JPH0282354A - Search strategy creation method - Google Patents

Abstract

(57) [Abstract] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Summary] Regarding a search strategy creation method in which a system automatically creates a database search strategy, the system accepts input that is more similar to natural language, the system creates a search strategy, and the system automatically creates a search strategy for the user. In a database system, the database system includes an input means for inputting data to be searched, an additional input means for inputting search data to be added during the search, and a combination of the input means and the powerful U input. A thesaurus conversion means adds search data from the search data and obtains a synonymous thesaurus, a compound thesaurus, and a concept thesaurus from the search data, and a search means searches with the data obtained from the thesaurus conversion means.

[Industrial application field]

The present invention relates to a database system, and more particularly to a search strategy creation method in which the system automatically creates a database search strategy.

[Conventional technology]

With the development of computer systems, database systems have come to be used in various fields.

Basically, in a database system, when inputting various information, a search term is also registered, and the desired information can be obtained by searching for the search term.

When there is a large amount of the above-mentioned registered data in a database system, even if a search is performed using a search term, the search term may not be input correctly.Therefore, it is desired to automatically generate an appropriate search term.

In order to make such database systems easier to use, conventional database search systems use two specifications for user input methods.

One is a natural language input method, in which the user inputs natural language, and a parser within the system parses the syntax using a dictionary and replaces it with a search word (a word that can search the database). The other method is a keyword input method using commands, in which the system receives the command name and input word (word input by the user) input by the user and replaces them with a search word.

[Problem to be solved by the invention]

When searching in the database mentioned above, it is desirable that the input method be natural language. However, in the case of natural language input, syntax analysis must be performed each time a search term is created.

Furthermore, in order to perform this syntactic analysis, a parser must be provided, which increases the scale of the system itself and requires the creation of a huge dictionary.

On the other hand, the keyword search method using commands has a problem in that the user must think about the search strategy and selection when there are multiple candidates. That is, although there is a demand for a method for accurately searching for further data, the creation of a search strategy is dependent on the user, which poses a problem in that it places a heavy burden on the user.

An object of the present invention is to accept input that is more similar to natural language, so that the system can create a search strategy and reduce the burden on the user.

[Means to solve the problem]

In the database system, input means 1 inputs data to be searched. The data to be searched is, for example, natural language.

Additional input means 2 inputs search data to be added during the search.

The thesaurus conversion means 3 obtains a synonym thesaurus, a compound thesaurus, and a concept thesaurus from the search data added from the input means 1 and the additional input means 2.

The search means 4 searches the database using the synonym thesaurus, compound thesaurus, concept thesaurus and search words obtained from the thesaurus conversion means 3.

[For production]

When the data to be searched is input by the input means 1, the thesaurus conversion means 3 converts the synonym thesaurus, compound thesaurus,
Find a concept thesaurus.

The search means 4 searches the database from the synonym thesaurus, compound thesaurus, concept thesaurus added from the thesaurus conversion means 3 and the search data.

If the search in the search means 4 is not accurate, the thesaurus conversion means 3 converts the search data input from the input means 1 and the search data to be added from the additional input means 2 into a synonym thesaurus. , composite thesaurus, and conceptual thesaurus are obtained again and added to the search means 4.

Similarly, the search means 4 searches the database system.

By repeating this sequentially, it is possible to accurately search the database system for the purpose.

〔Example] Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 2 is a processing flowchart 1 of the present invention in a data storage system. When flowchart 1 in FIG. 2 is activated in the data storage system, processing S1 is performed for input word reception. In response to this process, the operator inputs a keyword related to the matter desired to be searched. Note that this input may be a list of keywords (input words). The keyword input in step S1 is converted into a synonymous thesaurus (S2).

Here, the thesaurus will be explained.

A thesaurus for information retrieval organizes all concepts used in a specific field in a hierarchical manner, assigns standard terms to each concept, and establishes synonymous relationships between other terms and these standard terms. This is a terminology system based on There are synonymous thesauruses, compound thesauruses, and concept thesauruses, and in the following process, these thesauruses are used as needed to create a strategy, that is, a procedure.

