JP7234763B2 - SAME EVENT DETERMINATION PROGRAM, SAME EVENT DETERMINATION METHOD AND SAME EVENT DETERMINATION SYSTEM - Google Patents

Description

The present invention relates to a same event determination program, a same event determination method, and a same event determination system for determining whether or not different words are the same event.

A plurality of organizations have databases in which similar types of information are stored. For example, when integrating the databases of multiple organizations, it is possible to accurately check whether the words written in the master of the database of one organization indicate the same phenomenon (things, actions, etc.) as the words of the master of another organization. It is necessary to judge. For example, an electronic medical record used in a hospital is a database in which the same medical information is stored in multiple hospitals. Here, if the drug (drug name) described in the electronic medical record of the hospital 1 is different from the drug of the other hospital 2, it will be necessary to determine whether or not it is the same drug at the time of database integration.

As a conventional technique, there is a technique for estimating synonyms. In the estimation of synonyms, for example, words in which many common characters are used are estimated as synonyms. In addition, there is a technology called word2vec. This technique focuses on the fact that synonyms are used similarly in a sentence. Specifically, if words before and after appearing in sentences are the same, they are regarded as synonyms.

As a technique related to synonym extraction, a technique has been disclosed in which synonym candidates are extracted from a group of documents based on similarities such as co-occurrence relationships, and non-fixed phrases in the extracted fixed phrases are excluded from the synonym candidates (for example, See Patent Document 1 below.). In addition, short-term sessions of a certain word are extracted from the session information of the browsed product, and among the extracted words, synonym candidates whose co-occurrence frequency is equal to or higher than a predetermined value and whose word IDF (word importance) is equal to or lower than a threshold (For example, see Patent Document 2 below.).

JP 2014-132406 A JP 2013-164751 A

In the prior art, words that are used in the same way are judged to have the same meaning. However, even with the same usage, the actual phenomenon may differ. For example, assume that a certain hospital 1 prescribes a drug A to a patient diagnosed with hay fever, while another hospital 2 prescribes a drug B. FIG. In this example, the drugs A and B are used in common in that they are "prescribed to hay fever patients", but they are different drugs. Here, in the conventional technology, when the product names of the drug A and the drug B are similar, for example, a plurality of common characters are included, it is determined that the drugs are highly likely to be the same drug. When the database is integrated using such extraction results as they are, in the integrated database, drug A and drug B, which are actually different drugs, are registered as the same drug "prescribed for hay fever patients". problem arises.

In one aspect, an object of the present invention is to be able to accurately determine whether two different words are the same event.

According to one aspect of the present invention, each of the master databases owned by a plurality of organizations is accessed, and whether two arbitrary words exist in one of the master databases is checked by each of the master databases. Databases are collated, based on the results of the collation for each of the master databases , a same event index value indicating whether the two words are the same event is calculated, and the calculated same event index value is It is a requirement to record in the same storage unit.

According to one aspect of the present invention, it is possible to accurately determine whether or not two different words are the same event.

FIG. 1 is a diagram for explaining the outline of the identical event determination processing according to the present invention. FIG. 2 is a diagram of a hardware configuration example of the identical event determination device according to the embodiment; FIG. 3 is a block diagram of a configuration example 1 of the identical event determination device according to the embodiment. FIG. 4 is a block diagram of a configuration example 2 of the identical event determination device according to the embodiment. FIG. 5 is a flowchart showing a same event determination example 1 by the same event determination device according to the embodiment. FIG. 6 is a diagram for explaining a specific example 1 of data processing by the identical event determination device according to the embodiment; FIG. 7 is a flowchart of a same event determination example 2 by the same event determination device according to the embodiment. FIG. 8 is a diagram for explaining a specific example 2 of data processing by the identical event determination device according to the embodiment; FIG. 9 is a flowchart showing a same event determination example 3 by the same event determination device according to the embodiment. FIG. 10 is a diagram for explaining a specific example 3 of data processing by the identical event determination device according to the embodiment; FIG. 11 is a diagram for explaining a specific example 4 of data processing by the identical event determination device according to the embodiment;

Exemplary embodiments of the disclosed same-event determination program, same-event determination method, and same-event determination system will be described in detail below with reference to the drawings.

FIG. 1 is a diagram for explaining the outline of the identical event determination processing according to the present invention. As shown in FIG. 1, different organizations 1 to n have databases in which similar types of information are accumulated. For example, each organization 1-n has a master 1 of similar types of information. A master is static information that you record in a database. Similarly, each organization 1-n has a database of masters 2 of similar types of information. In this way, an organization has a master database of similar types of information, and an organization is a given collection, eg, a company, an institution, or the like.

The same-event determination system is configured such that the same-event determination device 100 accesses the master database of each organization 1 to n. For example, the identical event determination device 100 communicates with the servers of the organizations 1 to n and accesses the master database.

The identical event determination device 100, for example, accurately determines whether a word written in a master of a certain organization indicates the same event (object, action, etc.) as a word of a master of another organization. For example, the same-event determination device 100 determines whether word X described in master 1 of a certain organization 1 is the same event as word Y in masters of organizations 2 to n for two different words X and Y. do.

For example, if the organization is a hospital, master 1 is information on "medicine" used in electronic medical records used in each hospital, and master 2 is information on "disease name" used in electronic medical records used in each hospital. . An electronic medical record is a database in which the same medical information is accumulated in each of a plurality of hospitals (organizations 1 to n).

