JP2022144829A - Data correction method and data correction program - Google Patents

Abstract

To suggest appropriate correction candidates.SOLUTION: An information processing device detects any inconsistency between data in a data table and, if the inconsistency is detected, extracts a plurality of correction candidates for correcting the inconsistency. The information processing device calculates correction costs corresponding to the plurality of correction candidates, and displays the plurality of correction candidates and the correction costs corresponding to the plurality of correction candidates in association with each other.SELECTED DRAWING: Figure 3

Description

The present invention relates to a data correction method and the like.

Terminology used in various fields may not be unified. For example, "name" and "first name" have the same meaning, but different notations, so they are treated as completely different data on the system, which makes it impossible to link the data accurately.

By creating and using controlled vocabulary data that defines synonyms, hypernyms, and hyponyms, the ambiguity of terms can be absorbed, and the problems described above can be resolved. Controlled vocabulary data is a manually created dictionary summarizing the semantic relationships between multiple terms in order to prevent omissions in searches caused by ambiguity, homographs, and heteromorphic synonyms.

FIG. 15 is a diagram illustrating an example of the data structure of controlled vocabulary data. As an example, the data structure of controlled vocabulary data will be described in a tabular format. As shown in FIG. 15, the controlled vocabulary data 10 has a term name column 10a, a headline column 10b, and a hypernym column 10c.

The term name column 10a is set with terminology terms used in a specific field. Headings are set in the heading column 10b. Headings are names that represent multiple types of terms. For example, in the example shown in FIG. 15, "software" is set as a heading for the term names "software", "software", and "software". "OS" is set as the heading of the term name "os" and "OS".

Hypernym is set in the hypernym string 10c. A broader term indicates a heading in another column that is a broader concept of the heading in the corresponding row. For example, since "software" is set as the broader term of the heading "OS", the heading "software" is set as the broader concept of the heading "OS".

In the following description, a collection of synonymous terms from the values (terms) in the term name column 10a and the values (headings) in the heading column 10b is referred to as a "synonymous group". In the example shown in FIG. 15, the information (records) on the first to third lines belong to the same meaning group 10-1. Information on the fourth and fifth lines belongs to the same meaning group 10-2. In principle, each synonym group should have a unique heading and hypernym.

The synonym group 10-1 is given a unique heading "software". The synonym group 10-2 is set with a unique heading "OS" and a unique broader term "software".

Next, we define “inconsistency” regarding synonymous groups. FIG. 16 is a diagram for explaining mismatch. Mismatches A, B, C, and D will be described as an example.

"Inconsistency A" will be explained using the controlled vocabulary data 11a. Inconsistency A is the presence of two or more types of headings in the same synonymous group. For example, in the controlled vocabulary data 11a, two types of headings "soft" and "software" are set in the same synonym group 11a-1, which corresponds to inconsistency A.

"Inconsistency B" will be explained using the controlled vocabulary data 11b. Inconsistency B is the presence of two broader terms in the same synonym group. For example, in the controlled vocabulary data 11b, two broader terms "software" and "software" are set in the same synonym group 11b-1, which corresponds to inconsistency B.

"Inconsistency C" will be explained using the controlled vocabulary data 11c. Inconsistency C is that a hypernym set in a certain synonym group designates a term other than the heading of another synonym group. For example, controlled vocabulary data 11c includes synonym groups 11c-1 and 11c-2. The broader term "software" of the synonym group 11c-2 is different from the heading "software" of the synonym group 11c-1, which corresponds to inconsistency C.

"Inconsistency D" will be explained using the controlled vocabulary data 11d. Inconsistency D means that the hierarchy is cyclic among different synonymous groups. For example, controlled vocabulary data 11d includes synonym groups 11d-1 and 11d-2. The broader term "OS" of the synonym group 11d-1 specifies the heading "OS" of the synonym group 11d-2, but the broader term "software" of the synonym group 11d-2 specifies the heading of the synonym group 11d-1. Since "software" is specified, it is circular and falls under inconsistency D.

The amount of work required to manually correct the values of each cell of the controlled vocabulary data in order to correct the inconsistency described in FIG. 16 is referred to as "correction cost". FIG. 17 is a diagram for explaining correction costs.

In the controlled vocabulary data 12a-1, there are two or more types of headings in the same synonym group, resulting in inconsistency A. By correcting the heading "software" of the controlled vocabulary data 12a-1 to "software", the inconsistency A is resolved and the controlled vocabulary data 12a-2 is obtained. In this case, the correction cost is "1".

In the controlled vocabulary data 12b-1, there are two types of hypernyms in the same synonym group, which results in inconsistency B. FIG. By correcting the broader term "software" of the controlled vocabulary data 12b-1 to "software" and the broader term "software" of the controlled vocabulary data 12b-1 to "software", the inconsistency B is resolved and the controlled Vocabulary data 12b-2. In this case, the correction cost is "2".

Here, as described with reference to FIG. 17, when manually correcting the controlled vocabulary data, input errors or oversights may occur. In addition, correction work takes time, and the quality of the controlled vocabulary data may deteriorate.

There are the following conventional techniques for detecting inconsistencies and presenting correction candidates from controlled vocabulary data (tabular data).

