Government-enterprise customer large customer classification method and device, electronic equipment and medium

Abstract

The application provides a government and enterprise customer large customer classification method and device, electronic equipment and computer readable storage medium, and relates to the technical field of data processing. The method comprises the following steps: obtaining a government and enterprise customer name set and a large customer name set; respectively splitting a first government and enterprise customer name and a first large customer name to obtain a first government and enterprise customer name segmentation result and a first large customer name segmentation result, wherein the first large customer name segmentation result comprises N-level distance weight words; calculating the sum of the edit distances between the first government and enterprise customer name segmentation result and the N-level distance weight words of the first large customer name; and in response to the sum of the edit distances being less than a preset classification threshold, classifying the first government and enterprise customer name to the first large customer name. The application at least solves the problem of low accuracy of government and enterprise customer classification in the related art. The application is suitable for data analysis, customer marketing and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a large customer classification method and device for government and enterprise customers, electronic equipment and computer readable storage medium. BACKGROUND

[0002] At present, the industry classification of government and enterprise customers is mainly based on word frequency matrix similarity calculation, and the word frequency matrix similarity is only applicable to large text, such as word frequency matrix similarity calculation by word breaking according to enterprise detail information (such as enterprise introduction, intellectual property, company annual report, etc.), which is difficult to realize large customer classification only according to enterprise name.

[0003] In addition, the matching result in the word frequency matrix similarity is two results of yes or no, which has a coarse granularity, and since the names of group companies and subsidiary companies are often not related, the word frequency matrix similarity calculation cannot realize accurate classification of government and enterprise customers. SUMMARY

[0004] The technical problem to be solved by the present application is to solve the above-mentioned deficiencies of the prior art, and to provide a large customer classification method and device for government and enterprise customers, electronic equipment and computer readable storage medium, so as to at least solve the problem of low accuracy of government and enterprise customer classification in related technology.

[0005] In a first aspect, the present application provides a large customer classification method for government and enterprise customers, comprising: obtaining a government and enterprise customer name set and a large customer name set, wherein the government and enterprise customer name set at least includes a first government and enterprise customer name, and the large customer name set at least includes a first large customer name; breaking the first government and enterprise customer name and the first large customer name into words respectively to obtain a first government and enterprise customer name segmentation result and a first large customer name segmentation result,

[0006] Wherein, the first large customer name segmentation result includes N-level distance weight words, and N is a positive integer; calculating the sum of edit distances between the first government and enterprise customer name segmentation result and the N-level distance weight words of the first large customer name; in response to the sum of edit distances being less than a preset classification threshold, classifying the first government and enterprise customer name to the first large customer name.

[0007] Preferably, N=3, and the number of distance weight words in each level of N-level is greater than or equal to 1.

[0008] The calculation of the sum of edit distances between the first government and enterprise customer name segmentation result and the N-level distance weight words of the first large customer name specifically includes:

[0009] Calculating the product of the edit distance of the first government and enterprise customer name segmentation result and the first-level distance weight word and the first-level weight value to obtain a first-level edit distance;

[0010] The product of the edit distance between the first government and enterprise customer name segmentation result and the second distance weight word and the second weight value is calculated to obtain a second edit distance.

[0011] The product of the edit distance between the first government and enterprise customer name segmentation result and the third distance weight word and the third weight value is calculated to obtain a third edit distance.

[0012] The sum of the first edit distance, the second edit distance, and the third edit distance is calculated to obtain the sum of the edit distances between the first government and enterprise customer name segmentation result and the first large customer name and the third distance weight word.

[0013] Preferably, after the first government and enterprise customer name is classified to the first large customer name in response to the sum of the edit distances being less than the preset classification threshold, the method further comprises: in response to the first government and enterprise customer name being classified to M large customer names, selecting the large customer name corresponding to the minimum value of the sum of the edit distances as the final classified large customer name, where M is a positive integer greater than 1.

[0014] Preferably, after the first government and enterprise customer name and the first large customer name are segmented respectively, and before the sum of the edit distances between the first government and enterprise customer name segmentation result and the first large customer name and the N-level distance weight word is calculated, the method further comprises: obtaining a positive keyword in the first large customer name segmentation result; calculating the edit distance between the first government and enterprise customer name segmentation result and the positive keyword; and in response to the edit distance with the positive keyword being equal to 0, classifying the first government and enterprise customer name to the first large customer name.

[0015] Preferably, after the edit distance between the first government and enterprise customer name segmentation result and the positive keyword is calculated, and before the sum of the edit distances between the first government and enterprise customer name segmentation result and the first large customer name and the N-level distance weight word is calculated, the method further comprises: in response to the edit distance with the positive keyword not being equal to 0, obtaining a reverse keyword of the first large customer name; calculating the edit distance between the first government and enterprise customer name segmentation result and the reverse keyword; and in response to the edit distance with the reverse keyword being equal to 0, determining that the first government and enterprise customer name cannot be classified to the first large customer name.

[0016] Preferably, the first government and enterprise customer name and the first large customer name are segmented respectively to obtain the first government and enterprise customer name segmentation result and the first large customer name segmentation result, specifically comprising: segmenting the first government and enterprise customer name and the first large customer name respectively, and removing the interference words after segmentation to obtain the first government and enterprise customer name segmentation result and the first large customer name segmentation result.

[0017] After the first government and enterprise customer name is classified into the first large customer name in response to the sum of the edit distances being less than the preset classification threshold, the method further comprises: adjusting the forward keyword, the reverse keyword, the N-level distance weight word, and / or reducing the classification threshold to improve the accuracy of the classification result.

[0018] Preferably, the obtaining of the government and enterprise customer name set and the large customer name set comprises: obtaining the government and enterprise customer name set and the large customer name set from a database based on a distributed computing node.

[0019] In a second aspect, the present application further provides a large customer classification device for government and enterprise customers, comprising:

[0020] An obtaining module is configured to obtain a government and enterprise customer name set and a large customer name set, wherein the government and enterprise customer name set at least includes a first government and enterprise customer name, and the large customer name set at least includes a first large customer name.

[0021] A word splitting module is connected to the obtaining module and is configured to split the first government and enterprise customer name and the first large customer name respectively to obtain a first government and enterprise customer name split result and a first large customer name split result, wherein the first large customer name split result includes an N-level distance weight word, and N is a positive integer.

[0022] A calculating module is connected to the word splitting module and is configured to calculate the sum of the edit distances between the first government and enterprise customer name split result and the N-level distance weight word of the first large customer name.

[0023] A classifying module is connected to the calculating module and is configured to classify the first government and enterprise customer name into the first large customer name in response to the sum of the edit distances being less than a preset classification threshold.

[0024] In a third aspect, the present application further provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to implement the large customer classification method for government and enterprise customers according to the first aspect.

[0025] In a fourth aspect, the present application further provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the large customer classification method for government and enterprise customers according to the first aspect.

[0026] This invention provides a method, apparatus, electronic device, and computer-readable storage medium for classifying government and enterprise clients as large clients. It performs word segmentation on the names of government and enterprise clients and large clients, calculates the sum of edit distances between the segmented names of government and enterprise clients and the N-level distance weight words of the large client names, and then determines whether to classify the government and enterprise client names as large client names based on the sum of edit distances. By employing an edit distance algorithm to calculate the distance between names, fine-grained client name classification is achieved. Furthermore, based on the edit distances obtained from distance weight words at different levels, a more comprehensive dimensional distance calculation between names is realized, thereby further improving the accuracy of classification. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating a method for classifying large enterprise clients according to Embodiment 1 of the present invention.

[0028] Figure 2 This is a schematic diagram of the structure of a large customer classification device for government and enterprise customers according to Embodiment 2 of the present invention;

[0029] Figure 3 This is a schematic diagram of the structure of an electronic device according to Embodiment 3 of the present invention. Detailed Implementation

[0030] To enable those skilled in the art to better understand the technical solution of the present invention, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0031] It is understood that the specific embodiments and accompanying drawings described herein are merely for explaining the invention and are not intended to limit the invention.

[0032] It is understood that, without conflict, the various embodiments and features in the embodiments of the present invention can be combined with each other.

[0033] It is understood that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, while the parts unrelated to the present invention are not shown in the drawings.

[0034] It is understood that each unit or module involved in the embodiments of the present invention may correspond to only one entity structure, or may be composed of multiple entity structures, or multiple units or modules may be integrated into one entity structure.

[0035] It is understood that, without conflict, the functions and steps marked in the flowcharts and block diagrams of this invention may occur in a different order than that marked in the accompanying drawings.

[0036] It is understood that the flowcharts and block diagrams of this invention illustrate the possible architecture, functions, and operations of systems, apparatuses, devices, and methods according to various embodiments of this invention. Each block in the flowchart or block diagram may represent a unit, module, program segment, or code, containing executable instructions for implementing the specified function. Furthermore, each block or combination of blocks in the block diagram and flowchart can be implemented using a hardware-based system to achieve the specified function, or using a combination of hardware and computer instructions.

[0037] It is understood that the units and modules involved in the embodiments of the present invention can be implemented by software or by hardware. For example, the units and modules can be located in a processor.

[0038] Example 1:

[0039] like Figure 1 As shown, this embodiment provides a method for classifying large enterprise clients, including:

[0040] Step 101: Obtain the set of government and enterprise customer names and the set of major customer names. The set of government and enterprise customer names includes at least the first government and enterprise customer name, and the set of major customer names includes at least the first major customer name.

[0041] In this embodiment, an example application scenario is to categorize all enterprise and government customer names (such as Beijing United Network Communications Co., Ltd., Unicom (Zhejiang) Cloud Data Co., Ltd., China Unicom Beijing Branch, etc.) under the name "China United Network Communications Co., Ltd." in the system. Subsidiaries under SAIC Motor are categorized under "SAIC Motor," etc. The example of the first largest customer name is China United Network Communications Co., Ltd.

[0042] Optionally, obtaining the set of government and enterprise customer names and the set of major customer names specifically includes: obtaining the set of government and enterprise customer names and the set of major customer names from the database based on distributed computing nodes.

[0043] In this embodiment, to accelerate data acquisition efficiency and computation speed, a set of government and enterprise customer names and a set of major customer names are retrieved from the database using distributed computing nodes. For example, four computing nodes (i.e., host nodes) are simultaneously activated to retrieve and calculate the sets of government and enterprise customer names and major customer names from the database. For the government and enterprise customer name ID (IdentityDocument, identity identification number), the remainder is calculated using 3 (in the database, name IDs are numeric). Names with a remainder of 0 are placed on the first computing node for data acquisition and calculation, those with a remainder of 1 are placed on the second computing node, and so on. The number of computing nodes can be adjusted according to the number of government and enterprise customer names to improve computational efficiency and make rational use of resources.

[0044] In step 102, the first government and enterprise customer name and the first large customer name are respectively segmented, to obtain a first government and enterprise customer name segmentation result and a first large customer name segmentation result, wherein the first large customer name segmentation result includes N-level distance weight words, and N is a positive integer.

[0045] Specifically, the first government and enterprise customer name and the first large customer name are respectively segmented, and the interference words after segmentation are removed, to obtain the first government and enterprise customer name segmentation result and the first large customer name segmentation result.

[0046] In this embodiment, the interference words such as regional names, limited companies and other general words are removed after the first government and enterprise customer name and the first large customer name are segmented, so as to reduce the interference factors of the edit distance calculation and improve the calculation efficiency.

[0047] In step 103, the sum of the edit distances between the first government and enterprise customer name segmentation result and the N-level distance weight words of the first large customer name is calculated.

[0048] In this embodiment, the edit distance algorithm is used to obtain the distance between two words in the segmentation result. For example, the strings corresponding to two segmentation results are s1 and s2, and the edit distance between the two is defined as the minimum number of editing operations for converting s1 to s2 (equivalent to the minimum number of editing operations for converting s2 to s1). The fine-grained classification of the name is realized through the edit distance algorithm. In this embodiment, different levels of distance weight words are also defined, and multiple weight words can be configured at each level, so as to realize the comprehensiveness of the classification distance between the government and enterprise customer name and the large customer name, improve the classification accuracy, and solve the problem that the names of some group companies and subsidiaries are not associated.

[0049] Optionally, N=3, and the number of distance weight words in each level of N-level is greater than or equal to 1. Among them, the sensitivity of the first-level distance weight word is the highest, the sensitivity of the second-level distance weight word is the second, and so on.

[0050] Optionally, the calculating the sum of the edit distances between the first government and enterprise customer name segmentation result and the N-level distance weight words of the first large customer name comprises: calculating a first-level edit distance by multiplying the edit distance between the first government and enterprise customer name segmentation result and a first-level distance weight word by a first-level weight value; calculating a second-level edit distance by multiplying the edit distance between the first government and enterprise customer name segmentation result and a second-level distance weight word by a second-level weight value; calculating a third-level edit distance by multiplying the edit distance between the first government and enterprise customer name segmentation result and a third-level distance weight word by a third-level weight value; and calculating the sum of the first-level edit distance, the second-level edit distance and the third-level edit distance to obtain the sum of the edit distances between the first government and enterprise customer name segmentation result and the three-level distance weight words of the first large customer name, wherein the second-level weight value is greater than the first-level weight value and less than the third-level weight value.

[0051] In the embodiment, the first-level weight value is 0.2 (the embodiment configures the first-level distance weight word as "joint network"), the second-level weight value is 0.4 (the embodiment configures the second-level distance weight word as "united"), and the third-level weight value is 0.6 (the embodiment configures the third-level distance weight word as "network communication"). Therefore, the sum of the edit distances between the government and enterprise customer name segmentation result and the three-level distance weight words of the large customer name is equal to: first-level distance weight word distance*first-level weight+second-level distance weight word distance*second-level weight+third-level distance weight word distance*third-level weight.

[0052] Optionally, after the first government and enterprise customer name and the first large customer name are segmented respectively, and before the sum of the edit distances between the first government and enterprise customer name segmentation result and the N-level distance weight words of the first large customer name is calculated, the method further comprises: obtaining a positive keyword in the first large customer name segmentation result; calculating the edit distance between the first government and enterprise customer name segmentation result and the positive keyword; and in response to the edit distance between the first government and enterprise customer name segmentation result and the positive keyword being equal to 0, classifying the first government and enterprise customer name to the first large customer name.

[0053] In the embodiment, the positive keyword refers to the segmentation result that needs to be contained in the government and enterprise customer name. If the first government and enterprise customer name contains the positive keyword of the first large customer name, it is determined that the first government and enterprise customer name is classified into the first government and enterprise customer name. For example, the positive keyword in the embodiment is: joint network communication. Specifically, the government and enterprise customer name is iterated, and the edit distance between the segmentation result of the government and enterprise customer name and the positive keyword of each large customer is calculated. If the edit distance is 0, it is determined that the classification is successful, for example, the government and enterprise customer "Beijing joint network communication Co. Ltd.", after segmentation and removal of interference words, it is "joint network communication", which is completely the same as the positive keyword "joint network communication" of the large customer "China joint network communication Co. Ltd.", and the edit distance is 0, which is determined to be classified successfully. If the distance is not 0, the reverse keyword calculation or the calculation of the sum of the edit distances between the segmentation result of the government and enterprise customer name and the N-level distance weight words of the large customer name is performed. For example, the government and enterprise customer name set includes 100 government and enterprise customer names, and after the edit distance calculation of the positive keyword, 20 government and enterprise customer names can be classified into the corresponding large customer name. Then, the remaining 80 government and enterprise customer names need to be calculated and classified subsequently. Therefore, by calculating the edit distance with the positive keyword before calculating the sum of the edit distances between the N-level distance weight words, the calculation efficiency can be improved under the premise of ensuring the accuracy of the classification result.

[0054] Optionally, after calculating the edit distance between the first government and enterprise customer name segmentation result and the positive keyword, and before calculating the sum of the edit distances between the first government and enterprise customer name segmentation result and the N-level distance weight words of the first large customer name, the method further comprises: in response to the edit distance with the positive keyword not being equal to 0, obtaining the reverse keyword of the first large customer name; calculating the edit distance between the first government and enterprise customer name segmentation result and the reverse keyword; and in response to the edit distance with the reverse keyword being equal to 0, determining that the first government and enterprise customer name cannot be classified into the first large customer name.

[0055] In this embodiment, the reverse keyword refers to a segmented word that cannot be contained in the government and enterprise customer name. Since the government and enterprise customer name often contains special words, the corresponding special words are configured in the segmented word library according to the actual situation to support the segmentation of the government and enterprise customer name according to the special words. For example, the name of a large customer is Jingdongfang, and the segmented word "Jingdongfang" needs to be configured in the word library to avoid Jingdongfang being segmented into Jingdong. Therefore, the segmented word "Jingdong" can be configured as the reverse keyword of the sub-company classification of Jingdongfang. If the segmentation result of the first government and enterprise customer name matches the reverse keyword of the first large customer, that is, the edit distance is 0, the classification calculation of the first large customer is stopped, and the matching calculation of the next large customer is performed. If the edit distance with the reverse keyword is not 0, the next N-level distance weight word calculation is performed. By calculating the edit distance with the reverse keyword before calculating the sum of the edit distances between the N-level distance weight words, the calculation efficiency can be further improved.

