CN119379403A - A method, system and medium for screening target users - Google Patents
A method, system and medium for screening target users Download PDFInfo
- Publication number
- CN119379403A CN119379403A CN202411959241.9A CN202411959241A CN119379403A CN 119379403 A CN119379403 A CN 119379403A CN 202411959241 A CN202411959241 A CN 202411959241A CN 119379403 A CN119379403 A CN 119379403A
- Authority
- CN
- China
- Prior art keywords
- user
- vector
- target
- coding
- screening
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012216 screening Methods 0.000 title claims abstract description 125
- 238000000034 method Methods 0.000 title claims abstract description 47
- 239000013598 vector Substances 0.000 claims abstract description 345
- 238000004458 analytical method Methods 0.000 claims abstract description 14
- 230000006399 behavior Effects 0.000 claims description 92
- 238000012545 processing Methods 0.000 claims description 17
- 230000000694 effects Effects 0.000 claims description 15
- 238000006243 chemical reaction Methods 0.000 claims description 13
- 238000010276 construction Methods 0.000 claims description 10
- 238000004140 cleaning Methods 0.000 claims description 9
- 230000035945 sensitivity Effects 0.000 claims description 9
- 230000004044 response Effects 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 description 10
- 230000001737 promoting effect Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 2
- 238000009472 formulation Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 241000157593 Milvus Species 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the application provides a target user screening method, a system and a medium, wherein in the method, screening strategy analysis is firstly carried out according to a target marketing scene and target product attribute information so as to determine a target user screening strategy. And then, vector coding is carried out on a target user screening strategy based on a preset natural language model, and a target product coding vector is obtained. Further, similarity screening is performed according to the target product coding vector and a preset user vector database, so that multiple groups of first user coding vectors matched with the target product coding vector are determined from the preset user vector database. And finally, determining the target user according to the first user coding vector and the target product coding vector. Therefore, the product characteristics of the target product and shopping behavior preferences of different users can be more comprehensively represented, and the screening accuracy of the target user is improved.
Description
Technical Field
The present application relates to the field of data processing technologies, and in particular, to a method, a system, and a medium for screening a target user.
Background
In the current scene type marketing scheme, aiming at different types of products, a marketer needs to screen target users with potential willingness to know the products according to own experience, and the screening accuracy aiming at the target users is completely determined by the experience richness of the marketer, so that the screening accuracy of the target users is poor.
Disclosure of Invention
Based on the above problems, in order to improve screening accuracy for target users, the embodiments of the present application provide a target user screening method, system and medium.
The embodiment of the application discloses the following technical scheme:
in a first aspect, an embodiment of the present application provides a target user screening method, including:
screening strategy analysis is carried out according to the target marketing scene and the target product attribute information so as to determine a target user screening strategy;
Vector coding is carried out on the target user screening strategy based on a preset natural language model to obtain a target product coding vector, wherein the target product coding vector is used for representing product characteristics of the target product and marketing preference of the target marketing scene;
Performing similarity screening according to the target product coding vector and a preset user vector database to determine a plurality of groups of first user coding vectors matched with the target product coding vector from the preset user vector database, wherein the first user coding vectors are used for representing historical shopping behavior characteristics of users;
And determining a target user according to each first user coding vector and the target product coding vector.
In one possible implementation, the method further includes:
Performing data standardization processing on historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user;
based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
In one possible implementation, the historical shopping behavior information comprises a plurality of historical shopping behavior descriptions, and the user code vector comprises a shopping behavior code vector and a comprehensive preference code vector;
the method for constructing the preset user vector database by performing code vector conversion on each piece of historical shopping behavior information based on a preset natural language model to obtain a user code vector corresponding to each user independently comprises the following steps:
Vector coding is carried out on each historical shopping behavior description based on a preset natural language model to obtain a plurality of shopping behavior coding vectors, wherein the shopping behavior coding vectors correspond to the historical shopping behaviors one by one;
And carrying out vector merging processing on the shopping behavior coded vectors to obtain the comprehensive preference coded vector corresponding to each user independently so as to construct the preset user vector database.
In one possible implementation manner, the performing similarity screening according to the target product code vector and a preset user vector database to determine a plurality of groups of first user code vectors matched with the target product code vector from the preset user vector database includes:
calculating the vector similarity between the target product coding vector and each user coding vector;
and determining the user coding vector with the vector similarity higher than a preset first threshold value as the first user coding vector.
In one possible implementation manner, the determining the target user according to each of the first user code vector and the target product code vector includes:
Determining a second user-encoded vector from a first vector similarity ranking when a plurality of the first user-encoded vectors are present, the first vector similarity ranking being generated based on the vector similarity of each of the first user-encoded vectors;
Acquiring user liveness corresponding to the second user coding vector;
and determining the user with the user activity degree larger than a preset second threshold value as the target user aiming at the user activity degree corresponding to the second user coding vector.
