Method and device for mining customer group, electronic equipment and computer program product

Abstract

The invention provides a customer group mining method and device, electronic equipment and a computer program product, and the method comprises the steps: inputting the business information of a target business into a customer group portrait model, obtaining a customer group feature text of a to-be-marketed customer group outputted by the customer group portrait model, and based on the customer group feature text of the to-be-marketed customer group, obtaining a customer group feature text of the to-be-marketed customer group; matching user tags of different customers counted in a customer database to obtain customer group tags of the customer group to be marketed; and screening clients meeting a target condition from the client database to obtain a client group of the target business, the target condition including a screening condition formed based on the client group labels of the to-be-marketed client group. The clients are not mined by means of manual operation any more, so that the limits of business capability, personal preference and the like of the personnel can be broken through, and the stability of the mining result is improved. Especially for a client database with a huge user tag, the mining efficiency can be improved, and meanwhile, the quality of a mining result is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a method and device for mining a customer group, an electronic device and a computer program product. BACKGROUND

[0002] In a marketing scenario, it is often necessary to filter out customers who are most likely to be interested in a product or service and have purchase potential from a large number of customers, so as to carry out targeted marketing activities. The customer group composed of these filtered customers can be regarded as a product marketing customer group or a marketing customer group. At present, the customer database of most enterprises usually uses discrete user tags to record customer information, which makes it difficult to mine a suitable product marketing customer group according to discrete user tags.

[0003] At present, the work of mining a customer group is mostly performed by professional personnel through manual operation. For example, marketing personnel manually filter out a product marketing customer group based on their understanding of the product and the understanding of each user tag in the customer database. This method can to some extent mine some valuable customers, but the overall effect is not ideal. Due to the business level of personnel and the understanding of customer group tags, there are problems of unstable mining results and low mining efficiency. SUMMARY

[0004] Based on the above technical status, the present application proposes a method and device for mining a customer group, an electronic device and a computer program product.

[0005] According to a first aspect of an embodiment of the present application, a method for mining a customer group is provided, the method comprising:

[0006] inputting business information of a target business into a customer portrait model to obtain customer group feature text of a marketing customer group output by the customer portrait model, wherein the customer portrait model is used to generate customer group feature text of a suitable marketing customer group according to semantic understanding of the input business information;

[0007] matching user tags of different customers in a customer database based on the customer group feature text of the marketing customer group to obtain customer group tags of the marketing customer group;

[0008] filtering out customers meeting a target condition from the customer database to obtain a customer group of the target business, wherein the target condition comprises a filtering condition composed based on the customer group tags of the marketing customer group.

[0009] Optionally, the customer group feature text comprises multiple items of a customer group name, a demand of the customer group for the business, a self-description of the customer group, and an association degree between the customer group and the business.

[0010] Optionally, based on the customer group characteristic text of the to-be-marketed customer group, matching user tags of different customers in a customer database, a customer group tag of the to-be-marketed customer group is obtained, and the method comprises the following steps:

[0011] Target information is input into a tag matching model, and a customer group tag of the to-be-marketed customer group output by the tag matching model is obtained.

[0012] The target information comprises a tag library and the customer group characteristic text, and the tag library comprises user tags of different customers in the customer database.

[0013] Optionally, the target information further comprises business information of the target business.

[0014] Optionally, based on the customer group characteristic text of the to-be-marketed customer group, matching user tags of different customers in a customer database, a customer group tag of the to-be-marketed customer group is obtained, and the method comprises the following steps:

[0015] The similarity between each user tag and the customer group characteristic text of the to-be-marketed customer group is calculated respectively.

[0016] User tags with a similarity higher than a similarity threshold value are determined as the customer group tag of the to-be-marketed customer group.

[0017] Optionally, the similarity between each user tag and the customer group characteristic text of the to-be-marketed customer group is calculated respectively, and the method comprises the following steps:

[0018] Keywords in the customer group characteristic text of the to-be-marketed customer group are extracted.

[0019] The similarity between each user tag and the keywords of the to-be-marketed customer group is calculated respectively.

[0020] Optionally, after the customer group tag of the to-be-marketed customer group is obtained by matching the user tags of different customers in the customer database based on the customer group characteristic text of the to-be-marketed customer group, the method further comprises the following steps:

[0021] A business tag library corresponding to the target business is obtained, wherein the business tag library comprises user tags that must be possessed by the to-be-marketed customer group under the business rules of the target business.

[0022] The user tags in the business tag library are added to the customer group tag of the to-be-marketed customer group.

[0023] Optionally, after the customer group tag of the to-be-marketed customer group is obtained by matching the user tags of different customers in the customer database based on the customer group characteristic text of the to-be-marketed customer group, the method further comprises the following steps:

[0024] inputting the business information of the target business and the customer group characteristic text into a customer group attribute model to obtain attribute information of a to-be-marketed customer group output by the customer group attribute model, wherein the customer group attribute model is used to analyze attribute information possessed by a customer group of the target business according to the input customer group characteristic text and the business information;

[0025] filtering customer group tags of the to-be-marketed customer group based on the attribute information of the to-be-marketed customer group.

[0026] Optionally, the step of generating the target condition comprises:

[0027] assigning each customer group tag of the to-be-marketed customer group based on a preset tag assignment rule;

[0028] logically combining the assigned customer group tags to obtain the target condition.

[0029] According to a second aspect of an embodiment of the present application, a device for mining a customer group is provided, and the device comprises:

[0030] a customer group characteristic module configured to input business information of a target business into a customer group portrait model to obtain customer group characteristic text of a to-be-marketed customer group output by the customer group portrait model, wherein the customer group portrait model is used to generate customer group characteristic text of a customer group suitable for marketing according to semantic understanding of the input business information;

[0031] a customer group tag module configured to match user tags of different customers in a customer database based on customer group characteristic text of the to-be-marketed customer group to obtain customer group tags of the to-be-marketed customer group;

[0032] a mining module configured to filter customers meeting a target condition from the customer database to obtain a customer group of the target business, wherein the target condition comprises a filtering condition composed based on the customer group tags of the to-be-marketed customer group.

[0033] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory and a processor; the memory is connected with the processor and is used to store a program; the processor is used to realize the method for mining a customer group according to the first aspect by running the program in the memory.

