JP3635341B2 - How to join databases - Google Patents

Description

[0001]
[Industrial application fields]
The present invention relates to a database joining method, and particularly relates to individual records having different reliability of information or having different representation formats in any different independent databases, and cross-reference them. The present invention relates to a database joining method for accurately and automatically creating a possible joined database.
[0002]
[Prior art]
Conventionally, when there is a request to associate and join individual records in any different independent database, if any different independent database is created based on a uniform rule, the common database These fields can be easily joined by a JOIN (joining) operation prepared in a DBMS (database management system).
[0003]
The JOIN operation, which is a database join method, sorts both tables to the database by the item to be joined (or uses an index) and retrieves the same key of both tables 7 and between the tables in the database. There is a method in which one table is used as a reference and the other database table is brute-forced and a table satisfying the conditions is extracted.
[0004]
[Problems to be solved by the invention]
However, the above-mentioned conventional method of joining databases by JOIN operation is that, if any different independent database is created based on the respective rules, the information expression format is not uniform and arbitrarily different Some information in an independent database, for example, “Office” and “Office” are common fields, and even if they point to the same matter, there are often such fluctuations in the expression, In addition, since there may be a difference in reliability between common field information in the database, for example, address prefecture information and address block information, etc. The number of cases is small, and depending on the field of interest, the reliability is very low. So to create an accurate join database,
(1) Create a new combined database artificially from the beginning, including the contents of the other database based on one database;
(2) Using one database as a reference, artificially find the record to be combined with each record from the contents of the other database;
However, when these methods are adopted, there are drawbacks such that the combination cannot be realized unless a large amount of man-made operation, work time, and work cost are invested.
[0005]
The present invention has been made in view of the above points, solves the above-described conventional problems, and records individual records having a difference in reliability or expression in any different independent database. It is an object of the present invention to provide a database combining method capable of automatically creating a combined database that associates each other with each other and enables cross-references without accurately performing human work.
[0006]
[Means for Solving the Problems]
The present invention relates to a database in which records of a database having a common field and having different degrees of reliability or the representation format of the reliability of the common field information are individually associated by using the common field information. In the joining method:
For the common field information with different expression formats, an analysis process to unify the expression format is performed, and common field information after the analysis process is created,
The record of reliability degree or representation of a common field information are different databases, as compared to each common field, the success or failure of the matching, the processing contents of the analysis process and the reliability of every case of the common field information Based on the score, the score is set in the scoring table in advance, the score is assigned to each record according to the degree of reliability, and the different records are associated and the individual records to be combined are determined based on the score.
[0007]
[Action]
The present invention relates to common field information having different representation formats even if the reliability of the common field information is low or the representation format is different between any different databases having common fields. Processes using various analysis algorithms that unify the expression format, create common field information after the analysis process, compare each field, and process the analysis process according to the success or failure of the field data association database contents and, to impart common field information reliability degrees if either et created by scoring table to score each record, the score associated individual records of different databases, to allow cross-reference Can be created automatically without any human work.
[0008]
【Example】
Embodiments of the database combining method of the present invention will be described below with reference to the drawings.
[0009]
FIG. 1 shows a block diagram of a database combining method according to an embodiment of the present invention.
[0010]
The system shown in FIG. 1 includes a plurality of join target databases 1 such as a telephone directory database (DB1) or a map database (DB2), a join database (DB3) 2 representing the result of the join process, and a join target database 1. The temporary recording unit 4 and the temporary recording unit 4 for instructing the database control unit 3 and the database control unit 3 to retrieve the record or update the combined database 2 to request or store the record and store the record. The general character string processing unit 5 that applies conversion processing to the recorded character strings, the record groups recorded in the temporary recording unit 4 are compared for each field, and score processing is performed based on the comparison result. Based on the process of writing back to the recording unit 4 and the points accumulated in the temporary recording unit 4, the combination partner record is determined.較照 engaging portion 6, with respect to inquiries from the comparison unit 6, comprised of scoring table 7 to issue a number corresponding to the comparison result.
[0011]
FIG. 2 is a flowchart showing the operation of one embodiment of the present invention.
[0012]
In the following description, in the processing shown in the flowchart of FIG. 10, DB1 is a telephone directory database and DB2 is a map database.
[0013]
FIG. 3 shows a telephone directory database (DB1) according to an embodiment of the present invention, and FIG. 4 shows a map database (DB2) according to an embodiment of the present invention. FIG. 5 shows an example of a combined database according to an embodiment of the present invention.
[0014]
Each record in the database of FIGS. 3 and 4 has a field structure. In this embodiment, DB1 and DB2 are combined to construct a combined database DB3 as shown in FIG.
[0015]
Step 1) Check whether there is an unprocessed record in the telephone directory database (DB1) in the database 1 to be joined. If there is no DB1 record to be extracted, the process proceeds to the end process of Step 10. On the other hand, if there remains a record to be retrieved from DB1, the process proceeds to step 2.
[0016]
Step 2) The database control unit 3 takes out one record of the field structure as shown in FIG. The conversion process is performed, and the data after the conversion process is recorded from the database control unit 3 to the temporary recording unit 4.
[0017]
Step 3) As long as there is a map database (DB2) in the database 1 to be combined that has not been compared with one record extracted from DB1, the operations from Step 4 to Step 6 are repeated to extract from DB1. When the comparison of one record and all records in DB2 is completed, the process proceeds to step 7.
[0018]
Step 4) The database control unit 3 takes out one record of the field structure of DB2 as shown in FIG. 4 and applies a character string conversion process similar to that in step 2, and temporarily records the converted data from the database control unit 3 Part 4 is recorded.
[0019]
Step 5) The comparison and collation unit 6 performs comparison and collation on the address code of the DB1 record and the DB2 record recorded in the temporary recording unit 4 and the name and building name fields to which character string conversion processing has been added. The score is obtained by comparing the result of the above with the scoring table 7.
[0020]
Step 6) The score and the record obtained in Step 5 are stored in the temporary recording unit 4 and the process proceeds to Step 3.
[0021]
Step 7) When the comparison of all records in DB2 is completed, the record with the highest score among the scores of the records recorded in the temporary recording unit 4 is judged by the comparison / collation unit 6, and a flag indicating the combination is given. .
[0022]
Step 8) A join record having a field configuration as shown in FIG. 5 is generated from one record of DB1 and a record of DB2 to which a flag indicating join is assigned, and is written from the database control unit 3 to the join database 2 (DB3).
[0023]
Step 9) The contents of the temporary recording unit 4 are erased, and the process proceeds to Step 1.
[0024]
Step 10) The process is terminated.
[0025]
The combined database 2 can be created by the method as described above.
[0026]
The flowchart process shown above will be specifically described.
[0027]
First, as shown in FIG. 4, the fields of the DB 1 phonebook database include “address code”, “name”, “name”, “prefecture code”, “city code”, “town code”, and “chome” codes. There are “building name”, “phone number”, and “occupation code”. Further, as shown in FIG. 4, the fields of the map database of DB 2 include “address code”, “name”, “name” including “prefecture code”, “city code”, “town code”, and “chome” code. There are “building name”, “map coordinates”, “building ID”, and the like.
[0028]
In DB1 and DB2, the address code, which is a common field, is composed of hierarchized codes according to the address hierarchy such as prefecture code, city code, town code, and chome code. , Prefecture code, city code and town and village code are highly reliable, chome code is less reliable than those, the accuracy of the data varies depending on the characteristics of the field, on the other hand, The building name is described by a character string, and there is fluctuation in expression.
[0029]
Character conversion processing is performed in order to unify information having fluctuations in the expression of the character string.
[0030]
FIG. 6 shows an example of character conversion processing according to an embodiment of the present invention.
[0031]
The figure shows the contents of conversion processing for information expressed by character strings having fluctuations in expression such as names or building names which are common fields in DB1 and DB2.
[0032]
First, the clean-up process is a process of deleting the muddy sound or the muddy sound of the semi-turbid sound. For example, “button shobo” is set to “hotan shobo”. The stuttering / sound repelling normalization process is a process of converting a small character “tsu” into a large character “tsu”. For example, “cash finance” is set to “cache finance”. The corporation name deletion process is a process of deleting a word representing a corporation such as “corporation”. The occupation ending deletion process is a process of deleting occupation endings such as “station” and “temple”. For example, “JR Shinjuku Station” is set to “JR Shinjuku”. The unique information extraction process is a process of extracting only a word expressing the unique name by a certain logic. For example, “Tanaka Patent Office” is set to “Tanaka”.
[0033]
FIG. 7 shows a scoring table of one embodiment of the present invention. The scoring table shown in the figure describes the score processing determined by the processing contents of the character string conversion processing and the degree of matching of the address codes.
[0034]
The result of the above character string conversion process and the match of the address code are divided into complete match or different only at the end, and are scored. For example, when the address code is completely matched without performing the character string conversion process, the score is 20 points, and when only the end is different, the score is 10 points. If there is a complete match, the score is 13.
[0035]
A case where the combined database 2 is generated from the above-described DB1 and DB2 target database 1 will be described.
[0036]
(1) First, the database control unit 3 takes out one record of the phone book database (DB1) which is a database to be combined, and performs a character string conversion process for unifying the character strings by the general character string processing unit 5, The record of DB1 is stored in the temporary recording unit 4.
[0037]
(2) The database control unit 3 takes out one record of the map database (DB2) that is the database to be combined, performs character string conversion processing in the same manner as (1), and stores the record of DB2 in the temporary recording unit 4 .
[0038]
(3) In this embodiment, since the degree of coincidence of the address code is referred to, the comparison / collation unit 6 uses the scoring table 7 to store the addresses of the DB1 record and the DB2 record stored in the temporary recording unit 4. The codes are compared, and the compared DB2 records are stored in the temporary recording unit 4 together with the score. In this way, when the comparison of all DB2 records is completed, the record with the highest score among the records of DB2 assigned with the points stored in the temporary recording unit 4 and the temporary recording unit with (1) 4 and the record of DB1 stored in 4 are combined. In the case of the present embodiment, as the record for DB1 in FIG. 3, the address code of the record in DB2 shown in FIG. 4 is completely matched with the record in DB1 in the unprocessed state for the character string conversion process. The highest 20 points are awarded.
[0039]
(4) In this embodiment, the field structure of the combined database 2 is “phone number”, “occupation code”, “map coordinates”, and “building ID”. Therefore, the telephone number “03-3123-4567” and the occupation code “10242002” are transferred from DB1, the map coordinates “24-65-72-41” and the building ID “54202” are transferred from DB2, and these data are stored. It is written in the combined database 2.
[0040]
(5) When the writing of 2 to the combined database of 1 record is completed, the temporary recording unit 4 is cleared.
[0041]
As shown in the above-described embodiment, by using field data that does not fluctuate in the expression such as an address code, it is possible to create the combined database 2 at high speed without reducing reliability by narrowing down the records to be compared and collated. .
[0042]
Moreover, in the said Example, although the record of the database was linked | related by the score by the scoring table 7 produced from the degree of reliability of information and the combined database was produced | generated, you may provide a table in software and process. For common field information that differs only in expression format, various analysis algorithms that unify the expression format of common field information may be used.
[0043]
【The invention's effect】
As described above, according to the present invention, a combined database capable of cross-referencing by associating individual records having differences in reliability or expression formats with respect to any different independent database. It can be created automatically without compromising reliability and without human work.
[Brief description of the drawings]
FIG. 1 is a block diagram of a database combining method according to an embodiment of the present invention.
FIG. 2 is a flowchart showing the operation of an embodiment of the present invention.
FIG. 3 is a diagram showing a telephone directory database (DB1) according to an embodiment of the present invention.
FIG. 4 is a diagram showing a map database (DB2) according to an embodiment of the present invention.
FIG. 5 is a diagram showing an example of a combined database according to an embodiment of the present invention.
FIG. 6 is a diagram illustrating an example of character conversion processing according to an embodiment of the present invention.
FIG. 7 is a scoring table according to an embodiment of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Join object database 2 Join database 3 Database control part 4 Temporary recording part 5 General character string process part 6 Comparison collation part 7 Scoring table

Claims (1)

In a database combining method for associating records of databases having common fields and having different degrees of reliability that are the certainty of the common field information or different representation formats by using the common field information,
For the common field information with different expression formats, an analysis process to unify the expression format is performed, and common field information after the analysis process is created,
The record of the reliability degree or representation of a common field information are different databases, as compared to each common field, the success or failure of the matching, the processing contents of the analysis process and the reliability of the common field information characterized in that set in advance score based on the time if the scoring table, given to each record by the degree of reliability該得points, associate a different database based on該得point, to determine the individual records that binds Database join method.

Images