For example, the same event determination device 100 of the embodiment accesses the "medicine" master 1 of each organization 1 to n when performing the same event determination regarding "medicine". Then, the same event determination device 100 determines that the “medicine X” described in the master 1 of the organization 1 “hospital 1” and the “medicine Y” described in the master 1 of the organization 2 “hospital 2” are the same medicines. , and outputs the same event index as the determination result.

In the embodiment, synonyms are determined with high accuracy by utilizing the fact that the database master has predetermined characteristics. That is, in any organization 1-n, the same event (eg, drug) is not simultaneously described as separate words in the master within the organization (probability is low). This characteristic is due to the fact that databases are generally normalized (elimination of data duplication and consistent handling of data). In this case the master does not keep the same event as separate words. The same event determination device 100 focuses on such characteristics to determine whether two different words are the same event as follows.

1. If two different words are the same event, the two words will not appear at the same time in any organization's master (probability is low).
2. If two different words are not the same event, there is a high possibility that some organization has a master in which the two words are described at the same time.

For example, an organization 1 "hospital 1" prescribes "medicine A" to a patient diagnosed with hay fever, while an organization 2 "hospital 2" prescribes "medicine B". In such a case, hospitals 1 and 2 are common in that medicines are "prescribed to patients with hay fever", but "medicine A" and "medicine B" are different medicines. Here, in the conventional technology, it is determined that there is a high probability that "drug A" and "drug B" are the same drug.

On the other hand, the identical event determination device 100 of the embodiment determines whether two words X "drug A" and word Y "drug B" are the same event. For example, a "same event index" indicating whether or not two words X and Y of "medicine A" and "medicine B" are the same event is calculated. The “same event index” is an index that indicates whether or not two different words indicate the same event (object, action, etc.). According to the embodiment, the "same event index" indicates that the master 1 of an organization "hospital" has different notations of "drug A" and "drug B". Although two words X and Y are input from the outside in FIG.

In this way, by using the same-event determination device 100 of the embodiment, the utilization of data can be expanded, and for example, the data of different organizations 1 to n can be efficiently integrated. For example, it will be possible to integrate the masters (databases) of hospitals 1 to n, which are numerous throughout the country, to develop insurance products, improve the efficiency of pharmaceutical processes, develop diagnostic support AI, and the like.

Here, in an electronic medical record system within a certain organization 1 "Hospital 1", there is a tendency to use words that are easy to use in "Hospital 1". Therefore, even for the same event, words with different notations are often used in “Hospital 2” to “Hospital n”, which are different from “Hospital 1”. For example, "Hospital 1" manages events with formal names, while other "Hospital 2" manages events with abbreviated names. Therefore, in order to integrate and utilize the data of different "hospital 1" to "hospital n", it is necessary to align the notations so that "same event=same notation". When integrating this data, it is necessary to prepare a dictionary (corresponding to a synonym table) for judging whether or not words with different notations represent the same event. Since it takes a lot of man-hours to create this dictionary, there is a demand to automate it as much as possible. Therefore, there is a demand to automatically determine whether certain words are the same phenomenon with high accuracy.

In the embodiment, for example, for the purpose of integrating master data, it is automatically determined with high accuracy whether certain words are the same event. By using this same event index, it becomes possible to efficiently integrate data of different organizations 1 "hospital 1" to organization n "hospital n".

In order to make the above determination, the identical event determination device 100 has a master matching unit 101 that determines whether two words are written simultaneously in the master of each organization, and the matching result of the master matching unit 101 (whether or not it is written ), and calculates a “same event index” indicating whether two words are the same event.

The master collation unit 101 determines whether the word X "medicine A" and the word Y "medicine B" are described simultaneously in the master 1 of the organization 1 "hospital 1" (collation 1). Similarly, it is determined whether the word X "medicine A" and the word Y "medicine B" are simultaneously described in the master 1 of the organization 2 "hospital 2" (collation 2). The master collation unit 101 performs similar processing to determine whether or not the word X “drug A” and the word Y “drug B” are simultaneously described in the master 1 of the organization n “hospital n” (verification n ).

The same event index calculation unit 102 calculates a “same event index” indicating whether or not two words are the same event based on the N matching results “match 1” to “match n” collated by the master collation unit 101. do. At this time, the same event index calculation unit 102 calculates the number of masters in which words X and Y exist simultaneously (both_use) and the number of organizations in which either word X or Y exists (Share).

Further, the same-event index calculation unit 102 calculates the same-event index value based on only the process A, or the process B or the process C below. Processing A is a basic processing for calculating the same event index value. Processing B and processing C are advanced forms of the processing content of processing A, and the same event index calculation unit 102 performs processing B or processing C instead of processing A. FIG.

In process A, if the N matching results "matching 1" to "matching n" of the master matching unit 101 do not contain two words at the same time in all organizations, the value "1" is given as the same event index. , otherwise it outputs the value "0".

In process B, based on the N matching results “matching 1” to “matching n” of the master matching unit 101, as the number of tissues in which both two words exist in the same master increases, a small same event index is calculated. calculate. For example, the same event index=1−(number of organizations where both words exist in the same master)/(number of all organizations) is calculated. Here, in the above process A, if there is even one organization that describes the same event as a different word in the master, there is a case where the determination is erroneous, but according to the process B, this erroneous determination can be prevented. For example, it is possible to deal with a case where some organizations have duplicated management of words with the same meaning due to historical circumstances (system migration) or the like.