[0034] According to a fourth aspect of an embodiment of the present application, a storage medium is provided, and the storage medium stores a computer program; when the computer program is run by a processor, the method for mining a customer group according to the first aspect is realized.

[0035] According to a fifth aspect of the embodiments of the present application, a computer program product is provided, comprising: a computer program which, when executed by a processor, implements the method for mining a customer group as described in the first aspect.

[0036] In the embodiments of the present application, first, with the help of the ability of the customer portrait model, the customer group feature text of the to-be-marketed customer group is generated for the target business. The user tags in the customer database are matched with the customer group feature text as intermediate data, and the matched user tags are used as the customer group tags of the to-be-marketed customer group. Compared with the way of directly matching the user tags with the business information, the appropriate user tags can be matched more accurately, and thus the quality of the finally mined customer group is improved. Finally, the users meeting the conditions are screened from the customer database based on the screening conditions composed of the customer group tags of the to-be-marketed customer group, and the mined customer group can be obtained. Since the customer is no longer mined by means of manual operation, the business ability, personal preference and other limitations of the personnel can be broken through, and thus the stability of the mining result is improved. Especially for the customer database with a large number of user tags, the mining efficiency and the quality of the mining result can be improved at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] FIG. 1 One of the flowcharts of the method for mining a customer group provided by the embodiments of the present application;

[0039] FIG. 2 The second flowchart of the method for mining a customer group provided by the embodiments of the present application;

[0040] FIG. 3 The structural schematic diagram of the device for mining a customer group provided by the embodiments of the present application;

[0041] FIG. 4 The structural schematic diagram of the electronic device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, 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 are within the scope of protection of the present application.

[0043] SUMMARY

[0044] As described in the background, it is very important for enterprises to mine the to-be-marketed customer group / marketed customer group in the marketing scenario. In the face of discrete and irregular user tags in the customer database, the mining work can usually only be completed by professional personnel. Due to the differences in business capabilities and preferences of different personnel, even if they face the same data and products, different personnel may draw completely different mining results. At the same time, the low efficiency of manual mining is also a problem that cannot be ignored. For a customer database with a large number of user tags, personnel need to spend a lot of time and effort to understand all the user tags, and may not be able to fully understand all the user tags, which affects the timeliness of customer mining.

[0045] Since product information is usually complex and diverse, directly matching user tags with product information results in a large span of relationship between the two, so that most of the matched user tags are not suitable and cannot be used for subsequent mining work. With the rise of current large models, more and more marketing scenarios use large models to improve service quality. For example, large models can realize the automation of marketing processes and complete routine tasks such as customer service, email marketing, and search engine optimization. However, the discrete and irregular user tags in the customer database are of no use even if the large models are used to analyze and process the user tags.

[0046] In view of the above technical status, the embodiment of the present application proposes a method for mining a customer group. First, with the help of the ability of the customer group portrait model, the customer group feature text of the to-be-marketed customer group is generated for the target business. The user tags in the customer database are matched with the customer group feature text as intermediate data, and the matched user tags are used as the customer group tags of the to-be-marketed customer group. Compared with the method of directly matching user tags with business information, more accurate user tags can be matched, thereby improving the quality of the mined customer group. Finally, based on the screening conditions composed of the customer group tags of the to-be-marketed customer group, users meeting the conditions are screened from the customer database, and the mined customer group can be obtained. Since the customer is no longer mined by manual operation, the business capabilities, personal preferences, and other limitations of personnel can be broken through, thereby improving the stability of the mining results. Especially for a customer database with a large number of user tags, the mining efficiency can be improved while the quality of the mining results is improved.

[0047] Exemplary methods

[0048] Please refer to FIG. 1In an example embodiment, a method for mining a customer group is provided, which can be applied to an electronic device with data processing capability. For example, the electronic device can include a computer, a mobile phone, a server, a smart wearable device, etc. The method can include:

[0049] S101: inputting service information of a target service into a customer portrait model to obtain customer feature text of a customer group to be marketed output by the customer portrait model.

[0050] In this step, the customer portrait model is used to generate customer feature text of a customer group suitable for marketing according to semantic understanding of the input service information. The customer group suitable for marketing can be understood as a customer group suitable for marketing of the input service information, and the customer group suitable for marketing is the customer group to be marketed or the marketing customer group. For example, for a service A, it is more suitable for young people's preferences and is usually used in daily work process, and is not suitable for learning scenarios. Therefore, for the service A, the customer group suitable for marketing can be young people who have already worked. The old people and young people still in school can be regarded as a customer group unsuitable for marketing.

[0051] In some embodiments, the customer portrait model can be deployed in a cloud device, and the electronic device performing the method for mining a customer group provided in the embodiment has network communication function and can access the cloud device through the network, and then obtain the customer feature text by using the customer portrait model. Since the customer portrait model does not need to be deployed locally on the electronic device, the requirement and load of the electronic device can be reduced. For example, the method for mining a customer group provided in the embodiment can be performed on a computer in an enterprise, and the customer portrait model can be deployed on a public platform. When performing S101, the computer accesses the public platform to obtain the customer feature text by using the customer portrait model.

[0052] In some embodiments, the customer portrait model can also be deployed locally on the electronic device performing the method for mining a customer group provided in the embodiment, so that the customer feature text can be obtained at any time and anywhere, and quickly, without being affected by network factors. For example, the method for mining a customer group provided in the embodiment can be performed on a computer in an enterprise, and the customer portrait model is deployed on the computer. When performing S101, the customer feature text can be obtained at any time and anywhere by using the customer portrait model without accessing external network.

[0053] It should be noted that the customer group feature text output by the customer group portrait model can be a customer group feature text of a to-be-marketed customer group, or can be customer group feature texts of multiple different to-be-marketed customer groups. In some embodiments, the customer group portrait model will also output the degree of association of each to-be-marketed customer group with the target business, wherein the degree of association is positively correlated with the degree of adaptation or suitability of marketing. In the case where the customer group portrait model outputs multiple different customer group feature texts of to-be-marketed customer groups, to-be-marketed customer groups or their customer group feature texts with a degree of association lower than a target threshold can also be filtered out. Or only a specified number of to-be-marketed customer groups are retained, and the customer group feature texts of the retained to-be-marketed customer groups are used for the following subsequent steps.

