JPH11288416A - Method and device for excluding error of automatic connection of different kind of data having address information - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the reliability of connection by detecting and excluding the error of an automatic connected result in as small man-hours as possible at the time of automatically retrieving and connecting the same resident data by using a computer for two kinds of data having address information (town name, address, number, care of, and resident name or the like). SOLUTION: Two kinds of data 101 and 102 to be connected are connected by the plural kinds of connection processing systems whose connection evaluating methods are different (110 and 120). Those connected results and collated to each other, and the incoincident result is extracted (130), and any erroneous connection is excluded by using the extracted result as an object for judging the erroneous connection (140). Then, the connected results in the plural kinds of connection processing systems after the erroneous connection is excluded are combined so that a connected data base 200 can be prepared (150).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to two types of data including address-related information (town address, address, number, form, resident name, etc.) using a computer and referring to the address-related information. The present invention relates to a method and an apparatus for efficiently detecting and rejecting data of the same resident efficiently and automatically (combining them) and efficiently detecting and eliminating connection errors caused by automatic connection.

[0002]

2. Description of the Related Art For example, by combining customer data with detailed house map data having resident information for each building, it is possible to specify and display a building on a map from customer information such as a telephone number. Therefore, a process of combining different types of data based on address information is required.

[0003] However, the address information of different types of data created separately includes fluctuations, incompleteness, and inaccuracy such as address information. When the computerized house map DB is completely matched, only about 40% is automatically combined (Chapter 6 of Reference 1). If the remaining data is manually combined, for example, combining customer information and a map posted in a telephone directory (approximately 1 million) by occupation in the 23 wards of Tokyo would require a great deal of man-hours.

[Reference Document 1]: Yasuda, Matsumura, Mizumachi, Karasawa “Map guidance system using telephone and FAX”, Proc. Of the 3rd Symposium on Functional Graphic Information Systems, 1992
/ 4, pp.81-86. What is the fluctuation of the notation in the address information?
For example, it may be written in kanji or katakana, written as "stock company" or "(stock) ~"
May be abbreviated as follows,
It means that it may be written in full-width characters or half-width characters. There are various other fluctuations.

[0005] For this reason, conventionally, as a method for improving the automatic connection accuracy in a computer, two types of address information of a DB (machichome /
Address and number) are used only for narrowing down the comparison data without strict matching, and the emphasis is placed on comparison of resident information. A method of extracting information and determining the degree of coincidence has been proposed (see Reference 1 above). By using this method, the automatic connection rate by the computer can be increased to about 80% or more, and it has become possible to greatly reduce manual work.

Further, as disclosed in Japanese Patent Application No. 10-21617, "Method and Apparatus for Combining Different Types of Data Using Address Information", the degree of coincidence of characters is evaluated according to the degree of coincidence of addresses and combined. There are also methods. In this method, by changing the criteria for determining the degree of coincidence of the characters in the resident name information in consideration of the degree of coincidence of the address, the evaluation can be automatically combined by fine evaluation, and furthermore, the character string can be used as a set of words. Instead, it is regarded as a set of characters, processed by focusing only on the presence of common characters, and is resistant to fluctuations in notation. Since the processing does not require Japanese processing, it realizes very simple processing.

[0007]

However, in any of the methods for improving the connection ratio by allowing ambiguity, since the automatic connection is used, for example, "AA car rental car" and "AA car shop" are the same. It is inevitable that connection errors occur, such as being considered and combined. In particular, a user will not notice an error until they go to the site, for example, looking at a map combined based on this information.
Such a situation should be eliminated as a service as much as possible. To eliminate this error, conventionally,
There is no other choice but to manually review the entire validity of the combined result, and there is a problem that the number of data may be unrealistically large.

An object of the present invention is to provide a means for improving the reliability of connection by detecting and eliminating an error in the result of automatic connection with as few man-hours as possible.

[0009]

FIG. 1 is a block diagram showing a main configuration of the present invention. The processing device 100 is a CPU
And a memory comprising a plurality of connection processing units 1
10, 120, result matching classification unit 130, false result elimination unit 1
40, each processing means of the combining result unit 150.

Two types of databases (DB) DBa10
1 and a plurality of types of combination processing units 110 and 120 that combine attribute data for the same building from the address information (address, resident name, etc.) of the DBb 102, execute the respective combination processing, and perform the result matching classification. The processing is transferred to the unit 130, the combined results are compared, and the results are classified into unmatched ones and matched ones. The unmatched one is sent to the erroneous result elimination unit 140, and the display device 160 and the input device 170 are used to manually determine whether to use either of the results or not to use both results. In addition to eliminating erroneous combinations, those that can be determined to be successful are merged.

The join result merging unit 150 generates a final join DB 200 by combining the result with the result that has been matched by the process of the join matching unit 130. The operation according to the present invention is as follows.

In the processing apparatus 100 having the configuration shown in FIG. 1, two types of connection processing units 110 and 120 using different methods, such as focusing on proper words on the one hand and focusing only on the degree of coincidence of characters on the other hand, are used. , And the results are compared. For records where the name fluctuation or address fluctuation is large and automatic merging is likely to fail, it is easy for a combination error peculiar to each method to occur, and the results do not match. By manually checking only those of the automatic combination, it is possible to detect errors in the automatic combination result very efficiently.

It is to be noted that here two types of connection processing units 11
Although examples of 0 and 120 are shown, three or more types may be used.

[0014]

Next, embodiments of the present invention will be described with reference to the drawings. FIG. 2 is a diagram specifically showing an outline of a processing flow by the processing apparatus of the present invention shown in FIG.

DBa 101 and DB to be combined
As shown in FIG. 2, b102 has an address and other information for a certain name. Combination processing unit 11
0 and the combination processing unit 120 respectively access the DBa 101 and DBb 102, read out the respective records, perform the combination processing by their own methods, and output the intermediate combination results a103 and b104. As shown in FIG. 2, the intermediate combining result holds the combining result and the names and addresses of both the original DBa 101 and DBb 102. The combination flag in this data is "1" when the combination is successful and "0" when the combination fails. In the case of a combining failure, the intermediate combining result has only the information of DBa101 and no information of DBb102.

The result reconciliation classification unit 130
1 and storage means for a combined result mismatch data group 105, a combined result matched data group 106, and a combined failure data group 107. DBb1 combined with the intermediate combination results a103 and b104 based on the information of DBa101
02 is compared in the unit of record. If the combined results do not match, the information is stored in the combined result mismatch data group 105. If they match, the combined result is stored in the combined result matching data group 106. The data is stored in the unsuccessful data group 107.

The erroneous result elimination unit 140 is composed of a mismatch result selection unit 141 and a storage unit for the combination approval data group 108 and the combination denial data group 109. The mismatch result selection unit 141 reads from the combination result mismatch data group 105. The displayed data group is displayed on a display device 160 such as a display, and a human judges the validity of the combination by checking the result of the combination and the address and name information corresponding to the combination. Then, the input device 170 such as a keyboard notifies the user. The disagreement result selection unit 141 sorts the results based on the result. If one is OK, the result is stored in the combination approval data group 108, and if both are NG, the result is stored in the combination denial data group 109.

FIG. 3 shows an example in which the result is different between the processing A of the combining processing unit 110 and the processing B of the combining processing unit 120, and a human has adopted neither of them and determined that the combining has failed. For [5-2 / Tanaka Clinic] in [(address) block number-lot number / name] in DBa101, in process A of the combination processing unit 110, a matching DBb102 record at the same address could not be found, To find [6-10 / Ichiro Tanaka]
This was recognized as a target to be combined by matching the important proper noun "Tanaka".

On the other hand, in the processing B of the combining processing unit 120, since no matching record was found in the same way by address matching, the surrounding area was searched to find [5-12 / Nakai clinic].
Since 3 characters matched 4 characters from Tanaka Clinic, the degree of matching was determined to be high and the character was recognized as a combination target.

These results are respectively the intermediate combination result a
103, stored in the intermediate combination result b104. Both of the combination flags are “1”, indicating successful combination. The result matching unit 130 compares the intermediate combination result a103 and the intermediate combination result b104 with the result comparison unit 131, and stores them in the combination result mismatch data group 105 because they do not match.

Mismatch result selection unit 1 of false result elimination unit 140
41 displays on the display device 160 these combined results read from the combined result mismatch data group 105 to determine whether they are erroneous results. Here, the person who determines the result, after seeing the result of the combination of the two, judges that there is a high risk of combining the two, and selects the failure of the combination, so the two results are not adopted. Therefore, the result is stored in the connection denial data group 109.

In the join result merging unit 150, the results are totaled based on the result of the erroneous result elimination unit 140 and the data group of the join matching classifying unit 130, and a joint DB 200 is created.

[0023]

[Example] Specifically, a telephone directory DB (about 100
References 1 (Yasuda, Matsumura, Mizumachi, Karasawa "Map guidance system using telephone and FAX", 3rd Functional Graphic Information System Symposium, 1992/4, pp.81-86.) Specific information comparison method using Japanese processing (Processing A)
The result when combining automatically with is that the combining success rate is about 85%,
The connection failure was about 15%, and the estimated misconnection by sampling was 5% (connection reliability 95%).

On the other hand, Japanese Patent Application No. Hei 10-21617 discloses this.
As a result of performing the combining process by the method (Processing B) proposed in the above publication, the combining success was about 90% and the combining failure was about 10%. Was 20% (only treatment A had about 5% binding, only treatment B had about 10% binding, and both bindings had a mismatched result of 5%) and 5% had no binding.

Furthermore, as a result of manually checking only those that do not match both results, 2% can judge that either of the two methods is the correct answer, the combination succeeds by selection, and 3% (60% of the total errors) is a joint error. I was able to judge and I could eliminate it.

For example, in the case of using only the method of “processing A”, in order to improve the overall reliability by 3%, 10 errors are required.
Assuming that the data exists evenly in the data of 100,000, it is considered necessary to manually check the results of the connection of 600,000 data to eliminate the incorrect connection. In addition, this work must be performed while looking at the surrounding information on a house map or the like and judging whether there is something more appropriate, in order to judge whether or not the joining result is incorrect by looking at only one joining result. There is Tokyo 2
In the case of three wards, the efficiency was about 200 cases / person-day.

On the other hand, in the example in which the method of the present invention is employed and only the inconsistency (5%) as a result of the two methods is manually checked, 5
It is enough to check only 10,000 cases (5%), and it is necessary to compare the results of the two methods and exclude only those judged to be inappropriate (high risk of wrong connection) to be combined. , 3,000 jobs / person / day. As a result, the man-hour for eliminating 60% of the total misconnection can be reduced to 1/180.

[0028]

As described above, according to the present invention,
By comparing only the unmatched portions of the results of the plural types of join processing methods, it is possible to eliminate the automatic join error extremely efficiently and with a small number of man-hours, thereby improving the reliability of the entire join result.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a main configuration of the present invention.

FIG. 2 is a diagram specifically showing an outline of a processing flow by the processing apparatus of the present invention shown in FIG.

FIG. 3 is a diagram illustrating an example in which the combination results by a plurality of combination processing units (processing A and processing B) are different, and a human has adopted neither of them and has failed in the combination.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 101 Database to be combined (DBa) 102 Database to be combined (DBb) 110 Combination processing unit (Process A) 120 Combination processing unit (Process B) 130 Combined joining classification unit 140 False result elimination unit 150 Combined result united unit 160 Display device 170 Input device 200 Connection database (DB)

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Kensaku Fujii Nippon Telegraph and Telephone Co., Ltd., 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo

Claims (4)

[Claims] 1. A processing method for combining two different types of data having address information by a computer, wherein the two types of data to be combined are combined by a plurality of different types of combining processing methods having different combinations of evaluation methods. A process for eliminating the incorrect connection by comparing the combined results obtained by the plurality of types of combining processing, extracting a mismatched result, and judging whether or not the extracted result is a wrong connection, and eliminating the wrong connection. Combining the combined results of the plurality of types of combining processing to obtain a combined result to be obtained. 2. A method according to claim 1, further comprising: a first method for processing elements such as a resident name and a building name in the address information in Japanese and evaluating and combining important words; 2. A method according to claim 1, further comprising the step of: estimating the degree of coincidence of characters in accordance with the method. 3. A processing device for combining two different types of data having address information by a computer, wherein a plurality of combining processes are combined for each of the two types of data to be combined using different combining evaluation methods. And a result matching unit that matches the combined results obtained by the plurality of combining processing units and extracts a result having a mismatch, and a result extracted by the result matching unit that is determined to be an incorrect combination. An error result elimination unit having an input / output interface for eliminating a connection, and a connection result uniting unit that obtains a connection result obtained by combining the connection results by the plurality of connection processing units that eliminate the connection error. An error elimination processing device for automatically combining different types of data having address information. 4. A first combination processing unit for processing elements such as a resident name and a building name in the address information into Japanese and evaluating and combining important words in the plurality of types of combination processing units; 4. The apparatus according to claim 3, further comprising a second combination processing unit that evaluates the degree of coincidence of characters in accordance with the degree of coincidence of the address.