CN117893258A - User recommendation method, device and storage medium - Google Patents
User recommendation method, device and storage medium Download PDFInfo
- Publication number
- CN117893258A CN117893258A CN202410059592.8A CN202410059592A CN117893258A CN 117893258 A CN117893258 A CN 117893258A CN 202410059592 A CN202410059592 A CN 202410059592A CN 117893258 A CN117893258 A CN 117893258A
- Authority
- CN
- China
- Prior art keywords
- value
- index
- user
- weight
- users
- 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
- 238000000034 method Methods 0.000 title claims abstract description 116
- 238000003860 storage Methods 0.000 title claims abstract description 21
- 238000004891 communication Methods 0.000 claims abstract description 30
- 238000012545 processing Methods 0.000 claims description 55
- 230000015654 memory Effects 0.000 claims description 20
- 238000010606 normalization Methods 0.000 claims description 13
- 230000002159 abnormal effect Effects 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000012706 support-vector machine Methods 0.000 claims description 7
- 230000002596 correlated effect Effects 0.000 claims description 6
- 230000006870 function Effects 0.000 description 27
- 230000009286 beneficial effect Effects 0.000 description 10
- 239000011159 matrix material Substances 0.000 description 10
- 230000000875 corresponding effect Effects 0.000 description 9
- 230000008569 process Effects 0.000 description 6
- 238000011156 evaluation Methods 0.000 description 5
- 230000035945 sensitivity Effects 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000007787 long-term memory Effects 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000000926 separation method Methods 0.000 description 3
- 230000006403 short-term memory Effects 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 239000000470 constituent Substances 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000003064 k means clustering Methods 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 229920001690 polydopamine Polymers 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Abstract
The application provides a user recommending method, a device and a storage medium, relates to the technical field of communication, and can determine a service policy of each user based on indexes of multiple dimensions of the user. The method comprises the following steps: acquiring a plurality of first-level indexes of each user in a plurality of users, second-level indexes in each first-level index and index values of each second-level index; determining the weight of each secondary index and the index value and the weight of each primary index based on the index value of each secondary index; determining an explicit value score, an implicit value score and a user value score of each user based on the index value and the weight of each primary index; dominant value is the current value in the user's value; the implicit value is a desired value among the user values; a service policy for each user is determined based on the explicit value score and the implicit value score for each user.
Description
Technical Field
The present disclosure relates to the field of communications technologies, and in particular, to a user recommendation method, apparatus, and storage medium.
Background
In order to improve the advantages of the enterprise to users, the enterprise needs to accurately analyze and position the users, and a personalized service strategy is formulated for each user.
At present, most enterprises search target users to develop services through offline marketing channels, but sales personnel usually have the problem of insufficient information during marketing, so that the enterprises are difficult to accurately find target guest groups, the traditional marketing mode is single, personalized services cannot be formulated for different users, and the marketing success rate is low.
Disclosure of Invention
The application provides a user recommending method, a device and a storage medium, which can determine a service policy of each user based on indexes of multiple dimensions of the user.
In order to achieve the above purpose, the present application adopts the following technical scheme:
in a first aspect, the present application provides a user recommendation method, including: acquiring a plurality of first-level indexes of each user in a plurality of users, second-level indexes in each first-level index and index values of each second-level index; determining the weight of each secondary index and the index value and the weight of each primary index based on the index value of each secondary index; determining an explicit value score, an implicit value score and a user value score of each user based on the index value and the weight of each primary index; dominant value is the current value in the user's value; the implicit value is a desired value among the user values; a service policy for each user is determined based on the explicit value score and the implicit value score for each user.
In one possible implementation, a method for determining a weight of each secondary index and an index value and a weight of each primary index based on an index value of each secondary index includes: determining the weight of each secondary index by a combined weight method based on the index value of each secondary index; determining an index value of each primary index based on the weight and the index value of the secondary index in each primary index; the weight of each level index is determined by a combined weight method based on the index value of each level index.
In one possible implementation, the method for determining the weight of each secondary index by using the combined weight method based on the index value of each secondary index includes: determining a first weight of each secondary index by using an entropy weight method based on the index value of each secondary index; determining a second weight of each secondary index by using a CRITIC weighting method based on the index value of each secondary index; determining a third weight of each secondary index by using a coefficient of variation method based on the index value of each secondary index; determining coefficients of the first weight, the second weight and the third weight, and summing the first weight, the second weight and the third weight based on the coefficients to obtain the weight of each secondary index; the sum of the coefficients of the first weight, the second weight and the third weight is 1.
In one possible implementation, a method for determining an explicit value score, an implicit value score, and a user value score for each user based on an index value and a weight for each level one index includes: determining an explicit value score and a weight of the explicit value based on the index value and the weight of each level of index in the explicit value; determining a implicit value score and a weight of the implicit value based on the index value and the weight of each level of index in the implicit value; the user value score of the user is determined by weighted summation based on the explicit value score, the implicit value score, the weight of the explicit value, and the weight of the implicit value.
In one possible implementation, determining a service policy for each user based on the explicit value score and the implicit value score for each user includes: classifying the users based on the explicit value scores and the implicit value scores of the users to obtain classification results of the users; based on the classification result of the users, the service policy of each user is determined.
In one possible implementation manner, the method for classifying the user based on the explicit value score and the implicit value score of the user to obtain the classification result of the user includes: classifying the users by using k-means clusters based on the dominant value score and the recessive value score of the users to obtain a first type of the users; classifying the users by using a support vector machine model based on the explicit value score and the implicit value score of the users to obtain a second type of the users; classifying the users by using the long-short-term memory network model based on the explicit value score and the implicit value score of the users to obtain a third type of the users; if at least two types in the three types are consistent, determining the classification result with the consistent type as the classification result of the user.
In one possible implementation, obtaining the index value of each secondary index includes: acquiring an initial index value of each secondary index; carrying out data preprocessing and standardization processing on the initial index value of each secondary index to obtain an index value of each secondary index; the pretreatment comprises the following steps: processing the missing value and the abnormal value; the normalization processing comprises the normalization processing of positive indexes and negative indexes; the index value of the forward index is positively correlated with the user value; the index value of the negative index is inversely related to the user value.
In a second aspect, the present application provides a user recommendation device, the device comprising: a communication unit and a processing unit; a communication unit for acquiring a plurality of primary indexes of each user among the plurality of users, a secondary index in each primary index, and an index value of each secondary index; the processing unit is used for determining the weight of each secondary index and the index value and the weight of each primary index based on the index value of each secondary index; the processing unit is also used for determining the dominant value score, the recessive value score and the user value score of each user based on the index value and the weight of each first-level index; dominant value is the current value in the user's value; the implicit value is a desired value among the user values; and the processing unit is also used for determining the service strategy of each user based on the dominant value score and the implicit value score of each user.
In one possible implementation, the processing unit is specifically configured to: determining the weight of each secondary index by a combined weight method based on the index value of each secondary index; determining an index value of each primary index based on the weight and the index value of the secondary index in each primary index; the weight of each level index is determined by a combined weight method based on the index value of each level index.