[0054] S102: Based on the customer group feature text of the to-be-marketed customer group, match the user tags of different customers in the customer database to obtain the customer group tags of the to-be-marketed customer group.

[0055] In this step, the customer database can be a database in which an enterprise stores customer information. All users in the customer database are potential customers of the target business, but not every customer is truly suitable for the target business. This embodiment will filter out customers suitable for the target business in the customer database.

[0056] The customer database can store customer information in the form of user tags. In some embodiments, a user tag system can be designed to store customer information. The user tag system includes but is not limited to: personal basic attributes (such as user age, place of origin, occupation type), space-time attributes (the last online IP address, cell address, network age), business-related attributes (for example, in the operator scenario, such as the amount of traffic / voice usage in the past three months, the number of traffic / voice over-suits in the past three months, and the products subscribed), and the like. In addition, various user tags include but are not limited to date, numerical value, text, Boolean, enumeration, and other data types.

[0057] In some embodiments, all user tags in the customer database can be used to construct a tag library, and then the customer group feature text is matched with each user tag in the tag library.

[0058] In the case where the number of to-be-marketed customer groups is multiple, the customer group feature text of each to-be-marketed customer group can be matched with the user tags of different customers in the customer database to obtain the customer group tags of the to-be-marketed customer group, and finally obtain the customer group tags of each to-be-marketed customer group.

[0059] For the to-be-marketed customer group, the matched user tags will be used as the customer group tags. In some embodiments, the matched user tags can be considered as the target business customer group having a clear difference from other customers under the user tags. Continuing with the above example of business A, the to-be-marketed customer group is young people who have started working. Then, the matched user tags include the "work type" tag indicating whether the customer has started working and the "age" tag indicating whether the customer is young.

[0060] S103: Filtering customers meeting the target conditions from the customer database to obtain the customer group of the target business.

[0061] In this step, the target conditions include a filtering condition based on the customer group tags of the to-be-marketed customer group.

[0062] In some embodiments, an executable file (such as a script file) can be generated according to the target conditions. By automatically executing the script file, customers meeting the target conditions are filtered from the customer database to obtain the customer group of the target business.

[0063] In other embodiments, the target conditions can also be sent to designated personnel, and the designated personnel can perform filtering operations according to the received target conditions to obtain the customer group of the target business. For example, the target business can be a service provided by enterprise A. The electronic device performing the method for mining the customer group provided in this embodiment can be a public platform. After receiving the service information provided by enterprise A using the public platform and performing the above steps to obtain the target conditions, the target conditions are sent to enterprise A, and enterprise A can filter out the customer group of the target business by itself.

[0064] It should be noted that the group composed of customers meeting the target conditions in the customer database will be the customer group of the target business. In some embodiments, all or part of the information of each customer in the customer group of the target business can be directly output. For example, the unique identifier of each customer in the customer group of the target business in the customer database can be output.

[0065] In the embodiments of the present application, first, with the help of the customer portrait model, the customer group feature text of the target business is generated. The customer group feature text is used as intermediate data to match the user tags in the customer database, and the matched user tags are used as the customer group tags of the customer group to be marketed. Compared with the method of directly matching user tags with business information, more accurate user tags can be matched, thereby improving the quality of the customer group mined. Finally, based on the filtering conditions composed of the customer group tags of the customer group to be marketed, the users meeting the conditions are filtered from the customer database, and the mined customer group can be obtained. Since the customer is no longer mined by manual operation, the limitations of business ability, personal preference, etc. of the personnel can be broken through, thereby improving the stability of the mining results. Especially for the customer database with a large number of user tags, the mining efficiency and the quality of the mining results can be improved at the same time.

[0066] It is worth noting that a business in a marketing scenario may be a combination of multiple atomic businesses, such as a package, a benefit, a broadband, and a hardware terminal, etc. Therefore, one business may cover multiple different customer groups. If the customer groups are manually divided, they are easily limited by personal factors. The customer portrait model in the embodiments of the present application is a pre-trained network model. Through semantic understanding of the business by the customer portrait model, multiple customer groups to be marketed can be generated, and then the customer group feature text of each customer group to be marketed is output. For example, the input of the customer portrait model is business information, and the output is customer group feature text. The customer portrait model can well understand the input business information, thereby analyzing which customer groups are suitable for marketing the business indicated by the business information, and then generating the features of these suitable customer groups, i.e. the customer group feature text.

[0067] The customer portrait model can be any machine learning model. For example, it can be a large language model, but is not limited thereto. The process of training the customer portrait model will not be described in detail here. The model can be trained according to its function of generating customer group feature text based on business information. For example, the customer portrait model is obtained by supervised fine-tuning (SFT) based on a large language model.

[0068] In some embodiments of the present application, the customer group feature text includes multiple items of customer group name, customer group demand for the business, customer group self-description, and customer group association degree with the business.

[0069] It should be noted that the customer group name can be a simple summary of the customer group to be marketed. The customer group name can be used to distinguish different customer groups to be marketed and to simply understand some characteristics of the customer group to be marketed.

[0070] The demand of the customer group on the business can be a core demand of the customer group related to the business. The demand of the customer group can be used to preliminarily determine whether the customer group is suitable for the corresponding business.

[0071] The self-description of the customer group can be a personalized feature of the customer group, which can be obviously distinguished from other customers, and the content is also related to the corresponding business. The self-description of the customer group can be used to further determine whether the customer group is suitable for the corresponding business.

[0072] The association degree of the customer group and the business can be the degree of suitability of the customer group and the business. The association degree is positively correlated with the degree of adaptation or suitability of marketing.

[0073] For ease of understanding, the following is an example of customer group feature text. Assuming that the target business is: 18 yuan 30 GB exclusive directional traffic package. The customer groups to be marketed include four, and the customer group feature text of each customer group to be marketed is as follows:

[0074] Customer group to be marketed one: customer group name: high-frequency short video enthusiasts; demand of the customer group on the business: a large amount of directional traffic demand for watching short videos or live streaming on short video platforms. Self-description of the customer group: 18-35 years old, daily use of short video software more than 2 hours, and frequently use mobile network to watch videos.

[0075] Customer group to be marketed two: customer group name: student traffic supplement group; demand of the customer group on the business: low-cost directional traffic supplement is needed in addition to the basic package, especially to meet the entertainment needs when there is no local area network in the dormitory / classroom. Self-description of the customer group: college student group, monthly communication consumption is less than 50 yuan, and night activity is high.

[0076] Customer group to be marketed three: customer group name: commuter traffic dependent; demand of the customer group on the business: directional traffic is needed to stably watch short videos to kill time in subway / bus commuting scenarios. Self-description of the customer group: white-collar worker, one-way commuting time is more than 40 minutes, and there is more than 1 hour of fragmented entertainment time per day.

[0077] Customer group to be marketed four: customer group name: short video content producer; demand of the customer group on the business: stable traffic is needed to support short video shooting and uploading, live streaming and other creative activities. Self-description of the customer group: self-media practitioners, live streaming anchors, etc., more than 30 videos are uploaded per month, and rely on mobile network for content production.

[0078] In some embodiments, when the customer group feature text includes: customer group name, demand of the customer group on the business, self-description of the customer group, and association degree of the customer group and the business, the customer groups to be marketed can be sorted based on the association degree of the customer group and the business, and the customer group feature text of the customer group to be marketed with high association degree is preferentially output.

[0079] In the embodiments of the present application, the characteristics of the customer group can be interpreted from multiple dimensions such as customer group name, customer group demand for business, customer group self-description, and customer group association degree with business, so as to improve the accuracy of the customer group characteristic text.

[0080] In some embodiments of the present application, based on the customer group characteristic text of the customer group to be marketed, the user tags of different customers in the customer database are matched to obtain the customer group tags of the customer group to be marketed, including:

[0081] The target information is input into the tag matching model to obtain the customer group tags of the customer group to be marketed output by the tag matching model;

[0082] The target information includes: a tag library and a customer group characteristic text, and the tag library includes user tags of different customers in the customer database; and the tag matching model is used to determine the user tags associated with the customer group characteristic text from the tag library input into the model.

[0083] It should be noted that the tag matching model is used to determine the user tags with higher matching degree with the customer group characteristic text according to the semantic understanding of the input customer group characteristic text and user tags, and output the user tags with higher matching degree as the customer group tags. The higher the matching degree is, the more accurate the matched user tags are, and the more accurate the customer group filtered by the matched user tags is. Taking the business A as an example, the customer group to be marketed is young people who have started working. Then, the "work type" tag indicating whether the customer has started working and the "age" tag indicating whether the customer is young are considered as user tags with higher matching degree. The "name" tag has little association with young people who have started working, and therefore, the "name" tag can be considered as a user tag with lower matching degree.

[0084] In some embodiments, the deployment manner of the tag matching model can be the same as the deployment manner of the customer portrait model in the above embodiments. It can be deployed on a cloud device or locally on an electronic device that executes the method for mining customer groups provided by the present embodiment. The customer group characteristic text of the customer group to be marketed can be obtained by means of the customer portrait model, and the customer group tags of the customer group to be marketed can be obtained by means of the tag matching model. Although the data obtained by means of the two models is different, the communication process can be consistent. The process of obtaining the customer group tags of the customer group to be marketed by means of the tag matching model can be referred to the process of obtaining the customer group characteristic text of the customer group to be marketed by means of the customer portrait model in the above embodiments. To avoid repetition, it will not be described here.

[0085] It is worth noting that the tag matching model can also be any machine learning model. For example, it can be a large language model, but is not limited thereto. As for the process of the tag matching model, it will not be described in detail here. Model training can be performed according to its function of matching user tags based on target information. For example, a large language model is supervised and fine-tuned to obtain a tag matching model.

[0086] In some embodiments, the tag library and the customer group feature text are assembled into prompt words in natural language form as input of the tag matching model, and the tag matching model completes understanding of the two source information of the customer group and the tag, and outputs the user tags related to the customer group that should be retained.

[0087] In the embodiments of the present application, the trained tag matching model can be used to match user tags, and the user tags matched based on semantic understanding are more accurate.

[0088] It can be understood that whether the user tag is suitable for screening the customer group will also be affected by the business. Even if the customer group feature text is the same, the user tags used to screen the customer group should not be completely the same for different businesses. Therefore, in some embodiments of the present application, the target information further includes business information of a target business.

[0089] It should be noted that the model input of the tag matching model in this embodiment will include the tag library, the customer group feature text, and the business information of the target business. According to the semantic understanding of the input customer group feature text, user tags, and business information of the target business, the user tags with a higher matching degree with the customer group feature text are determined, and the user tags with a higher matching degree are output as customer group tags. The whole process is similar to the process of outputting user tags with a higher matching degree according to the customer group feature text and user tags in the above-mentioned embodiments. The difference is that the model input is more, and further includes the business information of the target business.

[0090] Correspondingly, when training the tag matching model, model training needs to be performed according to more model inputs. For example, the tag library, the customer group feature text, and the business information can be assembled into prompt words in natural language form as input of the tag matching model, and the tag matching model completes understanding of the three source information of the customer group, the tag, and the business, and outputs the user tags related to the customer group that should be retained. For similarities, see the above-mentioned embodiments, which will not be described here.

[0091] In the embodiments of the present application, the tag matching model will output the customer group tags based on the understanding of the customer group, the tag, and the business. Compared with the way of outputting the customer group tags only by the customer group and the tag, it is undoubtedly more accurate.

[0092] In some embodiments of the present application, based on the customer group characteristic text of the customer group to be marketed, the user tags of different customers in the customer database are matched to obtain the customer group tags of the customer group to be marketed, including:

[0093] The similarity between each user tag and the customer group characteristic text of the customer group to be marketed is calculated respectively;

[0094] The user tags with a similarity higher than a similarity threshold are determined as the customer group tags of the customer group to be marketed.

[0095] It should be noted that the customer group characteristic text can interpret the characteristics of a certain type of customer; similarly, the user tag can also interpret the characteristics of the customer. Therefore, the two have commonality to some extent.

[0096] For example, for a college student customer group, its customer group characteristic text can be "college student group". The user tag in the customer database also has a "whether a college student" tag. In matching the user tag, for the customer group characteristic text of "college student group", the "whether a college student" tag needs to be matched as a suitable tag for screening the customer group. At this time, with the help of similarity calculation between texts, the customer group tags can be quickly screened out.