In process C, considering the number of master registrations of two words, as the number of organizations in which both of the two words exist in the same master increases, the number of organizations in which either of the two words exists in the master increases. is smaller, a smaller same-event index is calculated. For example, the same event index=1−(the number of organizations in which both two words exist in the same master)/(the number of organizations in which either of the two words exists in the master) is calculated. In this case, when there are few organizations using the two words X and Y, it is possible to prevent the same event index from becoming high.

Note that the master matching unit 101 of the identical event determination device 100 determines that the two words X and Y are used in a similar manner using, for example, an existing technique (synonym search, etc.). It is also possible to use the words X and Y as

Here, when inputting the word similarity of two words X and Y to the identical event index calculation unit 102, the following processes A to C are performed.

In process A, if the N matching results “matching 1” to “matching n” of the master matching unit 101 do not include two words at the same time in all organizations, the same event index, for example, the word Output the value of the similarity, otherwise output the value "0". A value of "0" indicates that any organization contains two words X and Y at the same time.

In process B, based on the N matching results “matching 1” to “matching n” of the master matching unit 101, for example, same event index=word similarity×(1−(two words both exist The number of tissues that were treated)/(the number of all tissues)).

In process C, considering the number of master registrations of two words, for example, same event index = word similarity × (1 - (number of organizations where both two words existed in the same master) / (two words number of organizations for which any exist in the master).

FIG. 2 is a diagram showing a hardware configuration example of the same-event determination device of the present invention. The same-event determination device 100 can be configured by, for example, a general-purpose server made up of hardware shown in FIG.

The same event determination device 100 includes a CPU (Central Processing Unit) 201 , a memory 202 , a network interface (IF) 203 , a recording medium IF 204 and a recording medium 205 . A bus 200 connects each unit.

The CPU 201 is an arithmetic processing unit that functions as a control unit that controls the entire same-event determination device 100 . Memory 202 includes non-volatile memory and volatile memory. The nonvolatile memory is, for example, a ROM (Read Only Memory) that stores programs for the CPU 201 . The volatile memory is, for example, a DRAM (Dynamic Random Access Memory), an SRAM (Static Random Access Memory), etc. used as a work area for the CPU 201 .

A network IF 203 is a communication interface for a network 210 such as a LAN (Local Area Network), a WAN (Wide Area Network), and the Internet. Identical event determination device 100 is connected for communication to network 210 via network IF 203 . For example, the same-event determination device 100 accesses a server of an organization (hospital) holding a target master database via the network 210 .

The recording medium IF 204 is an interface for reading and writing information processed by the CPU 201 with the recording medium 205 . A recording medium 205 is a recording device that assists the memory 202, and may be a HDD (Hard Disk Drive), an SSD (Solid State Drive), a USB (Universal Serial Bus) flash drive, or the like.

The CPU 201 executes the program recorded in the memory 202 or the recording medium 205 to realize each function (master collation unit 101, same event index calculation unit 102) of the same event determination device 100 shown in FIG. Also, the memory 202 and the recording medium 205 record and hold information handled by the same event determination apparatus 100 .

FIG. 3 is a block diagram of a configuration example 1 of the identical event determination device according to the embodiment. The same-event determination device 100 shown in FIG. 3 includes the functions of the master matching unit 101 and the same-event index calculation unit 102 shown in FIG.

The master collating unit 101 determines whether the two input words X and Y are simultaneously described in the masters (for example, master 1) of each of the organizations 1 to n. Then, the same event index calculation unit 102 calculates the number of masters (both_use) in which the words X and Y exist at the same time. Then, by executing the process A, the same event index "0/1" indicating whether or not the two words are the same event is obtained from the N matching results (whether or not they are described) of each master matching unit. ,Output.

FIG. 4 is a block diagram of a configuration example 2 of the identical event determination device according to the embodiment. The same event determination device 100 shown in FIG. 4 has the function of a similar word extraction unit 401 in addition to the master matching unit 101 and the same event index calculation unit 102 having the same configuration as in FIG.

The similar word extraction unit 401 has a function of extracting candidates for a pair of words X and Y representing the same event at least from the word set and outputting the words X and Y to the master matching unit 101 . At this time, a combination of words X and Y is calculated from a set of words input from the outside.

Further, the similar word extraction unit 401 may calculate the degree of similarity between different words X and Y, and extract only the words X and Y with high degree of similarity. This eliminates the need for processing in the master collation unit 101 and the same event index calculation unit 102 for combinations of words X and Y that are unlikely to represent the same event, thereby improving the efficiency of the processing of the entire apparatus. become. For calculating the degree of similarity between the words X and Y, for example, the Levenshtein distance may be used, and the reciprocal of the Levenshtein distance may be used as the degree of similarity.

In addition to the words themselves, the similar word extraction unit 401 may also acquire feature amounts related to words (for example, sentences containing the words X and Y), etc., and use these feature amounts to calculate the degree of similarity. .

Also, in the above description, an example of extracting two different words X and Y has been described, but a set of two or more words may be used to calculate the same event index. In this case, the similar word extraction unit 401 extracts a set of multiple words. In this case, the master collation unit 101 may calculate the collation result×N for each pair, and the same event index calculation unit 102 may calculate the same event index for each pair. In this case, the same-event index calculation unit 102 calculates the same-event index in consideration of word similarity as well.

Also, the set of words to be externally input and the feature values related to the words may be acquired from the N "organizations 1 to n" holding masters and integrated. Also, it may be acquired regardless of the N "organizations 1 to n" holding the master. For example, sentences obtained from papers published on the Internet may be used. Furthermore, it is not limited to sentences, and may be time-series data. By using these sentences and time-series data, it is possible to acquire words that are used at the same time and situations that occur at the same time as or before and after the words as feature quantities (for example, time period). As a result, the similar word extraction unit 401 can accurately obtain words with "similar usage".

(Each judgment example of the same event)
Next, examples of determination of the same event for two words X and Y performed by the same event determination device 100 will be described. The control unit (CPU 201) of the same-event determination device 100 executes the same-event determination process by executing a program. At this time, the control unit inputs the two words X and Y for which the same event index is to be calculated to the master matching unit 101 based on the configuration example 1 or the configuration example 2 for the two words X and Y. FIG. For example, in the case of configuration example 1, the two words X and Y for which the same event index is calculated are manually input to the master matching unit 101 by user operation, for example, and in the case of configuration example 2, the similar word extraction unit 401 extracts and inputs to the master matching unit 101 . Then, the control unit performs the same event index calculation processing described in the above processing A to processing C for the two extracted words X and Y. FIG.

FIG. 5 is a flowchart showing a same event determination example 1 by the same event determination device according to the embodiment. First, the control unit (master collation unit 101) waits for input of a word pair (two words X and Y) (step S501: No loop). When a set of words is input (step S501: Yes), the control unit determines whether or not all necessary masters have been checked (step S502). This collation corresponds to the collation 1-n against the masters 1-n of the organizations 1-n shown in FIG.

If collation of all required masters has not been completed (step S502: No), the control unit acquires uncollated masters (step S503), and verifies whether word X and word Y exist simultaneously in the acquired masters. (step S504), and the process returns to step S502.

Then, when collation of all masters is completed (step S502: Yes), the control unit (same event index calculation unit 102) calculates the number of masters (both_use) in which words X and Y exist simultaneously (step S505).

In addition, the control unit (same event index calculation unit 102) calculates the same event index by executing the process A, and if there is no master number (both_use) in which the words X and Y exist simultaneously (when 0), the same event index "1" is output as the event index, otherwise "0" is output (step S506), and the above processing is terminated.

FIG. 6 is a diagram for explaining a specific example 1 of data processing by the identical event determination device according to the embodiment; A specific data processing example corresponding to the same event determination example 1 shown in FIG. 5 is shown. In FIG. 6, words X and Y for which the same event index is to be calculated are medicines according to configuration example 1 or 2, word X is "piperazine adipate" and word Y is "piperazine adipate". These word X "piperazine adipate" and word Y "piperazine adipate" are drugs that are similar to each other (used similarly).

Based on the input words X and Y, the master collating unit 101 of the identical event determination device 100 accesses the drug masters 600 of "hospitals 1 to n" corresponding to the respective organizations 1 to n. For example, it is assumed that words such as "piperazine adipate", "atropine sulfate hydrate", . . .

Then, the master collation unit 101 performs a collation process on whether or not both the word X “piperazine adipate” and the word Y “piperazine adipate” exist in the respective masters 600 of “hospitals 1 to n”, and performs collation. A result 601 is obtained. In this collation result 601, both the words X and Y do not exist in the master 600a of "Hospital 1" (symbol: x), and both the words X and Y exist in the master 600b of "Hospital 2". Not (symbol: x) is indicated.

Then, the same event index calculation unit 102 calculates the same event index calculation result 602 for the words X and Y based on the collation result 601 of the master collation unit 101 . In the example of FIG. 6, the same event index calculation unit 102 calculates that the number of organizations in which words X and Y are simultaneously used (both_use) is "0".

Further, the same-event index calculation unit 102 calculates the same-event index value “1” by executing the processing A. FIG. The same event determination device 100 (control unit) externally outputs the same event index calculation result 602 calculated by the same event index calculation unit 102 .

According to the above process, as the same event index calculation result 602, the number of masters (both_use) in which words X and Y exist simultaneously and the same event index value can be obtained. In the above example, out of all the masters 600 of "hospitals 1 to n", the words X and Y do not exist in all the hospital masters 600 at the same time.

In addition, as the same event index calculation result 602, the fact that the condition that "the two words X and Y are not included at the same time in all hospitals 1 to n" is satisfied is indicated by the same event index value "1". there is That is, for the same event index value "1", it can be clearly shown that the two words are not included in all hospitals 1-n at the same time. For example, when the collation result 601 is all hospital A=x, hospital B=x, and hospital C=x, the same event index value becomes "1" for the first time. If even one of the hospitals A to C has ◯ in the collation result, the same event index value will be "0". With this same event index calculation result 602, for example, when creating a common master that integrates all the masters of "Hospital 1 to n", or when creating a common dictionary between masters, it is possible to clearly present notes on words. become.

FIG. 7 is a flowchart of a same event determination example 2 by the same event determination device according to the embodiment. First, the control unit (master collation unit 101) waits for input of a word pair (two words X and Y) (step S701: No loop). When a set of words is input (step S701: Yes), the control unit determines whether all necessary masters have been checked (step S702).

If collation of all necessary masters has not been completed (step S702: No), the control unit acquires uncollated masters (step S703), and verifies whether word X and word Y exist simultaneously in the acquired masters. is performed (step S704), and the process returns to step S702.

Then, when collation of all masters is completed (step S702: Yes), the control unit (same event index calculation unit 102) calculates the number of masters (both_use) in which words X and Y exist simultaneously (step S705).

In addition, the control unit (same event index calculation unit 102) calculates the same event index as 1-(number of organizations in which both words exist in the same master)/(number of all organizations) (step S706), and the above processing ends.

FIG. 8 is a diagram for explaining a specific example 2 of data processing by the identical event determination device according to the embodiment; A specific data processing example corresponding to the same event determination example 2 shown in FIG. 7 is shown. FIG. 8 also shows an example in which the word X for which the same event index is to be calculated is "piperazine adipate" and the word Y is "piperazine adipate" according to configuration example 1 or 2.

Based on the input words X and Y, the master collating unit 101 of the identical event determination device 100 accesses the drug masters 600 of "hospitals 1 to n" corresponding to the respective organizations 1 to n. Then, the master collation unit 101 performs a collation process on whether or not both the word X “piperazine adipate” and the word Y “piperazine adipate” exist in the respective masters 600 of “hospitals 1 to n”, and performs collation. A result 601 is obtained. In the matching result 601, the master 600a of "Hospital 1" does not contain both the words X and Y (symbol: x), and the master 600b of "Hospital 2" contains both the words X and Y. (symbol: 〇) is indicated.

Then, the same event index calculation unit 102 calculates the same event index calculation result 602 for the words X and Y based on the collation result of the master collation unit 101 . In the example of FIG. 8, the same event index calculation unit 102 calculates that the number of organizations in which words X and Y are simultaneously used (both_use) is "1".

In addition, the same-event-index calculation unit 102 calculates the same-event-index value by performing process B based on 1-(the number of organizations in which both words exist in the same master)/(the number of all organizations). 0.95". The same event index value was calculated when N (total number of masters)=20. The same event determination device 100 (control unit) externally outputs the same event index calculation result 602 calculated by the same event index calculation unit 102 .

According to the above processing, as the same event index calculation result 602, the number of masters (both_use) in which the words X and Y exist simultaneously and the same event index can be obtained. In the above example, it can be shown that the words X and Y are present at the same time in the master 600b of some hospitals (Hospital 2) among all the masters 600 of "Hospital 1 to n".

In addition, as the same event index calculation result 602, the ratio that does not satisfy the condition that “two words X and Y are not included in all hospitals 1 to n at the same time” is calculated with the same event index value “0.95”. clearly show. The closer the value is to "1", the less it satisfies the condition "all hospitals 1 to n do not contain two words X and Y at the same time" ("all hospitals 1 to n contain two words X and Y are included at the same time”). By finely calculating the same event index value by processing B in this way, the same event index of all the masters 600 ("two words X and Y are not included in all hospitals 1 to n at the same time") can be more accurately calculated. be able to present it in detail.

FIG. 9 is a flowchart showing a same event determination example 3 by the same event determination device according to the embodiment. First, the control unit (master collation unit 101) waits for input of a word pair (two words X, Y) (step S901: No loop). When a set of words is input (step S901: Yes), the control unit determines whether all necessary masters have been checked (step S902).

If collation of all necessary masters has not been completed (step S902: No), the control unit acquires uncollated masters (step S903). Then, the control unit checks whether either word X or word Y exists in one master acquired this time (step S904), and whether word X and word Y exist at the same time in the master acquired this time. Collation is performed (step S905), and the process returns to step S902.

Then, when collation of all masters is completed (step S902: Yes), the control unit (same event index calculation unit 102) calculates the number of tissues (Share) in which either word X or Y exists ( step S906). Also, the number of masters (both_use) in which words X and Y exist simultaneously is calculated (step S907).

Then, the control unit (same event index calculation unit 102), by executing process B, calculates the same event index as 1-(the number of organizations in which both two words exist in the same master)/(the number of all organizations) Calculated based on In addition, the control unit (same event index calculation unit 102), by executing the process C, sets the same event index = 1 - (the number of organizations in which both of the two words exist in the same master) / (either of the two words number of tissues existing in the master) is calculated (step S908). With the above, the control unit terminates a series of processes.

FIG. 10 is a diagram for explaining a specific example 3 of data processing by the identical event determination device according to the embodiment; A specific data processing example 1 corresponding to the same event determination example 3 shown in FIG. 9 is shown. FIG. 10 also shows an example in which the word X for which the same event index is to be calculated is "piperazine adipate" and the word Y is "piperazine adipate" according to configuration example 1 or 2. FIG.

Based on the input words X and Y, the master collating unit 101 of the identical event determination device 100 accesses the drug masters 600 of "hospitals 1 to n" corresponding to the respective organizations 1 to n. Then, the master collation unit 101 obtains a collation result 601a by collating whether both the word X “piperazine adipate” and the word Y “piperazine adipate” exist simultaneously in one acquired master. In the collation result 601a, both the words X and Y do not exist at the same time in the master 600a of "Hospital 1" (symbol: x), and both the words X and Y exist in the master 600b of "Hospital 2". It is shown that they exist at the same time (symbol: ◯).

In addition, the master collation unit 101 obtains a collation result 601b by collating whether at least one of the word X “piperazine adipate” and the word Y “piperazine adipate” exists in the acquired master at the same time. In the collation result 601b, at least one of the words X and Y exists in the master 600a of "Hospital 1" (symbol: ◯), and at least one of the words X and Y exists in the master 600b of "Hospital 2". exists (symbol: 〇).

Then, the same event index calculation unit 102 calculates the same event index calculation result 602 for the words X and Y based on the matching results 601 a and 601 b of the master matching unit 101 . In the case of the example of FIG. 10, the same-event-index calculation unit 102 calculates the number of organizations (Share) that use either one of the words X and Y as "13" by performing the process C based on the collation result 601b. . Also, by performing the process B based on the collation result 601a, the number of concurrently using organizations (both_use) that simultaneously uses the words X and Y is calculated to be "1". In addition, the same event index calculation unit 102 calculates, as the same event index calculation result 602, the percentage of cases that do not satisfy the condition that “all the hospitals 1 to n do not contain the two words X and Y at the same time”. An index value of "0.92" is calculated. Then, the same event determination device 100 (control unit) outputs the same event index calculation result 602 calculated by the same event index calculation unit 102 to the outside.

According to the above process, as the same event index calculation result 602, the number of organizations using either word X or Y (Share), the number of masters in which words X and Y exist at the same time (both_use), and the same event You can get the index. In the above example, out of the masters 600 of all "hospitals 1 to n", the masters 600b of some hospitals (hospital 2) show that the words X and Y are present at the same time. It also shows that at least one of the words X and Y exists in the masters 600a and 600b of some hospitals (hospitals 1 and 2).

In addition, the same event index calculation unit 102 calculates, as the same event index calculation result 602, the percentage of cases that do not satisfy the condition that “all the hospitals 1 to n do not contain the two words X and Y at the same time”. This is clearly indicated by the index value "0.92". By finely calculating the same event index value by the processing B and the processing C in this way, the same event index of all the masters 600 ("two words X and Y are not included at the same time in all hospitals 1 to n") ) can be presented in more detail.

FIG. 11 is a diagram for explaining a specific example 4 of data processing by the identical event determination device according to the embodiment; A specific data processing example 2 corresponding to the same event determination example 3 shown in FIG. 9 is shown. Data processing example 1 shown in FIG. 10 described above was an input of a set (two words X, Y). On the other hand, in the data processing example 2 shown in FIG. 11, the similar word extraction unit 401 described in the configuration example 2 is provided, and the similar word extraction unit 401 extracts a plurality of pairs of words X and Y from a large number of word sets. It is configured to obtain and output to the master matching unit 101 .

In FIG. 11, the word set input to the same event determination apparatus 100 is the same drug with different notation, or a set of different drugs. The similar word extracting unit 401 obtains the degree of similarity between arbitrary words X and Y (obtained by round-robin, for example) from this set of words. In the example of FIG. 11, the degree of similarity between a pair of word X “piperazine adipate” and word Y “piperazine adipate” is calculated as 0.1. The degree of similarity of another set of word X "piperazine adipate" and word Y "atropine sulfate hydrate" is calculated as 0.1. The similarity between a pair of word X "atropine sulfate hydrate" and word Y "atropine sulfate" is calculated as 0.333.

The master matching unit 101 of the identical event determination device 100 compares each set of words X and Y extracted by the similar word extraction unit 401 to the drug master 600 of “hospitals 1 to n” corresponding to each organization 1 to n. to access.

Then, the master collation unit 101 obtains a collation result 601a by collating whether a pair of the word X “piperazine adipate” and the word Y “piperazine adipate” are simultaneously present in one acquired master. The collation result 601a shows that both the word X "piperazine adipate" and the word Y "piperazine adipate" do not exist simultaneously in the master 600a of "Hospital 1" (symbol: x). there is In addition, the master 600b of "Hospital 2" shows that both the word X "piperazine adipate" and the word Y "piperazine adipate" exist at the same time (symbol: ◯).

In addition, the master collation unit 101 collates whether or not another pair of the word X “piperazine adipate” and the word Y “atropine sulfate hydrate” exist at the same time in one acquired master. demand. In the collation result 601a, both the word X "piperazine adipate" and the word Y "atropine sulfate hydrate" exist simultaneously in the master 600a of "Hospital 1" and the master 600b of "Hospital 2" ( Symbol: ○) is indicated. Thereafter, similarly, the master collation unit 101 obtains a collation result 601a by collating whether the other pair of words X and Y both exist in the acquired master at the same time.

In addition, the master collation unit 101 obtains a collation result 601b by collating whether at least one of the set of words X “piperazine adipate” and word Y “piperazine adipate” exists in one acquired master. In the matching result 601b, at least one of the word X "piperazine adipate" and the word Y "piperazine adipate" exists in the master 600a of "Hospital 1" and the master 600b of "Hospital 2" ( Symbol: ○) is indicated.

In addition, the master collation unit 101 collates whether or not at least one of the word X “piperazine adipate” and the word Y “atropine sulfate hydrate” in another set of words exists in the acquired master, and returns the collation result 601b. demand. In the matching result 601b, at least one of the word X "piperazine adipate" and the word Y "atropine sulfate hydrate" exists in the master 600a of "Hospital 1" and the master 600b of "Hospital 2". (symbol: 〇). In the same way, the master matching unit 101 obtains a matching result 601b by checking whether at least one of the other pair of words X and Y exists in the acquired one master.

Then, the same event index calculation unit 102 calculates a same event index calculation result 602 for each pair of words X and Y based on the matching results 601 a and 601 b of the master matching unit 101 . In the case of the example of FIG. 11, the same-event-index calculation unit 102 performs the processing C based on the collation result 601b to determine the number of organizations using either one of the words X and Y for each pair of words X and Y. (Share) is calculated. For example, for a set of word X “piperazine adipate” and word Y “piperazine adipate”, the number of organizations using either word X or Y (Share) is calculated as “13”.

Also, the number of simultaneously using organizations (both_use) that use the words X and Y at the same time is calculated by executing the process B based on the collation result 601a. For example, for a set of word X “piperazine adipate” and word Y “piperazine adipate”, the number of concurrently using organizations (both_use) that use words X and Y at the same time is calculated as “1”. In addition, the same event index calculation unit 102 determines that the word X “piperazine adipate” and the word Y “piperazine adipate” are not included in all the hospitals 1 to n at the same time. The rate of not satisfying the conditions is calculated as the same event index value "0.092". This value is calculated by multiplying the similarity "0.1".

The same event index calculation unit 102 also calculates the same event index calculation result 602 for each of the other pairs of words X and Y based on the matching results 601 a and 601 b of the master matching unit 101 . Then, the same event determination device 100 (control unit) outputs the same event index calculation result 602 calculated by the same event index calculation unit 102 to the outside.

According to the above process, as the same event index calculation result 602, for each pair of words X and Y, the number of organizations using either one of the words X and Y (Share) and the words X and Y exist at the same time. It is possible to obtain the number of masters (both_use) to be used and the same event index. In addition, as the same event index calculation result 602, the rate that does not satisfy the condition "two words X and Y are not included in all hospitals 1 to n at the same time" is calculated as the same event index value for each set of words. can be clearly shown as

According to the embodiment described above, the masters 1 owned by a plurality of organizations 1 to n are accessed, and whether two arbitrary words X and Y exist in one master 1 of one organization 1 is checked. to match. Then, based on the collation result, the same event index value indicating whether or not the two words X and Y are included in each of the masters 1 of all the organizations 1 to n is calculated. For example, in any of the organizations 1 to n, one master 1 within one organization does not simultaneously describe the same event (for example, drug) as separate words. In this case, if two words do not appear in the master of any organization at the same time, it is determined that two different words are the same event. Also, if a master in which two words are described at the same time exists in the master 1 of any organization, it is determined that two different words are not the same event.

Also, acquire two words, access the masters 1 owned by a plurality of organizations 1 to n, and find that at least one of the plurality of masters 1 has both the two words X and Y. If so, it is determined that the two words X and Y have different meanings.

Thus, according to the embodiment, it becomes possible to accurately determine whether or not two different words are the same event. When integrating and utilizing master (database) data of a large number of organizations, it is possible to efficiently perform preparatory work for aligning notations so that "same event=same notation". For example, it becomes possible to easily create a dictionary (corresponding to a synonym table) for determining whether or not words with different notations represent the same event.

Also, the number of masters in which two words X and Y exist together may be calculated. This makes it possible to specifically indicate the number of masters in which two words X and Y are present among the masters of each of the multiple organizations.

In addition, based on the results of matching, the same event index value is calculated as "1" in all organizations when the two words X and Y are not included at the same time, and the value is calculated as "0" otherwise. good too. As a result, whether or not the two words X and Y are included in the master 1 of some of the target organizations 1 to n can be indicated by the simplest and clearest numerical value.

Also, based on the collation result, 1-(the number of organizations in which both words exist in the same master)/(the number of all organizations) may be calculated as the same event index value. As a result, the percentage of the total number of tissues in which the two words X and Y are present in the masters 1 of all the target tissues 1 to n can be expressed as a specific numerical value. In addition, it is possible to prevent erroneous determination when there is even one organization that describes the same phenomenon as different words in the master.

Also, the number of tissues 1 to n in which any of the two words X and Y exists may be calculated. This makes it possible to specifically indicate the number of organizations in which either of the two words X and Y exists among the plurality of organizations.

In addition, based on the collation results, as the same event index value, 1 - (number of organizations where both words exist in the same master) / (number of organizations where either of the two words exists in the master) can be calculated. As a result, for the masters 1 of all target organizations 1 to n, the ratio of the number of organizations in which either the two words X or Y exists to the number of organizations in which both the two words X and Y exist is calculated as a specific numerical value. can be shown as Also, when there are few organizations using the two words X and Y, it is possible to prevent the same event index from becoming high.

Alternatively, a set of two words similar to the two target words may be extracted from an arbitrary large number of word sets. A general-purpose technique can be used to calculate the similarity between two words, and by targeting two words in each set obtained using the similarity, it is possible to efficiently perform overall processing related to the same event index. Become. For example, by accessing a plurality of masters 1, a large number of word sets can be obtained, and a set of two similar words can be extracted from the large number of word sets. be able to do so efficiently.

The system for calculating the same event index (same event determination device 100) may also include a network interface 203 that communicates with the master 1 of the target organization. As a result, in the communication connection with the masters 1 of many organizations, the process of calculating the same event index value indicating whether or not two words are included in the masters 1 of many organizations can be efficiently performed. Become.

From these facts, according to the embodiment, it is possible to accurately determine whether two different words are described as the same event (object, action, etc.) in all masters of the same type (for example, medicine) in different organizations. be able to judge. In this respect, with the conventional synonym determination process alone, for example, it was erroneously determined that the two words "medicine A" and "medicine B" are highly likely to be the same drug. On the other hand, in the embodiment, it becomes possible to clearly present that these two different drug words are described as the same event in one master.

Note that the method for determining the same event described in the embodiment of the present invention can be realized by causing a processor such as a server to execute a program prepared in advance. This method is recorded in a computer-readable recording medium such as a hard disk, flexible disk, CD-ROM (Compact Disc-Read Only Memory), flash memory, etc., and is executed by being read from the recording medium by a computer. The method may also be distributed over a network such as the Internet.

Further, the following additional remarks are disclosed with respect to the above-described embodiment.

(Appendix 1) Access each master owned by multiple organizations,
Checking each said master whether any two words exist in one said master,
calculating, for each of the masters, a same-event index value indicating whether one of the masters contains the two words, based on the results of the matching;
recording the calculated same event index value in the same storage unit;
A same event determination program characterized by causing a computer to execute processing.

(Appendix 2) Further, calculating the number of the masters in which the two words are present together,
The identical event determination program according to Supplementary Note 1, characterized by:

(Appendix 3) Based on the result of the collation, as the same event index value,
Calculate a value according to the number of masters in which both words exist with respect to the total number of masters to be processed,
The identical event determination program according to Supplementary Note 1, characterized by:

(Appendix 4) As the two words, extract two similar words from an arbitrary large number of word sets.
The identical event determination program according to any one of Appendices 1 to 3, characterized by:

(Appendix 5) Access masters owned by multiple organizations,
Checking each said master whether any two words exist in one said master,
calculating, for each of the masters, a same-event index value indicating whether one of the masters contains the two words, based on the results of the matching;
recording the calculated same event index value in the same storage unit;
A same event determination method characterized in that the processing is executed by a computer.

(Appendix 6) Access masters owned by multiple organizations,
a master collating unit that collates each of the masters to determine whether two arbitrary words exist in one of the masters;
a same-event index calculation unit that calculates, for each of the masters, a same-event index value indicating whether or not the two words are included in one of the masters, based on the result of the collation;
a storage unit for recording the calculated identical event index value;
A same-event determination system characterized by comprising:

(Appendix 7) A similar word extraction unit that extracts a plurality of pairs of similar two words from a large number of word sets and outputs the two words of each pair to the master matching unit;
The same event determination system according to appendix 6, characterized by comprising:

(Appendix 8) The identical event determination system according to appendix 6 or 7, further comprising a network interface that communicates with the master of the organization.

(Appendix 9) Get two words,
Access masters owned by multiple organizations,
If both of the two words are present in at least one or more of the plurality of masters, it is determined that the two words have different meanings,
A same event determination program characterized by causing a computer to execute processing.

100 same event determination device 101 master collation unit 102 same event index calculation unit 201 CPU (control unit)
202 memory 203 network interface (IF)
205 recording medium 210 network 401 similar word extraction unit 600 master 601 (601a, 601b) matching result 602 same event index calculation result X, Y (a set of words)

Claims (8)

Accessing master databases owned by multiple organizations,
Checking each of the master databases whether any two words exist in one of the master databases ,
calculating a same-event index value indicating whether the two words are the same event based on the results of the matching for each of the master databases ;
recording the calculated same event index value in the same storage unit;
A same event determination program characterized by causing a computer to execute processing. Based on the results of the matching for each of the master databases, calculate the number of the master databases in which the two words are present together , and calculate the number of the master databases in which the two words are present together as the same event index value. Calculate a value according to the number of databases in
The identical event determination program according to claim 1, characterized by: Based on the results of the matching for each of the master databases , as the same event index value,
Calculating a value according to the ratio of the number of master databases in which both two words exist to the total number of the master databases to be collated,
The identical event determination program according to claim 1, characterized by: As the two words, extracting two similar words from an arbitrary large number of word sets,
The identical event determination program according to any one of claims 1 to 3, characterized in that: Access master databases owned by multiple organizations,
Checking each of the master databases whether any two words exist in one of the master databases ,
calculating a same-event index value indicating whether the two words are the same event based on the results of the matching for each of the master databases ;
recording the calculated same event index value in the same storage unit;
A same event determination method characterized in that the processing is executed by a computer. Access master databases owned by multiple organizations,
a master collation unit that collates each of the master databases to determine whether two arbitrary words exist in one of the master databases ;
a same-event index calculation unit that calculates a same-event index value indicating whether the two words are the same event based on the results of the collation for each of the master databases ;
a storage unit for recording the calculated identical event index value;
A same-event determination system characterized by comprising: a similar word extraction unit that extracts a plurality of sets of two similar words from a large number of word sets and outputs the two words of each set to the master matching unit;
7. The same event determination system according to claim 6, comprising: get two words,
Access master databases owned by multiple organizations,
If both of the two words exist in at least one or more master databases out of the plurality of master databases , the two words are determined not to be the same event ;
A same event determination program characterized by causing a computer to execute processing.

Images