In one possible implementation, the target marketing scene comprises a new product promotion scene and an old product warehouse cleaning scene;
When the target marketing scene is the new product promotion scene, the target product coding vector comprises new product preference characteristics, wherein the new product preference characteristics comprise new product sensitivity, new product attention and new product purchasing tendency;
When the target marketing scene is the old product warehouse-cleaning scene, the target product coding vector comprises promotion preference characteristics, wherein the promotion preference characteristics comprise promotion sensitivity, promotion participation degree and promotion response rate.
In a second aspect, an embodiment of the present application provides a target user screening system, including:
the strategy determining module is used for carrying out screening strategy analysis according to the target marketing scene and the target product attribute information so as to determine a target user screening strategy;
the first vector coding module is used for carrying out vector coding on the target user screening strategy based on a preset natural language model to obtain a target product coding vector, wherein the target product coding vector is used for representing the product characteristics of the target product and the marketing preference of the target marketing scene;
The first screening module is used for screening the similarity according to the target product coding vector and a preset user vector database so as to determine a plurality of groups of first user coding vectors matched with the target product coding vector from the preset user vector database, wherein the first user coding vectors are used for representing the historical shopping behavior characteristics of a user;
and the second screening module is used for determining a target user according to each first user coding vector and the target product coding vector.
In one possible implementation manner, the system further comprises a database construction module, wherein the database construction module is specifically used for:
Performing data standardization processing on historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user;
based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
In one possible implementation, the historical shopping behavior information comprises a plurality of historical shopping behavior descriptions, and the user code vector comprises a shopping behavior code vector and a comprehensive preference code vector;
the database construction module is specifically configured to:
Performing data standardization processing on historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user;
based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
In a third aspect, embodiments of the present application provide a computer readable storage medium having stored thereon a computer program which when executed by a processor implements any of the possible target user screening methods of the first aspect.
Compared with the prior art, the method, the system and the medium for screening the target users have the advantages that screening strategy analysis is needed according to the target marketing scene and the attribute information of the target products, so that the actual marketing requirements and the attributes of the target products are combined, the screening strategy of the target users in the actual marketing scene is determined to be more fit, and screening accuracy of the target users is improved. And then, vector coding is carried out on the target product attribute information and the target user screening strategy based on a preset natural language model, the natural language model can accurately capture the core semantics of the target product attribute and the target user screening strategy, and the core semantics features are converted into low-dimensional target product coding vectors so as to capture the similarity relationship between the historical shopping behaviors of the target user and the product features more accurately. Therefore, after the target product coding vector is determined, the target product coding vector is subjected to similarity screening with a preset user vector database, a plurality of groups of user coding vectors matched with the target product coding vector are determined, and target users are screened and determined according to the similarity between each user coding vector and the target product coding vector. Therefore, the similarity calculation is carried out on the code vectors generated through the natural language model, the target users are screened based on the similarity between the code vectors and the code vectors, the marketing features of the products and the historical behavior purchasing features of the users can be intuitively converted into semantic-level data, and the target users are screened based on the similarity degree of the data of the code vectors and the semantic-level data, so that the product features of the target products and shopping behavior preferences of different users can be more comprehensively represented, and the screening accuracy of the target users is improved.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the application, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a flow chart of a target user screening method according to an embodiment of the present application;
FIG. 2 is a schematic flow chart of constructing a database of preset user vectors according to an embodiment of the present application;
FIG. 3 is a flowchart of another method for screening target users according to an embodiment of the present application;
FIG. 4 is a schematic diagram of another method for screening target users according to an embodiment of the present application;
Fig. 5 is a schematic structural diagram of a target user screening system according to an embodiment of the present application.
Detailed Description
The present application will be further described in detail below with reference to specific embodiments and with reference to the accompanying drawings, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be specifically noted that the embodiments described in the embodiments of the present application are only some embodiments, but not all embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
It should be noted that unless otherwise defined, technical or scientific terms used in the embodiments of the present application should be given the ordinary meaning as understood by one of ordinary skill in the art to which the present application belongs. The terms "first," "second," and the like, as used in embodiments of the present application, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The word "comprising" or "comprises", and the like, means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof, but does not exclude other elements or items. The terms "connected" or "connected," and the like, are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "upper", "lower", "left", "right", etc. are used merely to indicate relative positional relationships, which may also be changed when the absolute position of the object to be described is changed.
As described above, in the current scenario-type marketing scheme, for different types of products, the marketer needs to screen the target users who potentially have knowledge about the products according to his own experience, and the screening accuracy for the target users is completely determined by the experience richness of the marketer, so that the screening accuracy of the target users is poor.