[0097] Regarding the similarity calculation between the user tag and the customer group characteristic text, the string matching method (such as the edit distance algorithm) can be used to calculate the similarity, or the vector space method can be used to calculate the similarity. Regarding the vector space method, the user tag and the customer group characteristic text are converted into vectors respectively, and then the cosine similarity or Euclidean distance algorithm is used to calculate the similarity between the two.

[0098] The similarity threshold can be manually set according to engineering experience. For example, the similarity threshold can be 0.5, but is not limited thereto.

[0099] In the embodiments of the present application, with the help of the similarity algorithm between texts, the customer group tags can be quickly matched from the tag library.

[0100] In some embodiments of the present application, the similarity between each user tag and the customer group characteristic text of the customer group to be marketed is calculated respectively, including:

[0101] Extracting the keywords in the customer group characteristic text of the customer group to be marketed;

[0102] The similarity between each user tag and the keywords of the customer group to be marketed is calculated respectively.

[0103] It should be noted that the text length of the customer group feature text can be relatively long. In some embodiments, the customer group feature text includes at least one sentence. Therefore, the customer group feature text inevitably includes some redundant words, such as conjunctions, modal particles, and the like. These redundant words not only increase a certain amount of calculation, but also form a certain interference in matching the customer group label. Therefore, in order to avoid the influence of the redundant words, in the embodiment, keywords in the customer group feature text are extracted, and the keywords are used to match the customer group label. The keywords are not redundant words. In some embodiments, the keywords in the customer group feature text can be extracted by any keyword extraction manner, which is not limited here.

[0104] In some embodiments, for the customer group feature text of a to-be-marketed customer group, if the number of keywords is multiple, all the keywords can be directly combined into a text, and the combined text is used to match the customer label. For example, the similarity between each user label and the combined text is calculated, which is used as the similarity between the customer group feature text and the user label for subsequent calculation.

[0105] In the embodiment of the application, the keywords in the customer group feature text are extracted, and then the similarity between the keywords and the user label is calculated, which can improve the calculation speed and reduce the influence of redundant words in the customer group feature text.

[0106] In order to further improve the quality of the customer group label, in some embodiments of the application, based on the customer group feature text of the to-be-marketed customer group, the user labels of different customers in the customer database are matched to obtain the customer group label of the to-be-marketed customer group, and then the method further includes:

[0107] Obtaining a business label library corresponding to the target business, wherein the business label library includes: user labels that the to-be-marketed customer group must have under the business rules of the target business;

[0108] Adding the user labels in the business label library to the customer group label of the to-be-marketed customer group.

[0109] It should be noted that there are usually some underlying business logics in the marketing process, such as the maximum difference between the price of the sold product and the current consumption of the user cannot be higher than a certain threshold, and there are certain specific premises when selling certain rights. These business logics are strongly related to the customer group label for screening the business customer group. Therefore, in the embodiment, the previously obtained customer group label is checked for missing and supplemented by using the business rules.

[0110] Here, the corresponding business label library can be established in advance for the target business, and the user labels are filtered from the customer database based on the business rules of the target business to construct the business label library corresponding to the target business. The business rules can be the underlying business logics that must be followed in the service process of the target business.

[0111] For example, for business B, the object of its service must be an adult. Then there is a "whether adult" label in the business tag library corresponding to business B. If the target business is business B, and the previously determined customer group tag does not exist "whether adult" label, then the "whether adult" label needs to be added to the customer group tag.

[0112] In the embodiments of the present application, the business rules can be used to check and supplement the customer group tags, and improve the quality of the customer group tags.

[0113] In some embodiments of the present application, after matching the user tags of different customers in the customer database based on the customer group feature text of the to-be-marketed customer group to obtain the customer group tags of the to-be-marketed customer group, the method further comprises:

[0114] inputting the business information of the target business and the customer group feature text into the customer group attribute model to obtain the attribute information of the to-be-marketed customer group output by the customer group attribute model, wherein the customer group attribute model is used to analyze the attribute information possessed by the customer group of the target business according to the input customer group feature text and business information;

[0115] filtering the customer group tags of the to-be-marketed customer group based on the attribute information of the to-be-marketed customer group.

[0116] It should be noted that the deployment mode of the customer group attribute model can be the same as the deployment mode of the customer portrait model in the above-mentioned embodiments. It can be deployed on a cloud device, or it can be deployed locally on an electronic device that executes the method of mining customer groups provided in the present embodiment. With the help of the customer portrait model, the customer group feature text of the to-be-marketed customer group can be obtained, and with the help of the customer group attribute model, the attribute information of the to-be-marketed customer group can be obtained. Although the data obtained by means of the two models is different, the communication process can be consistent. The process of obtaining the attribute information of the to-be-marketed customer group by means of the customer group attribute model can be referred to the process of obtaining the customer group feature text of the to-be-marketed customer group by means of the customer portrait model in the above-mentioned embodiments. To avoid repetition, it will not be repeated here.

[0117] It is worth noting that the customer group attribute model can also be any machine learning model. For example, it can be a large language model, but it is not limited thereto. The process of the customer group attribute model will not be described in detail here. For example, the customer group attribute model is obtained by supervised fine-tuning based on a large language model.

[0118] In some embodiments, the attribute information of the to-be-marketed customer group can be basic conditions or characteristics that the to-be-marketed customer group should meet. Then, the customer group tags in the to-be-marketed customer group that do not meet the basic conditions or characteristics are filtered out, and the remaining customer group tags enter the subsequent calculation link. The aforementioned basic conditions or characteristics can be conditions or characteristics set from the relevance of customer group tags and target business, customer group tag combination, and the like. For example, a large model can be used to summarize the core content of the target business according to the text form of the business information, and based on the description of the target business and the to-be-marketed customer group, the basic conditions or characteristics that the to-be-marketed customer group should meet are analyzed. The strength of the relevance of the customer group tags and the target business, the combination relationship of the customer group tag combination, and the like are further filtered out to remove irrelevant customer group tags.

[0119] In the embodiments of the present application, the customer group tags can be filtered, thereby further improving the quality of the customer group tags.

[0120] In some embodiments of the present application, the step of generating the target condition includes:

[0121] Based on the preset label assignment rule, each customer group tag of the to-be-marketed customer group is assigned a value;

[0122] The assigned customer group tags are logically combined to obtain the target condition.