(1) Synonymous thesaurus Synonymous thesaurus data is a thesaurus in which input words and search words are associated with each other in terms of synonyms and inflected forms. This synonymous thesaurus allows input from users to be accepted over a wide range of areas.

Furthermore, even if an input word is input that is not in the synonym thesaurus, it can be accepted by the registration function.

(2) Concept Thesaurus A concept thesaurus is created using only search terms from the system's synonym thesaurus. and a) superior/subordinate relationship (ISA relationship), b) overall/
It is expressed in terms of the concepts of relationship between parts (HAS A-PARTOF relationship), and C) type relationship (data with similar content can be searched).

In other words, the concept thesaurus represents three relationships.

This is useful when a user searches for an input word but there is no corresponding data, when the user is not satisfied with the searched data, or when the user requests to see related data to the searched data. The next selection or procedure is obtained from the conceptual relationship (name of data retrieved from the database).

The priority procedure is to use the upper/overall conceptual data of the original search term as the next search term.

ii) Use the conceptual data of the lower order/part of the base search term as the next search term.

iii) The next search term is replaced with conceptual data similar to the base search term.

It is.

If the search word is not found in 1) to iii) above, the next search word is used as the base search word and the search word is determined according to the above-mentioned priority order.

(3) Composite Thesaurus Composite thesaurus data is created using only search terms from the system's synonymous thesaurus, and includes elements for optimization after conversion using the synonymous thesaurus or concept thesaurus.

In other words, it has data on the relationship (optimal route) between specific search terms and search terms.

These data are related to the following.

b) Relationships in which the search term is used in multiple meanings.

口) A relationship in which the search term requires multiple steps.

c) A relationship in which the search term is frequently shared with a specific search term.

2) A relationship in which multiple search terms become a specific search term.

e) Relationships that include characteristic search terms.

f) - Correspondence between search terms used for boat fishing and search terms specific to the target field.

The synonym thesaurus, compound thesaurus, and concept thesaurus as described above establish relationships between search terms.

This will be further explained with reference to the processing flowchart shown in FIG.

For the input word input in the input word reception process (Sl), the synonymous thesaurus relationship described above is determined by the synchronous thesaurus conversion process (S2). Then, it is determined whether all the input words have the same relationship, that is, whether they can be converted (S3).

=9 In the determination (S3), there was an input that could not be converted (
If N), the input request is displayed again, and the operator inputs a similar word (S4). and synonym input processing (
Determining whether the synonyms input in S4) can be registered (S4)
Do 5). If registration is not possible (N), the process returns to the initial state and the input word reception process (Sl) is performed again.

On the other hand, if registration is possible (Y), similar synonyms are registered (S6).

FIG. 4 is a diagram explaining the principle of conversion and registration. A synonymous thesaurus has a user area and a system area, and conversion is performed in the user area and then further converted in the system area. Words that could not be converted using the synonym thesaurus can be registered in the user area.
Users are also asked to enter synonyms for the edicts that could not be registered and converted, so that they can be re-matched with the search term. Then, the search term is registered in the user area of the synonym thesaurus. This allows synonymous thesaurus conversion.

The registration in the process (S6) described above is a process for registering in this user area.

If all of the words can be converted in the above-mentioned determination (S3), and after registration (S6), compound thesaurus conversion is performed on those input words and the data subjected to synonymous thesaurus conversion (S7).

Then, the database is searched (S8) using the composite thesaurus obtained by the composite thesaurus conversion S7.

Search using synonymous thesaurus and compound thesaurus (S
8) In this case, the input of Itsuno words may not be correct. This is a case where the operator inputs an input word for a narrow field, for example, and it is next determined whether or not there is corresponding data based on the result of the search (S8) (S9).

As described above, if the data is narrow, it is determined in the determination (S9) that the data is not corresponding data (N). At this time, concept thesaurus conversion (310) is performed.

The keywords and the like input in the input word reception process (Sl) are just a list, and their hierarchical relationship, whole-part relationship, etc. are unclear. Therefore, when it is determined that there is no corresponding data (N) in determining whether there is corresponding data (S9), the conceptual relationship is changed by concept thesaurus conversion (SIO). i.e. i
- Find a concept thesaurus based on the priority order of iii.

After the concept thesaurus conversion (SIO) is performed, the same process is repeated again starting from thesaurus conversion (S7). By repeating this process, it is determined in the determination process (S9) that the corresponding data exists in the database (Y).

When it is found (Y) in the determination process (S9),
Next, it is determined whether there is only one corresponding data (Sl
l). When the number of corresponding data is one (Y), for example, the corresponding data is displayed on the display device of the database system (S
12) Do.

After displaying, it is determined whether or not there is data necessary for the operator's operation (S13). In the corresponding data display process (S12), one retrieved data is displayed, but it is unclear whether that data is the desired data. Therefore, when the operator inputs that necessary data is present (Y), that is, when the displayed data is necessary data, it is then determined whether the input operator's operation is a reference to related data (S44). do. In other words, if one item is displayed and it is determined that the item is necessary data, it is determined whether or not further related data is required. Does not require references to related data (
N) When the process ends (END). If further reference to related data is required, the concept thesaurus conversion (310) is performed again, and the same process as described above is repeated.

On the other hand, if the necessary data does not exist (N) in the determination (S13), it is determined whether the corresponding data is complete (S15), and if it is complete, execute the concept thesaurus conversion process (SIO) again do. Then, when the corresponding data is not completed (N), the display process is executed again from the display process (S12). In this case, there is no plurality of corresponding data, and the process is repeated after determination (317), which will be described later.

On the other hand, if there is not one corresponding data (N) in the determination (311), the corresponding data name is displayed. For example, if there are few keywords and the conditions are not favorable, there may be a large amount of display data.

Therefore, in the process (S16), only the corresponding data name is displayed. After this data name is displayed, when an instruction is added by an operator's operation, it is determined whether or not an input word is to be added to the data name (S17).

For example, if the corresponding data name is almost the desired name, the operator instructs not to add input words. Based on this instruction, the determination (S17) results in not adding (N), and the process (S12) is executed. In other words, when the input word is not added (S17) (N), data is displayed one by one, so first, 1 in the process (S12)
Similarly, it is determined whether or not there is necessary data (S13). If there is no necessary data (N), it is determined whether the data corresponding to the data is complete (S13).
S15) L, if not finished, process again (S12)
Display the following data from: When all are displayed, the corresponding data ends (Y) and the concept thesaurus conversion is performed again.

If a large amount of data is retrieved in the above search,
If an operation (Y) for adding an input word is performed in the determination (317), an additional input word is performed. This occurs when an input word is specified that covers a too wide range of fields, and the range is narrowed by the next additional input word. When the operator inputs an additional word by inputting an additional word (318), synonymous thesaurus conversion of the input word is performed for the additional input (S19). Then, it is determined whether the data obtained by the synonymous thesaurus conversion can be converted (S20
)I do. If the conversion is not possible (N) in the determination as to whether conversion is possible (S20), a similar synonym is input by synonym input (S21), and it is determined whether the input word of the similar synonym can be registered (S22). Then, it is determined that the registration is not possible (N) by determining whether the registration is possible (S22).
Occasionally, input from the operator is requested again by input word input (31B).

On the other hand, it was determined (S22) that it was possible to convert (Y)
If the time and determination of whether registration is possible (S22) shows that registration is possible (Y), a registration process (S23) is performed. Then, a search is performed again using the current search results and additional input words (324). As a result, the search is performed again, and the same process is repeated from the determination (S9) again to determine the result. By repeating this process, the target database can be searched. In particular, even if a simple input keyword, that is, a list of input words is input, the input word is converted into a synonymous thesaurus, and the result is further converted into a compound thesaurus to perform a search. Furthermore, if there is no corresponding data in the search, a concept thesaurus conversion is performed, the concept is changed based on its hierarchical relationship, the relationship between whole parts, similar relationships, etc., and then a compound thesaurus conversion is performed again. Since a post-search is performed, concepts are created simultaneously during the database search, and the operator can input words without having to worry about the input words as in the past.

Furthermore, the concept thesaurus conversion allows similar relationships to be entered for upper and lower positions, or for general public officials, without worrying about operations, which reduces the burden on the operator.