In order to solve the above problems, the embodiments of the present application provide a target user screening method, system and medium, in which screening policy analysis is first required according to a target marketing scene and attribute information of a target product, so as to combine actual marketing requirements with attributes of the target product, thereby determining a target user screening policy more fitting the actual marketing scene, and improving screening accuracy for the target user. And then, vector coding is carried out on the target product attribute information and the target user screening strategy based on a preset natural language model, the natural language model can accurately capture the core semantics of the target product attribute and the target user screening strategy, and the core semantics features are converted into low-dimensional target product coding vectors so as to capture the similarity relationship between the historical shopping behaviors of the target user and the product features more accurately. Therefore, after the target product coding vector is determined, the target product coding vector is subjected to similarity screening with a preset user vector database, a plurality of groups of user coding vectors matched with the target product coding vector are determined, and target users are screened and determined according to the similarity between each user coding vector and the target product coding vector. Therefore, the similarity calculation is carried out on the code vectors generated through the natural language model, the target users are screened based on the similarity between the code vectors and the code vectors, the marketing features of the products and the historical behavior purchasing features of the users can be intuitively converted into semantic-level data, and the target users are screened based on the similarity degree of the data of the code vectors and the semantic-level data, so that the product features of the target products and shopping behavior preferences of different users can be more comprehensively represented, and the screening accuracy of the target users is improved.
In order to make the present application better understood by those skilled in the art, the following description will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region.
Next, referring to fig. 1, a flowchart of a target user screening method provided by an embodiment of the present application will be described with reference to a specific embodiment of the present application, where the flowchart specifically includes the following steps:
s101, screening strategy analysis is carried out according to the target marketing scene and the target product attribute information so as to determine a target user screening strategy.
In an actual product marketing scene, aiming at the formulation of a target user screening strategy, if only through the attribute information (such as product name, product category, product characteristic description and the like) of a target product, the determined target user screening strategy often conforms to the actual product marketing requirement of a marketer.
Therefore, in the primary screening stage of the target user, the embodiment of the application needs to combine the target marketing scene of the marketer and the corresponding target product to carry out screening strategy analysis, thereby ensuring the comprehensiveness of the screening strategy of the target user. The target marketing scene can be roughly divided into a new product popularization scene and an old product warehouse cleaning scene, and the attribute information of the target product to be extracted also has differences under different marketing scenes. For example, when the target marketing scenario is a new product promotion scenario, the target product attribute information may be closer to the time of marketing with the target product and the product difference with other products that are new in the same time. When the target marketing scene is an old product warehouse-cleaning scene, the target product attribute information is more focused on the sales promotion activities, the sales promotion prices, the sales promotion time duration and the like related to the target products. Through classifying the target marketing scenes, the target product attribute information required by the screening strategy can be extracted more accurately, so that the target user screening strategy can cover key factors which possibly influence the purchasing decision of the user as comprehensively as possible, and the comprehensive formulation of the target user screening strategy is realized.
S102, vector coding is carried out on the target user screening strategy based on a preset natural language model to obtain a target product coding vector, wherein the target product coding vector is used for representing product characteristics of the target product and marketing preference of the target marketing scene.
The natural language model can capture deep semantic information in the text, and through the pre-trained preset natural language model, the nuances and complex relations between the target product and the target marketing scene can be captured, and the target user screening strategy of the text version is converted into the target product coding vector with high dimensionality (such as 768 dimensions), so that the specific characteristics of the product and the deviation of the marketing strategy are more accurately represented. For example, a textual description of attribute information of a target product may include aspects of the product's functionality, appearance, price, etc., while a description of a target marketing scenario may involve unstructured information of interest points, purchasing behavior, etc. of a target user, which can be efficiently encoded by natural language processing techniques into a target product encoding vector for the target product, thereby mathematically quantifying the product's characteristics and marketing preferences.
Specifically, in the process of vector encoding a target user screening policy by a preset natural language model, in order to ensure that the format of input data of the language model is accurate, the target user screening policy needs to be converted into standardized description, and the above-mentioned new product popularization scene and old product warehouse cleaning scene are taken as examples:
a new product promotion scene comprises new product preference characteristics, target product attribute information and target user characteristics;
the old product warehouse cleaning scene comprises promotion activity preference characteristics, target product attribute information and target user characteristics;
Therefore, the standardized description processing of the target user screening strategy can ensure the format standardization of the input data of the preset natural language model. And inputting the standardized description of the target user screening strategy into a preset natural language model, and generating a corresponding code vector by utilizing a code function in the language model. Therefore, the complex marketing scene requirements and the target product attribute information are converted into high-density target product coding vectors, so that the screening accuracy and the screening efficiency of target users are improved.
In one possible implementation, the preset natural language model may employ BERT, GPT, etc. as the underlying language model. The embodiment of the application further trains the language model by taking various products and corresponding user information as training data, thereby obtaining a pre-trained preset natural language model.
S103, screening the similarity according to the target product coding vector and a preset user vector database to determine a plurality of groups of first user coding vectors matched with the target product coding vector from the preset user vector database, wherein the first user coding vectors are used for representing historical shopping behavior characteristics of users.
The preset user vector database comprises a plurality of coded vectors of users. Wherein the user's coded vector is used as a characterization representation of the user's historical shopping behavior. Such as the user's merchandise type preferences, the degree of preference for new merchandise, the degree of preference for old merchandise promotions, and so forth. The user coding vectors of the users in the user vector database are preset, so that shopping behavior preferences of the users can be accurately represented, and comprehensive preferences formed based on various types of shopping behavior preferences can be accurately represented. The method is used as a data reference for screening the target user, and a similarity screening link between the user coding vector and the target product coding vector can be refined, so that the first user coding vector with certain similarity with the target product coding vector can be accurately screened.
Specifically, in the process of performing similarity screening based on the target product code vector and the preset user vector database, it is necessary to calculate the vector similarity between the target product code vector and each user code vector, and screen the user code vector with the vector similarity higher than the preset first threshold value from the vector similarity, and determine the user code vector as the first user code vector.
The method for calculating the vector similarity between the target product code vector and the user code vector includes, but is not limited to, a cosine similarity calculation method, a Euclidean distance calculation method, a Manhattan distance calculation method and a Hash similarity calculation method. In an actual application scenario, an adaptive similarity calculation method may be selected according to the computing resource condition and the efficiency requirement, and the mode of similarity calculation in this embodiment is not limited.
In addition, in one possible implementation manner, in order to ensure the accuracy of calculating the vector similarity, in the generation stage of the target product coding vector, besides the target user screening policy, a specific description for the target user may be used as a part of reference data, so as to generate a corresponding target product coding vector. In particular descriptions for the target user, historical shopping behavior preferences specific to the target user are included, as well as product attribute preferences. The specific feature descriptions aiming at the target user are used as basic data for generating the target product coding vector, so that the similarity comparison condition with higher accuracy can be provided when the vector similarity between the target product coding vector and each user coding vector is calculated, the calculation accuracy of the vector similarity is improved, and the screening accuracy of the target user is improved.
In step S103, it is known that, for the description of the preset user vector database, the user code vectors stored in the database about each user are the key points of the target user screening in the embodiment of the present application. In the user coding vector, the more accurate the description of the user historical shopping behavior characteristics is, the higher the screening accuracy of the user coding vector for the target user is, so that the establishment of a preset user vector database is particularly important. Next, a process of constructing the preset user vector database will be described with reference to the drawings of the specific embodiment.
Referring to fig. 2, the flow chart for constructing a preset user vector database according to an embodiment of the present application specifically includes the following steps:
S1031, performing data standardization processing on the historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user.
The user's historical shopping log is used to characterize the user's historical shopping behavior, as well as related product information that is involved in the historical shopping behavior. The user's historical shopping behavior may be collected from various data sources, such as an e-commerce platform database, user behavior logs, and the like. In the preliminary acquisition stage of the historical shopping log, data standardization processing is required to be performed on the historical shopping log so as to obtain historical shopping behavior information in a standard format, and the historical shopping behavior information of the user can be exemplified by the following form:
commodity information |channel|distance from the period of marketing|sales promotion force|equity form|;
Thus, through the historical shopping behavior information of the user, shopping behavior preferences of different users and preferences of the users on products can be characterized. On this basis, these data are converted into a format suitable for natural language model analysis to facilitate the conversion of the encoded vectors.
S1032, based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
In one possible implementation, the preset user vector database may be a relational database, a NoSQL database, or a specialized vector database (such as Faiss, milvus, etc.), and the specific choice depends on the size of the data, the query speed requirements, and the architectural design of the system, which is not limited in this embodiment.
Further, through a pre-trained preset natural language model, the historical shopping behavior information of each user is used as input data of the preset natural language model, and high-dimensional original data are converted into low-dimensional dense vector representations, so that the intrinsic characteristics of the user in the historical shopping behaviors are captured, and the user coding vector for each user is determined. And associating the user coding vector of each user with the identity ID of the user, and storing the user coding vector in a user vector database, so that the construction of the preset user vector database can be realized.
Specifically, in the process of constructing the preset user vector database, in order to ensure the data accuracy of each user coding vector in the database, the generation of the user coding vector needs to be obtained by carrying out vector coding analysis on each historical shopping behavior of the user. The process of constructing the preset user vector database in this way is realized by the following two steps:
Step one, carrying out vector coding on each historical shopping behavior description based on a preset natural language model to obtain a plurality of shopping behavior coding vectors, wherein the shopping behavior coding vectors are in one-to-one correspondence with the historical shopping behaviors.
In the historical shopping information of the user, a plurality of historical shopping behavior descriptions about the user are included. And shopping behavior code vectors and comprehensive preference code vectors are included in the user code vectors for the user. The shopping behavior coding vector is a coding vector corresponding to each historical shopping behavior description of the user, and the vector coding of each historical behavior description of the user can accurately reflect the behavior preference of the user represented by the historical shopping behavior under each specific historical shopping behavior, so that a user portrait can be constructed more accurately, and the corresponding user coding vector can be determined.
And secondly, carrying out vector combination processing on the shopping behavior coded vectors to obtain the comprehensive preference coded vectors which are independently corresponding to the users so as to construct the preset user vector database.
On the other hand, the integrated code vector is used to characterize the user's integrated preferences, which are not limited to the user's historical shopping behavior preferences, but rather favor the user's preference level for product information and various promotional campaigns. The multiple shopping behavior coding vectors of the user are subjected to vector combination processing, so that the shopping preference, shopping habit and other information of the user can be more comprehensively represented, more accurate user images are constructed, and the screening accuracy of the target user is improved.
In particular, the shopping behavior code vector and the comprehensive preference code vector of the user form a user code vector together, and the user code vector does not only include a single comprehensive preference code vector. The meaning of the arrangement is that the accuracy in the similarity screening is improved, when the similarity screening is performed, the shopping behavior code vector, the comprehensive preference code vector and the target product code vector of the user can be simultaneously compared, and when at least one code vector has certain similarity, the code vector is generalized into a corresponding first user code vector, so that the accuracy of the similarity screening in the step S103 is ensured.
S104, determining a target user according to each first user code vector and the target product code vector.
When screening out the first user coding vectors with the vector similarity larger than a preset first threshold, the final target users are screened out by combining the similarity ranking of the first user coding vectors and the liveness of all the affiliated users, so that the screening accuracy of the target users is further improved. Next, a specific screening process of a target user will be described with reference to a specific embodiment of the drawings, referring to fig. 3, which is a schematic flow chart of another screening method of a target user according to an embodiment of the present application, and specifically includes the following steps:
S1041, when a plurality of first user code vectors exist, determining a second user code vector according to a first vector similarity ranking, wherein the first vector similarity ranking is generated based on the vector similarity of each first user code vector.
When a plurality of first user coding vectors with vector similarity higher than a preset first threshold exist, ranking according to the first vector similarity of the plurality of coding vectors, screening out coding vectors with the rank higher than a certain value from the ranking, and determining the coding vectors as second user coding vectors.
S1042, obtaining the user activity corresponding to the second user coding vector.
S1043, determining the user with the user activity degree larger than a preset second threshold value as the target user according to the user activity degree corresponding to the second user coding vector.
In an actual application scenario, even if a user code vector corresponding to the user has a high similarity with a target product code vector, if the user is not active for a long time, a product recommendation for the user is likely to be unable to be responded. Therefore, on the basis of screening the second user coding vectors based on the first vector similarity ranking, user activity corresponding to each second user coding vector needs to be obtained, if the user activity of a certain user is lower than a preset second threshold value, the user is indicated to be in a dormant state for a long time, and the user is not determined to be a target user. In addition, the user with the user activity larger than the preset second threshold value in the plurality of second user coding vectors can be determined to be the target user, so that the accurate screening of the target user under the specific marketing scene and the target product is realized.
As can be seen from the foregoing description of the target marketing scenario, in the embodiment of the present application, the target marketing scenario may be divided into a new product promotion scenario and an old product warehouse cleaning scenario. Under different marketing scenarios, the corresponding marketing strategies will differ, and thus the vector features included in the target product encoding vector will also differ.
According to the new product promotion scene and the old product warehouse cleaning scene, vector features in the target product coding vector can be divided into new product preference features and promotion activity preference features. The new product preference features include new product sensitivity, new product attention and new product purchasing tendency, and the following is a specific introduction aiming at the three features:
New product sensitivity, which is the sensitivity degree of the user to the newly marketed goods, including browsing, collecting, purchasing and other actions.
The attention degree of the new product, namely the attention degree of the user to the new product, can be measured by the stay time of the user on the new product page, the frequency of browsing the new product for a plurality of times and the like.
The purchase trend of the new product, namely the purchase intention of the user in the initial stage of the new product marketing, can be reflected by the history record of the purchase of the new product by the user.
On the other hand, the promotion activity preference features include promotion sensitivity, promotion participation and promotion response rate, and the following are specific descriptions of these three features:
promotion sensitivity-the sensitivity of a user to a promotional program, including reactions to coupons, discounts, full reductions, and the like.
Promotion participation, the frequency and depth of user participation in a promotional program, such as the number of user browses, purchases and purchases during the promotional program.
Promotion response Rate-the rate of response of a user to promotional information, including clicking on a promotional link, participating in a promotional program, and proportion of final purchase.
Thus, when the target product code and the user code vector are subjected to similarity screening, the screening efficiency for the target user can be improved by comparing the characteristics determined by the target marketing scene with the characteristics in the user code vector. For example, when the target marketing scene is a new product promotion scene, the new product preference characteristics of each user are obtained from a preset user vector database, and are compared and screened with the new product preference characteristics in the target product coding vector in a similarity mode, so that the screening efficiency of the target user can be improved.
Further, reference may be made to a schematic diagram of a further target user screening method shown in fig. 4, in which embedding represents a correspondingly generated code vector. Through vector coding of commodities and users and similarity screening based on a vector database, corresponding target users can be accurately screened out of a plurality of users, and the effect of improving screening accuracy of the target users is achieved.
The embodiment of the application provides a target user screening method, which comprises the steps of firstly, carrying out screening strategy analysis according to target marketing scenes and target product attribute information so as to combine actual marketing requirements with the attribute of a target product, thereby determining a target user screening strategy which is more fit with the actual marketing scenes, and improving screening accuracy of target users. And then, vector coding is carried out on the target product attribute information and the target user screening strategy based on a preset natural language model, the natural language model can accurately capture the core semantics of the target product attribute and the target user screening strategy, and the core semantics features are converted into low-dimensional target product coding vectors so as to capture the similarity relationship between the historical shopping behaviors of the target user and the product features more accurately. Therefore, after the target product coding vector is determined, the target product coding vector is subjected to similarity screening with a preset user vector database, a plurality of groups of user coding vectors matched with the target product coding vector are determined, and target users are screened and determined according to the similarity between each user coding vector and the target product coding vector. Therefore, the similarity calculation is carried out on the code vectors generated through the natural language model, the target users are screened based on the similarity between the code vectors and the code vectors, the marketing features of the products and the historical behavior purchasing features of the users can be intuitively converted into semantic-level data, and the target users are screened based on the similarity degree of the data of the code vectors and the semantic-level data, so that the product features of the target products and shopping behavior preferences of different users can be more comprehensively represented, and the screening accuracy of the target users is improved.
The following describes a target user screening system provided in the embodiment of the present application, and a target user screening system described below and a target user screening method described above may be referred to correspondingly.
Referring to fig. 5, the structure diagram of a target user screening system provided by the embodiment of the present application specifically includes the following modules:
The policy determining module 100 is configured to perform screening policy analysis according to the target marketing scenario and the target product attribute information, so as to determine a target user screening policy;
The first vector encoding module 200 is configured to perform vector encoding on the target user screening policy based on a preset natural language model to obtain a target product encoding vector, where the target product encoding vector is used to characterize product features of the target product and marketing preference of the target marketing scene;
The first filtering module 300 is configured to perform similarity filtering according to the target product encoding vector and a preset user vector database, so as to determine a plurality of groups of first user encoding vectors matched with the target product encoding vector from the preset user vector database, where the first user encoding vectors are used for characterizing historical shopping behavior characteristics of a user;
And the second screening module 400 is configured to determine a target user according to each of the first user code vector and the target product code vector.
In one possible implementation manner, the system further comprises a database construction module, wherein the database construction module is specifically used for:
Performing data standardization processing on historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user;
based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
In one possible implementation, the historical shopping behavior information comprises a plurality of historical shopping behavior descriptions, and the user code vector comprises a shopping behavior code vector and a comprehensive preference code vector;
the database construction module is specifically configured to:
Performing data standardization processing on historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user;
based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
Based on the same inventive concept, corresponding to the method of any embodiment described above, an embodiment of the present application further provides a computer-readable storage medium storing computer instructions for causing the computer to perform the target user screening method according to any embodiment described above.
Computer readable media of embodiments of the application, including both permanent and non-permanent, removable and non-removable media, may be used to implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device.
The storage medium of the foregoing embodiment stores computer instructions for causing the computer to execute the method for screening a target user for a millimeter wave signal according to any one of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which are not described herein.
It should be noted that, in the present specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment is mainly described in a different point from other embodiments. In particular, the methods, systems, and media are described more simply as they are substantially similar to the method embodiments, as relevant to the description of the method embodiments. The methods, systems, and media described above are illustrative only, and elements described as separate elements may or may not be physically separate, and elements as hints for elements may or may not be physical elements, may be located in one place, or may be distributed over multiple network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
The foregoing is only one specific embodiment of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions easily contemplated by those skilled in the art within the technical scope of the present application should be included in the scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.
Claims (10)
1. A method for screening a target user, comprising:
screening strategy analysis is carried out according to the target marketing scene and the target product attribute information so as to determine a target user screening strategy;
Vector coding is carried out on the target user screening strategy based on a preset natural language model to obtain a target product coding vector, wherein the target product coding vector is used for representing product characteristics of the target product and marketing preference of the target marketing scene;
Performing similarity screening according to the target product coding vector and a preset user vector database to determine a plurality of groups of first user coding vectors matched with the target product coding vector from the preset user vector database, wherein the first user coding vectors are used for representing historical shopping behavior characteristics of users;
And determining a target user according to each first user coding vector and the target product coding vector.
2. The method according to claim 1, wherein the method further comprises:
Performing data standardization processing on historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user;
based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
3. The method of claim 2, wherein the historical shopping behavior information comprises a plurality of historical shopping behavior descriptions, and wherein the user-encoded vector comprises a shopping behavior-encoded vector and a comprehensive preference-encoded vector;
the method for constructing the preset user vector database by performing code vector conversion on each piece of historical shopping behavior information based on a preset natural language model to obtain a user code vector corresponding to each user independently comprises the following steps:
Vector coding is carried out on each historical shopping behavior description based on a preset natural language model to obtain a plurality of shopping behavior coding vectors, wherein the shopping behavior coding vectors correspond to the historical shopping behaviors one by one;
And carrying out vector merging processing on the shopping behavior coded vectors to obtain the comprehensive preference coded vector corresponding to each user independently so as to construct the preset user vector database.
4. The method of claim 2, wherein the performing similarity screening according to the target product code vector and a preset user vector database to determine a plurality of sets of first user code vectors matching the target product code vector from the preset user vector database comprises:
calculating the vector similarity between the target product coding vector and each user coding vector;
and determining the user coding vector with the vector similarity higher than a preset first threshold value as the first user coding vector.
5. The method of claim 4, wherein said determining a target user from each of said first user-encoded vector and said target product-encoded vector comprises:
Determining a second user-encoded vector from a first vector similarity ranking when a plurality of the first user-encoded vectors are present, the first vector similarity ranking being generated based on the vector similarity of each of the first user-encoded vectors;
Acquiring user liveness corresponding to the second user coding vector;
and determining the user with the user activity degree larger than a preset second threshold value as the target user aiming at the user activity degree corresponding to the second user coding vector.
6. The method of claim 1, wherein the targeted marketing scenarios include a new product promotion scenario and an old product clearance scenario;
When the target marketing scene is the new product promotion scene, the target product coding vector comprises new product preference characteristics, wherein the new product preference characteristics comprise new product sensitivity, new product attention and new product purchasing tendency;
When the target marketing scene is the old product warehouse-cleaning scene, the target product coding vector comprises promotion preference characteristics, wherein the promotion preference characteristics comprise promotion sensitivity, promotion participation degree and promotion response rate.
7. A target user screening system, comprising:
the strategy determining module is used for carrying out screening strategy analysis according to the target marketing scene and the target product attribute information so as to determine a target user screening strategy;
the first vector coding module is used for carrying out vector coding on the target user screening strategy based on a preset natural language model to obtain a target product coding vector, wherein the target product coding vector is used for representing the product characteristics of the target product and the marketing preference of the target marketing scene;
The first screening module is used for screening the similarity according to the target product coding vector and a preset user vector database so as to determine a plurality of groups of first user coding vectors matched with the target product coding vector from the preset user vector database, wherein the first user coding vectors are used for representing the historical shopping behavior characteristics of a user;
and the second screening module is used for determining a target user according to each first user coding vector and the target product coding vector.
8. The system of claim 7, further comprising a database construction module, wherein the database construction module is configured to:
Performing data standardization processing on historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user;
based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
9. The system of claim 8, wherein the historical shopping behavior information comprises a plurality of historical shopping behavior descriptions, and wherein the user-encoded vector comprises a shopping behavior-encoded vector and a comprehensive preference-encoded vector;
the database construction module is specifically configured to:
Performing data standardization processing on historical shopping logs of a plurality of users to obtain historical shopping behavior information of each user;
based on a preset natural language model, performing code vector conversion on each piece of historical shopping behavior information to obtain a user code vector corresponding to each user independently so as to construct the preset user vector database.