[0123] It should be noted that the preset label assignment rule is related to the content and type of each user label in the customer database. Different types of user labels have different value ranges and assignment rules. For example, for a Boolean type user label, its value range includes yes and no. For a date type user label, its value range includes a pre-set period of time. Here, they will not be listed one by one.

[0124] The assigned user labels can be shown in the following examples, for example, the user age label is assigned to: user age is greater than or equal to

[18] . The near-three-month traffic super set number label is assigned to: the near-three-month traffic super set number is equal to 【1 time】. The game preference user label is assigned to: game preference user is equal to

yes

[0125] Each assigned user label can be used as an atomic condition of the target condition, and each atomic condition is combined using a logical relationship to form the target condition. For example, a target condition of a to-be-marketed customer group:

[0126] User age is greater than or equal to

[30] and user age is less than or equal to

[50] and (near-three-month average Douyin APP time length is greater than or equal to

[3600] or last month DOU is greater than or equal to

[40] ) and whether a heavy short video live broadcast enthusiast is equal to

yes

[0127] In some embodiments, a large model can be pre-trained for assigning customer group tags and performing logical combination to generate target conditions. Regarding the training process of the large model, a traditional training method can be used, historical data collected during the mining of customer groups is used to make training data, and the large model is trained using the training data.

[0128] In the embodiments of the present application, the target conditions include not only the assigned customer group tags, but also the logical relationship between the customer group tags, so that the customer group of the target business can be accurately filtered from the customer database.

[0129] For ease of understanding, the method for mining customer groups provided by the present application will be described by way of example below. As shown in the following example, the method comprises the following steps: FIG. 2

[0130] S201: Tag system definition. The information of each user / customer in the customer database is recorded using a defined user tag. For example, the user tag system includes but is not limited to: personal basic attributes (such as user age, place of origin, occupation type), space-time attributes (the last online IP address, cell address, network age), business-related attributes (for example, in the operator scenario, such as the amount of traffic / voice usage in the past three months, the number of traffic / voice over the past three months, the number of products subscribed), personal derivative attributes (whether a family user, whether there are children in the family, whether a video / game preference user), etc. In addition, various user tags include but are not limited to date, numerical value, text, Boolean, enumeration, etc.

[0131] S202: Multi-customer group portrait generation. This process is the same as the process of obtaining the customer group feature text of the customer group to be marketed through the customer group portrait model in the above embodiments, and will not be described here. The customer group feature text can be regarded as the customer portrait of the customer group to be marketed.

[0132] S203: Customer group tag selection. The user tags matched in S201 based on the customer portrait generated in S202 are determined as customer group tags. The customer group tags can be obtained by directly selecting the large model according to the semantic information, or by recalling the customer group tags based on the similarity between the customer portrait and the user tag. The specific process can be referred to the related parts in the above embodiments, which will not be described here.

[0133] ​S204: Customer group combination generation. This step can use a long chain of thought method, step-by-step reasoning, to complete the output and explanation of the customer group combination. It includes: 1. Product and customer group analysis. Use the large model to summarize the core content of the product based on the text form of the product content, and analyze the basic conditions or characteristics that the customer group should meet based on the product and customer group description. 2. Tag filtering and combination logic. Consider the strength of the association, combination relationship, etc., and further filter out irrelevant tags with low relevance based on the basic conditions or characteristics that the customer group should meet. 3. Customer group combination generation. Use the large model to assign values to the customer group tags. Then, according to the different dimension categories of the customer group tags, use and, or, and priority brackets to combine them to get the final customer group combination

[0134] S205: Screening target customer group. The customer group obtained from the customer database using S204 is the target customer group.

[0135] In the embodiments of the present application, for a given product, potential marketable customer groups can be mined based on a discretized label system, filling the blank area in which existing technologies are located. And starting from product characteristics, first, the customer group portrait is mined, and then the semantic relevance between the customer group and the user label is used to realize accurate recall of the customer group label and generation of the customer group combination screening condition, automatically completing the marketing process.

[0136] Exemplary apparatus

[0137] Correspondingly, the embodiments of the present application also provide a device for mining customer groups, as shown in FIG. 3 The device includes:

[0138] The customer group feature module 301 is configured to input the business information of the target business into the customer group portrait model to obtain the customer group feature text of the customer group portrait model output, wherein the customer group portrait model is configured to generate the customer group feature text of the customer group suitable for marketing according to the semantic understanding of the input business information.

[0139] The customer group label module 302 is configured to match the user labels of different customers in the customer database based on the customer group feature text of the customer group to be marketed, to obtain the customer group label of the customer group to be marketed.

[0140] The mining module 303 is configured to filter the customers that meet the target conditions from the customer database to obtain the customer group of the target business, wherein the target conditions include a screening condition composed of the customer group label of the customer group to be marketed.

[0141] In some embodiments, the customer group feature text includes multiple items of customer group name, customer group demand for the business, customer group self-description, and customer group association degree with the business.

[0142] In some embodiments, the customer group label module 302 is specifically configured to input the target information into the label matching model to obtain a customer group label of the customer group to be marketed output by the label matching model.

[0143] The target information includes a label library and customer group characteristic text, and the label library includes user labels of different customers in a customer database.

[0144] In some embodiments, the target information further includes business information of the target business.

[0145] In some embodiments, the customer group label module 302 includes:

[0146] The similarity calculation unit is configured to calculate the similarity between each user label and the customer group characteristic text of the customer group to be marketed respectively.

[0147] The label determination unit is configured to determine a user label with a similarity higher than a similarity threshold as a customer group label of the customer group to be marketed.

[0148] In some embodiments, the similarity calculation unit is specifically configured to:

[0149] extract keywords in the customer group characteristic text of the customer group to be marketed;

[0150] calculate the similarity between each user label and the keywords of the customer group to be marketed respectively.

[0151] In some embodiments, the device further includes a supplement module configured to:

[0152] obtain a business label library corresponding to the target business, wherein the business label library includes user labels that the customer group to be marketed must have under the business rules of the target business;

[0153] add the user labels in the business label library to the customer group label of the customer group to be marketed.

[0154] In some embodiments, the device further includes an attribute screening module configured to:

[0155] input the business information of the target business and the customer group characteristic text into a customer group attribute model to obtain attribute information of the customer group to be marketed output by the customer group attribute model, wherein the customer group attribute model is configured to analyze attribute information possessed by a customer group of the target business according to the input customer group characteristic text and business information.

[0156] filter the customer group label of the customer group to be marketed based on the attribute information of the customer group to be marketed.

[0157] In some embodiments, the step of generating the target condition includes:

[0158] According to a preset label assignment rule, each customer group label of the customer group to be marketed is assigned;

[0159] The assigned customer group label is logically combined to obtain a target condition.

[0160] The device for mining customer groups provided in the embodiment belongs to the same application concept as the method for mining customer groups provided in the above embodiments of the application, can execute the method for mining customer groups provided in any of the above embodiments of the application, and has the corresponding function modules and beneficial effects of the execution method. Technical details not described in detail in the embodiment can be seen from the specific processing content of the method for mining customer groups provided in the above embodiments of the application, and will not be described here.

[0161] It should be understood that the modules in the above device can be implemented in the form of processor calling software. For example, the device includes a processor connected with a memory, the memory stores instructions, and the processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of the units of the device, wherein the processor can be a general processor such as a CPU or a microprocessor, and the memory can be an internal memory of the device or an external memory of the device. Alternatively, the units in the device can be implemented in the form of hardware circuit. The functions of part or all of the units can be implemented by designing the hardware circuit. The hardware circuit can be understood as one or more processors. For example, in one implementation, the hardware circuit is an ASIC, and the functions of part or all of the units are implemented by designing the logical relationship of elements in the circuit. For another example, in another implementation, the hardware circuit can be implemented by a PLD. Taking an FPGA as an example, it can include a large number of logic gate circuits, and the connection relationship between the logic gate circuits is configured by a configuration file, so as to implement the functions of part or all of the units. All the units of the above device can be implemented in the form of processor calling software, or all the units can be implemented in the form of hardware circuit, or part of the units are implemented in the form of processor calling software, and the remaining part is implemented in the form of hardware circuit.

[0162] In the embodiments of the present application, the processor is a circuit with signal processing capability. In one implementation, the processor can be a circuit with instruction reading and running capability, such as a CPU, a microprocessor, a GPU, or a DSP, etc. In another implementation, the processor can implement certain functions through a logic relationship of a hardware circuit, which is fixed or can be reconfigured. For example, the processor is a hardware circuit implemented by an ASIC or a PLD, such as an FPGA, etc. In the reconfigurable hardware circuit, the processor loads a configuration document to implement the hardware circuit configuration. It can be understood that the processor loads instructions to implement the functions of the above units.

[0163] It can be seen that each unit in the above apparatus can be one or more processors (or processing circuits) configured to implement the above methods, such as a CPU, a GPU, an NPU, a TPU, a DPU, a microprocessor, a DSP, an ASIC, an FPGA, or a combination of at least two of these processor forms.

[0164] In addition, each unit in the above apparatus can be integrated together or can be independently implemented. In one implementation, these units are integrated together to implement a SOC. The SOC can include at least one processor for implementing any of the above methods or the functions of the units of the apparatus. The at least one processor can be different, such as including a CPU and an FPGA, a CPU and an artificial intelligence processor, a CPU and a GPU, etc.

[0165] Exemplary electronic devices

[0166] The embodiments of the present application provide an electronic device, referring to FIG. 4 The device includes:

[0167] The memory 400 and the processor 410; wherein the memory 400 is connected with the processor 410, for storing programs; the processor 410 is used for realizing the method for excavating the guest group disclosed in any of the above embodiments by running the programs stored in the memory 400.

[0168] Specifically, the above electronic device can further include a bus, a communication interface 420, an input device 430, and an output device 440.

[0169] The processor 410, the memory 400, the communication interface 420, the input device 430, and the output device 440 are connected with each other through the bus. Among them:

[0170] The bus can include a channel for transmitting information between various components of a computer system.

[0171] The processor 410 can be a general processor, such as a general central processing unit (CPU), a microprocessor, or the like, or can be an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs of the present application. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a ready-to-use programmable gate array (FPGA), or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0172] The processor 410 can include a main processor and can further include a baseband chip, a modem, or the like.

[0173] The memory 400 stores programs for executing the technical solutions of the present application, and can also store operating systems and other key services. Specifically, the programs can include program codes, and the program codes include computer operation instructions. More specifically, the memory 400 can include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk memory, a flash, and the like.

[0174] The input device 430 can include a device that receives data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor, and the like.

[0175] The output device 440 can include a device that allows information to be output to a user, such as a display screen, a printer, a speaker, and the like.

[0176] The communication interface 420 can include a device using any transceiver, such as an Ethernet, a radio access network (RAN), a wireless local area network (WLAN), and the like, to communicate with other devices or communication networks.

[0177] The processor 410 executes the programs stored in the memory 400 and calls other devices, which can be used to implement each step of any one of the methods for mining a customer group provided by the above-mentioned embodiments of the present application.

[0178] The embodiments of the present application also propose a chip including a processor and a data interface, the processor reading and running programs stored on the memory through the data interface to execute the method for mining a customer group introduced in any of the above-mentioned embodiments. The specific processing process and its beneficial effects can be referred to the above-mentioned embodiments of the method for mining a customer group.

[0179] Exemplary computer program product and storage medium

[0180] In addition to the method and device described above, the embodiments of the present application can also be a computer program product, which includes computer program instructions, which, when executed by a processor, causes the processor to perform the steps in the method of mining a customer group according to various embodiments of the present application described in any of the embodiments of the present specification.

[0181] The computer program product can be written in any combination of one or more programming languages to perform the operations of the embodiments of the present application, including an object-oriented programming language, such as Java, C++, and the like, and a conventional procedural programming language, such as the "C" language or the like. Program code can be executed entirely on a user computing device, partially on a user device, as an independent software package, partially on a user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0182] In addition, the embodiments of the present application can also be a storage medium having a computer program stored thereon, which is executed by a processor to perform the steps in the method of mining a customer group according to various embodiments of the present application described in any of the embodiments of the present specification.

[0183] For each of the foregoing method embodiments, in order to simply describe, it is expressed as a combination of a series of actions, but those skilled in the art should know that the present application is not limited to the order of the actions described, because according to the present application, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0184] It should be noted that each of the embodiments in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the same and similar parts between each embodiment can be referred to each other. For device embodiments, since they are basically similar to method embodiments, they are described more simply, and the relevant parts refer to the part of the method embodiment.

[0185] The steps in the method of each embodiment of the present application can be adjusted, combined and reduced in sequence according to actual needs, and the technical features recorded in each embodiment can be replaced or combined.

[0186] The modules and sub-modules in the device and terminal of each embodiment of the present application can be combined, divided and reduced according to actual needs.

[0187] It should be understood that the disclosed terminal, device and method can be implemented in other ways. For example, the terminal embodiments described above are merely illustrative. For example, the division of modules or sub-modules is merely a logical function division. In actual implementation, another division manner can be used. For example, a plurality of sub-modules or modules can be combined or integrated into another module, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed modules can be indirect coupling or communication connection through some interfaces, devices or modules, and can be electrical, mechanical or other forms.

[0188] The modules or sub-modules described as separate components can or can not be physically separate, and the components of the modules or sub-modules can or can not be physical modules or sub-modules, i.e. can be located in one place or distributed on a plurality of network modules or sub-modules. Some or all of the modules or sub-modules can be selected according to actual needs to achieve the purpose of the embodiment.

[0189] In addition, the functional modules or sub-modules in each embodiment of the present application can be integrated into a processing module, or each module or sub-module can exist physically, or two or more modules or sub-modules can be integrated into one module. The integrated module or sub-module can be realized in the form of hardware or software functional module or sub-module.

[0190] The skilled person can further realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been described in general terms in the above description. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0191] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein can be directly implemented by hardware, software units executed by a processor, or a combination of both. The software units can be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0192] Finally, it should be noted that, in this document, the term "only" is used simply to set off from one entity or action to another in order to avoid the use of the term "and / or" or the like for the sake of clarity. In no way should the term "only" be interpreted as implying that there is an implied exclusion of any referenced entity or action. Moreover, the terms "comprising", "including", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises a... " does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the recited element.

[0193] The above description of disclosed embodiments provides enabling teaching for making or using the application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for mining customer groups, characterized in that: The method comprises: Inputting the business information of the target business into the customer group portrait model to obtain the customer group characteristic text of the target customer group output by the customer group portrait model, wherein the customer group portrait model is used to generate the customer group characteristic text of the target customer group suitable for marketing based on the semantic understanding of the input business information; Based on the customer group feature text of the customer group to be marketed, user tags of different customers in the customer database are matched to obtain the customer group tags of the customer group to be marketed; Customers meeting target conditions are screened from the customer database to obtain the customer group of the target business, wherein the target conditions include screening conditions based on customer group tags of the customer group to be marketed.

2. The method according to claim 1, characterized in that The customer group characteristic text includes multiple items of: customer group name, customer group's demand for the service, customer group's own description, and the correlation between the customer group and the service.

3. The method according to claim 1, characterized in that Based on the customer group feature text of the customer group to be marketed, user tags of different customers in the customer database are matched to obtain the customer group tags of the customer group to be marketed, including: Inputting the target information into a label matching model to obtain a customer group label of the target customer group output by the label matching model; The target information includes: a tag library and the customer group feature text, wherein the tag library includes user tags of different customers counted in the customer database; and the tag matching model is used to determine the user tags associated with the customer group feature text from the tag library input by the model.

4. The method according to claim 3, characterized in that The target information also includes: business information of the target business.

5. The method according to claim 1, wherein Based on the customer group feature text of the customer group to be marketed, user tags of different customers in the customer database are matched to obtain the customer group tags of the customer group to be marketed, including: Calculating the similarity between each user tag and the customer group feature text of the customer group to be marketed; The user tags with a similarity higher than a similarity threshold are determined as the customer group tags of the customer group to be marketed.

6. The method according to claim 5, characterized in that Calculating the similarity between each user tag and the customer group feature text of the customer group to be marketed, including: Extracting keywords from the customer group feature text of the target customer group; The similarity between each of the user tags and the keywords of the target customer group is calculated respectively.

7. The method according to any one of claims 1 to 6, characterized in that Based on the customer group feature text of the target customer group, after matching the user tags of different customers in the customer database to obtain the customer group tags of the target customer group, the method further includes: Obtaining a business tag library corresponding to the target business, wherein the business tag library includes: user tags that the target customer group must have under the business rules of the target business; The user tags in the business tag library are added to the customer group tags of the customer group to be marketed.

8. The method according to any one of claims 1 to 6, characterized in that Based on the customer group feature text of the target customer group, after matching the user tags of different customers in the customer database to obtain the customer group tags of the target customer group, the method further includes: Inputting the business information of the target business and the customer group characteristic text into a customer group attribute model to obtain attribute information of the target customer group output by the customer group attribute model, wherein the customer group attribute model is used to analyze the attribute information of the target business customer group based on the input customer group characteristic text and business information; The customer group tags of the customer group to be marketed are filtered based on the attribute information of the customer group to be marketed.

9. The method according to claim 1, characterized in that The steps of generating the target condition include: Assigning a value to each customer group label of the customer group to be marketed based on a preset label assignment rule; Perform logical combination on the assigned customer group labels to obtain the target conditions.

10. A device for mining customer groups, characterized in that: The device comprises: A customer profile module is configured to input the target business's business information into a customer profile model, and obtain a customer profile text of the target customer group output by the customer profile model. The customer profile model is configured to generate a customer profile text of the target customer group based on the semantic understanding of the input business information. A customer group labeling module is used to match the user labels of different customers in the customer database based on the customer group feature text of the customer group to be marketed, and obtain the customer group label of the customer group to be marketed; The mining module is used to filter customers that meet the target conditions from the customer database to obtain the customer group of the target business, wherein the target conditions include: filtering conditions based on customer group tags of the customer group to be marketed.

11. An electronic device, characterized in that: including memory and processor; The memory is connected to the processor and is used to store programs; The processor is configured to implement the method for mining customer groups according to any one of claims 1 to 9 by running the program in the memory.

12. A computer program product, characterized in that include: A computer program, which, when executed by a processor, implements the method for mining customer groups according to any one of claims 1 to 9.

Images