[0056] In step 104, in response to the sum of the edit distances being less than the preset classification threshold, the first government and enterprise customer name is classified into the first large customer name.

[0057] In this embodiment, when the sum of the edit distances is less than the preset classification threshold (the smaller the score, the higher the similarity of the segmented word), it is determined that the first government and enterprise customer name is successfully classified into the first large customer name. The classification relationship between the current government and enterprise customer name and the large customer and the corresponding sum of the edit distances are recorded, and then the sum of the edit distances of the next large customer is calculated. If the sum of the edit distances calculated with the next large customer is smaller than the sum of the edit distances of the already classified large customer, it is determined that the classification is to the large customer with the smallest sum of the edit distances. If the sum of the edit distances is greater than the preset classification threshold, the classification relationship is not recorded, and the sum of the edit distances of the next large customer is calculated.

[0058] Optionally, after the first government and enterprise customer name is classified into the first large customer name in response to the sum of the edit distances being less than the preset classification threshold, the method further includes: in response to the first government and enterprise customer name being classified into M large customer names, selecting the large customer name corresponding to the minimum value of the sum of the edit distances as the final classified large customer name, where M is a positive integer greater than 1.

[0059] In this embodiment, if the government and enterprise customer name is classified under multiple large customer names, it is finally classified into the large customer with the smallest distance score. The iteration process is essentially a process of finding the optimal classification.

[0060] Optionally, after the first government and enterprise customer name is classified into the first large customer name in response to the sum of the edit distances being less than the preset classification threshold, the method further includes: adjusting the positive keyword, the reverse keyword, the N-level distance weight word, and / or reducing the classification threshold to improve the accuracy of the classification result.

[0061] In this embodiment, China United Network Communications Corporation Limited needs to be classified under the name of the large customer of China United Digital Technology Co., Ltd. Therefore, the positive keywords "China United Digital Technology" and "China United Digital Technology" can be added. In this embodiment, the classification result accuracy is improved by adjusting the positive keywords, reverse keywords, N-level distance weight words, and / or reducing the classification threshold. The operation is simple and easy to implement.

[0062] The large customer classification method for government and enterprise customers provided in this embodiment includes the following steps: performing word segmentation on the government and enterprise customer name and the large customer name, calculating the sum of the edit distances between the N-level distance weight words of the government and enterprise customer name and the large customer name, and determining whether to classify the government and enterprise customer name into the large customer name according to the sum of the edit distances. The edit distance algorithm is used to calculate the distance between the names to achieve fine-grained customer name classification. Specifically, to speed up the efficiency and calculation speed of data acquisition, the distributed computing nodes are used to obtain the government and enterprise customer name set and the large customer name set from the database. After the first government and enterprise customer name and the first large customer name are segmented, the interfering words are removed to reduce the interference factors of the edit distance calculation, thereby improving the calculation efficiency. Further, the fine-grained classification of the names is realized by the edit distance algorithm, and different levels of distance weight words are defined, and multiple weight words can be configured at each level to realize the comprehensiveness of the classification distance between the government and enterprise customer name and the large customer name, improve the classification accuracy, and solve the problem that the names of some group companies and subsidiaries are not related. In addition, by calculating the edit distance with the positive keywords and / or the reverse keywords before calculating the sum of the edit distances between the N-level distance weight words, the calculation efficiency can be improved under the premise of ensuring the accuracy of the classification result. The classification result accuracy is improved by adjusting the positive keywords, reverse keywords, N-level distance weight words, and / or reducing the classification threshold. The operation is simple and easy to implement.

[0063] Embodiment 2:

[0064] As shown in Figure 2 The large customer classification device for government and enterprise customers provided in this embodiment includes:

[0065] The acquisition module 21 is configured to obtain a government and enterprise customer name set and a large customer name set, wherein the government and enterprise customer name set at least includes a first government and enterprise customer name, and the large customer name set at least includes a first large customer name.

[0066] The word segmentation module 22 is connected with the acquisition module 21 and is configured to segment the first government and enterprise customer name and the first large customer name respectively to obtain a first government and enterprise customer name segmentation result and a first large customer name segmentation result. The first large customer name segmentation result includes N-level distance weight words, and N is a positive integer.

[0067] The computing module 23 is connected with the word breaking module 22, and is configured to calculate a sum of edit distances between the first government and enterprise customer name word breaking result and N-level distance weight words of the first large customer name.

[0068] The categorizing module 24 is connected with the computing module 23, and is configured to, in response to the sum of edit distances being less than a preset categorizing threshold, categorize the first government and enterprise customer name into the first large customer name.

[0069] Optionally, N=3, and the number of distance weight words in each of the N levels is greater than or equal to 1.

[0070] The computing module is configured to calculate a product of an edit distance between the first government and enterprise customer name word breaking result and a first-level distance weight word and a first-level weight value to obtain a first-level edit distance, calculate a product of an edit distance between the first government and enterprise customer name word breaking result and a second-level distance weight word and a second-level weight value to obtain a second-level edit distance, and calculate a product of an edit distance between the first government and enterprise customer name word breaking result and a third-level distance weight word and a third-level weight value to obtain a third-level edit distance, and is configured to calculate a sum of the first-level edit distance, the second-level edit distance and the third-level edit distance to obtain a sum of edit distances between the first government and enterprise customer name word breaking result and the three-level distance weight words of the first large customer name, wherein the second-level weight value is greater than the first-level weight value and less than the third-level weight value.

[0071] Optionally, the apparatus further includes a selecting module.

[0072] The selecting module is connected with the categorizing module, and is configured to, in response to the first government and enterprise customer name being categorized into M large customer names, select a large customer name corresponding to a minimum value of the sum of edit distances as a final categorized large customer name, wherein M is a positive integer greater than 1.

[0073] Optionally, the apparatus further includes a forward computing module.

[0074] The forward computing module is connected with the word breaking module, and is configured to acquire a forward keyword in a first large customer name word breaking result, and calculate an edit distance between the first government and enterprise customer name word breaking result and the forward keyword, and is configured to, in response to the edit distance with the forward keyword being equal to 0, categorize the first government and enterprise customer name into the first large customer name.

[0075] Optionally, the apparatus further includes a reverse computing module.

[0076] The reverse calculation module is connected with the forward calculation module, and is configured to: in response to the edit distance with the forward keyword being unequal to 0, acquire a reverse keyword of the first large customer name, and calculate the edit distance between the first government and enterprise customer name segmentation result and the reverse keyword; and in response to the edit distance with the reverse keyword being equal to 0, determine that the first government and enterprise customer name cannot be classified into the first large customer name.

[0077] Optionally, the word breaking module is further configured to break the first government and enterprise customer name and the first large customer name respectively, and remove the interference words after the breaking, to obtain the first government and enterprise customer name segmentation result and the first large customer name segmentation result.

[0078] Optionally, the apparatus further comprises an adjustment module.

[0079] The adjustment module is connected with the classification module, and is configured to adjust the forward keyword, the reverse keyword, the N-level distance weight word, and / or reduce the classification threshold, to improve the classification result accuracy.

[0080] Optionally, the acquisition module is configured to acquire the government and enterprise customer name set and the large customer name set from the database based on a distributed computing node.

[0081] The government and enterprise customer large customer classification apparatus provided by the embodiment is configured to: perform word breaking processing on the government and enterprise customer name and the large customer name, calculate the sum of the edit distances between the government and enterprise customer name segmentation result and the large customer name N-level distance weight words, and then determine whether to classify the government and enterprise customer name into the large customer name according to the sum of the edit distances. The distance between the names is calculated by using the edit distance algorithm, so as to realize fine-grained customer name classification. Specifically, in order to accelerate the data acquisition efficiency and calculation speed, the government and enterprise customer name set and the large customer name set are acquired from the database based on a distributed computing node. The first government and enterprise customer name and the first large customer name are broken and the interference words are removed, so as to reduce the interference factors of the edit distance calculation and improve the calculation efficiency. Further, the fine-grained classification of the names is realized by using the edit distance algorithm, and different levels of distance weight words are defined, and multiple weight words can be configured at each level, so as to realize the comprehensiveness of the classification distance between the government and enterprise customer name and the large customer name, improve the classification accuracy, and solve the problem that the names of some group companies and subsidiary companies are not associated. In addition, the edit distance with the forward keyword and / or the reverse keyword is calculated before the sum of the edit distances between the N-level distance weight words is calculated, so as to improve the calculation efficiency under the premise of ensuring the classification result accuracy. The classification result accuracy is improved by adjusting the forward keyword, the reverse keyword, the N-level distance weight word, and / or reducing the classification threshold, which is simple to operate and easy to implement.

[0082] Embodiment 3:

[0083] As Figure 3 The embodiment provides an electronic device, including a memory 31 and a processor 32, the memory 31 stores a computer program, and the processor 32 is configured to run the computer program to realize the large customer classification method of the government and enterprise customer as described in the embodiment 1.

[0084] Embodiment 4

[0085] The embodiment provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the large customer classification method of the government and enterprise customer as described in the embodiment 1.

[0086] In summary, the large customer classification method, device, electronic device and computer readable storage medium of the government and enterprise customer provided by the embodiment of the present application are used for performing word segmentation processing on the government and enterprise customer name and the large customer name, calculating the sum of the edit distances between the word segmentation result of the government and enterprise customer name and the N-level distance weight words of the large customer name, and then determining whether to classify the government and enterprise customer name into the large customer name according to the sum of the edit distances. The edit distance algorithm is used to calculate the distance between the names to realize fine-grained customer name classification. Specifically, to speed up the efficiency and calculation speed of data acquisition, the distributed computing nodes are used to obtain the government and enterprise customer name set and the large customer name set from the database. The first government and enterprise customer name and the first large customer name are segmented and the interference words are removed to reduce the interference factors of the edit distance calculation and improve the calculation efficiency. Further, the fine-grained classification of the names is realized by the edit distance algorithm, and different levels of distance weight words are defined, and multiple weight words can be configured at each level to realize the comprehensiveness of the classification distance between the government and enterprise customer name and the large customer name, improve the classification accuracy, and solve the problem that the names of some group companies and subsidiaries are not associated. In addition, the edit distance between the positive keywords and / or the reverse keywords is calculated before calculating the sum of the edit distances between the N-level distance weight words to ensure the classification result accuracy under the premise of improving the calculation efficiency. The classification result accuracy is improved by adjusting the positive keywords, reverse keywords, N-level distance weight words and / or reducing the classification threshold in a simple and easy-to-implement manner.

[0087] It can be understood that the above embodiments are only exemplary embodiments adopted to illustrate the principles of the present application, and the present application is not limited thereto. Various modifications and improvements can be made by those skilled in the art without departing from the spirit and essence of the present application, and these modifications and improvements are also regarded as the protection scope of the present application.

Claims

1. A method for classifying large clients of government and enterprise customers, characterized in that, include: Obtain the set of government and enterprise customer names and the set of major customer names, wherein the set of government and enterprise customer names includes at least the first government and enterprise customer name, and the set of major customer names includes at least the first major customer name; The names of the first government and enterprise customer and the first largest customer are split into words to obtain the word segmentation results of the first government and enterprise customer name and the first largest customer name. The word segmentation result of the first largest customer name includes N-level distance weight words, where N is a positive integer. Calculate the sum of edit distances between the word segmentation result of the first government and enterprise customer name and the N-level distance weighted words of the first largest customer name; In response to the sum of the edit distances being less than a preset classification threshold, the first government and enterprise customer name is classified as the first major customer name. Where N=3, and the number of distance weight words in each of the N levels is greater than or equal to 1. The calculation of the sum of edit distances between the word segmentation result of the first government and enterprise customer name and the N-level distance weighted words of the first major customer name specifically includes: The first-level edit distance is obtained by multiplying the word segmentation result of the first government and enterprise customer name with the edit distance of the first-level distance weighted word and the first-level weight value; The second-level edit distance is obtained by multiplying the word segmentation result of the first government and enterprise customer name with the edit distance of the second-level distance weighted words and the second-level weight value; The third-level edit distance is obtained by multiplying the word segmentation result of the first government and enterprise customer name with the edit distance of the third-level distance weighted words and the third-level weight value; The sum of the first-level edit distance, second-level edit distance, and third-level edit distance is calculated to obtain the sum of the edit distances between the word segmentation result of the first government and enterprise customer name and the third-level distance weight words of the first major customer name. Among them, the second-level weight value is greater than the first-level weight value and less than the third-level weight value.

2. The method according to claim 1, characterized in that, After classifying the first government and enterprise customer name into the first major customer name in response to the sum of the edit distances being less than a preset classification threshold, the method further includes: In response to the first government and enterprise customer name being categorized into M major customer names, the major customer name corresponding to the minimum sum of the edit distances is selected as the final categorized major customer name, where M is a positive integer greater than 1.

3. The method according to claim 2, characterized in that, After the word segmentation of the first government and enterprise customer name and the first major customer name, respectively, and before the calculation of the sum of edit distances between the word segmentation results of the first government and enterprise customer name and the N-level distance weight words of the first major customer name, the method further includes: Obtain the positive keywords from the word segmentation results of the first largest customer's name; Calculate the edit distance between the word segmentation result of the first government and enterprise customer name and the positive keyword; In response to the edit distance from the positive keyword being equal to 0, the name of the first government and enterprise customer is classified as the name of the first major customer.

4. The method according to claim 3, characterized in that, After calculating the edit distance between the word segmentation result of the first government and enterprise customer name and the positive keyword, and before calculating the sum of the edit distances between the word segmentation result of the first government and enterprise customer name and the N-level distance weighted words of the first major customer name, the method further includes: In response to the fact that the edit distance from the positive keyword is not equal to 0, obtain the reverse keyword of the first major customer name; Calculate the edit distance between the word segmentation result of the first government and enterprise customer name and the reverse keyword; In response to the edit distance from the reverse keyword being equal to 0, it is determined that the name of the first government and enterprise customer cannot be classified as the name of the first major customer.

5. The method according to claim 4, characterized in that, The process of splitting the name of the first government and enterprise customer and the name of the first largest customer into words, respectively, to obtain the word segmentation results of the name of the first government and enterprise customer and the name of the first largest customer, specifically includes: The names of the first government and enterprise customer and the largest customer are segmented into words, and interfering words are removed after segmentation to obtain the word segmentation results for the first government and enterprise customer name and the largest customer name, respectively. After classifying the first government and enterprise customer name into the first major customer name in response to the sum of the edit distances being less than a preset classification threshold, the method further includes: Adjust positive keywords, negative keywords, N-level distance weight words, and / or reduce the classification threshold to improve the accuracy of classification results.

6. The method according to claim 5, characterized in that, The acquisition of the set of government and enterprise customer names and the set of major customer names specifically includes: The system uses distributed computing nodes to retrieve sets of government and enterprise customer names and sets of major customer names from the database.

7. A customer classification device for government and enterprise clients, characterized in that, include: The acquisition module is used to acquire a set of government and enterprise customer names and a set of major customer names. The set of government and enterprise customer names must include at least the name of the first government and enterprise customer, and the set of major customer names must include at least the name of the first major customer. The word segmentation module, connected to the acquisition module, is used to segment the names of the first government and enterprise customer and the first largest customer into words, respectively, to obtain the word segmentation results for the first government and enterprise customer name and the first largest customer name. The word segmentation result for the first largest customer name includes N-level distance-weighted words, where N is a positive integer. The calculation module, connected to the word segmentation module, is used to calculate the sum of edit distances between the word segmentation result of the first government and enterprise customer name and the N-level distance-weighted words of the first major customer name. The classification module, connected to the calculation module, is used to classify the first government and enterprise customer name into the first major customer name in response to the sum of the edit distances being less than a preset classification threshold. Where N=3, and the number of distance weight words in each of the N levels is greater than or equal to 1. The calculation module is used to calculate the first-level edit distance by multiplying the word segmentation result of the first government and enterprise customer name with the edit distance of the first-level distance weighted word and the first-level weight value. It is also used to calculate the second-level edit distance by multiplying the word segmentation result of the first government and enterprise customer name with the edit distance of the second-level distance weighted word and the second-level weight value. Furthermore, it is used to calculate the third-level edit distance by multiplying the word segmentation result of the first government and enterprise customer name with the edit distance of the third-level distance weighted word and the third-level weight value. Finally, it is used to calculate the sum of the first-level edit distance, the second-level edit distance, and the third-level edit distance to obtain the sum of the edit distances between the word segmentation result of the first government and enterprise customer name and the third-level distance weighted word of the first major customer name. The second-level weight value is greater than the first-level weight value and less than the third-level weight value.

8. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to implement the large customer classification method for government and enterprise customers as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for classifying large enterprise customers as described in any one of claims 1-6.

Images