10. A computer readable storage medium having stored thereon a computer program, characterized in that the program, when executed by a processor, implements the target user screening method of any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411959241.9A CN119379403A (en) | 2024-12-30 | 2024-12-30 | A method, system and medium for screening target users |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411959241.9A CN119379403A (en) | 2024-12-30 | 2024-12-30 | A method, system and medium for screening target users |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN119379403A true CN119379403A (en) | 2025-01-28 |
Family
ID=94340867
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411959241.9A Pending CN119379403A (en) | 2024-12-30 | 2024-12-30 | A method, system and medium for screening target users |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119379403A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107424043A (en) * | 2017-06-15 | 2017-12-01 | 北京三快在线科技有限公司 | A kind of Products Show method and device, electronic equipment |
| WO2022041979A1 (en) * | 2020-08-28 | 2022-03-03 | 腾讯科技（深圳）有限公司 | Information recommendation model training method and related device |
| CN117726359A (en) * | 2024-02-08 | 2024-03-19 | 成都纳宝科技有限公司 | Interactive marketing method, system and equipment |
| CN118069926A (en) * | 2024-01-19 | 2024-05-24 | 福建宜珊智能科技有限公司 | Visual intelligent decision system and method based on digital platform |
| CN118296209A (en) * | 2024-04-28 | 2024-07-05 | 零犀(北京)科技有限公司 | User quantitative screening method, device and storage medium |
| CN119151591A (en) * | 2024-09-06 | 2024-12-17 | 云南一舱科技文化有限公司 | Big data-based network live broadcast e-commerce marketing management system and method |
-
2024
- 2024-12-30 CN CN202411959241.9A patent/CN119379403A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107424043A (en) * | 2017-06-15 | 2017-12-01 | 北京三快在线科技有限公司 | A kind of Products Show method and device, electronic equipment |
| WO2022041979A1 (en) * | 2020-08-28 | 2022-03-03 | 腾讯科技（深圳）有限公司 | Information recommendation model training method and related device |
| CN118069926A (en) * | 2024-01-19 | 2024-05-24 | 福建宜珊智能科技有限公司 | Visual intelligent decision system and method based on digital platform |
| CN117726359A (en) * | 2024-02-08 | 2024-03-19 | 成都纳宝科技有限公司 | Interactive marketing method, system and equipment |
| CN118296209A (en) * | 2024-04-28 | 2024-07-05 | 零犀(北京)科技有限公司 | User quantitative screening method, device and storage medium |
| CN119151591A (en) * | 2024-09-06 | 2024-12-17 | 云南一舱科技文化有限公司 | Big data-based network live broadcast e-commerce marketing management system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101419504B1 (en) | System and method providing a suited shopping information by analyzing the propensity of an user | |
| RU2693184C2 (en) | Simulating session context for colloquial speech understanding systems | |
| US9348898B2 (en) | Recommendation system with dual collaborative filter usage matrix | |
| Selke et al. | Pushing the boundaries of crowd-enabled databases with query-driven schema expansion | |
| WO2018040069A1 (en) | Information recommendation system and method | |
| Xu et al. | Integrated collaborative filtering recommendation in social cyber-physical systems | |
| KR20160057475A (en) | System and method for actively obtaining social data | |
| US9269049B2 (en) | Methods, apparatus, and systems for using a reduced attribute vector of panel data to determine an attribute of a user | |
| CN103377443A (en) | Online trade platform and processing method thereof | |
| JP2010286910A (en) | Information processor, information processing method, and program | |
| CN112989169B (en) | Target object identification method, information recommendation method, device, equipment and medium | |
| CN111475744B (en) | Personalized position recommendation method based on ensemble learning | |
| CN110637317A (en) | Distributed node cluster for establishing digital contact points across multiple devices on a digital communications network | |
| CN116821475B (en) | Video recommendation method and device based on client data and computer equipment | |
| CN110069713A (en) | A kind of personalized recommendation method based on user's context perception | |
| CN106708871A (en) | Method and device for identifying social service characteristics user | |
| Zhang et al. | Matching state estimation scheme for content-based sensor search in the Web of things | |
| CN110633410A (en) | Information processing method and device, storage medium, and electronic device | |
| Jun et al. | A novel method of IP R&D using patent analysis and expert survey | |
| CN118250516B (en) | Hierarchical processing method for users | |
| Li et al. | A generative adversarial approach with social relationship for recommender systems | |
| Han et al. | Developing smart service concepts: morphological analysis using a Novelty-Quality map | |
| CN112036987B (en) | Method and device for determining recommended commodity | |
| CN114861079B (en) | A collaborative filtering recommendation method and system integrating product features | |
| Su | Accurate marketing algorithm of network video based on user big data analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |