JPH0836582A - Character string normalization rule creating device, character string normalizing rule creating method, information search device, and information search method - Google Patents

Abstract

(57) [Summary] [Object] The present invention provides a normalization rule for a character string that can be searched if the difference between the input character string and the character string included in the stored information is within a predetermined range. An object of the present invention is to provide a character string normalization rule creating device, a character string normalization rule creating method, an information search device, and an information search method that are created in a specific manner. According to the present invention, a conversion rule generating means 110 for generating a conversion rule, a conversion rule correcting means 120 for applying a conversion rule to a conversion rule to modify it, and an equality relation for applying a conversion rule to an equality relation to correct it. It is composed of correction means 130, control means 140, equation relation storage means 150, and conversion rule storage means 160.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character string normalization rule creating device, a character string normalization rule creating method, an information search device and an information search method, and stores the same character string as stored character string information. When it is difficult to input and search, it is possible to flexibly search within the range of the equation relation of a preset character string and output the search result. The present invention relates to a column normalization rule creating method, an information search device, and an information search method.

[0002] More specifically, the present invention is a character string normalization rule creating device and a character string normalization rule creating device, which makes it possible to search for information such as a person list having a kana character string by inputting a kana character string and obtain an appropriate search result The present invention relates to a column normalization rule creating method, an information search device, and an information search method.

[0003]

2. Description of the Related Art Generally, it is necessary to accurately enter a character string in information retrieval using a character string, but Japanese information includes "Yamasaki", "Yamazaki" or "Hozumi". Since there are many notational fluctuations such as "Hozumi", the input character string may differ from the stored character string information, and a correct search may not be possible in some cases. Conventionally, as a search method for obtaining a desired search result in such a case, there is the following method.

In the conventional search method, a conversion rule for normalizing a character string is manually created based on a method of normalizing a character string in advance, and the search character string and the stored information are normalized based on the created conversion rule. Search by converting. The conventional search method will be described below with reference to the drawings. FIG. 12 is a diagram illustrating a conventional search method. FIG. 12A is an example of a character string conversion rule. Conventionally, in order to normalize a character string, a normalization method is used in which "zu" is converted into a standard notation "zu", and then a dull sound is converted into a clean sound. This figure is an example of a conversion rule created manually based on the above-mentioned normalization method, where "zu" is "zu" and "za" is "sa".
In addition, "zu" represents converting to "su". FIG.
2 (B) is an example of a database to be searched,
FIG. 12C shows an example of a database converted by applying the above example of the conversion rule to the example of the database shown in FIG. 12B. In addition, in FIG. 12D, the search character string “Hozumi” that is input to search the database for the item “Name” is “Hozumi” based on the above conversion rule example.
Represents the process of converting “Hozumi” into “Hozumi” and further converting “Hozumi” into “Hosumi”. As described above, the conventional search method manually creates a conversion rule for character string normalization, applies the created conversion rule to convert the database, and converts the search character string based on the conversion rule when searching. It is possible to obtain a search result by performing a conversion and searching a database that has been converted by applying a conversion rule in advance.

[0005]

However, the above-mentioned conventional method has the following two problems.

The first problem is that when a character string has fluctuations of "Hozumi", "Hozumi", and "Hotsumi", if the character string "Hozumi" is normalized, it becomes "Hozumi". It is converted and further converted to "Hosumi". Also, if "Hozumi" is normalized, it becomes "Hosumi". However, if "hotsumi" is normalized, it becomes "hotsumi", so there are two types of strings "Hosumi" and "Hotsumi" in the normalized string. There is a problem in that the results of normalization of “,” “Hozumi”, and “Hosumi” are not uniquely determined, so that a search omission occurs when searching by inputting “Hosomi”.

The second problem is that the conversion rule for normalizing the character string has to be manually created, so that when adding a new rule, there should be no omission in the search for the newly created conversion rule. In addition, it is necessary to create a conversion rule while verifying the effect of adding a new rule, which requires a lot of labor, time, and cost to create a conversion rule, and there is a risk of making an error in creating a conversion rule. There is a point.

The present invention has been made in view of the above points, and normalizes a searchable character string if the difference between the input character string and the character string included in the stored information is within a predetermined range. An object of the present invention is to provide a character string normalization rule creating device, a character string normalization rule creating method, an information search device, and an information search method that automatically create rules.

Further, according to the present invention, a character string normalization rule creating apparatus capable of creating a conversion rule for normalizing a character string without omission based on an equality relation indicating that two character strings are equal. , A character string normalization rule creating method, an information search device, and an information search method.

Further, the present invention is a character string normalization rule creating device for creating a character string conversion rule capable of normalizing an input character string in one conversion process by eliminating redundant conversion rules. An object is to provide a character string normalization rule creating method, an information search device, and an information search method.

Further, a character string normalization rule creating device, a character string normalization rule creating method, an information retrieving device, and an information retrieving method capable of creating a conversion rule capable of uniquely converting a plurality of character strings having fluctuations. The purpose is to provide.

Further, the conversion rule for normalizing the character string is used to normalize and collate the input character string and the character string included in the information previously accumulated as a search target, whereby An object of the present invention is to provide a character string normalization rule creating device, a character string normalization rule creating method, an information searching device, and an information searching method that can be searched without omission within the range of the set equality relationship.

[0013]

FIG. 1 is a principle block diagram of a character string normalization rule creating device of the present invention.

The character string normalization rule creating device according to the present invention enables the search if the difference between the input character string and the character string included in the information to be searched is within a predetermined range. It has a character string conversion rule creating means 100 for creating a rule.

The character string conversion rule creating means 100 is also provided.
Is a plurality of equality relations 1 representing that two character strings are equal.
When 0 is given, it has a function of automatically creating a conversion rule for normalizing a character string based on the given equality relation 10.

Further, the character string conversion rule creating means 100 described above.
Is an equation relation storage means 1 for storing a plurality of equation relations 10.
50 and the equation relation 10 from the equation relation accumulating means 150 are compared, two character strings of the extracted relation 10 are compared, and a pre-conversion character string and a post-conversion character string are determined based on the comparison result. Conversion rule generation means 110 for generating conversion rules
And conversion rule accumulating means 160 for accumulating conversion rules, and the conversion rules accumulated in the conversion rule accumulating means 160 are applied to the converted character strings of the conversion rules accumulated in the conversion rule accumulating means 160 for conversion. The conversion rule correction means 120 for correcting the rule and the character string of the equation relation 10 stored in the equation relation storage means 150 are applied to the conversion rule stored in the conversion rule storage means 160 to apply the equation relation. When the equality relation modifying means 130 for correcting 10 and the equality relation accumulating means 150 have accumulated the equality relation 10, the equality relation accumulating means 1
It has a control means 140 that sequentially activates and repeats the conversion rule generation means 110, the conversion rule correction means 120, and the equality relationship correction means 130 until the equation relation 10 stored in 50 disappears.

Further, the conversion rule generation means 110 has a function of comparing the magnitudes of the two character strings of the extracted equality relation 10 in dictionary order.

Further, the conversion rule generating means 110 deletes the extracted equality relation 10 from the equality relation accumulating means 150 when the comparison results are equal, and the comparison results when the comparison results are not equal. Has a function of setting a character string having a large value as a pre-conversion character string and a character string having a small comparison result as a post-conversion character string.

FIG. 2 is an explanatory view of the principle of the character string normalization rule creating method of the present invention.

The method for creating a character string normalization rule of the present invention is 2
Equivalence relations are extracted from a set of equality relations indicating that two character strings are equal, two strings of the extracted equality relations are compared, and the equality relations are compared based on the result of the comparison. The conversion rule is deleted from the set, or the conversion rule is generated by determining the pre-conversion character string and the post-conversion character string, and added to the conversion rule set (step 1), and the conversion rule set added in step 1 above. The conversion rules are corrected by applying all the conversion rules included in the set of conversion rules to the converted character strings of all the conversion rules included in (step 2), and the equality relations included in the set of equality relations When the conversion rule included in the modified conversion rule set is applied to the two character strings of (1) to correct the equality relation (step 3), and the equality relation exists in the set of equality relations. To step 1 above, if there is no equation relation It ends the creation of the conversion rule (Step 4).

FIG. 3 is a block diagram showing the principle of the information retrieval apparatus of the present invention.

The information retrieving apparatus 300 of the present invention has an information accumulating means 310 for accumulating information having a character string. When the search character string 20 is input, the information accumulating means 310 is searched based on the search character string 20. The information retrieval apparatus 300 that outputs the retrieval result 30 as a result has the following means.

The information retrieving means 20 for converting the retrieval character string 20 and the character string accumulated in the information accumulating means 310 based on a predetermined conversion rule and performing a search based on the result of the conversion.
Has zero.

Further, the information retrieval means 200, the conversion rule accumulating means 160 accumulating a plurality of conversion rules for normalizing the character string, the retrieval character string 20 and the information accumulating means 31.
The normalization means 210 that applies the conversion rules stored in the conversion rule storage means 160 to the character string stored in 0 to normalize it, the search character string 20 normalized by the normalization means 210, and the information storage The character string stored in the means 310 is collated, and if the collation results match, the character string before normalization of the character string stored in the matched information storage means 310 is normalized. And a matching and matching unit 220 that outputs a column as a search result.

FIG. 4 is a diagram explaining the principle of the information retrieval method of the present invention.

In the information search method of the present invention, when a search character string is input (step 10), a conversion rule for normalizing the character string is applied to the search character string for normalization (step 1).
1) Apply a conversion rule for normalizing the character string to the character string included in the stored information to be searched for normalization (step 12), and the search character string normalized in step 11 and the step 12 The character string included in the normalized accumulated information is collated (step 13), and if the result of the matching is the same, the character string before being normalized of the character string included in the accumulated information that is matched is normalized. The column is output as a search result (step 14).

[0027]

The character string normalization rule creating device and the character string normalization rule creating method of the present invention are prepared by preparing in advance an equation relation representing that two character strings are equal as information on the fluctuation of the character strings. It is possible to automatically create a conversion rule for searchable character string normalization within the range of the equality relation.

Further, the character string normalization rule creating device and the character string normalization rule creating method of the present invention not only simply convert the given equality relations into conversion rules, but also a conversion obtained during the conversion rule creation. The rules are sequentially applied to the conversion rules to correct the conversion rules, and the conversion rules are applied to the equality relationships to correct the equality relationships to eliminate the redundancy of the conversion rules. This makes it possible to create a conversion rule that can normalize a character string in a single conversion process, and to prevent omission of creation of a conversion rule.

Further, in addition to the above, the character string normalization rule creating device and the character string normalization rule creating method of the present invention judge the size of a given equality-related character string in dictionary order. A conversion rule that allows unique conversion of multiple character strings with fluctuations and that can be searched without omission if the input character string is within the range of the given equality relation It is possible to create.

Further, the information search device and the information search method of the present invention use a character string conversion rule creating device and a conversion rule for character string normalization created by the character string conversion rule creating method to search for a search character string and By normalizing the stored information, even if there is a difference between the search character string and the character string included in the stored information, it is possible to search and output the result within a predetermined range.

[0031]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 5 is a block diagram of a character string normalization rule creating device according to an embodiment of the present invention. The character string normalization rule creation device 100 shown in the figure includes a conversion rule generation unit 110, a conversion rule correction unit 120, an equation relation correction unit 130, and a control unit 14.
0, an equation relation storage unit 150, and a conversion rule storage unit 160.

The equation relation accumulating unit 150 is a database for accumulating the equation relation 10 indicating that two character strings are equal, and is information that needs to be accumulated in advance before the conversion rule is created. FIG. 6 is a diagram showing a storage example of the equation relation storage unit according to the embodiment of the present invention. The example in the figure shows that “Hozumi” and “Hozumi” are equal ”,“ Hozumi ”and“ Hosumi ”are equal, and“ Hozumi ”and“ Hosumi ”are equal”. .

The conversion rule storage unit 160 is a database that stores conversion rules for normalizing character strings. Incomplete conversion rules are accumulated during the conversion rule creation process, but at the end of all conversion rule creation processes,
The conversion rules for normalizing the completed string are accumulated. FIG. 7 is a diagram showing a storage example of the conversion rule storage unit according to the embodiment of the present invention. In the example shown in the figure, "convert" Hozumi "to" Hosumi """convert" Hosumi "to" Hosumi """Convert" Hozumi "to" Hosumi "" It represents.

The conversion rule generation unit 110 extracts one equation relation 10 stored in the equation relation storage unit 150,
The magnitude relations of the two character strings of the extracted equality relation 10 are compared in lexicographical order, and if the comparison results are equal, the extracted equality relation 10 is deleted from the equality relation accumulating unit 150, and the comparison result is If they are not equal, the character string with a larger comparison result is the character string before conversion, and the character string with a smaller comparison result is the character string after conversion. "" Before character string "is converted to" after character string ". A conversion rule representing “Yes” is created and stored in the conversion rule storage unit 160.

The conversion rule correction unit 120 applies all the conversion rules stored in the conversion rule storage unit 160 to the converted character strings of all the conversion rules stored in the conversion rule storage unit 160 to apply the characters. Column conversion is performed and conversion rule storage unit 1
Make 60 corrections.

The equation relation correcting unit 130 adds the character strings of all the equation relations 10 accumulated in the equation relation accumulating unit 150 to
All the conversion rules stored in the conversion rule storage unit 160 are applied to convert the character string, and the equation relation storage unit 150 is corrected.

The control means 140 is means for controlling the operation of the character string normalization rule creating device 100 of the present invention, and in the state where the equation relation 10 is stored in the equation relation storage section 150 in advance, the conversion is performed. The rule generation unit 110 is activated, and after the operation of the conversion rule generation unit 110 ends, the conversion rule correction unit 120
And confirms whether or not the equality relation 10 is stored in the equality relation storage unit 150 after the operation of the conversion rule correction unit 120 is finished. If the equality relation 10 is registered, the equation is continued. Although the relation correction unit 130 is activated, when the equation relation is not registered in the equation relation storage unit 150, the conversion rule creation is completed and the operation ends. Further, even after the operation of the equation relation correction unit 130 is completed, the equation relation storage unit 1
Check if the equation relation 10 is stored in 50,
If the equation relation 10 is registered, the conversion rule generation unit 110 is activated. If the equation relation is not registered in the equation relation storage unit 150, it is assumed that the conversion rule has been created. To finish.

A series of operations of each of the above parts will be described below with reference to the drawings. FIG. 8 is a flowchart showing the operation of the character string normalization rule creating device according to the exemplary embodiment of the present invention.

(Step 50) In response to the instruction from the control unit 140, the conversion rule generation unit 110 extracts one equation relation 10 stored in the equation relation storage unit 150.

(Step 51) Conversion Rule Generation Unit 110
Checks whether the two character strings of the equation relation 10 extracted in the above step 50 are equal, and if they are not equal, moves to step 53.

(Step 52) Conversion rule generator 110
When the two character strings of the equation relation 10 are equal in step 51, the equation relation extracted in step 50 is deleted from the equation relation storage unit 150, and the process proceeds to step 50.

(Step 53) Conversion rule generator 110
When the two character strings of the equality relation 10 in step 51 are not equal to each other, the two character strings are compared in the dictionary order, and a large character string, that is, a character example that follows in the dictionary order is the character string before conversion, A conversion rule for converting a small character string into a converted character string is generated and stored in the conversion rule storage unit 160.

(Step 54) The conversion rule correction unit 120 takes out one conversion rule stored in the conversion rule storage unit 160 according to an instruction from the control unit 140.

(Step 55) Conversion Rule Modifying Unit 120
Applies all other conversion rules remaining in the conversion rule accumulating unit 160 to the converted character string of the conversion rule extracted in the above step 54, and accumulates the applied conversion rules in the conversion rule accumulating unit 160. And modify the conversion rules.

(Step 56) Conversion Rule Modifying Unit 120
Checks whether all the conversion rules stored in the conversion rule storage unit 160 have been subjected to the above step 55, and the above step 5 is executed for all the conversion rules.
When the process of 5 is not performed, the process proceeds to step 54.

(Step 57) The control unit 140 checks whether or not the equality relation 10 remains in the equality relation storage unit 150, and if it does not remain, the conversion rule creating process is completed.

(Step 58) In accordance with an instruction from the control unit 140, the equation relation correction unit 130 retrieves one equation relation 10 stored in the equation relation storage unit 150.

(Step 59) Equation relation correcting unit 130
Applies all the conversion rules stored in the conversion rule storage unit 160 to the two character strings of the equation relation 10 extracted in step 58, and stores the equation relation 10 resulting from the application of the equation relation 10. The equation relation 10 is corrected by accumulating in the unit 150.

(Step 60) Equation relation correcting unit 130
Checks whether the two character strings of the equation relation 10 applied as a result of the above step 59 are equal, and when they are equal, moves to step 61, and when they are not equal, moves to step 62.

(Step 61) Equality relation correcting unit 130
When the two character strings of the equality relation 10 are equal in step 60, the equality relation 10 is stored in the equality relation storage unit 1.
Delete from 50.

(Step 62) Equation relation correcting section 130
Checks whether all the equation relations 10 stored in the equation relation storage unit 150 have been subjected to the processing in step 59 above, and has not performed the processing in step 59 above for all equation relations 10. If step 5
Move to 8.

(Step 63) The control unit 140 checks whether or not the equality relation 10 is accumulated in the equality relation accumulating unit 15, and if so, shifts to Step 50 and is accumulated. If there is no conversion rule, all conversion rules are created.

A specific example of creating a character string normalization rule using the character string normalization rule creation device of one embodiment of the present invention will be described below. The character string normalization rule creating device in the following description is the same as the character string normalizing rule creating device described above. Also, the equation relation accumulating unit 150 in the processing process
9 shows the state of the conversion rule storage unit 160.

Prior to the start of processing, the equation relation storage unit 150
The three equality relations 10 shown in FIG. 6 are stored in FIG.
(State 1)]. The following description will be made based on this state.

According to the instruction from the control unit 140, the conversion rule generating unit 110 causes the equation relation accumulating unit 150 to store the equation relation 10, ““ Hozumi ”=“ Hozumi ”” ““ Hozumi ”=“ From "Hotsumi""" Hozumi "=" Hosumi "", take out one "" Hozumi "=" Hozumi "" (step 50) and check whether the two character strings are equal. But they are not equal (Step 5
(1, NO) When comparing two character strings in dictionary order, there is a relation of ““ Hozumi ”>“ Hozumi ””, so that the character string before conversion is “Hozumi” and the character string after conversion is “Hozumi”. A conversion rule of ““ Hozumi ”=>“ Hozumi ”” is created and stored in the conversion rule storage unit 160 (step 53) [FIG. 9 (state 2)].

According to the instruction from the control unit 140, the conversion rule correction unit 120 causes the conversion rule stored in the conversion rule storage unit 160, “conversion rule”, to “conversion rule,” “Zumi” => “Hozumi” ”is extracted (step 54), and the converted character string of the extracted conversion rule,“ Hozumi ”, is replaced with all other conversion rules remaining in the conversion rule accumulating unit 160. Although there is no other conversion rule left to be applied, the conversion rule extracted without any conversion is returned to the conversion rule accumulating unit 160 (step 55) [FIG. 9 (state 2)].

Since the conversion rule correction section 120 has completed the processing of step 55 for all conversion rules stored in the conversion rule storage section 160, the conversion rule correction processing ends (step 56, YES).

When the control unit 140 checks whether or not the equality relation 10 remains in the equality relation storage unit 150, ““ Hozumi ”=“ Hosumi ”” ““ Hozumi ”=“ Hosumi ” Since there are still two equality relations 10 of "," the equality relation correction unit 130 is instructed to start the processing (step 57,
Yes).

The equation relation correction unit 130 stores the equation relations 10 stored in the equation relation storage unit 150, ““ Hozumi ”=“ Hosumi ”” and ““ Hozumi ”=“ Hosumi ””. From the inside, take out ““ Hozumi ”=“ Hotsumi ”” (step 58) and take out the equation relation 10
Applying all the conversion rules stored in the conversion rule storage unit 160 to the two character strings, ““ Hozumi ”=>“ Hozumi ”” and replacing them with the character string of the application result, = “Hotsumi” ”, the equation relation 10 is corrected and accumulated in the equation relation accumulating unit 150 (step 59) [FIG. 9 (state 3)].

Since the two character strings of the equation relation 10 applied as a result of the above step 59 are not equal (step 6
0, NO), if it is checked whether or not the processing in step 59 has been performed for all the equation relations 10 stored in the equation relation storage unit 150, the equation relations 10 that have not been processed yet, ““ Since "Hozumi" = "Hosumi""exists (step 62, NO), the unprocessed equation relation 10,""Hozumi" = "Hosumi""is taken out (Step 58).

Equation relation 1 extracted in step 58 above
All the conversion rules stored in the conversion rule storage unit 160, ““ Hozumi ”=>“ Hozumi ””, are applied to the two character strings of 0, but they are not applicable conversion rules, and thus the extracted ones are extracted. Equation relation accumulator 1 without converting equation relation 10
It returns to 50 (step 59) [FIG. 9 (state 3)].

Since the two character strings of the equation relation 10 to which the conversion rule is applied in step 59 are not equal (step 60, NO), all equation relations stored in the equation relation storage unit 150 are Step 5 for 10
When it is checked whether or not the processing of step 9 has been performed, the processing of step 59 has been completed for all of the equation relations 10, so the equation relation correction processing ends (step 62, YE).
S).

When the control unit 140 checks whether or not the equation relation 10 is accumulated in the equation relation accumulating unit 15, "" Hozumi "=" Hotsumi """Hozumi" = "Hosumi" There are two equation relations 10 of "" remaining (step 6
(3, YES), the equality relation 10 of ““ Hozumi ”=“ Hotsumi ”” is taken out (step 50), and if it is checked whether the two character strings are equal, the two character strings are not equal ( (Step 51, NO) Compares the two character strings in dictionary order, and based on the relationship of "" hotsumi ">" hosumi "", the character string before conversion is "hotsumi" and the character string after conversion is "hosumi". Then, a conversion rule of ““ hotsumi ”=>“ hosomi ”” is created and stored in the conversion rule storage unit 160 (step 53) [FIG. 9 (state 4)].

In accordance with an instruction from the control unit 140, the conversion rule correction unit 120 causes the conversion rules stored in the conversion rule storage unit 160 to read "" Hozumi "=>" Hozumi """" Hotsumi "=>" Hosumi ""Hozumi"=>"Hozumi""is extracted from" Hozumi "(step 54), and the converted character string of the extracted conversion rule," Hozumi ", remains in the conversion rule accumulating unit 160. Tries to apply all other conversion rules that are "Hatsumi"=>"Hozumi", but do not convert any of the conversion rules that have been extracted because they are not applicable conversion rules. Return to the conversion rule storage unit 160 (step 55) [FIG.
(State 4)].

When the conversion rule correction unit 120 checks whether or not all the conversion rules stored in the conversion rule storage unit 160 have been subjected to the processing of step 55, the conversion rule that has not been processed, "" Since "tsumi"=>"hozumi""remains (step 56, NO), the remaining conversion rule""hotomi"=>"hozumi""is extracted (step 54), and the extracted conversion is performed. All other conversion rules remaining in the conversion rule accumulating unit 160, that is, "Hozomi", is applied to the converted character string "Hozomi" of the rule. Since it is not an applicable conversion rule, the extracted conversion rule is returned to the conversion rule accumulating unit 160 without any conversion (step 55) [FIG. 9 (state 4)].

When the conversion rule correction section 120 checks whether or not the processing in step 55 has been performed for all the conversion rules stored in the conversion rule storage section 160,
Since the processing of the above step 55 has been performed for all the conversion rules, the conversion rule correction processing ends (step 5).
6, YES).

When the control unit 140 checks whether or not the equality relation 10 remains in the equality relation storage unit 150, the equality relation 10 of ““ Hozumi ”=“ Hosumi ”” remains. Equation relation correction unit 130
To start processing (step 57, YE
S).

In accordance with an instruction from the control unit 140, the equation relation correcting unit 130 takes out the equation relation 10, ““ Hozumi ”=“ Hosumi ”” stored in the equation relation storing unit 150 (step 58). ), The extracted equation relation 1
Applying all the conversion rules stored in the conversion rule storage unit 160 to two character strings of 0, ““ Hozumi ”=>“ Hozumi ”” ““ Hotsumi ”=>“ Hozumi ”” I try to replace it with the string of the applied result,
Since there is no applicable conversion rule, the extracted equality relation is returned to the equality relation storage unit 150 without any conversion (step 5).
9) [FIG. 9 (state 4)].

Since the two character strings of the equation relation 10 applied as a result of the above step 59 are not equal (step 6
0, NO), if it is checked whether or not the process of step 59 has been performed for all equation relations 10 stored in the equation relation storage unit 150, the process of step 59 is performed for all equation relations 10. Since it has been performed, the equation relation correcting process is ended (step 62, YES).

When the control unit 140 checks whether or not the equality relation 10 is stored in the equality relation storage unit 15, the equality relation 10 of ““ Hozumi ”=“ Hosumi ”” remains. So (step 63, YE
S), the equality relation 10 of ““ Hozumi ”=“ Hosumi ”” is taken out (step 50) and it is checked whether the two character strings are equal, but they are not equal (step 51, NO), Compare the two strings in lexicographical order, and use the relationship "" Hozumi ">" Hosumi "" to define the pre-conversion character string as "Hozumi" and the post-conversion character string as "Hosumi". A conversion rule of "sumi"=>"hosumi""is created and stored in the conversion rule storage unit 160 (step 53) [Fig. 9 (state 5)].

In accordance with an instruction from the control unit 140, the conversion rule correction unit 120 causes the conversion rules stored in the conversion rule storage unit 160 to read "" Hozumi "=>" Hozumi """" Hotsumi "=>" Hosumi ""Humi""" Hozumi "=>" Hozumi "", the conversion rule of "" Hozumi "=>" Hozumi "" is extracted (step 54), and the converted character string of the extracted conversion rule is extracted. , "Hozumi", all other conversion rules remaining in the conversion rule accumulating unit 160, conversion rules of "" Hozumi "=>" Hozumi """Hozumi"=>"Hosumi"" Is applied, and the conversion rule resulting from conversion into ““ Hozumi ”=>“ Hosumi ”” is accumulated in the conversion rule accumulating unit 160 (step 55) [FIG. 9 (state 6)].

When the conversion rule correction section 120 checks whether or not the processing of step 55 has been performed for all conversion rules stored in the conversion rule storage section 160,
The conversion rules that have not yet been processed, ““ Hosumi ”=>“ Hozumi ”” and ““ Hozumi ”=>“ Hosumi ”” remain (step 56, NO). , The conversion rule of ““ Hotsumi ”=>“ Hosomi ”” is extracted (step 54), and the converted character string of the extracted conversion rule, “Hosomi”, remains in the conversion rule accumulating unit 160. Apply all the conversion rules of ““ Hozumi ”=>“ Hosumi ”” and “Hozumi” => “Hosumi” ”to convert“ Hosumi ”=>“ Hosumi ” The conversion rule obtained as a result of the conversion is stored in the conversion rule storage unit 160 (step 55) [FIG. 9 (state 7)].

When the conversion rule correction section 120 checks whether or not the processing of step 55 has been performed for all the conversion rules stored in the conversion rule storage section 160,
The conversion rule that has not yet been processed, "" Hozumi "=>" Hosumi "" remains (step 56, NO), so the remaining conversion rule, "" Hozumi "=>" Hosumi "" Is extracted (step 54), and the converted character string of the extracted conversion rule, “Hosumi”, is replaced with all the other conversion rules remaining in the conversion rule accumulating unit 160, ““ Hozumi ”=>“ Hosumi ”. ”” The conversion rule ““ hotsumi ”=>“ hosumi ”” is tried to be applied, but since there is no conversion rule that can be converted, the extracted conversion rule is returned to the conversion rule storage unit 160 without any conversion. (Step 55) [FIG. 9 (State 7)].

When the conversion rule correction section 120 checks whether or not the processing of step 55 has been performed for all the conversion rules stored in the conversion rule storage section 160,
Since the processing of the above step 55 has been performed for all the conversion rules, the conversion rule correction processing ends (step 5).
6, YES).

When the control unit 140 checks whether or not the equality relation 10 remains in the equality relation storage unit 150, there is no equality relation 10 in the equality relation storage unit 150. Then, all the processes for creating the are completed (step 57, NO).

From the above, the three equation relations 10 shown in FIG. 6, ““ Hozumi ”=“ Hozumi ””, “Hozumi” = “Hotsumi”, “Hozumi” = “Hosumi” From ", you can automatically create a conversion rule,""Hozumi"=>"Hosumi""""Hosumi"=>"Hosumi""" Hosumi "=>" Hosumi "" .

Although the embodiment of the character string normalization rule creating device has been described with reference to the example of creating the conversion rule based on three equality relationships, the number of equality relationships is significantly increased in actual use. Therefore, the manual conversion requires a great deal of labor, time, and cost, and at the same time, the risk of human error in the conversion rule creation process becomes extremely high. However, according to the present invention, it is possible to automatically create a conversion rule for string normalization without omission by only accumulating the range of fluctuation of the character string as an equality relation in advance, very efficiently, And it becomes possible to create a conversion rule accurately.

In the embodiment of the character string normalization rule creating apparatus, the size relation is determined in the dictionary order,
An example of creating a conversion rule is shown in which the character string determined to be large is the character string before conversion and the character string determined to be small is the character string after conversion, but the character string determined to be small is the character string before conversion, It is possible to use a character string that is determined to be large as the converted character string.Also, even if the determination is not based on the dictionary order, the relationship between the character strings, such as the size relationship of the character strings and the context, can be made consistent. Any other method can be used as long as it can be determined, and the method is not limited to the above-described embodiment.

Next, a description will be given of an embodiment of the information search device for performing information search using the conversion rule for character string normalization created by the above-mentioned character string normalization rule creating device.
FIG. 10 shows a block diagram of an information retrieval apparatus according to an embodiment of the present invention. The information search device 300 shown in the figure includes a database 310 that stores character string information and an information search unit 200.
Composed of and.

The information retrieval unit 200 includes the conversion rule storage unit 16
0, a normalization unit 210, and a collation matching unit 220.

The conversion rule storage unit 160 is a database that stores conversion rules for normalizing character strings, and is automatically created by the above-described character string normalization rule creating device.

The normalization unit 210 normalizes the search character string 20 input for the search using the conversion rules stored in the conversion rule storage unit 160.

The collation matching unit 220 collates the search character string normalized by the normalization unit 210 with the character string stored in the normalized database 310, and when the collation results match, The character string stored in the matched database 160 and normalized, which is originally stored in the database 160 before being normalized, is output as the search result 30.

The operation of the information retrieval device 300 will be described below. FIG. 11 is a flowchart showing the operation of the information search device according to the embodiment of the present invention.

(Step 70) The search character string 20 is input.

(Step 71) The normalization unit 210 normalizes the search character string 20 input in Step 70 by applying the conversion rule stored in the conversion rule storage unit 160.

(Step 72) The normalizing section 210 takes out one piece of stored information from the database 310 and normalizes it by applying the conversion rule stored in the conversion rule storage section 160 to the taken-out stored information.

(Step 73) The collation matching unit 220
The search character string 20 normalized in step 71 is collated with the stored information normalized in step 72. If they do not match, the process proceeds to step 75.

(Step 74) The collation matching unit 220
The record of the accumulated information that was originally accumulated in the database 310 before being normalized of the normalized accumulated information matched in step 73 is output as a search result.

(Step 75) The normalizing unit 210 takes out all the accumulated information in the database 310, normalizes and collates it after the step 74 or when the collation results do not match in the step 73. It is checked whether or not all of the extraction normalization has not been performed yet, and the process proceeds to step 72, and if all the extraction normalization and synthesis have been completed, all processing is terminated.

A specific example of performing information retrieval using the information retrieval apparatus according to the embodiment of the present invention will be described below. The conversion rules shown in FIG. 7 are accumulated in the conversion rule accumulating unit 160 of the following information retrieval device, and the database 310
Has the structure and accumulated information shown in FIG. 12 (B), and searches the character string 2 for the item “name” in the database 310.
An example of searching with 0 or "Hozumi" will be described.

When "Hozumi" is input as the search character string 20 (step 70), the normalization unit 210 adds the conversion rule stored in the conversion rule storage unit 160 to the input search character string 20, "" Hozumi "=>" Hosumi "" is applied to normalize the search character string 20 to obtain "Hosumi" (step 71).

The normalization unit 210 retrieves one piece of stored information "Hozumi" of the item "name" from the database 310, and converts the retrieved stored information into the conversion rule """Hozumi" is applied and normalized to obtain "Hosumi" (step 72).

Since the matching unit 220 matches the search character string 20, “Hosumi”, which is normalized in the above step 71, with the accumulated information, “Hosumi”, which is normalized in the above step 72, there is a match. (Step 73, YE
S), a record of the accumulated information that was originally accumulated in the database 310 before being normalized of the normalized accumulated information that matched in step 73 above, "surname: Yamazaki""first name: hozumi""blood type: A Is output as the search result 30 (step 74).

Since the normalizing section 210 takes out all the stored information in the database 310, normalizes it, and does not collate it (step 75, NO), the stored information of the item "name", "Hanako", is set to 1 from the database 310. Then, the conversion rule stored in the conversion rule storage unit 160 is applied to the extracted storage information and normalized. Actually, since there is no applicable conversion rule, the accumulated information, "Hanako", is the normalization result as it is (step 72).

The collating / matching unit 220 first executes step 71.
The search character string 20, “Hosumi”, which has been normalized by the above, is collated with the accumulated information, “Hanako”, which is normalized in the above step 72, but since they do not match, the process is normalized by the normalizing unit 21.
0 (step 73, NO), the normalization unit 210
When it is checked whether all the stored information in the database 310 has been taken out and normalized and the collation has been performed, all the retrieved information has been taken out and normalized, and the collation has also been completed, so the search processing ends (step 75, YES).

According to the information retrieving apparatus of the above-described embodiment, in the case of retrieving a character string having a fluctuation such as "hotsumi", "hozumi", "hozumi", it is possible to perform a search considering the fluctuation. It is possible to prevent omission of search.

In the above embodiment of the information retrieval apparatus, an example has been shown in which the stored information in the database 310 is retrieved and normalized one by one at the time of retrieval. Set up,
It is possible to perform normalization in advance before the search, or only the accumulated information that matches the length of the character string of the normalization result of the search character string 20 can be the target of collation.
It is not limited to sequentially normalizing and collating all the information stored in the.

[0100]

As described above, according to the present invention, by giving an equality relation indicating that two character strings are equal to each other as information on the fluctuation of the character string, within the range of the given equality relationship. It is possible to automatically create conversion rules for searchable string normalization.

Further, not only the given equality relation is converted into the conversion rule, but also the conversion rule obtained in the process of creating the conversion rule is successively applied to the conversion rule to correct the conversion rule. Redundancy of conversion rules can be eliminated by applying the conversion rules to expression relationships and modifying the equality relationships, and it is possible to normalize character strings in one conversion process. It is possible to create conversion rules, and it is possible to prevent omission of creation of conversion rules.

Further, by automatically creating the conversion rule for normalizing the character string, it is possible to prevent human error from being mixed into the conversion rule, and at the same time, the time and effort required to create the conversion rule can be reduced. It is also possible to reduce the cost.

Further, since the size of the character string of the given equality relation is judged according to the dictionary order, it becomes possible to uniquely convert a plurality of character strings having fluctuations, and the given It is possible to create a conversion rule for a character string that can be searched without exception for an input character string within the range of expression relation.

Further, since the search character string and the stored information are normalized by using the conversion rule of the character string normalization created by the character string conversion rule creating device and the character string conversion rule creating method, the information searching device and the information are stored. In the search method, even if there is a difference between the search character string and the character string included in the stored information, it is possible to search without exception within the predetermined range and output the result.

In addition, since the search character string and the character string included in the stored information are converted by applying the sequential conversion rule and the conversion result is collated and searched, the stored information is searched without normalizing in advance. It becomes possible to

[Brief description of drawings]

FIG. 1 is a principle configuration diagram of a character string normalization rule creating device of the present invention.

FIG. 2 is a diagram for explaining the principle of the character string normalization rule creating method of the present invention.

FIG. 3 is a principle configuration diagram of an information retrieval device of the present invention.

FIG. 4 is a diagram explaining the principle of the information retrieval method of the present invention.

FIG. 5 is a configuration diagram of a character string normalization rule creating device according to an embodiment of the present invention.

FIG. 6 is a diagram showing a storage example of an equation relation storage unit according to an embodiment of the present invention.

FIG. 7 is a diagram showing a storage example of a conversion rule storage unit according to an embodiment of the present invention.

FIG. 8 is a flowchart showing an operation of the character string normalization rule creating device according to the exemplary embodiment of the present invention.

FIG. 9 is a diagram showing an example of an execution process of a character string normalization rule creating device according to an exemplary embodiment of the present invention.

FIG. 10 is a configuration diagram of an information search device according to an embodiment of the present invention.

FIG. 11 is a flowchart showing the operation of the information search device according to the embodiment of the present invention.

FIG. 12 is a diagram illustrating a conventional search method.

[Explanation of symbols]

10 Equation Relation 20 Search Character String 30 Search Result 100 Character String Normalization Rule Creating Means, Character String Normalization Rule Creating Device 110 Conversion Rule Generating Means, Conversion Rule Generating Unit 120 Conversion Rule Modifying Means, Conversion Rule Modifying Unit 130 Equations Relationship correction means, equation relationship correction section 140 control means, control section 150 equation relationship storage means, equation relationship storage section 160 conversion rule storage section, conversion rule storage section 200 information search means, information search section 210 normalization means, Normalization unit 220 Collation matching unit, collation matching unit 300 Information retrieval device 310 Information storage unit, database

Claims (13)

[Claims] 1. A character string conversion rule creating means for creating a character string conversion rule that enables a search if a difference between an input character string and a character string included in information to be searched is within a predetermined range. An apparatus for creating a character string normalization rule, comprising: 2. The character string conversion rule creating means, when a plurality of equality relations indicating that two character strings are equal to each other are given, a transformation for normalizing a character string based on the given equality relations. The character string normalization rule creating device according to claim 1, wherein the rule is created automatically. 3. The character string conversion rule creating means includes an equality relation accumulating means for accumulating the plurality of equality relations, and the equality relation taken out from the equality relation accumulating means. A conversion rule generating means for comparing the two character strings, and determining a pre-conversion character string and a post-conversion character string based on the comparison result to generate a conversion rule; and a conversion rule accumulating means for accumulating the conversion rule. Conversion rule correction means for correcting the conversion rules by applying the conversion rules stored in the conversion rule storage means to the converted character strings of the conversion rules stored in the conversion rule storage means, An equation relation correction means for correcting the equation relation by applying the conversion rule stored in the conversion rule storage means to the equation relation character string stored in the equation relation storage means; If the above equations are stored in the storage means 3. The control means for activating and repeating the conversion rule generation means, the conversion rule correction means, and the equality relation correction means in order until the equality relation stored in the equation relation storage means disappears. The described character string normalization rule creating device. 4. The character string normalization rule creating device according to claim 3, wherein the conversion rule generating means compares the magnitudes of the two extracted character strings in the equation relation in dictionary order. 5. The conversion rule generation means deletes the extracted equality relations from the equality relation accumulating means when the comparison results are equal to each other, and the comparisons when the comparison results are not equal to each other. 5. The character string normalization rule creating device according to claim 3, wherein a character string having a large result of is the pre-conversion character string, and a character string having a small result of the comparison is the post-conversion character string. 6. A character string conversion rule creating step for creating a character string conversion rule that enables a search if the difference between the input character string and the character string included in the search target information is within a predetermined range. A method for creating a character string normalization rule, comprising: 7. The character string conversion rule creating step normalizes a character string based on the given equality relations when a plurality of equality relations indicating that two character strings are equal are given. The method for creating a character string normalization rule according to claim 6, wherein the rule is created automatically. 8. The character string conversion rule creating step extracts the equality relation from the set of equality relations, compares two character strings of the extracted equality relations, and converts based on a result of the comparison. A conversion rule generation step of determining a pre-character string and a post-conversion character string, generating a conversion rule, and adding the conversion rule to the set of conversion rules; and the converted character strings of all the conversion rules included in the set of conversion rules, A conversion rule correction step of correcting all the conversion rules included in the set of conversion rules to modify the conversion rules; An equality relation correcting step of correcting the equality relation by applying the included conversion rule, and if the equality relation set includes the equality relation, the equality relation correction step is included in the equality relation set. The conversion rule until the equation relationship Generating step, the conversion rule modification step, according to claim 7 string normalization rules creation method according repeating running the equal expression related correction steps in order. 9. The character string normalization rule creating method according to claim 8, wherein the conversion rule generating step compares the magnitudes of the two extracted character strings in the equation relation in dictionary order. 10. The conversion rule generating step deletes the extracted equality relations from the set of equality relations when the results of the comparison are equal, and the comparison when the results of the comparison are not equal. The character string normalization rule creating method according to claim 8 or 9, wherein a character string having a large result of is a pre-conversion character string, and a character string having a small result of the comparison is a post-conversion character string. 11. An information search which has an information storage means for storing information having a character string, and when a search character string is input, searches the information storage means based on the search character string and outputs a search result. The information, characterized in that the device has an information retrieval means for converting the retrieval character string and the character string accumulated in the information accumulating means based on a predetermined conversion rule and performing a search based on the result of the conversion. Search device. 12. The information retrieving means stores the conversion rule accumulating means for accumulating a plurality of conversion rules for normalizing a character string, and the character string accumulated in the retrieval character string and the information accumulating means. Normalization means for applying and normalizing the conversion rules stored in the conversion rule storage means, the search character string normalized by the normalization means, and the character string stored in the information storage means If the results of the collation match, the collating and matching means for outputting the character string before normalization of the character string obtained by normalizing the matched character string stored in the information storage means as the search result. The information search apparatus according to claim 11, further comprising: 13. An information search method for, when a search character string is input, searching for information stored in advance with the search character string and outputting a search result, wherein the character string is normalized to the search character string. Applying a plurality of conversion rules for normalization, and collating the normalized search character string with the character string normalized by applying the conversion rules to the character string included in the accumulated information. , Information that is characterized in that when the result of collation matches, the character string before normalization of the character string that is obtained by normalizing the character string included in the stored information that matches is output as the search result. retrieval method.