input unit 1 to the next search process in step S4. Input the command. If there is no contradiction in the search conditions, it is possible to display on the display device 4 that the search conditions have been correctly input.

When a command to proceed to the search process is input, the process proceeds from step S4 to step S5, and the material database 6
Is searched (step S5). As this search processing, various search methods can be used. The detailed contents of one example will be described later. And as a result of the search,
The obtained data is displayed on the display device 4 and further printed if necessary, and the search process ends.

In the database search device using the knowledge base described above, since the contradiction between the search conditions is eliminated, a sufficient number of hits can be obtained. Next, detailed contents of the search condition check process and the display process performed in steps S2 and S3 will be described with reference to the flowcharts of FIGS. When the search condition regarding the physical property is input, the check unit 2 counts the number n of input search conditions, and determines whether the number is 2 or more. When this number is 1, that is, when one search condition is not input, it is not necessary to check the contradiction between the search conditions, so the check processing is exited (No in step S11), and the process proceeds to step S4 in FIG.

When two or more search conditions are input, the process proceeds to step S12, and the unit of each condition value of the search condition is aligned with the unit on the knowledge base 3. Detailed description of this process is omitted. Next, as described above, in this embodiment, the physical property interrelationship knowledge base is composed of a combination of two physical properties. Therefore, in the following step S13 and subsequent steps, two combinations of physical properties are created from each search condition, and each combination is sequentially compared with the knowledge base.

Here, FIG. 5 shows an example of the physical property mutual relation knowledge base. The physical property interrelationship knowledge base is provided with a first physical property column, a second physical property column, and a message column. In the columns of physical property 1 and physical property 2, an ID that is a number that specifies the physical property, a condition value of the physical property, and an inequality sign that represents the magnitude relationship between the physical property and the condition value are provided. For example, in the first row of the knowledge base, dense ones (> 3.5g)
/ Cm ³ ) has a low Young's modulus (<3000 kg / m
Data representing the knowledge of m ² ) is recorded. That is, as the physical property 1, "10" representing the ID of the density, "3.5" of the condition value of the density, and ">" of the inequality sign indicating that the density is larger than 3.5 kg / mm ³ are recorded, Also,
As the physical properties 2, “40” representing the Young's modulus ID, “3000” as the Young's modulus condition value, and the Young's modulus of 3000 k
An inequality sign "<" is recorded indicating less than g / mm ² . In the message column, in the relationship between physical property 1 and physical property 2, a message "high density is displayed" is a message displayed on the display device 4 when each condition value in the input search condition does not satisfy this condition. The message is "Young's modulus is low".

In step S13, 1 is set as the initial value in the first register i and 1 is set as the initial value in the second register j. Next, in step S14, the type of physical property in the order indicated by the contents of the register i (first is 1) and the type of physical properties indicated by the sum of the contents of the register i and the contents of register j (the first is 2). Combinations are created.
Then, as the i = 1st physical property, for example, ID10
Is specified, and the Young's modulus of ID40, for example, is specified as j + i = the second physical property. Then, it is searched whether or not the combination (density and Young's modulus) is recorded in the knowledge base 3.

When the combination of relevant physical properties exists in the knowledge base (Yes in step S15), the condition value of the relevant physical property of the search condition is compared with the condition value of the relevant physical property of the knowledge base (step S16). As a result, when the search conditions do not satisfy any of the conditions (Yes in step S17)
s), the relevant message of the knowledge base 3 is displayed on the display device 4 as inconsistent information (step S18).

Next, in step S19, 1 is added to the content of the register j. If there is no combination of physical properties corresponding to the knowledge base in step S15, and step S15
If all of the combinations of the physical properties satisfy the condition values in 17, the intermediate step is skipped and the process goes to step S19. Then, in step S20, i
It is determined whether or not + j> n. If this condition is not satisfied, the process returns to step S14.

In step S14, this time, i = 1, j
+ I = 3, and, for example, a combination of density and refractive index is designated. In the same manner, the processes up to step S9 are performed. Hereinafter, if the process returns to step S14 again, i
The first physical property in the search condition is fixed as physical property 1 and the other physical property is changed as physical property 2, such as a combination of = 1 and j + i = 4. Then, when the search for the combinations of all other physical properties of the first physical property is completed (Yes in step S20), step S2.
Go to 1.

In step S21, 1 is set in the second register j and 1 is added to the content of the first register i. Then, in step S22, it is checked whether or not the content of the second register j has reached the number n of search conditions. If not, the process returns to step S14. Step S1
In step 4, i = 2, i + j = 3 this time, and the steps after step S15 are repeated, and the second physical property is sequentially searched for a combination with the other physical properties. When it is determined in step S22 that the search has been completed for all combinations of the physical properties of the search conditions (i + j> n), the knowledge base 3 exits the search process and proceeds to step S4 in FIG. During this time, if a condition that conflicts with the knowledge recorded in the knowledge base 3 is set between the physical properties of the input search conditions, the conflict information is displayed on the display device 4 in step S18.

Next, details of an example of the contents of the search process of step S5 of FIG. 2 will be described with reference to the flowcharts of FIGS. First, the structure of the material database will be described. The material database is a physical property ID database,
It consists of three databases, a composition ID database and a glass database. Each of the three databases is composed of an index file and a data file. Physical property ID database, composition ID database,
The record structure of the data file of the glass database is shown in FIG.

In the physical property ID database (a), one or two or more files are composed of physical property IDs for specifying one physical property. One file contains 20 records. When the number of glass data including a certain physical property is 20 or less, one file is prepared, and when it exceeds 20, two or more files are prepared. For each record,
A glass ID representing the glass data number including the physical properties is recorded.

The composition ID database of (b) is a physical property ID.
Configured like a database. In the glass database of (c), the physical properties of one piece of glass data include 40 records. Forty records,
20 physical property tables and 20 composition tables are recorded. In each physical property table and composition table, as shown in (d), a physical property ID and a physical property value corresponding to the physical property, and a composition ID and a composition ratio corresponding to the composition are recorded.

Referring back to FIGS. 6 and 7, the operation will be described. In step S31, the physical property ID database corresponding to the physical property is specified by the first physical property in the search conditions. Next, one record (glass ID) is read from the identified physical property ID database (step S32).
If the content of this record is EOF (End of File) or a record other than the relevant physical properties (step S
33), the search process ends.

When the read record is a record having the corresponding physical property, 1 is set in the register i as an initial value (step S34). Next, the glass database is searched by the glass ID recorded in the i-th (first is the first) record of the physical property ID database (step S35), and the hit record is read from the glass database (step S36).

Then, in step S37, it is determined whether or not each physical property of the search condition and the physical property value of the physical property ID corresponding to the physical property in the glass database match. When the condition values are met for all physical properties (Ye
s), the record is added to the search result table (step S38). If even one condition value does not match (N
o), step S38 is skipped. In step S39, 1 is added to the content of the register i.

Then, from step S35 to step S3
The processes up to 9 are repeated, the records of the physical property ID database are sequentially read, and the same steps are repeated through No in step S40 until one file (20 records) is completed. When the processing for one file is completed, the process returns from Yes in step S40 to step S32, and one record is read from the physical property ID database of the next file. And the read record is E
If it is a record other than OF or the corresponding physical property, the process is terminated (Yes in step S33). On the other hand, if the read record is the record having the first physical property, the processes after step S34 are continuously executed.

During the above processing, glass data that meets all the search conditions are sequentially recorded in the search result table. After the search processing is completed, the glass data recorded in the search result table is displayed on the display device and further printed out by a printing device (not shown). In the above, the example which searches a composition for the material database of glass using physical properties as a search condition has been described. As for the glass material database, as described above, the physical properties can be searched by using the composition as a search condition. In this case, the knowledge base is prepared by storing empirical or theoretical knowledge about the composition of glass. Regarding the composition of glass, there are three-component system, four-component system, etc.
Regarding the combination of compositions, three composition combinations, four composition combinations, and the like are created according to the number of components.

Further, the present invention is not applicable only to the search of the material database of glass, and other
It is applicable to general database searches.
Further, the number of search conditions to be combined is not limited to the combination of two search conditions, and may be the combination of three or more search conditions.

[0034]

According to the present invention, in a database search device, a combination of search conditions that reduces the number of hits can be automatically found in advance by using a knowledge base. It is possible to make corrections to correct ones so that even non-specialists can easily improve the hit rate.

[Brief description of drawings]

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

FIG. 2 is a flowchart for explaining the operation of the apparatus shown in FIG.

FIG. 3 is a flowchart (part 1) for explaining the operation of the check unit in FIG.

FIG. 4 is a flowchart (part 2) for explaining the operation of the check unit in FIG. 1.

5 is a diagram showing the structure of the knowledge base shown in FIG. 1. FIG.

FIG. 6 is a flowchart (part 1) for explaining the operation of the search unit in FIG.

FIG. 7 is a flowchart (part 2) for explaining the operation of the search unit in FIG. 1.

FIG. 8 is a diagram for explaining the contents of the material database of FIG.

[Explanation of symbols]

 1 ... Input part 2 ... Check part 3 ... Knowledge base 4 ... Display part 5 ... Search part 6 ... Material database

Claims (1)

[Claims] 1. A database search apparatus comprising a database, an input section for inputting search conditions, and a search section for searching the database according to the input search conditions, wherein a condition for making an event true or a condition for making an event false is provided. A database search apparatus using a knowledge base, comprising: a knowledge base in which a combination is described; and a check unit for determining whether or not the input search condition satisfies the description of the knowledge base.