FIG. 3 is a thesaurus conversion path diagram. It represents the conversion centered on three types of thesaurus data in the operation described above. It is written in the form of a synonymous thesaurus (input word → search word), and ■=1 or 1z many.

In addition, the input word has 1 in the system area of the synonym thesaurus.
Only one exists.

A compound thesaurus is written in the form (search word → search word), 1:1, 1:many, many:1. Many: Many times is fine.

If a search term passed from a synonym/concept thesaurus meets multiple conditions, it is converted by applying the strongest constraint.

The concept thesaurus has three types of search terms (upper/lower, overall/
(partial, similar) links. That is, when the input word inputted by the user is input from the synonym thesaurus TNX, a relationship of 1=1 or 1:many is determined by the synonym thesaurus SB. Then, based on this relationship, the composite thesaurus is 1:1, and further, 1:many, many:1, and many:many, that is, the relationship between search terms is established.

Then, the database is searched and DFed using the obtained composite thesaurus. In this search, if one piece of data is obtained or not obtained as shown in Figure 2, the concept thesaurus SG will perform links such as upper/lower, whole/part, similar, etc. depending on the result. In other words, the composite thesaurus is obtained again after making changes. Search is performed using the composite thesaurus.

The procedure described above is used to search for target data.

FIG. 5 is an explanatory diagram of synonymous thesaurus conversion.

When the user inputs "display", it first tries to see if thesaurus conversion can be performed using the data in the user area.

Next, perform thesaurus conversion using the data in the system area.

In this case, the data in the system area is converted into a thesaurus and marked as "display".

If the user inputs "represent" and the synonym thesaurus conversion cannot be performed, a synonym is input. Synonyms include “
When "Display" is entered, this edict can be converted into a thesaurus, so synonymous thesaurus data such as [Ozuzu → Display] is created and added to the user area.

FIG. 6 is an explanatory diagram of composite thesaurus conversion.

If the search term converted using the synonymous thesaurus or concept thesaurus is "Window Label Display J", use the data of [[Window Display → ACTIVATE or 5HOIJ:] and [Label Display → SET TITLE] for conversion. The two conditions have the same strength, so when converted, [Window ACTIVATE
, window 5HOW, window SET TIT
Search for either LE J.

If the search term converted using the synonym thesaurus or concept thesaurus is "menu scroll" [menu scroll → WITHMΔRGIN and A].

FIG. 7 is an explanatory diagram of concept thesaurus conversion.

If the search term is rltllTHLABEL, in order to create a strategy, first find the top or overall "satushi class" or "window". Next, find the lower level data. Finally, similar data is sought.

As described above, in the present invention, there is no need to input the hierarchical relationship of input words using keywords or other commands, and by simply inputting a list, the operator can intentionally think about the search strategy such as the search procedure. In other words, it searches the database for data that the user may need and displays it. In addition, when multiple data are obtained from a search, the data is automatically narrowed down by simply adding a keyword without having to re-enter the data, and even if a keyword is entered that is not in the system. Since keyword input is registered, keywords can be used in a wide range of areas as the search progresses.

The database can be searched by simply listing the input words without having to think about it, reducing the burden on the operator.

[Brief explanation of drawings]

Fig. 1 is a functional block diagram of the present invention, Fig. 2 is a processing flowchart of the present invention, Fig. 3 is a conversion path diagram, Fig. 4 is a diagram explaining the principles of conversion and registration, and Fig. 5 is an explanation of synonym thesaurus conversion. 6 is an explanatory diagram of compound thesaurus conversion, and FIG. 7 is an explanatory diagram of conceptual thesaurus conversion. 1... Input means, 2... Additional input means, 3... Thesaurus conversion means, 4... Search means. 〔Effect of the invention〕

Claims (1)

[Claims] In a database system, an input means (1) for inputting data to be searched, an additional input means (2) for inputting search data to be added during the search, and the input means (1). Search data is added from the additional input means (2), and a synonym thesaurus,
A search strategy creation method comprising: a thesaurus conversion means (3) for obtaining a composite thesaurus or a concept thesaurus; and a search means (4) for searching using data obtained from the thesaurus conversion means (3).