FIG. 18 is a diagram for explaining a conventional technique. In FIG. 18, the tabular data 13 is used for explanation. In tabular data 13, the data in the first to third rows belong to synonym group 13-1. The tabular data 13 includes a term name column 13a and a heading column 13b.

In the prior art, the degree of similarity between character strings for each column of tabular data 13 is calculated using an edit distance or the like, character strings with close distances are divided into clusters, and character strings are presented for each cluster. The operator can check the character strings for each cluster and collectively correct spelling variations.

For example, using the heading column 13b, there are two types of headings, "software" and "software", in the same synonym group 13-1. In the prior art, it is possible to determine that "software" and "software" are in the same cluster for the heading column 13b, and present a correction candidate with the lowest correction cost. For example, in the conventional technology, the correction cost is "2" when software is replaced with software, and the correction cost is "1" when software is replaced with software, so "soft" is presented as the value after replacement. . The worker corrects the tabular data 13 to the tabular data 14 by modifying the software with reference to the presented values after replacement. Inconsistency A is resolved in tabular data 14 .

JP 2008-293532 A JP-A-6-266769

However, the conventional technology described above has a problem that it is not possible to present appropriate correction candidates when there is a hierarchical relationship between synonym groups.

FIG. 19 is a diagram for explaining the problem of the conventional technology. In FIG. 19, the controlled vocabulary data 15 will be used for explanation. The controlled vocabulary data 15 includes a term name column 15a, a headline column 15b, and a hypernym column 15c. It is also assumed that the information on the 1st to 3rd rows of the controlled vocabulary data 15 belongs to the synonym group 15-1, and the information on the 4th and 5th rows belongs to the synonym group 15-2.

Focusing on the heading column 15b of the synonym group 15-1, the conventional technology presents "soft" as the post-replacement value by executing the processing described with reference to FIG. The worker modifies the controlled vocabulary data 15 to the controlled vocabulary data 16 by modifying the software with reference to the presented values after replacement. The controlled vocabulary data 16 includes a term name column 16a, a headline column 16b, and a hypernym column 16c. Information on the first to third lines of the controlled vocabulary data 16 belongs to the synonym group 16-1, and information on the fourth and fifth lines belongs to the synonym group 16-2.

Here, in the controlled vocabulary data 16, the hypernym contained in the hypernym string 16c of the synonym group 16-2 is different from the heading contained in the heading string 16b of the synonym group 16-1, and inconsistent C is included. ing. On the other hand, in the conventional technology, in the synonym group 16-2 of the controlled vocabulary data 16, only "software" is detected as a cluster detected from the hypernym string. cannot be presented.

In one aspect, an object of the present invention is to provide a data correction method and a data correction program capable of presenting suitable correction candidates.

In the first alternative, the computer performs the following processes. The computer detects inconsistencies between data in the data table. The computer extracts a plurality of correction candidates for correcting the inconsistency when the inconsistency is detected. A computer calculates a correction cost corresponding to each of the plurality of correction candidates. The computer associates and displays a plurality of correction candidates and correction costs corresponding to the plurality of correction candidates.

Appropriate correction candidates can be presented.

FIG. 1 is a diagram for explaining the reference technology. FIG. 2 is a diagram for explaining other correction candidates. FIG. 3 is a diagram for explaining the processing of the information processing apparatus according to the embodiment. FIG. 4 is a functional block diagram showing the configuration of the information processing apparatus according to this embodiment. FIG. 5 is a diagram showing an example of the data structure of controlled vocabulary data. FIG. 6 is a diagram showing an example of the data structure of the intra-group correction candidate table. FIG. 7 is a diagram showing an example of the data structure of a correction candidate table. FIG. 8 is a diagram (1) showing an example of a correction result. FIG. 9 is a diagram (2) showing an example of the correction result. FIG. 10 is a diagram (3) showing an example of the correction result. FIG. 11 is a diagram (4) showing an example of the correction result. FIG. 12 is a diagram illustrating an example of screen information generated by a display control unit; FIG. 13 is a flow chart showing the processing procedure of the information processing apparatus according to this embodiment. FIG. 14 is a diagram illustrating an example of a hardware configuration of a computer that implements functions similar to those of the information processing apparatus. FIG. 15 is a diagram illustrating an example of the data structure of controlled vocabulary data. FIG. 16 is a diagram for explaining mismatch. FIG. 17 is a diagram for explaining correction costs. FIG. 18 is a diagram for explaining a conventional technique. FIG. 19 is a diagram for explaining the problem of the conventional technology.

Hereinafter, embodiments of the data correction method and data correction program disclosed in the present application will be described in detail based on the drawings. In addition, this invention is not limited by this Example.

Before describing the present embodiment, a reference technique devised by the inventor will be described. FIG. 1 is a diagram for explaining the reference technology.

In FIG. 1, the controlled vocabulary data 15 is used for explanation. The controlled vocabulary data 15 includes a term name column 15a, a headline column 15b, and a hypernym column 15c. Terms are set in the term name column 15a. Headings are set in the heading column 15b. Hypernym is set in the hypernym string 15c. It is also assumed that the information on the 1st to 3rd rows of the controlled vocabulary data 15 belongs to the synonym group 15-1, and the information on the 4th and 5th rows belongs to the synonym group 15-2.