In one possible implementation, the processing unit is specifically configured to: determining a first weight of each secondary index by using an entropy weight method based on the index value of each secondary index; determining a second weight of each secondary index by using a CRITIC weighting method based on the index value of each secondary index; determining a third weight of each secondary index by using a coefficient of variation method based on the index value of each secondary index; determining coefficients of the first weight, the second weight and the third weight, and summing the first weight, the second weight and the third weight based on the coefficients to obtain the weight of each secondary index; the sum of the coefficients of the first weight, the second weight and the third weight is 1.
In one possible implementation, the processing unit is specifically configured to: determining an explicit value score and a weight of the explicit value based on the index value and the weight of each level of index in the explicit value; determining a implicit value score and a weight of the implicit value based on the index value and the weight of each level of index in the implicit value; the user value score of the user is determined by weighted summation based on the explicit value score, the implicit value score, the weight of the explicit value, and the weight of the implicit value.
In one possible implementation, the processing unit is specifically configured to: classifying the users based on the explicit value scores and the implicit value scores of the users to obtain classification results of the users; based on the classification result of the users, the service policy of each user is determined.
In one possible implementation, the processing unit is specifically configured to: classifying the users by using k-means clusters based on the dominant value score and the recessive value score of the users to obtain a first type of the users; classifying the users by using a support vector machine model based on the explicit value score and the implicit value score of the users to obtain a second type of the users; classifying the users by using the long-short-term memory network model based on the explicit value score and the implicit value score of the users to obtain a third type of the users; if at least two types in the three types are consistent, determining the classification result with the consistent type as the classification result of the user.
In one possible implementation, the communication unit is specifically configured to: acquiring an initial index value of each secondary index; the processing unit is specifically used for: carrying out data preprocessing and standardization processing on the initial index value of each secondary index to obtain an index value of each secondary index; the pretreatment comprises the following steps: processing the missing value and the abnormal value; the normalization processing comprises the normalization processing of positive indexes and negative indexes; the index value of the forward index is positively correlated with the user value; the index value of the negative index is inversely related to the user value.
In a third aspect, the present application provides a user recommendation device, including: a processor and a communication interface; the communication interface is coupled to a processor for running a computer program or instructions to implement the user recommendation method as described in any one of the possible implementations of the first aspect and the first aspect.
In a fourth aspect, the present application provides a computer readable storage medium having instructions stored therein which, when run on a terminal, cause the terminal to perform a user recommendation method as described in any one of the possible implementations of the first aspect and the first aspect.
In a fifth aspect, the present application provides a computer program product comprising instructions which, when run on a user recommendation device, cause the user recommendation device to perform the user recommendation method as described in any one of the possible implementations of the first aspect and the first aspect.
In a sixth aspect, the present application provides a chip comprising a processor and a communication interface, the communication interface and the processor being coupled, the processor being for running a computer program or instructions to implement the user recommendation method as described in any one of the possible implementations of the first aspect and the first aspect.
In particular, the chip provided in the present application further includes a memory for storing a computer program or instructions.
According to the user recommendation method provided by the embodiment of the application, the user recommendation equipment can acquire the indexes of the plurality of dimensions of the user, determine the weights of the indexes, and analyze the user from the plurality of dimensions more comprehensively and accurately to determine the user value of the user. The user recommendation equipment determines the dominant value and the invisible value in the user values based on the indexes of multiple dimensions of the users, not only considers the current value of the users, but also considers the expected value of the users, and is beneficial to the enterprises to more comprehensively and deeply understand the user values, so that the enterprises can provide service strategies which are more in line with each user.
Drawings
Fig. 1 is a schematic diagram of a framework of a user recommendation system according to an embodiment of the present application;
fig. 2 is a schematic hardware structure of a user recommendation device according to an embodiment of the present application;
fig. 3 is a flowchart of a user recommendation method according to an embodiment of the present application;
fig. 4 is a second flowchart of a user recommendation method according to an embodiment of the present application;
fig. 5 is a flowchart illustrating a user recommendation method according to an embodiment of the present application;
Fig. 6 is a flowchart of a user recommendation method according to an embodiment of the present application;
fig. 7 is a flowchart fifth of a user recommendation method provided in the embodiment of the present application;
fig. 8 is a flowchart of a user recommendation method provided in the embodiment of the present application;
fig. 9 is a flowchart of a user recommendation method provided in the embodiment of the present application;
fig. 10 is a schematic structural diagram of a user recommendation device according to an embodiment of the present application.
Detailed Description
The following description of the technical solutions in the embodiments of the present application will be made clearly and completely with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be noted that, in the embodiments of the present application, words such as "exemplary" or "such as" are used to mean serving as an example, instance, or illustration. Any embodiment or design described herein as "exemplary" or "for example" should not be construed as preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "such as" is intended to present related concepts in a concrete fashion.
In order to clearly describe the technical solutions of the embodiments of the present application, in the embodiments of the present application, the terms "first", "second", and the like are used to distinguish the same item or similar items having substantially the same function and effect, and those skilled in the art will understand that the terms "first", "second", and the like are not limited in number and execution order.
Current user recommendation methods generally include the following:
the method comprises the following steps: service channel preferences for different power customers specify more targeted drainage policies. The method predicts the service channel preference of the power customer by collecting the power customer information containing the data of gender, age, region, payment frequency, payment channel and the like and using the XGBoost model.
However, this approach is mainly applicable to power customers, where the target customer has limitations and the collected customer information is of a small dimension, resulting in a potentially inaccurate final prediction result.
The second method is as follows: and fully mining user-product interaction information and user-product behavior information by utilizing a graph calculation method of dynamic relation embedding and relation path exploration, predicting the dynamic relation between a user and a product, constructing an intelligent marketing recommendation model, and accurately recommending marketing business products for the user.
However, the method is extremely time-consuming in the association rule mining part of the core step, complexity is greatly increased along with the increase of data items, large-scale actual production requirements are not met, user behavior attribute characteristics are changed irregularly, and model accuracy is affected.
And a third method: screening according to marketing requirements of merchants in corresponding products to obtain top-mounted products, and screening according to product attributes in corresponding products to obtain exclusive products; displaying exclusive products, common products, top-mounted products, recommended products and other products in a home page; the client is not taken as a unique main body, the demands of the client are considered, personalized product recommendation is performed by combining marketing demands of merchants, the client and the server are convenient to cooperatively operate, and the experience of both parties is improved.
However, the method simply recommends products according to the intention demands of customers, and the factors of the customer demands are few.
By combining the analysis, the prior art has no accurate system grouping of the customer value, single dimension of evaluation when the customer is evaluated, no preference of the user is analyzed by combining the industry characteristics, and the complexity of the data mining algorithm is higher.
In view of this, the embodiment of the application provides a user recommendation method, where a user recommendation device may obtain indexes of multiple dimensions of a user, determine weights of the indexes, and analyze the user from the multiple dimensions more comprehensively and accurately to determine a user value of the user. The user recommendation equipment determines the dominant value and the invisible value in the user values based on the indexes of multiple dimensions of the users, not only considers the current value of the users, but also considers the expected value of the users, and is beneficial to the enterprises to more comprehensively and deeply understand the user values, so that the enterprises can provide service strategies which are more in line with each user.
Illustratively, the user recommendation method is used for a user recommendation system, and fig. 1 shows a schematic frame diagram of the user recommendation system. In fig. 1, the user recommendation device needs to determine the weight of the secondary index and the index value and weight of the primary index through the index value of the secondary index, and then determine the dominant value score, the recessive value score, the weight of the dominant value and the weight of the recessive value based on the index value and weight of the primary index. The user recommendation device determines the user value and determines the service policy of the user based on the explicit value score, the implicit value score, the weight of the explicit value, and the weight of the implicit value.
Wherein the secondary indicators include, but are not limited to, at least one of: product income, value-added service income, enterprise scale, enterprise belonging industry, enterprise home market, price sensitivity, user grade, satisfaction, network access time, ordered product number, agreement constraint, arrearage number, arrearage amount and enterprise key person information.
The primary metrics include, but are not limited to, at least one of: revenue contribution index, user attribute index, loyalty index, credit index, and growth potential index. Wherein, the income contribution index comprises: product revenue and value added service revenue. The user attribute index includes: enterprise scale, industry to which the enterprise belongs, enterprise home market, price sensitivity. The loyalty index includes: user level, satisfaction, time of network access, number of ordered products, and protocol constraint. The confidence index includes: the number of arrearages and the amount of arrearages. The growth potential index includes: enterprise key person information.
Among the secondary indicators, revenue-contributing indicators, user attribute indicators have explicit value, loyalty indicators, credit indicators, and growth potential indicators have implicit value. The explicit value and implicit value constitute the user value.
In one example, the user recommendation device may be a handheld device with wireless connectivity, or a wireless terminal connected to other processing devices of a wireless modem, or a wired terminal. For example, smart devices such as cell phones, personal computers (personal computer, PCs), desktop computers, tablet computers, notebook computers, netbooks, personal digital assistants (personal digital assistant, PDAs), and the like, which are not limited in this embodiment.
In addition, the user recommendation system described in the embodiments of the present application is for more clearly describing the technical solution of the embodiments of the present application, and does not constitute a limitation on the technical solution provided in the embodiments of the present application, and as a person of ordinary skill in the art can know, with evolution of the network architecture and appearance of the new user recommendation system, the technical solution provided in the embodiments of the present application is applicable to similar technical problems.
In particular, the user recommendation device may employ the constituent structure shown in fig. 2 or include the components shown in fig. 2. Fig. 2 is a schematic hardware structure of a user recommendation device provided in an embodiment of the present application, where the user recommendation device may be a chip or a system on a chip in the user recommendation device 101. As shown in fig. 2, the user recommendation device may include a processor 201 and a communication line 202.
Further, the user recommendation device may further comprise a communication interface 203 and a memory 204. The processor 201, the memory 204, and the communication interface 203 may be connected through a communication line 202.
The processor 201 is a CPU, general-purpose processor, network processor (network processor, NP), digital signal processor (digital signal processing, DSP), microprocessor, microcontroller, programmable logic device (programmable logic device, PLD), or any combination thereof. The processor 201 may also be other devices with processing functions, such as, without limitation, circuits, devices, or software modules.
A communication line 202 for communicating information between the components included in the user recommendation device.
Communication interface 203 for communicating with other devices or other communication networks. The other communication network may be an ethernet, a radio access network (radio access network, RAN), a wireless local area network (wireless local area networks, WLAN), etc. The communication interface 203 may be a module, a circuit, a communication interface, or any device capable of enabling communication.
Memory 204 for storing instructions. Wherein the instructions may be computer programs.
The memory 204 may be, but is not limited to, a read-only memory (ROM) or other type of static storage device capable of storing static information and/or instructions, a random access memory (random access memory, RAM) or other type of dynamic storage device capable of storing information and/or instructions, an EEPROM, a CD-ROM (compact disc read-only memory) or other optical disk storage, an optical disk storage (including compact disk, laser disk, optical disk, digital versatile disk, blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, etc.
It should be noted that the memory 204 may exist separately from the processor 201 or may be integrated with the processor 201. Memory 204 may be used to store instructions or program code or some data, etc. The memory 204 may be located within the user recommendation device or may be located outside the user recommendation device, without limitation. The processor 201 is configured to execute the instructions stored in the memory 204 to implement the user recommendation method provided in the following embodiments of the present application.
In one example, processor 201 may include one or more CPUs, e.g., CPU0 and CPU1.
As an alternative implementation, the user recommendation device comprises a plurality of processors.
As an alternative implementation, the user recommendation device further comprises an output device and an input device. The output device is illustratively a display screen, speaker (spaker) or the like, and the input device is a keyboard, mouse, microphone or joystick or the like.
It should be noted that the user recommendation device may be a desktop, a laptop, a web server, a mobile phone, a tablet, a wireless terminal, an embedded device, a chip system or a device with a similar structure as in fig. 2. Furthermore, the constituent structures shown in fig. 2 do not constitute limitations on the respective apparatuses in fig. 2, and the respective apparatuses in fig. 2 may include more or less than those illustrated, or may combine some of the components, or may be arranged differently, in addition to the components shown in fig. 2.
In the embodiment of the application, the chip system may be formed by a chip, and may also include a chip and other discrete devices.
Further, actions, terms, etc. referred to between embodiments of the present application may be referred to each other without limitation. In the embodiment of the present application, the name of the message or the name of the parameter in the message, etc. interacted between the devices are only an example, and other names may also be adopted in the specific implementation, and are not limited.
The user recommendation method provided in the embodiment of the present application is described below with reference to the user recommendation system shown in fig. 1. In which, the terms and the like related to the embodiments of the present application may refer to each other without limitation. In the embodiment of the present application, the name of the message or the name of the parameter in the message, etc. interacted between the devices are only an example, and other names may also be adopted in the specific implementation, and are not limited. The actions involved in the embodiments of the present application are just an example, and other names may be used in specific implementations, for example: the "included" of the embodiments of the present application may also be replaced by "carried on" or the like.
The following describes the user recommendation method provided in the embodiment of the present application in detail with reference to the accompanying drawings.
The user recommending method provided by the embodiment of the application can be applied to user recommending equipment. As shown in fig. 3, the user recommendation method includes:
s301, the user recommendation device acquires a plurality of first-level indexes of each user in a plurality of users, second-level indexes in each first-level index and index values of each second-level index.
It should be understood that, in order to ensure accuracy of user recommendation, the secondary indexes of the user, including the characteristics of the user, the characteristics of the industry and the characteristics of the product, obtained by the method and the device, analyze the user value from multiple dimensions, so that accuracy of user recommendation can be improved.
The user recommending device obtains the first-level index of each user, and obtains the second-level index in each first-level index, for example, the user recommending device obtains the income contribution index of each user, the product income and the value-added service income in the income contribution index, the user recommending device obtains the user attribute index of each user, the enterprise scale, the enterprise affiliated industry, the enterprise affiliated market and the price sensitivity in the user attribute index, the user recommending device obtains the loyalty index of each user, the user grade, the satisfaction degree, the network access time, the number of ordered products and the protocol constraint degree in the loyalty index, the user recommending device obtains the credit index of each user, the arrearage number and the arrearage amount in the credit index, and the enterprise key person information in the growth potential index.
For a specific understanding of the primary index and the secondary index, reference may be made to the description of the primary index and the secondary index in fig. 1, which is not described in detail in this application.
