Value range matching method and device, storage medium and electronic device

Abstract

The application discloses a value range matching method and device, a storage medium and an electronic equipment. The method comprises the following steps: obtaining source value meanings corresponding to each source value range in at least two source value ranges and a target value meaning corresponding to a target reference value range; performing an extraction operation under a phrase dimension on the source value meanings corresponding to each source value range to obtain source phrases corresponding to each source value range; calculating first similarity information between the source phrases corresponding to each source value range and the target value meaning, and performing a first recall to obtain a first recall result; calculating second similarity information between the source value meanings matched by the first recall result and the target value meaning, and performing a second recall on the first recall result based on the second similarity information to obtain a second recall result; and performing alignment matching on the source value ranges matched by the second recall result and the target reference value range, which can be applied to an artificial intelligence scene. The application solves the technical problem of low value range matching efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the computer field, in particular to a value range matching method and device, storage medium and electronic equipment. BACKGROUND

[0002] In a value range matching scenario, all to-be-matched value ranges are usually matched by traversal, such as first matching calculation between different value ranges and matching calculation between value meanings of different value ranges. The overall complexity corresponding to such direct matching calculation manner increases exponentially, resulting in a large waste of redundant computing power and further causing a problem of low value range matching efficiency. Therefore, there is a problem of low value range matching efficiency.

[0003] At present, no effective solution has been proposed for the above problem. SUMMARY

[0004] Embodiments of the present application provide a value range matching method, device, storage medium and electronic equipment to at least solve the technical problem of low value range matching efficiency.

[0005] According to an aspect of an embodiment of the present application, a value range matching method is provided, including: obtaining source value meanings corresponding to each source value range in at least two source value ranges and a target value meaning corresponding to a target reference value range; performing an extraction operation under a word group dimension on the source value meanings corresponding to each source value range to obtain source word groups corresponding to each source value range; calculating first similarity information between the source word groups corresponding to each source value range and the target value meaning, and performing a first recall on the source value meanings corresponding to each source value range based on the first similarity information to obtain a first recall result; calculating second similarity information between the source value meanings matched by the first recall result and the target value meaning, and performing a second recall on the first recall result based on the second similarity information to obtain a second recall result, wherein the source value meanings corresponding to each source value range include the source value meanings matched by the first recall result; and performing alignment matching on the source value ranges matched by the second recall result and the target reference value range to map the source value ranges matched by the second recall result to the target reference value range, wherein the at least two source value ranges include the source value ranges matched by the second recall result.

[0006] According to another aspect of the embodiments of the present application, a value range matching device is also provided, which comprises: a first acquisition unit configured to acquire source value meanings corresponding to each of at least two source value ranges and target value meanings corresponding to a target reference value range; an extraction unit configured to perform extraction operation on the source value meanings corresponding to each of the at least two source value ranges in a phrase dimension to obtain source phrases corresponding to each of the at least two source value ranges; a first recall unit configured to calculate first similarity information between the source phrases corresponding to each of the at least two source value ranges and the target value meanings, and perform first recall on the source value meanings corresponding to each of the at least two source value ranges based on the first similarity information to obtain a first recall result; a second recall unit configured to calculate second similarity information between source value meanings matched by the first recall result and the target value meanings, and perform second recall on the first recall result based on the second similarity information to obtain a second recall result, wherein the source value meanings corresponding to each of the at least two source value ranges include the source value meanings matched by the first recall result; and a matching unit configured to perform alignment matching on source value ranges matched by the second recall result and the target reference value range, so that the source value ranges matched by the second recall result are aligned and mapped onto the target reference value range, wherein the at least two source value ranges include the source value ranges matched by the second recall result.

[0007] As an optional solution, the first recall unit comprises: a first calculation module configured to calculate the first similarity information and select at least one candidate value meaning from the source value meanings corresponding to each of the at least two source value ranges based on the first similarity information; the second recall unit comprises: a second calculation module configured to calculate the second similarity information and select at least one second candidate value range from first candidate value ranges corresponding to the at least one candidate value meaning based on the second similarity information, wherein the at least two source value ranges include the first candidate value ranges corresponding to the at least one candidate value meaning; and the matching unit comprises: a first matching module configured to perform alignment matching on each of the at least one second candidate value range and the target reference value range.

[0008] As an optional solution, the first calculation module comprises: a calculation submodule configured to calculate phrase similarity information between the source phrases corresponding to each of the at least two source value ranges and target phrases in the target value meanings; and a selection submodule configured to select the at least one candidate value meaning with phrase similarity greater than or equal to a preset similarity from the source value meanings corresponding to each of the at least two source value ranges based on the phrase similarity information.

[0009] As an optional solution, the extraction unit comprises: a first extraction module, configured to perform the extraction operation in the phrase dimension on the source phrase corresponding to each source value domain according to the target format, to obtain the source phrase in the target format corresponding to each source value domain; and the device further comprises: a second extraction module, configured to perform the extraction operation in the phrase dimension on the target phrase corresponding to the target value meaning according to the target format, to obtain the target phrase in the target format, before the calculation of the phrase similarity information between the source phrase corresponding to each source value domain and the target phrase in the target value meaning.

[0010] As an optional solution, the device further comprises: a second acquisition unit, configured to acquire a plurality of source value domains to be matched and the target reference value domain, before the acquisition of the source value meaning corresponding to each source value domain in the at least two source value domains and the target value meaning corresponding to the target reference value domain, wherein the plurality of source value domains to be matched comprises the at least two source value domains; and a third recall unit, configured to perform a similar recall on the plurality of source value domains to be matched based on the target reference value domain, to obtain a plurality of source value domains recalled similarly, and determine the plurality of source value domains recalled similarly as the at least two source value domains, before the acquisition of the source value meaning corresponding to each source value domain in the at least two source value domains and the target value meaning corresponding to the target reference value domain.

[0011] As an optional solution, the third recall unit comprises: a first acquisition module, configured to acquire the source value domain length corresponding to each source value domain to be matched in the plurality of source value domains to be matched and the target value length corresponding to the target reference value domain; a first recall module, configured to calculate third similarity information between the source value domain length corresponding to each source value domain to be matched and the target value length, and perform a recall on each source value domain to be matched based on the third similarity information, to obtain a third recall result; a second recall module, configured to calculate fourth similarity information between the source value meaning matched by the third recall result and the target value meaning, and perform a recall on the third recall result based on the third similarity information, to obtain a fourth recall result, wherein the source value domain matched by the fourth recall result comprises the plurality of source value domains recalled similarly; and a determination module, configured to determine the source value domain matched by the fourth recall result as the at least two source value domains.

[0012] As an optional solution, the device further comprises at least one of the following: a first determining unit configured to calculate first similarity information between the source word group corresponding to each source value domain and the target value meaning, perform a recall on the source value meaning corresponding to each source value domain based on the first similarity information, and determine the source value domain matched with the first recall result and count the recall times of the source value domain matched with the first recall result after obtaining the first recall result.

[0013] As an optional solution, the matching unit comprises: a second obtaining module configured to obtain the cumulative recall times of the source value domain matched with the second recall result; a filtering module configured to filter the source value domain with the recall times less than a preset threshold from the source value domain matched with the second recall result, and reserve the target source value domain with the recall times greater than or equal to the preset threshold from the source value domain matched with the second recall result; and a second matching module configured to perform alignment matching on the target source value domain and the target reference value domain.

[0014] As an optional solution, the extracting unit comprises: a third extracting module configured to perform an extracting operation in the word group dimension on the source value meaning corresponding to each source value domain to obtain an initial source word group corresponding to each source value domain; and a counting module configured to count the occurrence frequency of the initial source word group in the source value meaning corresponding to each source value domain, filter the initial source word group with the occurrence frequency less than a preset frequency, and reserve the source word group corresponding to each source value domain with the occurrence frequency greater than or equal to the preset frequency.

[0015] According to yet another aspect of the embodiments of the present application, a computer program product or computer program is provided, which comprises computer instructions stored in a computer readable storage medium. A processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to enable the computer device to perform the value domain matching method.

[0016] According to yet another aspect of the embodiments of the present application, an electronic device is also provided, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor performs the value domain matching method through the computer program.

[0017] In the embodiment of the present application, the source value meanings corresponding to each of the at least two source value domains and the target value meaning corresponding to the target reference value domain are obtained; the extraction operation under the phrase dimension is performed on the source value meanings corresponding to each of the source value domains to obtain the source phrases corresponding to each of the source value domains; the first similarity information between the source phrases corresponding to each of the source value domains and the target value meaning is calculated, and the source value meanings corresponding to each of the source value domains are recalled once based on the first similarity information to obtain a first recall result; the second similarity information between the source value meanings matched with the first recall result and the target value meaning is calculated, and the first recall result is recalled twice based on the second similarity information to obtain a second recall result, wherein the source value meanings corresponding to each of the source value domains include the source value meanings matched with the first recall result; the source value domains matched with the second recall result are aligned and matched with the target reference value domain, so that the source value domains matched with the second recall result are aligned and mapped onto the target reference domain, wherein the at least two source value domains include the source value domains matched with the second recall result; the first recall is performed in the phrase dimension, then the second recall is performed in the value meaning dimension on the result of the first recall, and the value domain matching is further performed by using the result of the second recall, so that the total calculation amount in the recall process is reduced, the complexity of the value domain matching is reduced, and the overall time consumption of the value domain matching is compressed, thereby realizing the technical effect of improving the value domain matching efficiency, and further solving the technical problem of low value domain matching efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0018] The accompanying drawings, which are included to provide a further understanding of the present application, constitute a part of this application, illustrate embodiments of the present application and specific embodiments thereof, and serve to explain the present application, and do not limit the present application in any manner. In the drawings:

[0019] Figure 1 is a schematic diagram of an application environment of an optional value domain matching method according to an embodiment of the present application;

[0020] Figure 2 is a schematic diagram of a flow of an optional value domain matching method according to an embodiment of the present application;

[0021] Figure 3 is a schematic diagram of an optional value domain matching method according to an embodiment of the present application;

[0022] Figure 4 is a schematic diagram of another optional value domain matching method according to an embodiment of the present application;

[0023] Figure 5 is a schematic diagram of another optional value domain matching method according to an embodiment of the present application;

[0024] Figure 6 is a schematic diagram of another optional value range matching method according to an embodiment of the present application;

[0025] Figure 7 is a schematic diagram of another optional value range matching method according to an embodiment of the present application;

[0026] Figure 8 is a schematic diagram of another optional value range matching method according to an embodiment of the present application;

[0027] Figure 9 is a schematic diagram of another optional value range matching method according to an embodiment of the present application;

[0028] Figure 10 is a schematic diagram of another optional value range matching method according to an embodiment of the present application;

[0029] Figure 11 is a schematic diagram of another optional value range matching method according to an embodiment of the present application;

[0030] Figure 12 is a schematic diagram of an optional value range matching apparatus according to an embodiment of the present application;

[0031] Figure 13 is a structural schematic diagram of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to make the personnel in the art better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the scope of protection of the present application.

[0033] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] For the convenience of understanding, the following terms are explained:

[0035] Artificial Intelligence (AI) is the use of digital computers or digital computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology of computer science, which tries to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that machines have the functions of perception, reasoning and decision-making.

[0036] Artificial intelligence technology is a comprehensive discipline, involving a wide range of fields, both hardware and software technologies. Artificial intelligence basic technologies generally include technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0037] With the research and progress of artificial intelligence technology, artificial intelligence technology has been researched and applied in many fields, such as common smart home, smart wearable devices, virtual assistants, smart speakers, smart marketing, unmanned vehicles, autonomous vehicles, drones, robots, smart medical care, smart customer service, etc. It is believed that with the development of technology, artificial intelligence technology will be applied in more fields and play an increasingly important role.

[0038] According to an aspect of an embodiment of the present application, a value range matching method is provided. Optionally, as an optional implementation, the value range matching process of the above-mentioned value range matching method can be applied in an environment as shown in Figure 1 The specific process can be executed as follows until the value range matching process is ended:

[0039] Step S102, determining a current source value range (such as source value range 104) from at least two source value ranges 102;

[0040] Step S104, performing an extraction operation under the phrase dimension on the source value meaning (such as source value meaning 106) corresponding to the current source value range to obtain a current source phrase (such as source phrase 108);

[0041] Step S106, similarity information between the current source word group and the target value meaning 110 (such as the value meaning corresponding to the target reference value domain 112) is calculated, and a first recall is performed to determine whether the source value meaning corresponding to the current source value domain can be used as the current source value meaning. If not, the next source value domain is taken as the current source value domain. If yes, the source value meaning corresponding to the current source value domain is determined as the current source value meaning.

[0042] Step S108, similarity information between the current source value meaning and the target value meaning is calculated, and a second recall is performed to determine whether the current source value meaning meets the second recall condition. If not, the next source value domain is taken as the current source value domain. If yes, a matching operation is performed.

[0043] Step S110, a matching operation is performed on the current source value domain and the target reference value domain, so that the current source value domain is aligned and mapped to the target reference value domain.

[0044] Step S112, it is determined whether all source value domains have been subjected to value domain matching. If yes, the value domain matching process is ended. If not, the next source value domain is taken as the current source value domain.

[0045] Optionally, as an optional implementation, as shown in Figure 2 the value domain matching method comprises:

[0046] S202, source value meanings corresponding to each source value domain in at least two source value domains and a target value meaning corresponding to a target reference value domain are obtained;

[0047] S204, an extraction operation in the word group dimension is performed on the source value meanings corresponding to each source value domain, to obtain source word groups corresponding to each source value domain;

[0048] S206, first similarity information between the source word groups corresponding to each source value domain and the target value meaning is calculated, and a first recall is performed on the source value meanings corresponding to each source value domain based on the first similarity information, to obtain a first recall result;

[0049] S208, second similarity information between the source value meanings matched by the first recall result and the target value meaning is calculated, and a second recall is performed on the first recall result based on the second similarity information, to obtain a second recall result, wherein the source value meanings corresponding to each source value domain include the source value meanings matched by the first recall result;

[0050] S210, the source value domains matched by the second recall result are aligned and matched with the target reference value domain, so that the source value domains matched by the second recall result are aligned and mapped to the target reference domain. At least two source value domains include the source value domains matched by the second recall result.

[0051] Optionally, in the embodiment, the value range matching method described above can be applied to value range mapping, such as mapping the value range (referred to as source value range) in a certain original data standard to the value range (referred to as reference value range) in a certain reference data standard. In general, in a business scenario, a set of value ranges is used as a reference value range to align and map different value ranges. The basis for determining whether different value ranges can be aligned and mapped can be, but is not limited to, the value range matching result of the value range matching method described above.

[0052] For further illustration, optionally, two value ranges from two data standards are mapped, for example, the value range SZCV01.02.001-patient source code table contains multiple value meanings, which are 1-local, 2-outside (domestic range other than the city), 3-overseas, etc. At the same time, the value range GD99.02.350-outside marker code table also includes multiple value meanings, which are 1-local, 2-outside, etc. The value range mapping task needs to give at least two matching results: matching result 1) value range mapping, i.e. mapping of the value range SZCV01.02.001-patient source code table to the value range GD99.02.350-outside marker code table; matching result 2) value meaning mapping, i.e. mapping of 2-outside (domestic range other than the city) to 2-outside. The matching result 2) will affect the matching result 1), i.e. the two value ranges can be matched successfully only if the value meanings are matched.

[0053] For the value range mapping task described above, traversal matching and calculation of edit distance can be used, such as one-to-one matching and calculation of all value ranges from data standard A and data standard B. Therefore, m(A)×m(B) times of calculation are required, where m represents the number of value ranges in the data standard. In each value range calculation, n(Ai)×n(Bj) times of value meaning matching results need to be calculated. The overall calculation complexity is The above method has a large amount of redundant calculation waste, which also causes unacceptable time consumption and reduces the value range matching efficiency. In the embodiment, the matching calculation dimension is reduced from the value range to the value meaning and from the value meaning to the phrase. The matching range is expanded, the recall scale in the matching process is reduced, the useless redundant calculation is reduced, the overall time consumption of the engine is compressed, and the value range matching efficiency is improved.

[0054] Optionally, in the embodiment, the value range can refer to data standard metadata defined and used in a certain field in actual application. The metadata defines the fields of the database and needs to comply with certain practical application rationality. A value range can include multiple value meanings, which are used to explain the colloquial meaning of the value range. A phrase can be understood as a component of the value meaning.

[0055] For further illustration, optionally, for example Figure 3 As shown, value range 302 "CT01.00.030 - Surgical operation code", and the corresponding value meanings 304 of value range 302 include "local anesthesia", "anesthetic medication", "anesthetic physiology". Among them, each value meaning 304 is split and extracted into corresponding phrases 306. For example, value meaning 304 "local anesthesia" corresponds to phrases 306 "local" and "anesthesia", value meaning 304 "anesthetic medication" corresponds to phrases 306 "anesthesia" and "medication", value meaning 304 "anesthetic physiology" corresponds to phrases 306 "anesthesia" and "physiology". Furthermore, phrase 306 "anesthesia" has a relatively high similarity with value meanings 304 "local anesthesia", "anesthetic medication", and "anesthetic physiology". That is, using the phrase can recall more value meanings (one-time recall), and at the same time, some obviously mismatched value meanings are filtered out, providing a more streamlined set of value meanings to be matched for the subsequent recall of value ranges (secondary recall), reducing the recall scale in the matching process, reducing unnecessary redundant calculations, thus compressing the overall time consumption of the engine and improving the value range matching efficiency.

[0056] Optionally, in this embodiment, the relationship between the source value range and the target reference value range can be but is not limited to being understood as that it is necessary to unify the value range standards of the source value range, and the unified standard is the value range standard of the target reference value range. Furthermore, first, it is necessary to determine the value ranges and value meanings that match between the source value range and the target reference value range through value range matching, and then perform standard alignment on the matched value ranges and value meanings. Furthermore, the source value range and the target reference value range after unifying the standard can be but is not limited to being displayed and subsequently managed under the same standard.

[0057] Optionally, in this embodiment, calculate the first similarity information between the source phrases corresponding to each source value range and the target value meanings, and perform a one-time recall on the source value meanings corresponding to each source value range based on the first similarity information to obtain the first recall result;

[0058] For further illustration, optionally based on Figure 3 the scenario shown, continue for example Figure 4 As shown, assume that value range 302 is a value range in at least two source value ranges, and value range 402 is the target reference value range. Furthermore, calculate the first similarity information between the phrase 306 (source phrase) corresponding to value range 302 and the value meaning 404 (target value meaning) corresponding to value range 402, and determine whether to recall value range 302 once based on the first similarity information. For example, value meaning 404 "local" contains phrase 302 "local", value meaning 404 "anesthetic drug" contains phrase 302 "anesthesia", value meaning 404 "anesthesiology" contains phrase 302 "anesthesia", etc.

[0059] Optionally, in the embodiment, second similarity information between the source value meaning matched by the first recall result and the target value meaning is calculated, and the first recall result is secondly recalled based on the second similarity information to obtain a second recall result;

[0060] Further, as shown in the scenario of FIG. 4B, the value domain 302 is determined as the first recall result through the first recall, and then the second similarity information between the value meaning 304 (source value meaning) corresponding to the value domain 302 and the value meaning 404 (target value meaning) corresponding to the value domain 402 is calculated, and it is determined whether to secondly recall the value domain 302 based on the second similarity information. For example, the value meaning 304 “local anesthesia” is similar to the value meaning 404 “local”, the value meaning 304 “anesthetic drug” is similar to the value meaning 404 “anesthetic drug”, and the value meaning 304 “anesthesia physiology” is similar to the value meaning 404 “anesthesiology”. Figure 4 Figure 5 Further, as shown in the scenario of FIG. 4B, the value domain 302 is determined as the first recall result through the first recall, and then the second similarity information between the value meaning 304 (source value meaning) corresponding to the value domain 302 and the value meaning 404 (target value meaning) corresponding to the value domain 402 is calculated, and it is determined whether to secondly recall the value domain 302 based on the second similarity information. For example, the value meaning 304 “local anesthesia” is similar to the value meaning 404 “local”, the value meaning 304 “anesthetic drug” is similar to the value meaning 404 “anesthetic drug”, and the value meaning 304 “anesthesia physiology” is similar to the value meaning 404 “anesthesiology”.

[0061] Optionally, in the embodiment, the source value domain matched by the second recall result is matched with the target reference value domain;

[0062] Further, as shown in the scenario of FIG. 4B, the value domain 302 is determined as the first recall result through the first recall, and then the second similarity information between the value meaning 304 (source value meaning) corresponding to the value domain 302 and the value meaning 404 (target value meaning) corresponding to the value domain 402 is calculated, and it is determined whether to secondly recall the value domain 302 based on the second similarity information. For example, the value meaning 304 “local anesthesia” is similar to the value meaning 404 “local”, the value meaning 304 “anesthetic drug” is similar to the value meaning 404 “anesthetic drug”, and the value meaning 304 “anesthesia physiology” is similar to the value meaning 404 “anesthesiology”. Figure 5 Figure 6 Further, as shown in the scenario of FIG. 4B, the value domain 302 is determined as the first recall result through the first recall, and then the second similarity information between the value meaning 304 (source value meaning) corresponding to the value domain 302 and the value meaning 404 (target value meaning) corresponding to the value domain 402 is calculated, and it is determined whether to secondly recall the value domain 302 based on the second similarity information. For example, the value meaning 304 “local anesthesia” is similar to the value meaning 404 “local”, the value meaning 304 “anesthetic drug” is similar to the value meaning 404 “anesthetic drug”, and the value meaning 304 “anesthesia physiology” is similar to the value meaning 404 “anesthesiology”.

[0063] It should be noted that the first recall is performed in the phrase dimension, the second recall is performed in the value meaning dimension on the result of the first recall, and the value domain matching is further performed by using the result of the second recall, so that the total calculation amount in the recall process is reduced, the complexity of the value domain matching is reduced, and the overall time consumption of the value domain matching is compressed, thereby improving the value domain matching efficiency.

[0064] Further, as shown in the scenario of FIG. 4B, the value domain 302 is determined as the first recall result through the first recall, and then the second similarity information between the value meaning 304 (source value meaning) corresponding to the value domain 302 and the value meaning 404 (target value meaning) corresponding to the value domain 402 is calculated, and it is determined whether to secondly recall the value domain 302 based on the second similarity information. For example, the value meaning 304 “local anesthesia” is similar to the value meaning 404 “local”, the value meaning 304 “anesthetic drug” is similar to the value meaning 404 “anesthetic drug”, and the value meaning 304 “anesthesia physiology” is similar to the value meaning 404 “anesthesiology”. Figure 6 Figure 7 ​​​As shown, at least two source value domains 702 are obtained, and a first recall is performed on the at least two source value domains 702 in the phrase dimension, and then a second recall is performed on the result of the first recall in the value meaning dimension, and further, the value domain matching is performed by using the result of the second recall, so that the total amount of calculation in the recall process is reduced, the complexity of the value domain matching is reduced, and the overall time cost of the value domain matching is compressed, thereby improving the value domain matching efficiency; for example, the value domain 302 is determined from the at least two source value domains 702, the source phrase corresponding to the source value meaning of the value domain 302 is extracted, and the source phrase corresponding to the value domain 302 is obtained; the first similarity information between the source phrase corresponding to the value domain 302 and the target value meaning is calculated, and the source value meaning corresponding to the value domain 302 is recalled once based on the first similarity information, and the first recall result is obtained; the second similarity information between the source value meaning matched with the first recall result and the target value meaning is calculated, and the first recall result is recalled twice based on the second similarity information, and the second recall result is obtained; and the source value domain matched with the second recall result is aligned and matched with the target reference value domain.

[0065] According to the embodiments provided in the present application, the source value meaning corresponding to each source value domain in the at least two source value domains and the target value meaning corresponding to the target reference value domain are obtained; the extraction operation in the phrase dimension is performed on the source value meaning corresponding to each source value domain, and the source phrase corresponding to each source value domain is obtained; the first similarity information between the source phrase corresponding to each source value domain and the target value meaning is calculated, and the source value meaning corresponding to each source value domain is recalled once based on the first similarity information, and the first recall result is obtained; the second similarity information between the source value meaning matched with the first recall result and the target value meaning is calculated, and the first recall result is recalled twice based on the second similarity information, and the second recall result is obtained, wherein the source value meaning corresponding to each source value domain includes the source value meaning matched with the first recall result; and the source value domain matched with the second recall result is aligned and matched with the target reference value domain, so that the source value domain matched with the second recall result is mapped onto the target reference value domain, wherein the at least two source value domains include the source value domain matched with the second recall result; the first recall is performed in the phrase dimension, and then the second recall is performed on the result of the first recall in the value meaning dimension, and further, the value domain matching is performed by using the result of the second recall, so that the total amount of calculation in the recall process is reduced, the complexity of the value domain matching is reduced, and the overall time cost of the value domain matching is compressed, thereby achieving the technical effect of improving the value domain matching efficiency.

[0066] As an optional solution, the first similarity information between the source phrase corresponding to each source value domain and the target value meaning is calculated, and a first recall is performed on the source value meaning corresponding to each source value domain based on the first similarity information to obtain a first recall result, including: calculating the first similarity information, and screening at least one candidate value meaning from the source value meaning corresponding to each source value domain based on the first similarity information;

[0067] As an optional solution, the second similarity information between the source value meaning matched with the first recall result and the target value meaning is calculated, and a second recall is performed on the first recall result based on the second similarity information to obtain a second recall result, including: calculating the second similarity information, and screening at least one second candidate value domain from the first candidate value domain corresponding to the at least one candidate value meaning based on the second similarity information, wherein the at least two source value domains include the first candidate value domain corresponding to the at least one candidate value meaning;

[0068] As an optional solution, the source value domain matched with the second recall result and the target reference value domain are aligned and matched, including: aligning and matching each second candidate value domain in the at least one second candidate value domain with the target reference value domain.

[0069] Optionally, in actual business scenarios, the data standard entered may have only ten thousand value domains and millions of value meanings, which will cause a large amount of unacceptable engine time overhead. In order to improve the overall usability of the engine, a multi-level group recall technology can be proposed to optimize and accelerate such extreme cases, such as first constructing two indexes by the engine, wherein the first index is an index of value meanings to value domains in a certain data standard, and the second index is an index of phrases extracted from the value meanings in the first index to value meanings;

[0070] Further, as an optional example Figure 8 As shown, first, an index table of value meanings to value domains in the data standard is constructed, and all keys in the constructed first index are extracted into phrases, such as the value meaning "local anesthesia", which can extract three phrases "local", "departmental anesthesia", and "anesthesia".

[0071] Further, all extracted phrases are constructed into a second index, that is, an index of phrases to value meanings, such as the "anesthesia" index to "local anesthesia", "anesthesia medication", "anesthesia physiology", and more value meanings. It is noted that the second index also has the feature of many-to-many, that is, the same phrase can index to multiple value meanings, and one value meaning can be indexed by multiple phrases, and the scale of the second index is generally much smaller than that of the first index;

[0072] And in the recall stage based on the cosine similarity threshold screening, firstly, a recall (first recall) is performed on the secondary index, since the scale of the first recall calculation is small, and the word length of the index is limited, so the recall of the secondary index can be completed in a short time, and the recall result is a plurality of candidate value meaning pairs; then a recall (second recall) is performed on the primary index, and the search space in the recall of the primary index is no longer all value meaning combinations, but the recall result of the first recall, and the result of the recall of the primary index is a plurality of candidate value domain pairs.

[0073] Further, after the recall stage ends, the recall result is normally input to the subsequent precision sorting and rearrangement module, and then the value domain mapping result is obtained, the multi-level index is established to perform hierarchical and group recall, which can effectively compress the calculation amount of the recall stage, reduce useless redundant calculation, and thus compress the overall time consumption of the engine.

[0074] According to the embodiments provided in the application, the first similarity information is calculated, and at least one candidate value meaning is screened from the source value meanings corresponding to each source value domain based on the first similarity information; the second similarity information is calculated, and at least one second candidate value domain is screened from the first candidate value domain corresponding to the at least one candidate value meaning based on the second similarity information, wherein the at least two source value domains include the first candidate value domain corresponding to the at least one candidate value meaning; each second candidate value domain in the at least one second candidate value domain is aligned and matched with the target reference value domain, thereby achieving the purpose of compressing the overall time consumption of the engine, and thus realizing the technical effect of improving the value domain matching efficiency.

[0075] As an optional solution, the first similarity information is calculated, and at least one candidate value meaning is screened from the source value meanings corresponding to each source value domain based on the first similarity information, including:

[0076] S1, calculating the word group similarity information between the source word groups corresponding to each source value domain and the target word groups in the target value meaning;

[0077] S2, screening at least one candidate value meaning with a word group similarity greater than or equal to a preset similarity from the source value meanings corresponding to each source value domain.

[0078] Optionally, in the present embodiment, the word vector similarity technology can be used to calculate the word group similarity information between the source word groups corresponding to each source value domain and the target word groups in the target value meaning, and a pre-trained language model can be used to perform string coding of the source word groups corresponding to each source value domain and the target word groups in the target value meaning, so that the corresponding fixed dimension vector can be obtained, such as h i =PLM(n i); wherein, for each source n-gram corresponding to a source value domain and target n-gram in the target value meaning, a same language model is used for encoding to avoid semantic drift, and then Faiss and other tools can be used for dense vector calculation, and normalized vector dot product (i.e., cosine similarity) can be used for similarity measurement, and pairs of value meanings with similarity greater than a threshold are taken out, and the greater the threshold, the more stringent the constraint in the similarity calculation stage, and the fewer the number of candidates meeting the recall condition.

[0079] As an optional solution, the extraction operation in the n-gram dimension is performed on each source value meaning corresponding to a source value domain to obtain a source n-gram corresponding to each source value domain, including: performing the extraction operation in the n-gram dimension on each source value meaning corresponding to a source value domain according to the target format to obtain a source n-gram in the target format corresponding to each source value domain.

[0080] As an optional solution, before calculating the n-gram similarity information between each source n-gram corresponding to a source value domain and a target n-gram in the target value meaning, the method further includes: performing the extraction operation in the n-gram dimension on the target value meaning according to the target format to obtain a target n-gram in the target format.

[0081] Optionally, in this embodiment, the target format can be but is not limited to the unit magnitude of n-grams, data format, etc. For example, if the unit magnitude of the source n-gram is set to a fixed word length, the unit magnitude of the target n-gram can also be set to a fixed word length, and setting the fixed word length to a smaller word length can also reduce the size of the recall stage and further improve the overall matching efficiency of the value domain.

[0082] According to the embodiments provided in the present application, the extraction operation in the n-gram dimension is performed on each source value meaning corresponding to a source value domain according to the target format to obtain a source n-gram in the target format corresponding to each source value domain, and the extraction operation in the n-gram dimension is performed on the target value meaning according to the target format to obtain a target n-gram in the target format, thereby achieving the purpose of unifying the formats of the target n-gram and the source n-gram, and realizing the technical effect of improving the overall matching efficiency of the value domain.

[0083] As an optional solution, before obtaining each source value meaning corresponding to each source value domain in the at least two source value domains, and the target value meaning corresponding to the target reference value domain, the method further includes:

[0084] S1, obtaining a plurality of source value domains to be matched and a target reference value domain, wherein the plurality of source value domains to be matched include at least two source value domains;

[0085] S2, perform identical recall on the plurality of source value domains to be matched based on the target reference value domain, to obtain a plurality of identically recalled source value domains, and determine the plurality of identically recalled source value domains as the at least two source value domains.

[0086] Optionally, in the embodiment, the identical recall can be, but is not limited to, value domain identical check. For example, for a value domain in a certain data standard, there is repetition, identity, inconsistency, etc. in its own definition, which is usually found in some more complex value domains.

[0087] It should be noted that, since the source value domain and the reference value domain input by the engine are the same, their value domains and value meanings are usually all consistent. For such identical value domains, if the two recalls in the above value domain matching method are still repeatedly performed, the calculation amount is not compressed, which may cause the problem of reduced matching efficiency of the value domains. Therefore, for such identical value domains, the embodiment performs identical recall on the plurality of source value domains to be matched based on the target reference value domain, and determines the plurality of identically recalled source value domains as the at least two source value domains. That is, the identical value domains need to be pre-filtered before the two recalls in the above value domain matching method are performed, so as to avoid the problem of reduced matching efficiency of the value domains.

[0088] According to the embodiments provided in the present application, the plurality of source value domains to be matched and the target reference value domain are obtained, wherein the plurality of source value domains to be matched include at least two source value domains; identical recall is performed on the plurality of source value domains to be matched based on the target reference value domain, to obtain a plurality of identically recalled source value domains, and the plurality of identically recalled source value domains are determined as the at least two source value domains, thereby achieving the purpose of avoiding the problem of reduced matching efficiency of the value domains, and realizing the technical effect of improving the overall matching efficiency of the value domains.

[0089] As an optional solution, the identical recall is performed on the plurality of source value domains to be matched based on the target reference value domain, to obtain a plurality of identically recalled source value domains, and the plurality of identically recalled source value domains are determined as the at least two source value domains, including:

[0090] S1, obtaining a source value domain length corresponding to each source value domain to be matched in the plurality of source value domains to be matched, and a target value length corresponding to the target reference value domain;

[0091] S2, calculating third similarity information between the source value domain length corresponding to each source value domain to be matched and the target value length, and performing recall on each source value domain to be matched based on the third similarity information, to obtain a third recall result;

[0092] S3, calculating fourth similarity information between the source value meaning matched by the third recall result and the target value meaning, and performing recall on the third recall result based on the third similarity information, to obtain a fourth recall result, wherein the source value domain matched by the fourth recall result includes a plurality of identically recalled source value domains.

[0093] S4, determine the source value domain matched with the fourth recall result as at least two source value domains.

[0094] Optionally, in the embodiment, the value domain length index-based similar value domain fast recall technique can be used for similar value domains, but is not limited to this. For example, the value domain length index-based similar value domain fast recall technique can be used for similar value domains, and the value domain length index-based similar value domain fast recall technique can be used for similar value domains. Figure 9 As shown in the figure, first, an index table 1 of value meanings to value domains in the data standard is constructed, and an index table 2 of value domain lengths to value domains is constructed, such as the index to the two value domains of the “anesthesia mode code table” and the “drug type code” with a value domain length of 6 (wherein there are 6 value meanings);

[0095] Further, ratio threshold calculation and reverse value domain length combination are performed, such as reuse of the ratio threshold filtering in the normal recall stage, but in reverse, that is, it is calculated in advance which combination of value domain lengths can be filtered by the ratio threshold; assuming that the calculation of value domains A / B, by default, as many value meanings as possible are mapped to be found in the recall stage, then the maximum value is min(L A , L B ), that is, the shorter length in the value domains A / B, and this default means that all value meanings in the short value domain can be found in the value meanings in the long value domain. Therefore, all value domain length combinations that satisfy the following formula (1) are traversed; in addition, for lengths of 6 and 50, they cannot pass the following formula because when σ is equal to 1.2, 50 / / 1.2=41 is not smaller than 6, and thus it is not true. As can be seen, this step of operation can skip a large number of candidate value domain pairs that do not match in value domain length, thereby reducing redundant calculation.

[0096] f(L A )≤min(L A , L B ), f(L B )≤min(L A , L B ) (1)

[0097] Further, for the recall of the index table 2, the matching result based on the value domain length is used to select the candidate value domain pairs; and for the recall of the index table 1, a recall is performed in the recall stage based on the cosine similarity threshold screening, and the search space in this recall is no longer all value domain combinations, but is limited in the selection of the above candidate value domain pairs, and the recall result for the index table 1 is a plurality of candidate value domain pairs.

[0098] Further, after the recall stage ends, the recall result is normally input to the subsequent fine arrangement and rearrangement modules, and then the value range checking result is obtained. It can be seen that, by constructing the value range length and the pre-filtering linkage with the ratio threshold, the recall implemented in this way can greatly reduce the candidate value range pairs that do not match in value range length (because these value range pairs will not pass the recall stage in the first place, and the word vector recall and more calculations performed on these value range pairs are all useless work), thereby reducing the time overhead of the recall stage.

[0099] According to the embodiments provided in the present application, the source value range length corresponding to each of the plurality of source value ranges to be matched and the target value length corresponding to the target reference value range are obtained; the third similarity information between the source value range length corresponding to each of the plurality of source value ranges to be matched and the target value length is calculated, and each of the plurality of source value ranges to be matched is recalled based on the third similarity information to obtain a third recall result; the fourth similarity information between the source value meaning matched by the third recall result and the target value meaning is calculated, and the third recall result is recalled based on the third similarity information to obtain a fourth recall result, wherein the source value range matched by the fourth recall result includes a plurality of similar recalled source value ranges; and the source value range matched by the fourth recall result is determined as at least two source value ranges, thereby achieving the purpose of reducing the time overhead of the recall stage, and realizing the technical effect of improving the overall matching efficiency of the value range.

[0100] As an optional solution, after the first similarity information between the source word group corresponding to each of the plurality of source value ranges to be matched and the target value meaning is calculated, and each of the plurality of source value ranges to be matched is recalled once based on the first similarity information to obtain a first recall result, the method further includes at least one of the following:

[0101] S1, determining the source value range matched by the first recall result, and counting the recall times of the source value range matched by the first recall result;

[0102] S2, determining the source value range matched by the second recall result, and counting the recall times of the source value range matched by the second recall result.

[0103] As an optional solution, the alignment matching of the source value range matched by the second recall result and the target reference value range includes:

[0104] S1, obtaining the cumulative recall times of the source value range matched by the second recall result;

[0105] S21, filtering the source value range matched by the second recall result and having a recall time less than a preset threshold, and retaining the target source value range matched by the second recall result and having a recall time greater than or equal to the preset threshold;

[0106] S3, performing alignment matching on the target source value range and the target reference value range.

[0107] Optionally, in the embodiment, to further improve the matching efficiency, the number of recalls of the value domain is counted after the recall (primary recall and / or secondary recall), and it is determined whether it is necessary to perform a subsequent matching operation based on the counted number of recalls, thereby reducing some unnecessary matching operations and improving the execution effectiveness of the matching operation.

[0108] According to the embodiments provided in the application, the number of recalls of the source value domain matched by the second recall result is obtained; the source value domain with the number of recalls less than the preset threshold in the source value domain matched by the second recall result is filtered, and the target source value domain with the number of recalls greater than or equal to the preset threshold in the source value domain matched by the second recall result is retained; and the target source value domain and the target reference value domain are aligned and matched, thereby achieving the purpose of improving the execution effectiveness of the matching operation, and realizing the technical effect of improving the matching efficiency of the value domain.

[0109] As an optional solution, the extraction operation in the phrase dimension is performed on the source value meaning corresponding to each source value domain, to obtain the source phrase corresponding to each source value domain, including:

[0110] S1, the extraction operation in the phrase dimension is performed on the source value meaning corresponding to each source value domain, to obtain the initial source phrase corresponding to each source value domain;

[0111] S2, the frequency of occurrence of the initial source phrase in the source value meaning corresponding to each source value domain is counted, and the initial source phrase with the frequency of occurrence less than the preset frequency is filtered, and the source phrase corresponding to each source value domain with the frequency of occurrence greater than or equal to the preset frequency is retained.

[0112] Optionally, in the embodiment, to improve the effectiveness of subsequent phrase recall, the word frequency filtering of the phrase is increased, such as counting the frequency of the phrase before the primary recall stage and removing the low-frequency phrases, such as the phrase "part of the anesthesia" in the value meaning "local anesthesia" is filtered because of low frequency, so that only the phrases "local" and "anesthesia" are retained.

[0113] According to the embodiments provided in the application, the extraction operation in the phrase dimension is performed on the source value meaning corresponding to each source value domain, to obtain the initial source phrase corresponding to each source value domain; the frequency of occurrence of the initial source phrase in the source value meaning corresponding to each source value domain is counted, and the initial source phrase with the frequency of occurrence less than the preset frequency is filtered, and the source phrase corresponding to each source value domain with the frequency of occurrence greater than or equal to the preset frequency is retained, thereby achieving the purpose of improving the effectiveness of subsequent phrase recall, and realizing the technical effect of improving the matching efficiency of the value domain.

[0114] As an optional solution, for the convenience of understanding, the above value range matching method is applied in the medical business scenario. The medical data standard is the cornerstone of medical informatization, and the value range in the data standard is a key element in the application of medical big data and medical artificial intelligence. Value range mapping (matching) is a key step to connect multiple data standards. It realizes the automatic mapping of source value range to benchmark value range through algorithm technology, helps to reduce the time of a large number of manual alignment, and thus achieves the purpose of reducing cost and increasing efficiency. Among them, the value range mapping function mainly provides automatic mapping from the source value range to the benchmark value range, such as the interface 1002 of the data standard platform shown in Figure 10 . The automatic mapping from the source value range to the benchmark value range, after selecting the "SZCV01.02.001-Patient Source Code Table" source value range, entering the intelligent mapping function and selecting the "GD99.02.350-Local Sign Code Table" benchmark value range, after calculation by the value range matching engine, the product outputs the following automatic mapping results, where the left side is the value range mapping result, and the right side is the value meaning mapping result of a pair of mapping value ranges (such as "1 local" in Figure 10 , etc.). When the user clicks on other mapping value ranges, the right side area will be automatically replaced with the corresponding value meaning matching result;

[0115] In addition, in this embodiment, in addition to the value range mapping function, the data standard platform also provides a value range internal duplicate checking function. This function reuses the value range matching engine, only inputs the same set of value ranges as the source value range and the benchmark value range. In this way, by mapping itself, some repeated / duplicate / inconsistent value meaning definitions within the same value range can be mined, such as in "CT01.00.030-Surgical Operation Site Code", there are two repeated value meanings left upper arm (although their values are different). Such duplicate value meaning mining can help users and standard administrators discover inconsistencies in value ranges in a timely manner, and the completely automated way also greatly reduces the manpower and time consumed by manual checking.

[0116] It should be noted that in the value range mapping and value range duplicate checking functions, the recall, precision, and rearrangement stages of the standard version value range matching engine can effectively realize the alignment of a large number of value ranges, but the content included in the value range of some specific versions or local standards is extremely large, which poses a huge challenge to the overall time consumption of the engine. And because the engine needs to read all candidate value range pairs into the server cache, when the candidate quantity is very large, it may even have a negative impact on server operation. Further to solve this problem and reduce invalid redundant calculations, this embodiment proposes a value range matching acceleration technology based on index grouping, which includes a multi-level grouping recall technology and a duplicate value range fast recall technology based on value range length index.

[0117] For further illustration, the optional recall stage based on cosine similarity threshold screening is, for exampleFigure 11 As shown, first, a double index table is established, assuming that b represents a value meaning, such as "1-local", and B represents a value domain, such as "SZCV01.02.001-patient source code table". The purpose of establishing the double index table is to index map all value meanings in the source value domain to the value domain in the reference value domain; wherein the same value meaning can appear in multiple value domains, and similarly, multiple value meanings can exist in one value domain, and thus the index here is a many-to-many relationship. Moreover, the index table is established independently, that is, the index tables of the source value domain and the reference value domain are not shared. The establishment of the double index table can greatly reduce the overall operation amount of the engine, because a value meaning such as "99-other" can appear in a large number of different value domains, and after constructing the index, it only needs to be calculated once;

[0118] Furthermore, after obtaining the recalled value meaning pair, the embodiment also inverts the index to obtain the corresponding value domain pair as the input of the subsequent step; further, for the b2 value meaning, corresponding to the B1 / B2 value domains, and for the a1 value meaning, corresponding to the A3 value domain, if the b2 and a1 value meaning pair can be recalled, then the scores of the B1-A3 and B2-A3 value domain pairs can be obtained, respectively, and the scores are added by one, indicating that there is a same / similar value meaning pair on the mapping. After traversing the entire double index table, the scores of all candidate value domain pairs are obtained, for example, A1-B1=3, A1-B6=3, and the like, and then the embodiment performs score / frequency filtering, that is, only the value domain pairs with a score greater than a certain threshold value can be accepted.

[0119] In addition, considering that the value domain length (i.e., the number of value meanings contained in the value domain) often differs greatly in actual business scenarios, if a certain threshold value is uniformly determined, it can lead to the recall of some super-long value domain pairs, for example, when "yes", "no", and "unknown" value meanings exist in two value domains with hundreds of rows, this situation is particularly obvious. Therefore, the embodiment proposes a way of automatically calculating the threshold value using the ratio (referred to as the ratio threshold method). For a group of candidate value domains A / B, the value domain lengths are L_A / L_B, and among the X value meaning pairs, the threshold value of 3.2.1.2 is calculated, and then two ratios are calculated, as shown in the following formula (2):

[0120] θ A =f(L A )=L A / / τ,θ B =f(L B )=L B / / τ (2)

[0121] Wherein, / / is the integer division sign, for example, when σ is equal to 1.2, the calculation example is as shown in the following formula (3):

[0122] θA = f(12) = 12 / / τ = 10, θ B = f(7) = 7 / / τ = 5 (3)

[0123] At this time, only when X satisfies both greater than the first preset threshold (such as 10) and greater than the second preset threshold (such as 5), the candidate value range A / B can be recalled. It can be seen that such a two-way value range calculation can ensure that there is no problem of candidate number surge due to an excessively long value range, and further reduces the calculation amount of the subsequent steps.

[0124] Through the embodiments provided in the present application, invalid redundant calculations are reduced, and the overall time overhead of the engine is compressed. In order to verify the gain of the engine overall time consumption of the fast recall technology based on the value range length index for verifying the similar value range, the present embodiment tests the value range checking function time consumption of the initial version engine and the accelerated version engine on two data standards respectively, and the experimental results are shown in the following table:

[0125]

[0126]

[0127] As can be seen, under the relatively small-scale data standard of SZ, the fast recall technology can accelerate the engine time consumption from 0.466s to 0.216s, reducing by 53.6%. Under the larger-scale data standard of GD, although the same proportion of reduction is 60.3%, the acceleration technology compresses the engine time consumption from 9.047s to 3.596s, which is nearly 6s of time consumption reduction, which is very obvious in actual use, which significantly improves the user experience. In addition, the engine can maintain a relatively consistent time compression ratio under different data standards, which also shows the universality of the present embodiment, which can be compatible with more real-world data standards and business scenarios.

[0128] It can be understood that in the specific embodiments of the present application, data related to user information is involved. When the above embodiments of the present application are applied to specific products or technologies, user permission or consent needs to be obtained, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0129] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited to the action order described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0130] According to another aspect of the embodiments of the present application, there is also provided a value range matching apparatus for implementing the above-mentioned value range matching method. As shown in Figure 12 The apparatus comprises:

[0131] A first obtaining unit 1202 is configured to obtain source value meanings corresponding to respective source value ranges in the at least two source value ranges, and a target value meaning corresponding to the target reference value range;

[0132] An extracting unit 1204 is configured to perform an extracting operation in the phrase dimension on the source value meanings corresponding to the respective source value ranges, to obtain source phrases corresponding to the respective source value ranges.

[0133] A first recalling unit 1206 is configured to calculate first similarity information between the source phrases corresponding to the respective source value ranges and the target value meaning, and perform a first recall on the source value meanings corresponding to the respective source value ranges based on the first similarity information, to obtain a first recall result.

[0134] A second recalling unit 1208 is configured to calculate second similarity information between the source value meanings matched by the first recall result and the target value meaning, and perform a second recall on the first recall result based on the second similarity information, to obtain a second recall result, wherein the source value meanings corresponding to the respective source value ranges include the source value meanings matched by the first recall result.

[0135] A matching unit 1210 is configured to perform alignment matching on the source value ranges matched by the second recall result and the target reference value range, so that the source value ranges matched by the second recall result are mapped onto the target reference value range, wherein the at least two source value ranges include the source value ranges matched by the second recall result.

[0136] The specific embodiments can refer to the examples shown in the above-mentioned value range matching apparatus, which will not be described herein again.

[0137] As an optional solution, the first recalling unit 1206 comprises a first calculating module configured to calculate the first similarity information, and filter at least one candidate value meaning from the source value meanings corresponding to the respective source value ranges based on the first similarity information.

[0138] The second recalling unit 1208 comprises a second calculating module configured to calculate the second similarity information, and filter at least one second candidate value range from the first candidate value ranges corresponding to the at least one candidate value meaning based on the second similarity information, wherein the at least two source value ranges include the first candidate value ranges corresponding to the at least one candidate value meaning.

[0139] The matching unit 1210 comprises a first matching module configured to perform alignment matching on respective second candidate value ranges in the at least one second candidate value range and the target reference value range.

[0140] The specific embodiments can refer to the examples shown in the above value range matching method, and details are not described herein in this example.

[0141] As an optional solution, the first calculation module comprises:

[0142] The calculation sub-module is configured to calculate word group similarity information between the source word group corresponding to each source value range and the target word group in the target value meaning.

[0143] The screening sub-module is configured to screen at least one candidate value meaning with a word group similarity greater than or equal to a preset similarity from the source value meaning corresponding to each source value range.

[0144] The specific embodiments can refer to the examples shown in the above value range matching method, and details are not described herein in this example.

[0145] As an optional solution, the extraction unit 1204 comprises a first extraction module configured to perform an extraction operation in the word group dimension on the source value meaning corresponding to each source value range according to the target format, to obtain a source word group in the target format corresponding to each source value range.

[0146] The device further comprises a second extraction module configured to perform an extraction operation in the word group dimension on the target value meaning according to the target format, to obtain a target word group in the target format, before calculating the word group similarity information between the source word group corresponding to each source value range and the target word group in the target value meaning.

[0147] The specific embodiments can refer to the examples shown in the above value range matching method, and details are not described herein in this example.

[0148] As an optional solution, the device further comprises:

[0149] The second acquisition unit is configured to acquire a plurality of source value ranges to be matched and a target reference value range before acquiring the source value meaning corresponding to each source value range in the at least two source value ranges and the target value meaning corresponding to the target reference value range, wherein the plurality of source value ranges to be matched include the at least two source value ranges.

[0150] The third recall unit is configured to perform a similar recall on the plurality of source value ranges to be matched based on the target reference value range before acquiring the source value meaning corresponding to each source value range in the at least two source value ranges and the target value meaning corresponding to the target reference value range, to obtain a plurality of similar recalled source value ranges, and to determine the plurality of similar recalled source value ranges as the at least two source value ranges.

[0151] The specific embodiments can refer to the examples shown in the above value range matching method, and details are not described herein in this example.

[0152] As an optional solution, the third recall unit comprises:

[0153] The first obtaining module is configured to obtain a source value domain length corresponding to each of a plurality of to-be-matched source value domains and a target value length corresponding to a target reference value domain;

[0154] The first recalling module is configured to calculate third similarity information between the source value domain length corresponding to each of the to-be-matched source value domains and the target value length, and perform recalling on each of the to-be-matched source value domains based on the third similarity information to obtain a third recalling result;

[0155] The second recalling module is configured to calculate fourth similarity information between a source value meaning matched by the third recalling result and a target value meaning, and perform recalling on the third recalling result based on the third similarity information to obtain a fourth recalling result, wherein the source value domain matched by the fourth recalling result includes a plurality of homologous recalled source value domains.

[0156] The determining module is configured to determine the source value domain matched by the fourth recalling result as at least two source value domains.

[0157] The specific embodiments can refer to the examples shown in the above value domain matching method, which will not be described herein again in this example.

[0158] As an optional solution, the apparatus further includes at least one of the following:

[0159] The first determining unit is configured to, after calculating the first similarity information between the source word group corresponding to each of the source value domains and the target value meaning and performing one-time recalling on the source value meaning corresponding to each of the source value domains based on the first similarity information to obtain a first recalling result, determine the source value domain matched by the first recalling result and perform counting of the number of times of recalling on the source value domain matched by the first recalling result.

[0160] The second determining unit is configured to, after calculating the first similarity information between the source word group corresponding to each of the source value domains and the target value meaning and performing one-time recalling on the source value meaning corresponding to each of the source value domains based on the first similarity information to obtain a first recalling result, determine the source value domain matched by the second recalling result and perform counting of the number of times of recalling on the source value domain matched by the second recalling result.

[0161] The specific embodiments can refer to the examples shown in the above value domain matching method, which will not be described herein again in this example.

[0162] As an optional solution, the matching unit 1210 includes:

[0163] The second obtaining module is configured to obtain the number of times of recalling accumulated by the source value domain matched by the second recalling result.

[0164] The filtering module is configured to filter the source value domain in the second recall result matching source value domain with the recall times less than the preset threshold, and keep the target source value domain in the second recall result matching source value domain with the recall times greater than or equal to the preset threshold.

[0165] The second matching module is configured to perform alignment matching on the target source value domain and the target reference value domain.

[0166] The specific embodiments can refer to the examples shown in the above value domain matching method, which will not be repeated here in this example.

[0167] As an optional solution, the extraction unit 1204 includes:

[0168] The third extraction module is configured to perform extraction operation on the source value meaning corresponding to each source value domain in the phrase dimension, to obtain the initial source phrase corresponding to each source value domain.

[0169] The statistical module is configured to count the occurrence frequency of the initial source phrase in the source value meaning corresponding to each source value domain, filter the initial source phrase with the occurrence frequency less than the preset frequency, and keep the source phrase corresponding to each source value domain with the occurrence frequency greater than or equal to the preset frequency.

[0170] The specific embodiments can refer to the examples shown in the above value domain matching method, which will not be repeated here in this example.

[0171] According to another aspect of the embodiments of the present application, an electronic device for implementing the above value domain matching method is also provided, as shown in the figure, the electronic device includes a memory 1302 and a processor 1304, the memory 1302 stores a computer program, and the processor 1304 is configured to execute the steps in any of the above method embodiments through the computer program. Figure 13

[0172] Optionally, in the present embodiment, the above electronic device can be located in at least one network device of a plurality of network devices of a computer network.

[0173] Optionally, in the present embodiment, the above processor can be configured to execute the following steps through the computer program:

[0174] S1, obtaining the source value meaning corresponding to each source value domain in the at least two source value domains, and the target value meaning corresponding to the target reference value domain;

[0175] S21, performing extraction operation on the source value meaning corresponding to each source value domain in the phrase dimension, to obtain the source phrase corresponding to each source value domain;

[0176] ​S3, calculate the first similarity information between the source word group corresponding to each source value domain and the target value meaning, and perform a recall on the source value meaning corresponding to each source value domain based on the first similarity information to obtain the first recall result;

[0177] S4, calculate the second similarity information between the source value meaning and the target value meaning matched by the first recall result, and perform a second recall on the first recall result based on the second similarity information to obtain the second recall result, wherein the source value meaning corresponding to each source value domain includes the source value meaning matched by the first recall result;

[0178] S5, perform alignment matching between the source value domain matched by the second recall result and the target reference value domain, so that the source value domain matched by the second recall result is aligned and mapped onto the target reference value domain, wherein at least two source value domains include the source value domain matched by the second recall result.

[0179] Alternatively, as those skilled in the art will understand, Figure 13 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones (such as Android phones, iOS phones, etc.), tablets, PDAs, mobile internet devices (MIDs), PADs, and other terminal devices. Figure 13 This does not limit the structure of the aforementioned electronic devices. For example, the electronic device may also include components that are more... Figure 13 The more or fewer components shown (such as network interfaces, etc.), or having the same Figure 13 The different configurations shown.

[0180] The memory 1302 can be used to store software programs and modules, such as the program instructions / modules corresponding to the value range matching method and apparatus in this embodiment. The processor 1304 executes various functional applications and data processing by running the software programs and modules stored in the memory 1302, thereby implementing the aforementioned value range matching method. The memory 1302 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1302 may further include memory remotely located relative to the processor 1304, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Specifically, the memory 1302 may be used, but is not limited to, to store information such as the meaning of source values, the meaning of target values, source phrases, first recall results, and second recall results. As an example, such as Figure 13As shown, the memory 1302 can include, but is not limited to, the first obtaining unit 1202, the extraction unit 1204, the first recalling unit 1206, the second recalling unit 1208 and the matching unit 1210 in the value range matching apparatus. In addition, other module units in the value range matching apparatus can also be included, which will not be described herein.

[0181] Optionally, the transmission device 1306 is configured to receive or send data via a network. The network can include a wired network and a wireless network. In an example, the transmission device 1306 includes a network adapter (NIC) which can be connected to other network devices and routers through a network cable to communicate with the Internet or a local area network. In an example, the transmission device 1306 is a radio frequency (RF) module which is configured to communicate with the Internet in a wireless manner.

[0182] In addition, the electronic device further includes a display 1308 configured to display the source value meaning, the target value meaning, the source phrase, the first recall result and the second recall result, and a connection bus 1310 configured to connect various module components in the electronic device.

[0183] In other embodiments, the terminal device or the server can be a node in a distributed system, and the distributed system can be a blockchain system. The blockchain system can be a distributed system formed by the plurality of nodes communicating through a network. The nodes can form a peer-to-peer (P2P) network, and any computing device, such as a server or a terminal, can become a node in the blockchain system by joining the P2P network.

[0184] According to an aspect of the present application, a computer program product is provided, which includes computer programs / instructions containing program codes for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit, various functions provided by the embodiments of the present application are performed.

[0185] The serial numbers of the embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0186] It should be noted that the computer system of the electronic device is only an example, and should not limit the functions and use range of the embodiments of the present application.

[0187] The computer system includes a central processing unit (CPU) which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) or loaded from a storage section into a random access memory (RAM). In the random access memory, various programs and data required for the operation of the system are also stored. The central processing unit, the read-only memory, and the random access memory are connected to each other through a bus. An input / output interface (I / O interface) is also connected to the bus.

[0188] Connected to the input / output interface are an input section including a keyboard, a mouse, etc.; an output section including a display such as a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section including a hard disk, etc.; and a communication section including a network interface card such as a local area network card, a modem, etc. The communication section performs communication processing via a network such as the Internet. A drive is also connected to the input / output interface as necessary. A removable recording medium such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is attached to the drive as necessary, so that a computer program read out therefrom is installed into the storage section as necessary.

[0189] In particular, according to embodiments of the present application, the processes described in the various method flowcharts can be implemented as a computer software program. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program code for executing the methods illustrated in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section, and / or installed from a removable recording medium. When the computer program is executed by the central processing unit, various functions defined in the system of the present application are performed.

[0190] According to an aspect of the present application, there is provided a computer readable storage medium, from which a processor of a computer device reads computer instructions, and the processor executes the computer instructions to cause the computer device to perform the method provided in the various optional implementations described above.

[0191] Optionally, in the present embodiment, the computer readable storage medium described above can be configured to store a computer program for executing the following steps:

[0192] S1, obtaining source value meanings corresponding to each source value domain in at least two source value domains, and target value meanings corresponding to a target reference value domain;

[0193] S21, performing an extraction operation under a phrase dimension on the source value meaning corresponding to each source value domain to obtain a source phrase corresponding to each source value domain;

[0194] S3, calculating first similarity information between the source phrase corresponding to each source value domain and the target value meaning, and performing a first recall on the source value meaning corresponding to each source value domain based on the first similarity information to obtain a first recall result;

[0195] S4, calculating second similarity information between the source value meaning matched by the first recall result and the target value meaning, and performing a second recall on the first recall result based on the second similarity information to obtain a second recall result, wherein the source value meaning corresponding to each source value domain includes the source value meaning matched by the first recall result;

[0196] S5, performing alignment matching on the source value domain matched by the second recall result and the target reference value domain, so that the source value domain matched by the second recall result is aligned and mapped onto the target reference value domain, wherein the at least two source value domains include the source value domain matched by the second recall result.

[0197] Optionally, in the embodiment, all or part of the steps in the various methods of the above-described embodiments can be completed by a program instructing the hardware related to the terminal device, and the program can be stored in a computer-readable storage medium, which can include a flash disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0198] The serial numbers of the above-described embodiments of the present application are only for description, and do not represent the advantages or disadvantages of the embodiments.

[0199] The integrated units in the above-described embodiments, if realized in the form of software function units and sold or used as independent products, can be stored in the above-described computer-readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the parts that make contributions to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing one or more computer devices (which can be personal computers, servers or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0200] In the above-described embodiments of the present application, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0201] In several embodiments provided in the present application, it should be understood that the disclosed client can be implemented in other manners. Of course, the described apparatus embodiments are merely schematic, and the division of units is merely logical function division, and there can be other division manners in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, units or modules, and can be in electrical, mechanical or other forms.

[0202] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0203] In addition, each functional unit in the various embodiments of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0204] The above is only the preferred embodiment of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, and these improvements and refinements should be considered as the protection scope of the present application.

Claims

1. A value range matching method, characterized by, The method comprises: obtaining source value meanings corresponding to each of at least two source value domains and a target value meaning corresponding to a target reference value domain; performing an extraction operation in a phrase dimension on the source value meanings corresponding to each of the source value domains to obtain source phrases corresponding to each of the source value domains; calculating first similarity information between the source phrases corresponding to each of the source value domains and the target value meaning, and performing a first recall on the source value meanings corresponding to each of the source value domains based on the first similarity information to obtain a first recall result; calculating second similarity information between source value meanings matching the first recall result and the target value meaning, and performing a second recall on the first recall result based on the second similarity information to obtain a second recall result, wherein the source value meanings corresponding to each of the source value domains include the source value meanings matching the first recall result; aligning and matching the source value domains matching the second recall result with the target reference value domain to map the source value domains matching the second recall result to the target reference value domain, wherein the at least two source value domains include the source value domains matching the second recall result.

2. The method of claim 1, wherein: the calculating the first similarity information between the source phrases corresponding to each of the source value domains and the target value meaning, and performing the first recall on the source value meanings corresponding to each of the source value domains based on the first similarity information to obtain the first recall result comprises: calculating the first similarity information and screening at least one candidate value meaning from the source value meanings corresponding to each of the source value domains based on the first similarity information; the calculating the second similarity information between the source value meanings matching the first recall result and the target value meaning, and performing the second recall on the first recall result based on the second similarity information to obtain the second recall result comprises: calculating the second similarity information and screening at least one second candidate value domain from a first candidate value domain corresponding to the at least one candidate value meaning based on the second similarity information, wherein the at least two source value domains include the first candidate value domain corresponding to the at least one candidate value meaning; the aligning and matching the source value domains matching the second recall result with the target reference value domain comprises: aligning and matching each of the at least one second candidate value domain with the target reference value domain.

3. The method of claim 2, wherein, The calculating the first similarity information and screening at least one candidate value meaning from the source value meanings corresponding to each of the source value domains based on the first similarity information comprises: calculating phrase similarity information between the source phrases corresponding to each of the source value domains and target phrases in the target value meaning; screening the at least one candidate value meaning with a phrase similarity degree corresponding to the phrase similarity information greater than or equal to a preset similarity degree from the source value meanings corresponding to each of the source value domains.

4. The method of claim 3, wherein: The extracting operation under the phrase dimension is performed on the source value meaning corresponding to each source value domain according to the target format, and source phrases in the target format corresponding to each source value domain are obtained. Before the calculating of the phrase similarity information between the source phrases corresponding to each source value domain and the target phrase in the target value meaning, the method further comprises:

5. The method of claim 1, wherein, Before the obtaining of the source value meaning corresponding to each source value domain in the at least two source value domains and the target value meaning corresponding to the target reference value domain, the method further comprises: Obtaining a plurality of source value domains to be matched and the target reference value domain, wherein the plurality of source value domains to be matched comprises the at least two source value domains; Performing a similar recall on the plurality of source value domains to be matched based on the target reference value domain, obtaining a plurality of similar-recalled source value domains, and determining the plurality of similar-recalled source value domains as the at least two source value domains.

6. The method of claim 5, wherein, The performing of the similar recall on the plurality of source value domains to be matched based on the target reference value domain, the obtaining of the plurality of similar-recalled source value domains, and the determining of the plurality of similar-recalled source value domains as the at least two source value domains comprise: Obtaining a source value domain length corresponding to each source value domain to be matched in the plurality of source value domains to be matched and a target value length corresponding to the target reference value domain; Calculating third similarity information between the source value domain length corresponding to each source value domain to be matched and the target value length, and performing a recall on each source value domain to be matched based on the third similarity information to obtain a third recall result; Calculating fourth similarity information between the source value meaning matched by the third recall result and the target value meaning, and performing a recall on the third recall result based on the third similarity information to obtain a fourth recall result, wherein the source value domain matched by the fourth recall result comprises the plurality of similar-recalled source value domains; Determining the source value domain matched by the fourth recall result as the at least two source value domains.

7. The method of claim 1, wherein, After the calculating of the first similarity information between the source phrase corresponding to each source value domain and the target value meaning, and the performing of a first recall on the source value meaning corresponding to each source value domain based on the first similarity information to obtain a first recall result, the method further comprises at least one of the following: Determining the source value domain matched by the first recall result, and counting the number of recalls on the source value domain matched by the first recall result; Determining the source value domain matched by the second recall result, and counting the number of recalls on the source value domain matched by the second recall result.

8. The method of claim 7, wherein, The aligning and matching of the source value domain matched by the second recall result and the target reference value domain comprise: Obtaining the number of recalls accumulated by the source value domain matched by the second recall result; filtering source value domains in which the number of recalls of the second recall result matches is less than a preset threshold, and retaining target source value domains in which the number of recalls of the second recall result matches is greater than or equal to the preset threshold; aligning and matching the target source value domains and the target reference value domains.

9. The method according to any one of claims 1 to 8, characterized in that, The extracting operation in the phrase dimension is performed on the source value meanings corresponding to the respective source value domains to obtain source phrases corresponding to the respective source value domains, including: The extracting operation in the phrase dimension is performed on the source value meanings corresponding to the respective source value domains to obtain initial source phrases corresponding to the respective source value domains; The frequency of occurrence of the initial source phrases in the source value meanings corresponding to the respective source value domains is counted, and initial source phrases in which the frequency of occurrence is less than a preset frequency are filtered, and source phrases corresponding to the respective source value domains in which the frequency of occurrence is greater than or equal to the preset frequency are retained.

10. A range matching apparatus, characterized by comprising: including: The first obtaining unit is configured to obtain source value meanings corresponding to respective source value domains in at least two source value domains, and a target value meaning corresponding to a target reference value domain; The extracting unit is configured to perform an extracting operation in a phrase dimension on the source value meanings corresponding to the respective source value domains to obtain source phrases corresponding to the respective source value domains; The first recalling unit is configured to calculate first similarity information between the source phrases corresponding to the respective source value domains and the target value meaning, and perform a first recall on the source value meanings corresponding to the respective source value domains based on the first similarity information to obtain a first recall result; The second recalling unit is configured to calculate second similarity information between source value meanings matched by the first recall result and the target value meaning, and perform a second recall on the first recall result based on the second similarity information to obtain a second recall result, wherein the source value meanings corresponding to the respective source value domains include the source value meanings matched by the first recall result; The matching unit is configured to perform alignment matching on source value domains matched by the second recall result and the target reference value domain, so that the source value domains matched by the second recall result are aligned and mapped onto the target reference value domain, wherein the at least two source value domains include the source value domains matched by the second recall result.

11. A computer readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein the program can be run by a terminal device or a computer to execute the method described in any one of claims 1 to 9.

12. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instructions are executed by the processor to implement the steps of the method described in any one of claims 1 to 9.

13. An electronic device comprising a memory and a processor, characterized in that The memory stores a computer program, and the processor is configured to execute the method described in any one of claims 1 to 9 by using the computer program. The memory stores a computer program, and the processor is configured to execute the method described in any one of claims 1 to 9 by using the computer program.

Images