In the reference technique, "soft" is presented as the post-replacement value in the same manner as the processing described with reference to FIG. The worker modifies the controlled vocabulary data 15 to the controlled vocabulary data 16 by modifying the software with reference to the presented values after replacement. The controlled vocabulary data 16 includes a term name column 16a, a headline column 16b, and a hypernym column 16c. Information on the first to third lines of the controlled vocabulary data 16 belongs to the synonym group 16-1, and information on the fourth and fifth lines belongs to the synonym group 16-2.

By correcting the controlled vocabulary data 15 to the controlled vocabulary data 16, the inconsistency A is corrected. Also, the correction cost for correcting the controlled vocabulary data 15 to the controlled vocabulary data 16 is "1".

Here, when consistency is checked between the synonym group 16-1 and the synonym group 16-2, the value corresponding to "software" included in the hypernym string 16c of the synonym group 16-2 is Inconsistency C is detected because it does not exist in the heading column of synonym group 16-1.

In the reference technique, "soft" is presented as a post-substitution value for the broader term "software" of the synonym group 16-1. The worker modifies the controlled vocabulary data 16 to the controlled vocabulary data 17 by modifying the two pieces of software with reference to the presented values after replacement. The controlled vocabulary data 17 includes a term name column 17a, a headline column 17b, and a hypernym column 17c. Information on the first to third lines of the controlled vocabulary data 17 belongs to the synonym group 17-1, and information on the fourth and fifth lines belongs to the synonym group 17-2.

By correcting the controlled vocabulary data 16 to the controlled vocabulary data 17, the inconsistency C is corrected. Also, the correction cost for correcting the controlled vocabulary data 16 to the controlled vocabulary data 17 is "2".

As described with reference to FIG. 1, in the reference technology, a total correction cost of "3" is required to correct inconsistencies in the controlled vocabulary data 15. FIG.

With the above-described reference technology, it is possible to correct inconsistencies not only within the same synonymous group but also between different synonymous groups. However, the reference technique may not be able to present a correction candidate with the lowest correction cost.

FIG. 2 is a diagram for explaining other correction candidates. In FIG. 2, the controlled vocabulary data 15 is used for explanation. The explanation regarding the controlled vocabulary data 15 is the same as the explanation given in FIG.

For example, a correction candidate for correcting the two headings "soft" included in the heading column 15b of the synonym group 15-1 to "software" is conceivable. The correction cost of this correction candidate is "2", and the controlled vocabulary data 15 is corrected to the controlled vocabulary data 18 by this correction candidate. The controlled vocabulary data 18 includes a term name column 18a, a headline column 18b, and a hypernym column 18c. Information on the 1st to 3rd rows of the controlled vocabulary data 18 belongs to the synonym group 18-1, and information on the 4th and 5th rows belongs to the synonym group 18-2. There are no inconsistencies in the controlled vocabulary data 18 .

In the reference technique described with reference to FIG. 1, the total correction cost is "3", but in the correction candidate described with reference to FIG. 2, the total correction cost is "2". In other words, if the reference technique is used for correction, the correction cost is not minimized.

Next, processing of the information processing apparatus according to the present embodiment will be described. FIG. 3 is a diagram for explaining the processing of the information processing apparatus according to the embodiment. In FIG. 3, the controlled vocabulary data 20 will be used for explanation. The controlled vocabulary data 20 includes a term name string 20a, a heading string 20b, and a hypernym string 20c. The information processing device classifies the records included in the controlled vocabulary data 20 into a plurality of synonym groups based on the terms and headings of the controlled vocabulary data 20 . The example shown in FIG. 3 shows a synonymous group 20-A and a synonymous group 20-Z, but other synonymous groups are also included. When the information processing device detects inconsistency from the controlled vocabulary data 20, it performs the following processing.

The information processing device identifies a plurality of correction candidates for each synonym group and calculates a correction cost for each of the correction candidates. In the following description, a correction candidate specified for each synonym group will be referred to as an "intra-group correction candidate" as appropriate. When the synonym group includes a plurality of headings, the information processing device extracts a correction candidate to be unified with one of the headings. When a synonym group includes a plurality of hypernyms, the information processing device extracts a correction candidate that unifies to one of the hypernyms. For example, if there are n ₁ types of headings and n ₂ types of hypernyms in one synonym group, n ₁ ×n ₂ in-group correction candidates are extracted from one synonym group. be.

Intra-group correction candidates identified from the synonym group 20-A will be described. The information processing device extracts in-group correction candidates ga1, ga2, ga3, and ga4 from the synonym group 20-A. The in-group correction candidate ga1 is "unify the item to A1 and unify the hypernym to E1". The correction cost of the in-group correction candidate ga1 is "2". The in-group correction candidate ga2 is "unify items to A1 and unify broader terms to E2". The correction cost of the in-group correction candidate ga1 is "2".

The in-group correction candidate ga3 is "unify items to A2 and unify broader words to E1". The correction cost of the in-group correction candidate ga3 is "2". The in-group correction candidate ga4 is "unify items to A2 and unify hypernyms to E2". The correction cost of the in-group correction candidate ga4 is "2".

Intra-group correction candidates identified from the synonym group 20-Z will be described. The information processing device identifies in-group correction candidates gz1, gz2, gz3, and gz4 from the synonym group 20-Z. The in-group correction candidate gz1 is "unify items to Z1 and unify broader terms to H1". The correction cost of the in-group correction candidate gz1 is "2". The in-group correction candidate gz2 is "unify the item to Z1 and unify the hypernym to H2". The correction cost of the in-group correction candidate gz2 is "2".

The in-group correction candidate gz3 is "unify items to Z2 and unify broader words to H1". The correction cost of the in-group correction candidate gz3 is "2". The in-group correction candidate gz4 is "unify items to Z2 and unify broader words to H2". The correction cost of the in-group correction candidate gz4 is "2".

Although illustration is omitted, the information processing device extracts in-group correction candidates from each synonym group by executing the above process for other synonym groups included in the controlled vocabulary data 20 as well.

Subsequently, the information processing apparatus generates correction candidates by combining the intra-group correction candidates extracted from the synonymous groups. One correction candidate includes one intra-group correction candidate of a different synonym group. In the example shown in FIG. 3, correction candidate (1) and correction candidate (2) are shown, but other correction candidates are also included. Correction candidate (1) includes in-group correction candidates ga1 and gz1, and other in-group correction candidates are omitted from illustration. The correction candidate (2) includes the intra-group correction candidates ga1 and gz2, and the other intra-group correction candidates are omitted.

The information processing device calculates the total correction cost of the correction candidate (1) by totaling the correction costs of the correction candidates within the group included in the candidate (1). Further, the information processing device determines whether or not there is inconsistency when the controlled vocabulary data 20 is corrected for the correction candidate (1).

The information processing device calculates the total correction cost of the correction candidate (2) by totaling the correction costs of the correction candidates within the group included in the candidate (2). Further, the information processing device determines whether or not there is an inconsistency when the controlled vocabulary data 20 is corrected with the correction candidate (2).

The information processing device repeats the above process for other correction candidates.

Based on the total correction cost of the plurality of correction candidates and the existence or non-existence of inconsistency, the information processing device identifies the correction candidate with the lowest total correction cost and in which no inconsistency is detected. The information processing device associates and displays the specified correction candidate and the total correction cost of this correction candidate. When specifying a plurality of correction candidates with the lowest total correction cost, the information processing device displays the plurality of correction candidates in association with the total correction cost.

As described above, the information processing apparatus according to the present embodiment extracts a plurality of correction candidates for correcting the inconsistency when there is inconsistency in the controlled vocabulary data 20. It is displayed in association with the correction cost. This allows the operator to select a correction candidate in consideration of the correction cost.

Further, the information processing device identifies and displays a correction candidate that minimizes the total correction cost, thereby presenting a correction candidate that minimizes the burden on the operator and does not include inconsistency. be able to.

Next, the configuration of the information processing apparatus according to this embodiment will be described. FIG. 4 is a functional block diagram showing the configuration of the information processing apparatus according to this embodiment. As shown in FIG. 4 , information processing apparatus 100 includes communication section 110 , input section 120 , display section 130 , storage section 140 and control section 150 .

Communication unit 110 receives various data from an external device via a network. Communication unit 110 is an example of a communication device. For example, communication unit 110 may receive controlled vocabulary data 141, which will be described later, from an external device.

The input unit 120 is an input device that inputs various types of information to the control unit 150 of the information processing device 100 . The input unit 120 corresponds to a keyboard, mouse, touch panel, or the like. The operator may operate the input unit 120 to input data regarding the controlled vocabulary data 141 .

The display unit 130 is a display device that displays information output from the control unit 150 . For example, the display unit 130 displays information on correction candidates.

The storage unit 140 has controlled vocabulary data 141 , an intra-group correction candidate table 142 , and a correction candidate table 143 . The storage unit 140 corresponds to semiconductor memory devices such as RAM (Random Access Memory) and flash memory, and storage devices such as HDD (Hard Disk Drive).

The controlled vocabulary data 141 is dictionary data that summarizes semantic relationships between a plurality of terms in order to prevent omissions in retrieval caused by ambiguity of terms, homographs, and heteromorphic synonyms. The controlled vocabulary data 141 corresponds to the controlled vocabulary data 20 described with reference to FIG.

FIG. 5 is a diagram showing an example of the data structure of controlled vocabulary data. As shown in FIG. 5, this controlled vocabulary data 50 has a term name column 50a, a headline column 50b, and a hypernym column 50c. Records of the controlled vocabulary data 141 are classified into a synonym group 141-1 and a synonym group 141-2 by the processing of the control unit 150, which will be described later.

The intra-group correction candidate table 142 is a table that holds intra-group correction candidates. The data structure of the intra-group correction candidate table 142 will be described later.

The correction candidate table 143 is a table that holds correction candidates. The data structure of the correction candidate table 143 will be described later.

The control unit 150 has a detection unit 151 , an extraction unit 152 and a display control unit 153 . The control unit 150 is implemented by hardwired logic such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an ASIC (Application Specific Integrated Circuit), and an FPGA (Field Programmable Gate Array).

The detection unit 151 is a processing unit that detects inconsistencies in the controlled vocabulary data 141 . Processing of the detection unit 151 will be described with reference to FIG. The detection unit 151 classifies the records of the controlled vocabulary data 141 into synonymous groups based on the terms in the term name column 141a and the headings in the heading column 141b of the controlled vocabulary data 141 .

For example, the detection unit 151 detects a connected component between the term “software” and the heading “soft” in the second line of the controlled vocabulary data 141 . As a result, the detection unit 151 sets records in which "software" and "software" are set in terms or headings in the same synonym group (synonym group 141-1).

The detection unit 151 detects a connected component between the term “os” and the heading “OS” in the fourth line of the controlled vocabulary data 141 . As a result, the detection unit 151 sets records in which "os" and "OS" are set in terms or headings in the same synonym group (synonym group 141-2).

After classifying the records of the controlled vocabulary data 141 into synonymous groups, the detection unit 151 determines whether or not there is inconsistency in the controlled vocabulary data 141 . In the example shown in FIG. 6, the detection unit 151 detects inconsistency A because there are two or more types of headings included in the heading string 141b of the synonym group 141-1. Detecting unit 151 outputs to extracting unit 152 that inconsistency has been detected from controlled vocabulary data 141 .

Although the description is omitted in FIG. 5, the detection unit 151 outputs to the extraction unit 152 the fact that the mismatch is detected from the controlled vocabulary data 141 even when other mismatches B to D are detected. The detection unit 151 also outputs the classification result of synonymous groups to the detection unit 151 .

The extraction unit 152 is a processing unit that, when inconsistency is detected from the controlled vocabulary data 141, extracts a plurality of correction candidates for correcting the inconsistency. The extraction unit 152 extracts an in-group correction candidate for each synonym group, and registers the extracted in-group correction candidate in the in-group correction candidate table 142 .

FIG. 6 is a diagram showing an example of the data structure of the intra-group correction candidate table. As shown in FIG. 6, it has synonymous group identification information, in-group correction candidate identification information, and correction cost. The synonymous group identification information is information for identifying a synonymous group. For example, synonym group identification information “SG1” indicates synonym group 141-1 in FIG. The synonymous group identification information “SG2” indicates the synonymous group 141-2 in FIG. The in-group correction candidate identification information is information for identifying an in-group correction candidate. The in-group correction candidate indicates the content of the in-group correction candidate. The correction cost is the correction cost of the intra-group correction candidate.

First, the extraction unit 152 extracts in-group correction candidates based on the synonym group 141-1 of the controlled vocabulary data 141 shown in FIG.

The extraction unit 152 extracts in-group correction candidates that unify the headings of the synonym group 141-1 to "soft". When the headings of the synonym group 141-1 are unified by software, there is only one correction point, so the correction cost is "1". The extraction unit 152 assigns unique intra-group correction candidate identification information to the intra-group correction candidate, and assigns the intra-group correction candidate identification information (for example, ga-1), the intra-group correction candidate “unify the heading to “soft””, A correction cost of “1” is registered in the intra-group correction candidate table 142 . In the following description, the intra-group correction candidate of the intra-group correction candidate identification information ga-1 is referred to as "in-group correction candidate ga-1".

The extraction unit 152 extracts in-group correction candidates that unify the headings of the synonym group 141-1 to "software". Since there are two points to be corrected when the headings of the synonym group 141-1 are integrated into the software, the correction cost is "2". The extraction unit 152 assigns unique in-group correction candidate identification information to the in-group correction candidate, and assigns the in-group correction candidate identification information (for example, ga-2), the in-group correction candidate 'unify the heading to 'software'', The correction cost “2” is registered in the intra-group correction candidate table 142 . In the following description, the intra-group correction candidate of the intra-group correction candidate identification information ga-2 is referred to as "in-group correction candidate ga-2".

Subsequently, the extraction unit 152 extracts in-group correction candidates based on the synonym group 141-2 of the controlled vocabulary data 141 shown in FIG. Extraction unit 152 extracts "software" from "software" as correction candidates for broader terms of synonym group 141-2 based on term names in controlled vocabulary data 141 as a whole and connected components with headings (software and software connected components). Identifies candidates for correction to .

The extraction unit 152 extracts in-group correction candidates that unify the hypernyms of the synonym group 141-2 to "soft". Since there are two points to be corrected when unifying the broader terms of the synonym group 141-2 into software, the correction cost is "2". The extraction unit 152 assigns unique in-group correction candidate identification information to the in-group correction candidate, and extracts the in-group correction candidate identification information (for example, gb-1) and the in-group correction candidate ``unify the broader terms into ``soft''''. , the correction cost “2” is registered in the intra-group correction candidate table 142 . In the following description, the intra-group correction candidate of the intra-group correction candidate identification information gb-1 is referred to as "in-group correction candidate gb-1".

The extraction unit 152 extracts in-group correction candidates that unify the hypernyms of the synonym group 141-2 to "software". Since there are 0 points to be corrected when unifying the higher terms of the synonym group 141-2 into software, the correction cost is "0". The extraction unit 152 assigns unique in-group correction candidate identification information to the in-group correction candidate, and extracts the in-group correction candidate identification information (for example, gb-2) and the in-group correction candidate ``unify the broader terms into ``software''''. , the correction cost “0” is registered in the intra-group correction candidate table 142 . In the following description, the intra-group correction candidate of the intra-group correction candidate identification information gb-2 is referred to as "in-group correction candidate gb-2".

Data is registered in the intra-group correction candidate table 142 shown in FIG. 6 by the extraction unit 152 executing the above process.

Subsequently, the extraction unit 152 generates correction candidates by combining the intra-group correction candidates included in the intra-group correction candidate table 142 , and registers them in the correction candidate table 143 .

FIG. 7 is a diagram showing an example of the data structure of a correction candidate table. As shown in FIG. 7, the correction candidate table 143 has correction candidate identification information, correction candidate configuration, total correction cost, and presence/absence of inconsistency. The correction candidate identification information is information that identifies a correction candidate. The correction candidate configuration indicates identification information of in-group correction candidates (in-group correction candidate identification information) included in the correction candidates. The total correction cost indicates the total correction cost of the correction candidates within the group included in the correction candidates. The presence or absence of inconsistency indicates whether or not there is inconsistency when the controlled vocabulary data 141 is corrected by the corresponding correction candidate. If there is an inconsistency, it will be "present", and if there is no inconsistency, it will be "absent".

The extraction unit 152 generates a correction candidate consisting of the intra-group correction candidates ga-1 and gb-1. Let "Am1" be the correction candidate identification information of the correction candidate consisting of the intra-group correction candidates ga-1 and gb-1. Further, the correction candidate of the correction candidate identification information "Am1" is written as "correction candidate Am1". As shown in FIG. 6, since the correction cost of the in-group correction candidate ga-1 is "1" and the correction cost of the in-group correction candidate gb-1 is "2", the total correction cost of the correction candidate Am1 is "3 ”.

FIG. 8 is a diagram (1) showing an example of a correction result. When the extraction unit 152 corrects the correction candidate Am1 to the controlled vocabulary data 141 shown in FIG. 5, the controlled vocabulary data 141-Am1 shown in FIG. 8 is obtained. The correction of the correction candidate Am1 includes a correction to unify the headings of the synonym group 141-1 to "soft" and a correction to unify the hypernyms of the synonym group 141-2 to "soft". There is no inconsistency in the controlled vocabulary data 141-Am1.

By executing the above process, the extraction unit 152 corrects the correction candidate identification information “Am1”, the correction candidate configuration “ga-1, gb-1”, the total correction cost “3”, and the presence/absence of inconsistency “none”. Register in the candidate table 143 .

The extraction unit 152 generates a correction candidate consisting of the intra-group correction candidates ga-1 and gb-2. Let "Am2" be the correction candidate identification information of the correction candidate consisting of the intra-group correction candidates ga-1 and gb-2. Further, the correction candidate of the correction candidate identification information "Am2" is written as "correction candidate Am2". As shown in FIG. 6, since the correction cost of the in-group correction candidate ga-1 is "1" and the correction cost of the in-group correction candidate gb-2 is "0", the total correction cost of the correction candidate Am2 is "1". ”.

FIG. 9 is a diagram (2) showing an example of the correction result. When the extraction unit 152 corrects the correction candidate Am2 to the controlled vocabulary data 141 shown in FIG. 5, the controlled vocabulary data 141-Am2 shown in FIG. 9 is obtained. The correction of the correction candidate Am2 includes a correction to unify the headings of the synonym group 141-1 to "software" and a correction (no change) to unify the hypernyms of the synonym group 141-2 to "software". In the controlled vocabulary data 141-Am2, inconsistency C is detected because the hypernym "software" of the synonym group 141-2 is different from the heading "soft" of the synonym group 141-1.

By executing the above process, the extraction unit 152 corrects the correction candidate identification information “Am2”, the correction candidate configuration “ga-1, gb-2”, the total correction cost “1”, and the presence or absence of mismatch “present”. Register in the candidate table 143 .

The extraction unit 152 generates a correction candidate consisting of the intra-group correction candidates ga-2 and gb-1. Let "Am3" be the correction candidate identification information of the correction candidate consisting of the intra-group correction candidates ga-2 and gb-1. Further, the correction candidate of the correction candidate identification information "Am3" is written as "correction candidate Am3". As shown in FIG. 6, since the correction cost of the in-group correction candidate ga-2 is "2" and the correction cost of the in-group correction candidate gb-1 is "2", the total correction cost of the correction candidate Am3 is "4 ”.

FIG. 10 is a diagram (3) showing an example of the correction result. When the extraction unit 152 corrects the correction candidate Am3 to the controlled vocabulary data 141 shown in FIG. 5, the controlled vocabulary data 141-Am3 shown in FIG. 10 is obtained. The correction of the correction candidate Am3 includes a correction to unify the headings of the synonym group 141-1 to "software" and a correction to unify the hypernyms of the synonym group 141-2 to "soft". In the controlled vocabulary data 141-Am3, inconsistency C is detected because the hypernym "soft" of the synonym group 141-2 is different from the heading "software" of the synonym group 141-1.

By executing the above process, the extraction unit 152 corrects the correction candidate identification information “Am3”, the correction candidate configuration “ga-2, gb-1”, the total correction cost “4”, and the presence or absence of mismatch “present”. Register in the candidate table 143 .

The extraction unit 152 generates a correction candidate consisting of the intra-group correction candidates ga-2 and gb-2. Let "Am4" be the correction candidate identification information of the correction candidate consisting of the intra-group correction candidates ga-2 and gb-2. Further, the correction candidate of the correction candidate identification information "Am4" is written as "correction candidate Am4". As shown in FIG. 6, since the correction cost of the in-group correction candidate ga-2 is "2" and the correction cost of the in-group correction candidate gb-2 is "0", the total correction cost of the correction candidate Am4 is "2 ”.

FIG. 11 is a diagram (4) showing an example of the correction result. When the extraction unit 152 corrects the correction candidate Am4 to the controlled vocabulary data 141 shown in FIG. 5, the controlled vocabulary data 141-Am4 shown in FIG. 11 is obtained. The correction of the correction candidate Am4 includes a correction to unify the headings of the synonym group 141-1 to "software" and a correction (no correction) to unify the broader terms of the synonym group 141-2 to "software". There is no inconsistency in the controlled vocabulary data 141-Am4.

By executing the above process, the extraction unit 152 corrects the correction candidate identification information “Am4”, the correction candidate configuration “ga-2, gb-2”, the total correction cost “2”, and the presence or absence of inconsistency “none”. Register in the candidate table 143 .

Data is registered in the correction candidate table 143 shown in FIG. 7 by the extraction unit 152 executing the above process.

Returning to the description of FIG. Based on the intra-group correction candidate table 142 and the correction candidate table 143, the display control unit 153 generates screen information in which a plurality of correction candidates and total correction costs corresponding to the plurality of correction candidates are associated with each other. Information is displayed on the display unit 130 .

For example, based on the correction candidate table 143, the display control unit 153 identifies a correction candidate with inconsistency “no” from a plurality of correction candidates, a total correction cost corresponding to this correction candidate, and a correction candidate configuration. do. The display control unit 153 identifies an intra-group correction candidate corresponding to the correction candidate based on the identified correction candidate configuration and the intra-group correction candidate table 142 .

The display control unit 153 generates screen information based on the information specified by the above process. FIG. 12 is a diagram illustrating an example of screen information generated by a display control unit; As shown in FIG. 12, the screen information 60 includes an area 60a containing information on the correction candidate Am4 and an area 60b containing information on the correction candidate Am1.

The display control unit 153 may set each correction candidate in the screen information 60 in descending order of the total correction cost, or may set only the correction candidate with the lowest total correction cost in the screen information 60 . Display control unit 153 outputs screen information 60 to display unit 130 .

The operator refers to the screen information 60 displayed on the display unit 130 , operates the input unit 120 , and corrects the controlled vocabulary data 141 . Note that the information processing apparatus 100 may automatically correct the controlled vocabulary data 141 based on correction candidates that do not contain inconsistencies, according to the operator's settings.

Next, an example of the processing procedure of the information processing apparatus 100 according to this embodiment will be described. FIG. 13 is a flow chart showing the processing procedure of the information processing apparatus according to this embodiment. As shown in FIG. 13, the detection unit 151 of the information processing device 100 acquires controlled vocabulary data 141 (step S101).

The detection unit 151 classifies the controlled vocabulary data into synonymous groups based on terms and headings (step S102). The detection unit 151 identifies whether or not there is an inconsistency based on a plurality of synonymous groups (step S103).

If the detection unit 151 does not detect any inconsistency (step S104, No), the process ends. When the detection unit 151 detects mismatch (step S104, Yes), the process proceeds to step S105.

The extraction unit 152 of the information processing apparatus 100 extracts in-group correction candidates for each synonym group, and calculates a calculation cost (step S105).

The extraction unit 152 generates a plurality of correction candidates by combining intra-group correction candidates between a plurality of synonymous groups (step S106). The extraction unit 152 identifies the presence or absence of inconsistency when the controlled vocabulary data 141 is corrected based on the correction candidates (step S107).

The extraction unit 152 calculates a total correction cost by summing the correction costs of the correction candidates within the group included in the correction candidates (step S108). The extraction unit 152 registers the information of the correction candidate in the correction candidate table 143 (step S109).

The display control unit 153 generates screen information based on correction candidates that do not have inconsistency (step S110). The display control unit 153 causes the display unit 130 to display screen information (step S111).

Next, the effects of the information processing apparatus 100 according to this embodiment will be described. When the information processing apparatus 100 detects inconsistency in the controlled vocabulary data 141, it extracts a plurality of correction candidates for correcting the inconsistency, and calculates a total correction cost for the plurality of correction candidates. The information processing apparatus 100 associates and displays a plurality of correction candidates and the total correction cost. This allows the operator to select a correction candidate in consideration of the correction cost.

The information processing apparatus 100 identifies and displays a correction candidate that minimizes the total correction cost, thereby presenting a correction candidate that minimizes the burden on the operator and that does not include inconsistency. can be done.

The information processing apparatus 100 classifies the controlled vocabulary data 141 into a plurality of synonymous groups based on the terms and headings included in the controlled vocabulary data 141, and detects inconsistencies between the plurality of synonymous groups. This makes it possible to detect a mismatch (mismatch C) in which the value set in the hypernym of the synonym group is not set in the heading of another synonym group.

The information processing apparatus 100 calculates a total correction cost by totaling the correction costs of a plurality of correction candidates within a group included in the correction candidates. This makes it possible to calculate the total correction cost of each correction candidate and to quantitatively identify the labor of the operator.

Next, an example of the hardware configuration of a computer that implements the same functions as the information processing apparatus 100 shown in the above embodiment will be described. FIG. 14 is a diagram illustrating an example of a hardware configuration of a computer that implements functions similar to those of the information processing apparatus.

As shown in FIG. 14, a computer 200 has a CPU 201 that executes various arithmetic processes, an input device 202 that receives data input from a user, and a display 203 . The computer 200 also has a communication device 204 that receives data from an external device, and an interface device 205 that connects with various devices. The computer 200 has a RAM 206 that temporarily stores various information and a hard disk device 207 . Each device 201 - 207 is then connected to a bus 208 .

The hard disk device 207 has a detection program 207a, an extraction program 207b, and a display control program 207c. The CPU 201 reads out the extraction program 207b and the display control program 207c and develops them in the RAM 206. FIG.

Detection program 207a functions as detection process 206a. Extraction program 207b functions as extraction process 206b. The display control program 207c functions as a display control process 206c.

The processing of the detection process 206 a corresponds to the processing of the detection unit 151 . The processing of the extraction process 206 b corresponds to the processing of the extraction unit 152 . Processing of the display control process 206 c corresponds to processing of the display control unit 153 .

Note that the programs 207a to 207c do not necessarily have to be stored in the hard disk device 207 from the beginning. For example, each program is stored in a “portable physical medium” such as a flexible disk (FD), CD-ROM, DVD disk, magneto-optical disk, IC card, etc. inserted into the computer 200 . Then, the computer 200 may read and execute each program 207a to 207c.

The following additional remarks are disclosed regarding the embodiments including the above examples.

(Appendix 1) A computer-executed data modification method comprising:
detect inconsistencies between data in data tables,
extracting a plurality of correction candidates for correcting the inconsistency when the inconsistency is detected;
calculating a correction cost corresponding to each of the plurality of correction candidates;
A data correction method, comprising: displaying the plurality of correction candidates and correction costs corresponding to the plurality of correction candidates in association with each other.

(Supplementary Note 2) A process of identifying a correction candidate with the lowest correction cost among the correction costs corresponding to the plurality of correction candidates is further executed, and the display process includes selecting the correction candidate with the lowest correction cost. The data correction method according to appendix 1, characterized by displaying.

(Appendix 3) The detecting process classifies the records of the data table into a plurality of synonymous groups based on terms and headings contained in the data table, and detects inconsistencies among the plurality of synonymous groups. The data correction method according to appendix 1 or 2, characterized by:

(Appendix 4) The extracting process extracts a plurality of correction candidates for the plurality of synonymous groups, and the calculating process calculates a correction cost for a combination of the correction candidates of the plurality of synonymous groups. A data correction method according to appendix 3.

(Appendix 5) to the computer,
detect inconsistencies between data in data tables,
extracting a plurality of correction candidates for correcting the inconsistency when the inconsistency is detected;
calculating a correction cost corresponding to each of the plurality of correction candidates;
A data correction program for executing a process of displaying the plurality of correction candidates and correction costs corresponding to the plurality of correction candidates in association with each other.

(Supplementary Note 6) A process of identifying a correction candidate with the lowest correction cost among the correction costs corresponding to the plurality of correction candidates is further executed, and the display process includes selecting the correction candidate with the lowest correction cost. The data correction program according to appendix 5, characterized by displaying.

(Appendix 7) The detecting process classifies the records of the data table into a plurality of synonymous groups based on terms and headings included in the data table, and detects inconsistencies among the plurality of synonymous groups. The data correction program according to appendix 5 or 6, characterized by:

(Appendix 8) The extracting process extracts a plurality of correction candidates for the plurality of synonymous groups, and the calculating process calculates a correction cost for a combination of the correction candidates of the plurality of synonymous groups. A data correction program according to appendix 7, characterized in that:

100 information processing device 110 communication unit 120 input unit 130 display unit 140 storage unit 141 controlled vocabulary data 142 in-group correction candidate table 143 correction candidate table 150 control unit 151 detection unit 152 extraction unit 153 display control unit

Claims (5)

A computer implemented data modification method comprising:
detect inconsistencies between data in data tables,
extracting a plurality of correction candidates for correcting the inconsistency when the inconsistency is detected;
calculating a correction cost corresponding to each of the plurality of correction candidates;
A data correction method, comprising: displaying the plurality of correction candidates and correction costs corresponding to the plurality of correction candidates in association with each other. Further executing a process of identifying a correction candidate with the lowest correction cost among the correction costs corresponding to the plurality of correction candidates, and the displaying process includes displaying the correction candidate with the lowest correction cost. 2. The data correction method according to claim 1. The detecting process classifies the records of the data table into a plurality of synonymous groups based on terms and headings contained in the data table, and detects inconsistencies among the plurality of synonymous groups. 3. The data correction method according to claim 1 or 2. The extracting process extracts a plurality of correction candidates for each of the plurality of synonymous groups, and the calculating process calculates a correction cost for a combination of the correction candidates of the plurality of synonymous groups. Item 3. The data correction method according to item 3. to the computer,
detect inconsistencies between data in data tables,
extracting a plurality of correction candidates for correcting the inconsistency when the inconsistency is detected;
calculating a correction cost corresponding to each of the plurality of correction candidates;
A data correction program for executing a process of displaying the plurality of correction candidates and correction costs corresponding to the plurality of correction candidates in association with each other.

Images