S302, the user recommendation device determines the weight of each secondary index and the index value and the weight of each primary index based on the index value of each secondary index.
In one possible implementation manner, the user recommendation device may determine the weight of each secondary index by adopting methods such as an entropy weight method, a CRITIC weighting method, a coefficient of variation method, and the like, and then determine the index value and the weight of the primary index to which each secondary index belongs based on the weight of each secondary index and the index value of each secondary index.
S303, the user recommendation equipment determines an explicit value score, an implicit value score and a user value score of each user based on the index value and the weight of each level index.
Wherein the explicit value is the current value of the user values; the implicit value is a desired value among the user values.
In one implementation, the user recommendation device may determine a user value score for the user based on the indicator value and the weight of each level one indicator in the user.
As an example, user value score F for user b b The following equation 1 is satisfied:
wherein W is m Weights representing the mth primary index, X' bm An index value indicating the mth primary index of the user b.
S304, the user recommendation equipment determines a service strategy of each user based on the explicit value score and the implicit value score of each user.
It should be understood that the explicit value score is a quantized data obtained by evaluating the user based on the index value and the weight of the index having the explicit value by the user recommendation apparatus. The implicit value score is a quantized data obtained by evaluating the user based on the index value and the weight of the index with the implicit value of the user by the user recommendation equipment. The index may be a primary index or a secondary index, which is not limited in this application.
In one possible implementation, for important value users, such as users with long online time and high enterprise loyalty, the user recommendation device can provide a higher service level for such users, so as to improve the satisfaction of the users. For potential value users, such as users with stable behaviors and large development space of user values, the user recommendation equipment can provide specific preferential service for the users to improve the consumption of the users, so that the user values are improved. For users with ordinary value, such as users with low consumption and stable habit and low value, the user recommendation equipment can perform ordinary service for the users.
The technical scheme provided by the embodiment at least brings the following beneficial effects: as can be seen from S301-S304, the user recommendation device can obtain indexes of multiple dimensions of the user, determine weights of the indexes, and analyze the user from the multiple dimensions more comprehensively and accurately to determine the user value of the user. The user recommendation equipment determines the dominant value and the invisible value in the user values based on the indexes of multiple dimensions of the users, not only considers the current value of the users, but also considers the expected value of the users, and is beneficial to the enterprises to more comprehensively and deeply understand the user values, so that the enterprises can provide service strategies which are more in line with each user.
In an alternative embodiment, as shown in fig. 4 in conjunction with fig. 3, the process of determining the weight of each secondary index and the index value and the weight of each primary index based on the index value of each secondary index in S302 may be implemented specifically by the following S401-S403:
s401, the user recommendation device determines the weight of each secondary index through a combined weight method based on the index value of each secondary index.
It should be appreciated that to improve the accuracy of the user recommendation device in determining the secondary index weights, the user recommendation device may determine the weight of each secondary index using a combined weighting method. The user recommendation device uses more than two weight methods to determine the weight of the secondary index, and then determines the coefficient corresponding to each weight method, wherein the sum of the coefficients of the weight methods is 1.
S402, the user recommendation device determines an index value of each primary index based on the weight and the index value of the secondary index in each primary index.
As an example, for a primary index revenue contribution index, the primary index includes a secondary index product revenue, a value added service revenue, and an index value c of the primary index revenue contribution index satisfies the following equation 2:
c＝d×W d +e×W e Equation 2
Wherein d represents an index value of product income, e represents an index value of value added service income, W d Representing the income weight of the product, W e And the weight of the income of the value added service is represented.
For example, the index value of the user attribute index may be an index value of an enterprise scale multiplied by a weight of the enterprise scale, an index value of an industry to which the enterprise belongs multiplied by a weight of the industry to which the enterprise belongs, and an index value of price sensitivity multiplied by a weight of price sensitivity.
S403, the user recommendation equipment determines the weight of each level index through a combined weight method based on the index value of each level index.
In one possible implementation, after determining the index value c of the primary index revenue contribution index, the user recommendation device may determine the index values of all the primary indexes by the same method, and then determine the weight of each primary index by using a combined weight method.
The technical scheme provided by the embodiment at least brings the following beneficial effects: as can be seen from S401-S403, the user recommendation device in the present application adopts a combined weight method to determine the weight of the secondary index and the weight of the primary index, and the combined weight method can improve the accuracy of the weight and the objectivity of the weight result, and can improve the accuracy of the dominant value and the implicit value when determining the dominant value and the implicit value of the user based on the primary index later, so as to further more objectively evaluate and classify the user.
In an alternative embodiment, as shown in fig. 5 in conjunction with fig. 4, the above-mentioned step S401 of determining the weight of each secondary index by using the combined weight method based on the index value of each secondary index may be implemented specifically by the following steps S501 to S504:
s501, the user recommendation equipment determines the first weight of each secondary index by using an entropy weight method based on the index value of each secondary index.
As an example, when the user recommendation device determines the first weight of each secondary index by using the entropy weight method, the user recommendation device needs to determine the specific gravity of the a-th secondary index of the user b in all the a-th secondary indexes of the user, that is, the probability P used in the subsequent information entropy calculation ab Probability P ab The following equation 3 is satisfied:
wherein X 'is' ab And an index value representing the second-level index of the item a of the user b, wherein m is a positive integer greater than 1.
As an example, the user recommendation device is based on probability P ab Determining the information entropy e of the second-level index of the a-th item a Information entropy is a measure of the degree of disorder of a system. Information entropy e a The following equation 4 is satisfied:
where k is a constant.
As an example, the user recommendation device determines a first weight W for the a-th secondary indicator a1 . First weight W of a second level index of a user b a1 The following equation 5 is satisfied:
wherein n is a positive integer greater than 1.
S502, the user recommendation equipment determines the second weight of each secondary index by using a CRITIC weighting method based on the index value of each secondary index.
It should be understood that when the user recommendation device determines the second weight of each secondary index by using the CRITIC weighting method, the user recommendation device calculates the information volatility of the secondary index first, and the larger the information volatility of the secondary index is, the stronger the evaluation strength of the secondary index is, and the larger the weight of the secondary index is.
As an example, the information volatility S of the item a secondary index a The following equation 6 is satisfied:
wherein X is ′ ab An index value representing the a-th secondary index of the b-user,the average value of index values representing the second level index of the a-th item of all users.
It should be appreciated that when the user recommendation device determines the information conflict of the secondary indicators, the information conflict of the secondary indicators may be represented by a correlation coefficient, where the correlation coefficient represents a correlation between the secondary indicators. The stronger the correlation between the secondary index and other secondary indexes, the smaller the collision, which indicates that the evaluation content of the secondary index and other secondary indexes is repeated, the weaker the evaluation strength of the secondary index is, and the smaller the weight of the secondary index is.
As an example, the correlation matrix R of the secondary index satisfies the following formula 7:
wherein X 'is' kb An index value representing the kth secondary index of the b-user,the average of index values representing the kth secondary index of all users.
As an example, information conflict R of the a-th secondary index a The following equation 8 is satisfied:
wherein r is ab And the correlation coefficient of the second-level index of the a-th item and the second-level index of the b-th item is represented.
As an example, the second weight W of the a-th secondary index a2 The following equation 9 is satisfied:
s503, the user recommendation device determines a third weight of each secondary index by using a coefficient of variation method based on the index value of each secondary index.
In one possible implementation, when the user recommendation device determines the third weight of each secondary index by using the coefficient of variation method, the user recommendation device determines the coefficient of variation of each secondary index first. The user recommendation device then determines a third weight of the secondary indicators based on a sum of the coefficients of variation of the plurality of secondary indicators.
As an example, the average of index values of the user b item a secondary indexThe following common formulas are satisfied
Formula 10:
wherein X 'is' ab An index value indicating the a-th secondary index of the user b.
As an example, the standard deviation S of the user b item a secondary index a The following equation 11 is satisfied:
as an example, the coefficient of variation V of the user b item a secondary index a The following equation 12 is satisfied:
as an example, the third weight W of the user b item a secondary index a3 The following equation 13 is satisfied:
wherein m is a positive integer greater than 1.
S504, the user recommendation equipment determines coefficients of the first weight, the second weight and the third weight, and sums the first weight, the second weight and the third weight based on the coefficients to obtain the weight of each secondary index.
As an example, the weight W of the user b item a secondary index a The following equation 14 is satisfied:
wherein W is a1 First weight, W, representing second level index of item a a2 Second weight representing second level index of item aHeavy, W a3 And a third weight representing the second level index of item a, and β is a coefficient of the weight.
Illustratively, if the user recommendation device determines β to be 0.5, then the coefficients of the first weight, the second weight, and the third weight may be determined to be 0.5, 0.25, and 0.25, respectively, and the weight of the second level index of item a may be determined.
The technical scheme provided by the embodiment at least brings the following beneficial effects: as can be seen from S501-S504, the user recommendation device in the present application combines the three methods, namely, the entropy weighting method, the CRITIC weighting method and the coefficient of variation method, to determine the weight of the secondary index, and by combining these three methods, the characteristics and interrelationships of the secondary index can be more comprehensively considered, so that more accurate and comprehensive weights are obtained, and the result of the weights is more objective.
In an alternative embodiment, as shown in fig. 6 in conjunction with fig. 3, the process of determining the explicit value score, the implicit value score, and the user value score of each user based on the index value and the weight of each level of index in S303 may be implemented specifically by the following S601-S603:
s601, the user recommendation equipment determines an explicit value score and a weight of the explicit value based on the index value and the weight of each level of index in the explicit value.
In one possible implementation, the revenue contribution index and the user attribute index in the first-level index have explicit values, the user recommendation device multiplies the index value of the revenue contribution index by the weight of the revenue contribution index plus the index value of the user attribute index by the weight of the user attribute index to obtain the explicit value score of the user, and then uses a combination weight method to determine the weight of the explicit value.
S602, the user recommendation equipment determines a implicit value score and a weight of the implicit value based on the index value and the weight of each level of index in the implicit value.
In one possible implementation, the loyalty index, the credit index and the growth potential index in the first-level index have implicit values, the user recommending device multiplies the weight of the loyalty index by the index value of the loyalty index, adds the weight of the credit index by the index value of the credit index, adds the weight of the growth potential index by the index value of the growth potential index, obtains the implicit value score of the user, and then uses a combination weight method to determine the weight of the implicit value.
S603, the user recommendation equipment determines the user value score of the user through weighted summation based on the explicit value score, the implicit value score, the weight of the explicit value and the weight of the implicit value.
In one possible implementation, the user recommendation device multiplies the explicit value score by the weight of the explicit value, and adds the implicit value score by the weight of the implicit value to obtain the user value score of the user.
The technical scheme provided by the embodiment at least brings the following beneficial effects: as can be seen from S601-S603, the user recommendation device in the present application evaluates the user value of the user through multiple dimensions, comprehensively considers the index value and the weight of the first-level index, the explicit value score, the implicit value score, and the weight of the explicit value and the implicit value, and can provide a more comprehensive and accurate method for determining the user value, so that the user recommendation device is more accurate in providing services for the user in the following steps.
In an alternative embodiment, as shown in fig. 7 in connection with fig. 3, the process of determining the service policy of each user based on the explicit value score and the implicit value score of each user in S304 may be implemented specifically by the following S701-S702:
S701, classifying the users by the user recommendation equipment based on the explicit value score and the implicit value score of the users to obtain classification results of the users.
In one possible implementation manner, the method and the device can classify the users based on the dominant value score and the implicit value score of the users, divide the users with high dominant value and high implicit value into A-class high-value users, divide the users with medium dominant value and high implicit value into B-class potential users, divide the users with low dominant value and high implicit value into C-class sleep users, divide the users with high dominant value and low implicit value into D-model users, and divide the users with low dominant value and low implicit value into E-class low-value users. The user value classification result table is shown in table 1:
TABLE 1 user value Classification results Table
| User classification | Dominant value | Implicit value |
| Class a-high value user | High height | High height |
| Class B-potential user | In (a) | High height |
| Class C-sleeping user | Low and low | High height |
| Class D-model user | High height | Low and low |
| Class E-low value user | Low and low | Low and low |
S702, the user recommending device determines a service strategy of each user based on the classification result of the user.
It should be understood that the enterprise should offer more service to the high-value user, and for the low-value user, the implicit value of the low-value user should also be considered, and if the low-value user has a higher implicit value, the enterprise may select a certain service to cultivate the product habit of the user, so that the implicit value of the user appears.
The technical scheme provided by the embodiment at least brings the following beneficial effects: as can be seen from S701-S702, the user recommendation device in the application can classify the user more carefully and accurately by comprehensively considering the explicit and implicit values of the user. Therefore, the enterprise is helped to better understand the demands and the behavior modes of the users, specific service recommendation is carried out for the users, more reliable basis is provided for service recommendation, the resource utilization efficiency of the enterprise is helped to be improved, and the operation benefit of the enterprise or the platform is improved.
In an alternative embodiment, referring to fig. 7, as shown in fig. 8, the process of classifying the users based on the explicit value score and the implicit value score of the users in S701, to obtain the classification result of the users may be specifically implemented by the following S801 to S804:
s801, the user recommendation equipment classifies the users by using k-means clustering based on the dominant value score and the implicit value score of the users to obtain a first type of the users.
It should be appreciated that the user may be categorized using integrated decisions based on the explicit and implicit value scores of the user. The integrated decision is made by adopting a plurality of classifiers, and the accuracy is higher than that of a single classifier, so that the user recommendation equipment can classify the user by adopting a plurality of classification methods, and then the integrated decision is made on the classification result, namely when the number of classification results with the same type of the user is greater than the preset number, the classification result with the same type is determined as the classification result of the user.
In one possible implementation, the user recommendation device refers to a two-dimensional array of explicit and implicit value scores of a user as user value data. The user recommendation device selects at least one user value data from the plurality of user value data as a first cluster center point. The user recommendation device determines first similarity between each user value data in the plurality of user value data and a first clustering center point, and divides target user value data with the first similarity being greater than preset similarity into the same first cluster to obtain a plurality of first clusters.
The user recommendation device determines the average value of the target user value data in each first cluster as a second cluster center point. And the user recommending device divides the target user value data with the second similarity larger than the preset similarity into the same second cluster according to the second similarity between each user value data and the second aggregation center point to obtain a plurality of second clusters.
And the user recommendation equipment determines the average value of the target user value data in each second cluster as a third class center point. When the third cluster center point is the same as the second cluster center point, the user recommendation equipment determines the second cluster center point or the third cluster center point as a target cluster center point, and the user recommendation equipment determines a cluster corresponding to the target cluster center point as a target cluster of each target user value in the clusters. The user recommendation device determines the target clusters as the first type of the user corresponding to each target user value data.
In one possible implementation manner, the user recommendation device classifies the users by using k-means clusters, and when the first type of the users is obtained, the first type of the users can be evaluated by combining two evaluation indexes of CH indexes and profile coefficients.
In one possible implementation, the user recommendation device measures the closeness of the data within the clusters by calculating the sum of the squares of the distances of the points in each cluster from the center point of the cluster to which the point belongs. The user recommendation device measures the degree of separation between clusters by calculating the sum of squares of the distances of the center points of the respective clusters and the center points of the dataset. And obtaining the CH index based on the ratio of the separation degree to the compactness. Therefore, the larger the CH index is, the more compact the class is, the more the classes are dispersed, and the clustering result is better.
In one possible implementation, the user recommendation device evaluates the clustering results by analyzing both factors of the degree of aggregation and the degree of separation. The closer the point distances of the same category are, the more the point distances of different categories are, the higher the profile coefficient is determined by the user recommendation device, and the clustering effect is better.
S802, the user recommendation equipment classifies the users by using a support vector machine model based on the explicit value score and the implicit value score of the users to obtain a second type of the users.
As an example, when the user recommendation device classifies the user by using the support vector machine model to obtain the second type of the user, an objective function f (x) needs to be constructed based on the explicit value score and the implicit value score of the user, and the weight of the explicit value and the weight of the implicit value. The objective function f (x) satisfies the following equation 14:
/>
wherein,is a nonlinear function, ω is a weight coefficient matrix, and b is a bias coefficient.
As an example, we show overcoming the overfitting problem, completing the strict classification, improving the generalization ability of the model, and the user recommendation device transforms the problem to represent a quadratic programming problem, satisfying the following equation 15:
wherein Q represents an optimization target, W represents a weight coefficient, C represents a regularization constant, and beta 1 、β 2 Representing the relaxation variable.
As an example, the user recommendation device transforms the quadratic programming problem into a dual form through a lagrangian function, maps the data to a high dimension using a kernel function to achieve an optimal segmentation hyperplane, and the objective function f (x) satisfies the following equation 16:
k(x i ,x j )＝exp(-c|x i -x j | 2 ) Equation 16
Wherein alpha is i ，Represents the Lagrangian factor, K (x i ,x j ) Representing a gaussian kernel function. And the user recommending equipment classifies the users by using the model to obtain a second type of the users.
S803, the user recommendation equipment classifies the users by using the long-term and short-term memory network model based on the dominant value score and the implicit value score of the users to obtain a third type of the users.
In one possible implementation, the user recommendation device classifies the user using the long-short term memory network model to obtain a third type of user. The long-term and short-term memory model consists of three gating units: forget gate, input gate and output gate. At time t, two transmission states exist in the long-short-term memory model network: cell state matrix C t The transmission is the selective memory of history information before the moment t, and the state matrix h is hidden t Stored is the overall information at the current time t. The long-term memory model has four stages when working at the time t: forget phase, store phase, update phase and output phase.
As an example, in the forget phase, a forget gateCan control the last state C t Information retention ratio, forget gate->The following equation 17 is satisfied:
wherein W is f Is the overall information h at time t-1 t-1 And input x t Weight matrix at forgetting gate, b f Representing intercept, σ represents a sigmoid function, mapping the output to [0-1 ]]And the selection ratio of the forgotten portion is shown.
As an example, in the storage phase, the user recommendation device needs to determine the information saved in the cell state, first output the gateInformation requiring to be updated is determined-> The following equation 18 is satisfied:
wherein W is i Represents h t-1 And x t Weight matrix at input gate, b i Representing the intercept.
As an example, the user recommendation device may create a substantive information vector z through a tanh layer t ，z t The following formula 19 is satisfied:
z t ＝tanh(W C ·[h t-1 ,x t ]+b C ) Equation 19
Wherein W is C Represents h t-1 And x t And z t Corresponding weight matrix, b C Representing the intercept, tanh represents the hyperbolic tangent function, and the activation function is represented in the long-short-term memory model.
As an example, during the update phase, the user recommendation device will C t-1 The information input at time t is added to the information stored, thereby updating C t ，C t The following equation 20 is satisfied:
wherein,representing Hadamard operations, C t-1 Substantial information indicating time t-1, < +.>And z t Information representing the actual input is combined.
As an example, in the output phase, it is the output gateThe following equation 21 is satisfied:
wherein W is O Represents h t-1 And x t The weight matrix at the output gate, b O Representing the intercept.
As one example, the user recommendation device willParenchymal information C selectively memorized with the current time t Combining to obtain the output h of the current state t Output h of the current state t The following equation 22 is satisfied:
wherein the user recommending device uses the tanh function to C t Is scaled to obtain output y t ，y t The following equation 23 is satisfied:
y t ＝W 2 h t +b 2 equation 23
Wherein W is 2 And y is t Corresponding weight matrix, b 2 Representing the intercept. And the user recommending equipment classifies the users by using the model to obtain a third type of the users.
S804, if at least two types in the three types are consistent, the user recommending device determines the classification result with the consistent type as the classification result of the user.
In one possible implementation manner, after obtaining the first type, the second type and the third type of the user, the user recommending device may be based on the idea of voting in the ensemble learning, and if and only if at least two types of the three types are consistent, the user recommending device determines the classification result with the consistent type as the classification result of the user.
The technical scheme provided by the embodiment at least brings the following beneficial effects: as can be seen from S801-S804, the user recommendation device classifies users by using k-means clustering, a support vector machine model and a long-term and short-term memory network model, and can verify the effectiveness and accuracy of user classification from different angles and models. In this way, the reliability and stability of the classification result can be increased. The users are classified through multiple models, so that misjudgment risks brought by a single model can be reduced, and the robustness of the user classification model is improved.
In an alternative embodiment, as shown in fig. 9 in conjunction with fig. 3, the process of obtaining the index value of each secondary index in S301 may be implemented specifically by the following S901-S902:
s901, the user recommending device acquires initial index values of each secondary index.
S902, the user recommendation equipment performs data preprocessing and standardization processing on the initial index value of each secondary index to obtain an index value of each secondary index.
Wherein the preprocessing comprises the following steps: and (5) processing the missing value and the abnormal value. The normalization processing comprises normalization processing of positive indexes and negative indexes. The index value of the forward index is positively correlated with the user value; the index value of the negative index is inversely related to the user value.
In one possible implementation, the user recommendation device obtains an initial index value of each secondary index, and performs missing value processing on the initial index value. The user recommending device performs the missing value processing on the initial index value by the following two methods: the user recommending device deletes the information of the user with the missing value in the initial index value; the user recommendation device fills in the missing values with constants, such as filling in the missing values with 0, the average value of the initial index values, mode, median, etc.
It should be understood that outliers are values that are relatively most of the initial index values that are significantly inconsistent, and that can be identified using the 3sigma principle when identifying outliers. The 3sigma principle is that assuming that a set of initial index values only contains random errors, the user recommendation device needs to calculate the initial index values to obtain standard deviation. The user recommending device determines a section according to the preset probability, and for the error exceeding the section, the user recommending device does not belong to random error but coarse error, and eliminates the data containing the coarse error. For example, the interval is set to (μ -3σ, μ+3σ), and values having an error exceeding the interval of (μ -3σ, μ+3σ) are regarded as outliers.
In one possible implementation, the user recommendation device performs preliminary detection on the outliers by using the 3sigma principle, and after the detection, further determines the true outliers based on experience. For the real abnormal value, the user recommending device can delete the abnormal value by using a drop function, and can replace the abnormal value by using a replace function method by using the mean value, the mode, the median and the like of the initial index value.
In one possible implementation, since the units of different indexes are different, it is not possible to directly calculate the matrix of all the secondary index values, and thus it is necessary to perform normalization processing on the initial index values. The user recommending device continues to perform standardization processing on the preprocessed data, and the maximum and minimum values can be adopted for standardization.
As an example, in normalization, for a secondary index X 'that is positively correlated with user value' ab If the product income, the following formula 24 is satisfied:
wherein X is ab An initial index value, min { X { representing a secondary index of each user b The minimum value of the secondary index among a plurality of users, max { X } is represented by b And represents the maximum value of the secondary index among the plurality of users.
As an example, in normalization, for a secondary index X 'that is inversely related to user value' ab If the arrearage number, the following formula 25 is satisfied:
wherein X is ab An initial index value, min { X { representing a secondary index of each user b The minimum value of the secondary index among a plurality of users, max { X } is represented by b And represents the maximum value of the secondary index among the plurality of users.
The technical scheme provided by the embodiment at least brings the following beneficial effects: as can be seen from S901-S902, the user recommendation device can process the missing value and the abnormal value in the initial value through data preprocessing, and the user recommendation device performs standardized processing on the data, so that errors caused by different units of the secondary index can be eliminated, and further the accuracy of determining the user value subsequently can be improved.
The foregoing description of the solution provided in the embodiments of the present application has been mainly presented in terms of a method. To achieve the above functions, it includes corresponding hardware structures and/or software modules that perform the respective functions. Those of skill in the art will readily appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be implemented as hardware or combinations of hardware and computer software. Whether a function is implemented as hardware or computer software driven hardware depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
According to the embodiment of the application, the function modules of the user recommendation device can be divided according to the method example, for example, each function module can be divided corresponding to each function, and two or more functions can be integrated in one processing module. The integrated modules may be implemented in hardware or in software functional modules. Optionally, the division of the modules in the embodiments of the present application is schematic, which is merely a logic function division, and other division manners may be actually implemented.
Fig. 10 is a schematic structural diagram of a user recommendation device according to an embodiment of the present application. The user recommendation device may be used to perform the method of data processing shown in fig. 3 to 9. The user recommendation device 100 shown in fig. 10 includes: a communication unit 1001 and a processing unit 1002.
A communication unit 1001 configured to acquire a plurality of first-level indexes of each of a plurality of users, second-level indexes in each of the first-level indexes, and index values of each of the second-level indexes;
a processing unit 1002, configured to determine a weight of each secondary index and an index value and a weight of each primary index based on the index value of each secondary index;
The processing unit 1002 is further configured to determine an explicit value score, an implicit value score, and a user value score of each user based on the index value and the weight of each level one index; dominant value is the current value in the user's value; the implicit value is a desired value among the user values;
the processing unit 1002 is further configured to determine a service policy of each user based on the explicit value score and the implicit value score of each user.
In one possible implementation, the processing unit 1002 is specifically configured to: determining the weight of each secondary index by a combined weight method based on the index value of each secondary index; determining an index value of each primary index based on the weight and the index value of the secondary index in each primary index; the weight of each level index is determined by a combined weight method based on the index value of each level index.
In one possible implementation, the processing unit 1002 is specifically configured to: determining a first weight of each secondary index by using an entropy weight method based on the index value of each secondary index; determining a second weight of each secondary index by using a CRITIC weighting method based on the index value of each secondary index; determining a third weight of each secondary index by using a coefficient of variation method based on the index value of each secondary index; determining coefficients of the first weight, the second weight and the third weight, and summing the first weight, the second weight and the third weight based on the coefficients to obtain the weight of each secondary index; the sum of the coefficients of the first weight, the second weight and the third weight is 1.
In one possible implementation, the processing unit 1002 is specifically configured to: determining an explicit value score and a weight of the explicit value based on the index value and the weight of each level of index in the explicit value; determining a implicit value score and a weight of the implicit value based on the index value and the weight of each level of index in the implicit value; the user value score of the user is determined by weighted summation based on the explicit value score, the implicit value score, the weight of the explicit value, and the weight of the implicit value.
In one possible implementation, the processing unit 1002 is specifically configured to: classifying the users based on the explicit value scores and the implicit value scores of the users to obtain classification results of the users; based on the classification result of the users, the service policy of each user is determined.
In one possible implementation, the processing unit 1002 is specifically configured to: classifying the users by using k-means clusters based on the dominant value score and the recessive value score of the users to obtain a first type of the users; classifying the users by using a support vector machine model based on the explicit value score and the implicit value score of the users to obtain a second type of the users; classifying the users by using the long-short-term memory network model based on the explicit value score and the implicit value score of the users to obtain a third type of the users; if at least two types in the three types are consistent, determining the classification result with the consistent type as the classification result of the user.
In one possible implementation, the communication unit 1001 is specifically configured to: acquiring an initial index value of each secondary index; the processing unit 1002 is specifically configured to: carrying out data preprocessing and standardization processing on the initial index value of each secondary index to obtain an index value of each secondary index; the pretreatment comprises the following steps: processing the missing value and the abnormal value; the normalization processing comprises the normalization processing of positive indexes and negative indexes; the index value of the forward index is positively correlated with the user value; the index value of the negative index is inversely related to the user value.
The present application also provides a computer-readable storage medium, where the computer-readable storage medium includes computer-executable instructions that, when executed on a computer, cause the computer to perform the user recommendation method provided in the above embodiments.
The embodiment of the application also provides a computer program which can be directly loaded into a memory and contains software codes, and the computer program can realize the user recommendation method provided by the embodiment after being loaded and executed by a computer.
Those skilled in the art will appreciate that in one or more of the examples described above, the functions described in the present invention may be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, these functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer-readable storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.
From the foregoing description of the embodiments, it will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of functional modules is illustrated, and in practical application, the above-described functional allocation may be implemented by different functional modules according to needs, i.e. the internal structure of the apparatus is divided into different functional modules to implement all or part of the functions described above.
In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and, for example, the division of modules or units is merely a logical function division, and other manners of division may be implemented in practice. For example, multiple units or components may be combined or may be integrated into another device, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form. The units described as separate parts may or may not be physically separate, and the parts shown as units may be one physical unit or a plurality of physical units, may be located in one place, or may be distributed in a plurality of different places. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units. The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a readable storage medium. Based on such understanding, the technical solution of the embodiments of the present application may be essentially or a part contributing to the general technology or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium, where the software product includes several instructions to cause a device (may be a single-chip microcomputer, a chip or the like) or a processor (processor) to execute all or part of the steps of the methods of the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk, etc.
The present invention is not limited to the above embodiments, and any changes or substitutions that can be easily understood by those skilled in the art within the technical scope of the present invention are intended to be included in the scope of the present invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.
Claims (10)
1. A method of user recommendation, the method comprising:
acquiring a plurality of first-level indexes of each user in a plurality of users, a second-level index in each first-level index and an index value of each second-level index;
determining the weight of each secondary index and the index value and the weight of each primary index based on the index value of each secondary index;
determining an explicit value score, an implicit value score and a user value score of each user based on the index value and the weight of each primary index; dominant value is the current value in the user's value; the implicit value is a desired value of the user values;
and determining the service strategy of each user based on the explicit value score and the implicit value score of each user.
2. The method of claim 1, wherein the determining the weight of each secondary indicator and the index value and weight of each primary indicator based on the index value of each secondary indicator comprises:
determining the weight of each secondary index by a combined weight method based on the index value of each secondary index;
Determining an index value of each primary index based on the weight and the index value of the secondary index in each primary index;
and determining the weight of each primary index by a combined weight method based on the index value of each primary index.
3. The method of claim 2, wherein determining the weight of each secondary index by a combined weight method based on the index value of each secondary index comprises:
determining a first weight of each secondary index by utilizing an entropy weight method based on the index value of each secondary index;
determining a second weight of each secondary index by using a CRITIC weighting method based on the index value of each secondary index;
determining a third weight of each secondary index by using a coefficient of variation method based on the index value of each secondary index;
determining coefficients of the first weight, the second weight and the third weight, and summing the first weight, the second weight and the third weight based on the coefficients to obtain the weight of each secondary index; and the sum of coefficients of the first weight, the second weight and the third weight is 1.
4. The method of claim 1, wherein determining an explicit value score, an implicit value score, and a user value score for each of the users based on the index value and the weight of each of the primary indexes comprises:
Determining an explicit value score and a weight of the explicit value based on the index value and the weight of each of the primary indexes in the explicit value;
determining a implicit value score and a weight of the implicit value based on an index value and the weight of each primary index in the implicit value;
determining a user value score for the user by weighted summation based on the explicit value score, the implicit value score, the weight of the explicit value, and the weight of the implicit value.
5. The method of claim 1, wherein the determining a service policy for each of the users based on the explicit value score and the implicit value score for each of the users comprises:
classifying the users based on the explicit value score and the implicit value score of the users to obtain classification results of the users;
and determining the service strategy of each user based on the classification result of the user.
6. The method of claim 5, wherein classifying the user based on the explicit value score and the implicit value score of the user to obtain the classification result of the user comprises:
Classifying the users by using k-means clusters based on the dominant value score and the recessive value score of the users to obtain a first type of the users;
classifying the users by using a support vector machine model based on the explicit value score and the implicit value score of the users to obtain a second type of the users;
classifying the users by using a long-short-term memory network model based on the explicit value score and the implicit value score of the users to obtain a third type of the users;
if at least two types in the three types are consistent, determining the classification result with the consistent types as the classification result of the user.
7. The method according to any one of claims 1 to 6, wherein the obtaining an index value of each of the secondary indexes includes:
acquiring an initial index value of each secondary index;
carrying out data preprocessing and standardization processing on the initial index value of each secondary index to obtain an index value of each secondary index; the pretreatment comprises the following steps: processing the missing value and the abnormal value; the normalization processing comprises the normalization processing of positive indexes and negative indexes; the index value of the forward index is positively correlated with the user value; the index value of the negative index is inversely related to the user value.
8. A user recommendation device, the device comprising: a communication unit and a processing unit;
the communication unit is used for acquiring a plurality of first-level indexes of each user in a plurality of users, second-level indexes in each first-level index and index values of each second-level index;
the processing unit is used for determining the weight of each secondary index and the index value and the weight of each primary index based on the index value of each secondary index;
the processing unit is further used for determining an explicit value score, an implicit value score and a user value score of each user based on the index value and the weight of each primary index; dominant value is the current value in the user's value; the implicit value is a desired value of the user values;
the processing unit is further configured to determine a service policy of each user based on the explicit value score and the implicit value score of each user.
9. A user recommendation device, comprising: a processor and a communication interface; the communication interface is coupled to the processor for running a computer program or instructions to implement the user recommendation method as claimed in any one of claims 1 to 7.
10. A computer readable storage medium having instructions stored therein, characterized in that when executed by a computer, the computer performs the user recommendation method as claimed in any one of the preceding claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410059592.8A CN117893258A (en) | 2024-01-15 | 2024-01-15 | User recommendation method, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410059592.8A CN117893258A (en) | 2024-01-15 | 2024-01-15 | User recommendation method, device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117893258A true CN117893258A (en) | 2024-04-16 |
Family
ID=90650620
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410059592.8A Pending CN117893258A (en) | 2024-01-15 | 2024-01-15 | User recommendation method, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117893258A (en) |
-
2024
- 2024-01-15 CN CN202410059592.8A patent/CN117893258A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11188935B2 (en) | Analyzing consumer behavior based on location visitation | |
| Chen et al. | Selecting critical features for data classification based on machine learning methods | |
| US11238473B2 (en) | Inferring consumer affinities based on shopping behaviors with unsupervised machine learning models | |
| AU2022204241A1 (en) | Machine learning classification and prediction system | |
| Lingras et al. | Rough cluster quality index based on decision theory | |
| US11610085B2 (en) | Prototype-based machine learning reasoning interpretation | |
| US9536201B2 (en) | Identifying associations in data and performing data analysis using a normalized highest mutual information score | |
| WO2018200996A1 (en) | Method and system of managing item assortment based on demand transfer | |
| WO2021135562A1 (en) | Feature validity evaluation method and apparatus, and electronic device and storage medium | |
| Jain et al. | A novel scheme for employee churn problem using multi-attribute decision making approach and machine learning | |
| US20150248630A1 (en) | Space planning and optimization | |
| US11295328B2 (en) | Intelligent prospect assessment | |
| US11182841B2 (en) | Prospect recommendation | |
| CN110532429B (en) | Online user group classification method and device based on clustering and association rules | |
| CN110909222A (en) | User portrait establishing method, device, medium and electronic equipment based on clustering | |
| WO2020150611A1 (en) | Systems and methods for entity performance and risk scoring | |
| Chen et al. | A context-aware recommendation approach based on feature selection | |
| CN113742492A (en) | Insurance scheme generation method and device, electronic equipment and storage medium | |
| Ahmed et al. | An enhanced ensemble classifier for telecom churn prediction using cost based uplift modelling | |
| CN116318989A (en) | System, method and computer program product for user network activity anomaly detection | |
| CN115983900A (en) | Method, apparatus, device, medium, and program product for constructing user marketing strategy | |
| CN109937421A (en) | For predicting two category classification methods of specific project generic and utilizing the calculating equipment of this method | |
| Urkup et al. | Customer mobility signatures and financial indicators as predictors in product recommendation | |
| Lewaaelhamd | Customer segmentation using machine learning model: an application of RFM analysis | |
| US11977565B2 (en) | Automated data set enrichment, analysis, and visualization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |