Retail platform user portrait generation method and system based on cloud computing

Abstract

The invention discloses a retail platform user portrait generation method and system based on cloud computing, relates to the technical field of retail platforms, and solves the problems that the accuracy of user portraits is not high and the use experience of target users is affected due to the fact that historical browsing information is not screened in detail in sequence. According to the method, the historical browsing information of the target user is locked according to confirmation of the target user, then based on specific characteristics of the historical browsing information, preliminary screening is performed firstly, then secondary screening is performed, screening modes of each time are inconsistent, finally reserved real and effective information is determined, and then according to specific parameters in the real and effective information, the real and effective information is obtained. In order to enable the generated user portraits to be more accurate, the ranking of the corresponding categories is determined by adopting an assignment summation processing mode, and then the user portraits needing to be pushed are generated, so that the processing mode is more accurate, the efficiency is higher, and the accuracy is higher.

Description

Technical Field

[0001] The present invention relates to the technical field of retail platforms, and in particular to a method and system for generating user portraits of retail platforms based on cloud computing. Background Art

[0002] Retail platforms refer to e-commerce platforms used to sell products or services online. These platforms provide a convenient way for consumers to browse and purchase goods, while providing sellers with an online sales channel. Common retail platforms include Amazon, Taobao, JD.com, etc. They provide consumers and sellers with transaction, payment, logistics and other services, greatly facilitating e-commerce transactions; Patent publication number CN114943573A discloses a construction method, system, storage medium and electronic device based on user portrait of retail SaaS platform, obtains product information and user information, profiles user attribute behavior, establishes user feature matrix, establishes product feature matrix, calculates product association and user interest, generates a first recommendation list, calculates the similarity between the feature value in the product feature matrix of each product with user interest ranking within the threshold range and the feature value in the user feature matrix of the current user, sorts the products according to the similarity, and selects products with similarity meeting the threshold for display. The present invention performs multi-layer convolution operation on product association and then calculates user interest, closely associates product association with user interest, and improves the accuracy of user interest product search.

[0003] When generating user portraits for corresponding users, retail platforms generally confirm the user habits of corresponding users based on the usage records of corresponding users, and then generate user portraits for corresponding users. However, this portrait generation method does not conduct detailed screening of historical browsing information in sequence, resulting in low accuracy of user portraits, which affects the user experience of target users. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a method and system for generating user portraits for a retail platform based on cloud computing, which solves the problem that the historical browsing information is not screened in detail in sequence, resulting in low accuracy of user portraits and affecting the user experience of target users.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A retail platform user portrait generation system based on cloud computing, comprising: The target user confirmation end determines the target user from the specific purchase amount of the corresponding user in the retail platform; The browsing information acquisition terminal directly extracts the historical browsing information of the target user from the retail platform based on the confirmed target user, which also includes the specific browsing time of the historical browsing information; The information screening end performs a preliminary screening of the historical browsing information generated by the target user, removes invalid historical browsing information, and retains valid historical browsing information. The specific method is as follows: Average the browsing time of several groups of different historical browsing information, confirm that the average browsing time of several groups of historical browsing information is JS, and generate the browsing time interval of this target user [JS-X1, JS+X1], where X1 is ⅓ of JS; Based on the browsing time of each group of browsing information, the browsing information belonging to this browsing time interval [JS-X1, JS+X1] is marked as valid historical browsing information, otherwise no marking is performed; The information analysis terminal re-analyzes the valid historical browsing information and determines whether the corresponding browsing information is real valid information or assumed valid information by confirming the browsing speed. The specific method is as follows: Obtain the browsing speed of each set of valid historical browsing information from the retail platform, and mark each set of browsing speed as SS i , where i represents different valid historical browsing information; Set several groups of browsing speed SS i Perform clustering processing, sort several groups of browsing speeds from small to large, determine the middle value at the sorting position, and mark it as the center value. If there are two groups of middle values, randomly select one group of values; Perform difference processing on different browsing speeds and the central value, then perform mean processing on several groups of differences to determine a group of means, mark the valid historical browsing information with a difference greater than the mean as assumed valid information, otherwise, mark it as real valid information, and transmit the processed groups of real valid information to the assignment processing end; The assignment processing end confirms the total browsing time and click count of the corresponding information based on the real and effective information of the corresponding target user, and then locks the information feature value according to the confirmed two sets of parameters, and then checks the information feature value with the preset interval to confirm the corresponding assignment of the real and effective information. The processing method is: The total browsing time of different real and effective information is marked as SC k , mark the number of clicks as DC k , where k represents different real and effective information; Using TZ k =SC k ×C1+DC k ×C2 determines the corresponding information characteristic value TZ k , where C1 and C2 are preset fixed coefficient factors, and the determined information characteristic value TZ kCompare with the preset intervals respectively, wherein the preset intervals include three groups, namely, interval 0 [Y1, Y2], interval 1 (Y2, Y3], interval 2 (Y3, Y4], wherein Y1, Y2, Y3 and Y4 are all preset interval end values; Confirm information characteristic value TZ k To which group of intervals it belongs, the corresponding interval naming value is directly selected to assign the real and valid information, and the real and valid information after the assignment is completed is transmitted to the portrait generation end. If the interval is not found, no assignment is performed; The portrait generation end receives the real and valid information after the value assignment, analyzes and compares it with the sales categories of the retail platform, confirms the specific attribution of the real and valid information, and then generates the user portrait of the target user based on the value assignment. The specific process is as follows: Determine the multiple major categories divided by the retail platform, and then determine the group of major categories with the largest number of subordinate categories based on the subordinate categories of real and effective information; Then, determine several corresponding subcategories in this large category, and then confirm the subcategory based on the specific IP ownership of real and effective information; The values ​​of several real and effective information in each different small category are summed up, and the specific value obtained by the summation is calibrated as the ranking value of the corresponding small category; Based on the numerical values ​​of the sorting values, several small categories are sorted from large to small, and the top three small categories are extracted and three groups of user portraits are generated. The three groups of user portraits generated are transmitted to the push center.

[0006] Preferably, the push center includes a time period parameter acquisition unit, a parameter analysis unit and a push unit; The time period parameter acquisition unit acquires different time periods when the target user uses the retail platform, and transmits the acquired different time periods to the parameter analysis unit; The parameter analysis unit performs a comprehensive ratio analysis on the different time periods obtained and the preset pushable time periods, selects the maximum ratio, and thus determines the best push time period, and the determination method is: According to the rest time in the morning, noon and evening, three groups of push time periods are determined. The push time periods are all planned in advance and operated by the operators themselves; Based on the different time periods obtained, the total usage time of the target users corresponding to each group of different pushable time periods is confirmed, and then the usage ratio value is determined, where: usage ratio value = total usage time ÷ total time of pushable time periods; From the three groups of usage ratio values, determine the maximum ratio, and mark the push time period to which the maximum ratio belongs as the optimal push time period; The push unit randomly pushes the three groups of user portraits to the corresponding target user terminals during the optimal push time period and when the target users are using the retail platform.

[0007] Preferably, a method for generating a user portrait of a retail platform based on cloud computing comprises: Determine target users based on the specific purchase volume of registered users; Extract historical browsing information generated by target users from the retail platform and ensure that the historical browsing information meets the standards; Perform a preliminary screening of historical browsing information, remove invalid historical browsing information by confirming the average browsing time, and retain valid historical browsing information; Cluster the viewing speed of valid historical browsing information, remove valid historical browsing information with large clustering deviation, and retain real and valid information; Based on the total browsing time and click count of the real and effective information, determine the information feature value, and based on the information feature value, determine the specific value of the real and effective information; Based on the specific assignment and classification of different real and valid information, the specific ownership of the real and valid information is confirmed and a user portrait of the corresponding target user is generated.

[0008] The present invention provides a method and system for generating user portraits of a retail platform based on cloud computing. Compared with the prior art, the method has the following beneficial effects: The present invention locks the historical browsing information of the target user according to the confirmation of the target user, and then performs preliminary screening based on the specific characteristics of the historical browsing information, and then performs secondary screening. The screening method is different each time, and the final retained real and valid information is determined. Then, according to the specific parameters inside the real and valid information, the assigned value is determined. In order to make the generated user portrait more accurate, the assignment summation processing method is adopted to determine the sorting of the corresponding category, and then the user portrait to be pushed is generated. The processing method is more accurate, more efficient, and more accurate. Confirm the corresponding rest periods of the target users, and then determine the optimal duration of their use based on the specific duration of the corresponding users using the corresponding platform during the corresponding rest periods, thereby determining a time period with the longest usage time. Pushing during this time period will have a better overall push effect and can be pushed to the corresponding target users in a more timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a schematic diagram of the principle framework of the present invention; Figure 2 This is a schematic diagram of the internal principles of the push center of the present invention. DETAILED DESCRIPTION

[0010] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0011] Embodiment 1

[0012] See also Figure 1 , the present application provides a retail platform user portrait generation system based on cloud computing, including a target user confirmation terminal, a browsing information acquisition terminal, an information screening terminal, an information analysis terminal, an assignment processing terminal, a portrait generation terminal and a push center; The target user confirmation end, the browsing information acquisition end, the information screening end, the value assignment processing end, the portrait generation end and the push center are electrically connected in sequence from the output node to the input node, and the information analysis end and the information screening end are bidirectionally connected; The target user confirmation end determines the specific purchase amount of the corresponding user from the retail platform. When the number of purchases of the corresponding user on this retail platform exceeds ten, the corresponding user is marked as the target user, and the marked target user is transmitted to the browsing information acquisition end. Specifically, the users are all users who have completed registration on the retail platform, and the purchase information of the registered users can be directly obtained from the retail platform. The browsing information acquisition end directly extracts the historical browsing information of the target user from the retail platform based on the confirmed target user, including the specific browsing time of the historical browsing information: When the target user's historical browsing information obtained does not exceed 30 groups, the historical browsing information obtained will be transmitted to the information screening end. If it does not exceed 30 groups, monitoring will continue until the historical browsing information obtained exceeds 30 groups and then it will be transmitted. For example: when the historical browsing information obtained is not enough, the user screen analyzed will not be standard enough and the error will be large, so it is necessary to ensure the total amount of information.

[0013] The information screening end performs a preliminary screening of the historical browsing information generated by the target user, removes invalid historical browsing information, and retains valid historical browsing information. The valid historical browsing information is then re-analyzed by the information analysis end, and the browsing speed is confirmed to determine whether the corresponding browsing information is real valid information or assumed valid information. The specific method of preliminary screening of historical browsing information is as follows: The browsing time of several groups of different historical browsing information is averaged, and the average time of several groups of historical browsing information is determined to be JS, and the browsing time interval of this target user is generated [JS-X1, JS+X1], where X1 is a preset value, and X1 is generally ⅓ of JS, and its specific value is determined by the operator based on experience; Based on the browsing time of each group of browsing information, the browsing information belonging to this browsing time interval [JS-X1, JS+X1] is marked as valid historical browsing information, otherwise no marking is performed; The specific method of reanalyzing the valid historical browsing information at the information analysis end is as follows: Obtain the browsing speed of each set of valid historical browsing information from the retail platform, and mark each set of browsing speed as SS i , where i represents different valid historical browsing information; Set several groups of browsing speed SS i Perform clustering processing, sort several groups of browsing speeds from small to large, determine the middle value at the sorting position, and mark it as the center value. If there are two groups of middle values, randomly select one group of values. For example: if the browsing speed is 1, 2, 3, 4, 4, 5, then the middle value is 3 or 4, and the center value is one of 3 or 4. If the browsing speed is 1, 2, 3, 4, 5, then the middle value is 3, and the center value is 3. Perform difference processing on different browsing speeds and the central value, and then perform average processing on several groups of differences to determine a group of means, and mark the valid historical browsing information with a difference greater than the mean as assumed valid information, otherwise, mark it as real valid information, and transmit the processed groups of real valid information to the assignment processing end. Specifically, the so-called assumed valid information is that when the corresponding interface is opened, the corresponding historical browsing information is generated, but no one views it, and after reaching the specified time, it is closed. This type of historical browsing information should be eliminated, which will affect the subsequent confirmation of the corresponding target user portrait.

[0014] The assignment processing end confirms the total browsing time and click times of the corresponding information based on the real and effective information of the corresponding target user, and then locks the information feature value according to the confirmed two sets of parameters, and then checks the information feature value with the preset interval to confirm the corresponding assignment of the real and effective information, and performs the assignment processing. The specific processing method is as follows: The total browsing time of different real and effective information is marked as SC k , mark the number of clicks as DC k , where k represents different real and effective information; Using TZ k =SC k ×C1+DC k×C2 determines the corresponding information characteristic value TZ k , where C1 and C2 are preset fixed coefficient factors, and their specific values ​​are determined by the operator based on experience. k The preset intervals are compared with the preset intervals respectively, wherein the preset intervals include three groups, namely, interval 0 [Y1, Y2], interval 1 (Y2, Y3], interval 2 (Y3, Y4], wherein Y1, Y2, Y3 and Y4 are all preset interval end values, which are formulated by the operator based on experience; Confirm information characteristic value TZ k If the interval is not found, the corresponding interval naming value is directly selected to assign the real and valid information, and the real and valid information after the assignment is completed is transmitted to the image generation end. If the interval is not found, no assignment is performed, which may cause data abnormality due to accidental touch; Example: Different real and valid information has different total browsing time and number of clicks. Excluding accidental touches, when the total browsing time and number of clicks of the corresponding information are valid values, the larger the value, the longer this information is viewed. In this case, the corresponding information feature value generated is also larger, so the assigned value is larger, that is, "2". On the contrary, when the total browsing time and number of clicks are smaller, the assigned value is smaller, that is, "0".

[0015] The portrait generation end receives the real and valid information after the value assignment, and analyzes and compares it with the sales categories of the retail platform to confirm the specific attribution of the real and valid information. Then, based on the value assignment, a user portrait of the target user is generated. The specific generation process is as follows: Determine the multiple major categories divided by the retail platform, and then determine the group of major categories with the largest number of subordinate categories based on the subordinate categories of real and effective information; Then, determine several corresponding subcategories in this large category, and then confirm the subcategory based on the specific IP ownership of real and effective information; The values ​​of several real and effective information in each different small category are summed up, and the specific value obtained by the summation is calibrated as the ranking value of the corresponding small category; Based on the numerical values ​​of the sorting values, several small categories are sorted from large to small, and the top three small categories are extracted and three groups of user portraits are generated, and the generated three groups of user portraits are transmitted to the push center; For example, the major category is pants, but the minor categories are different styles of pants, such as shorts, trousers, jeans, etc. The corresponding categories can be selected and confirmed on the retail platform. The retail platforms have specific divisions. In order to make the generated user portrait more accurate, the assignment and summation processing method is used to determine the order of the corresponding categories, and then the user portrait to be pushed is generated. The processing method is more accurate, more efficient, and more precise.

[0016] Combination Figure 2 The push center includes a time period parameter acquisition unit, a parameter analysis unit, and a push unit. The main function of the push center is to randomly push the three groups of user portraits to the target user's terminal at the appropriate time. The time period parameter acquisition unit acquires different time periods when the target user uses this retail platform, and transmits the acquired different time periods to the parameter analysis unit; The parameter analysis unit performs a comprehensive ratio analysis on the different time periods obtained and the preset pushable time periods, selects the maximum ratio, and thus determines the best push time period, and the determination method is as follows: According to the rest time in the morning, noon and evening, three groups of push time periods are determined. The push time periods are all planned in advance and operated by the operators themselves; Based on the different time periods obtained, the total usage time of the target users corresponding to each group of different pushable time periods is confirmed, and then the usage ratio value is determined, where: usage ratio value = total usage time ÷ total time of pushable time periods; The maximum ratio is determined from the three groups of usage ratio values, and the push time period to which the maximum ratio belongs is marked as the optimal push time period.

[0017] The push unit randomly pushes three groups of user portraits to the corresponding target user terminals during the optimal push time period and when the target users are using the retail platform.

[0018] Specifically, in order to ensure that the pushed user portrait can be pushed to the target user more appropriately, we must first confirm the corresponding rest period of the target user, and then determine the optimal duration of the usage process based on the specific duration of the corresponding user using the corresponding platform during the corresponding rest period, thereby determining a time period with the longest usage time. Pushing within this time period will have a better overall push effect and can be pushed to the corresponding target user more timely. The corresponding target user will also have sufficient time to view and place an order for such information.

[0019] Embodiment 2

[0020] In the specific implementation process of this embodiment, compared with the first embodiment, this embodiment mainly operates on the method for generating a user portrait; A method for generating a user portrait of a retail platform based on cloud computing, comprising: Determine target users based on the specific purchase volume of registered users; Extract historical browsing information generated by target users from the retail platform and ensure that the historical browsing information meets the standards; Perform a preliminary screening of historical browsing information, remove invalid historical browsing information by confirming the average browsing time, and retain valid historical browsing information; Cluster the viewing speed of valid historical browsing information, remove valid historical browsing information with large clustering deviation, and retain real and valid information; Based on the total browsing time and click count of the real and effective information, determine the information feature value, and based on the information feature value, determine the specific value of the real and effective information; Based on the specific assignment and classification of different real and valid information, the specific ownership of the real and valid information is confirmed and a user portrait of the corresponding target user is generated.

[0021] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0022] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A retail platform user portrait generation system based on cloud computing, characterized in that: include: The target user confirmation end determines the target user from the specific purchase amount of the corresponding user in the retail platform; The browsing information acquisition terminal directly extracts the historical browsing information of the target user from the retail platform based on the confirmed target user, which also includes the specific browsing time of the historical browsing information; The information screening end performs preliminary screening on the historical browsing information generated by the target user, removes invalid historical browsing information, and retains valid historical browsing information; The information analysis terminal re-analyzes the valid historical browsing information and determines whether the corresponding browsing information is real valid information or assumed valid information by confirming the browsing speed; The assignment processing end confirms the total browsing time and click count of the corresponding information based on the real and effective information of the corresponding target user, and then locks the information feature value according to the confirmed two sets of parameters, and then checks the information feature value with the preset interval to confirm the corresponding assignment of the real and effective information; The portrait generation end receives the real and valid information after the assignment processing, analyzes and compares it with the sales categories of the retail platform, confirms the specific ownership of the real and valid information, and then generates a user portrait of the target user based on the assignment.

2. A retail platform user portrait generation system based on cloud computing according to claim 1, characterized in that: The target user confirmation terminal shown, when the corresponding user purchases more than ten times in this retail platform, the corresponding user will be marked as a target user.

3. The system for generating user portraits of retail platforms based on cloud computing according to claim 1, characterized in that: The browsing information acquisition end transmits all the acquired historical browsing information to the information screening end when the acquired historical browsing information of the target user does not exceed 30 groups. If it does not exceed 30 groups, the end continues to monitor until the acquired historical browsing information exceeds 30 groups and then transmits it.

4. The system for generating user portraits of a retail platform based on cloud computing according to claim 1, characterized in that: The specific method of the information screening end for preliminary screening of historical browsing information is as follows: Average the browsing time of several groups of different historical browsing information, confirm that the average browsing time of several groups of historical browsing information is JS, and generate the browsing time interval of this target user [JS-X1, JS+X1], where X1 is ⅓ of JS; Based on the browsing time of each group of browsing information, the browsing information belonging to this browsing time interval [JS-X1, JS+X1] is marked as valid historical browsing information, otherwise no marking is performed.

5. The system for generating user portraits of a retail platform based on cloud computing according to claim 1, characterized in that: The specific method by which the information analysis terminal analyzes the valid historical browsing information is as follows: Obtain the browsing speed of each set of valid historical browsing information from the retail platform, and mark each set of browsing speed as SS i , where i represents different valid historical browsing information; Set several groups of browsing speed SS i Perform clustering processing, sort several groups of browsing speeds from small to large, determine the middle value at the sorting position, and mark it as the center value. If there are two groups of middle values, randomly select one group of values; Different browsing speeds are processed by difference with the central value, and then several groups of differences are processed by average to determine a group of means. The valid historical browsing information with a difference greater than the mean is marked as assumed valid information, otherwise, it is marked as real valid information, and the processed groups of real valid information are transmitted to the assignment processing end.

6. The cloud computing-based retail platform user portrait generation system according to claim 1, characterized in that: The assignment processing end performs assignment processing, and its specific processing method is as follows: The total browsing time of different real and effective information is marked as SC k , mark the number of clicks as DC k , where k represents different real and effective information; Using TZ k =SC k ×C1+DC k ×C2 determines the corresponding information characteristic value TZ k , where C1 and C2 are preset fixed coefficient factors, and the determined information characteristic value TZ k Compare with the preset intervals respectively, wherein the preset intervals include three groups, namely, interval 0 [Y1, Y2], interval 1 (Y2, Y3], interval 2 (Y3, Y4], wherein Y1, Y2, Y3 and Y4 are all preset interval end values; Confirm information characteristic value TZ k Depending on which group of intervals it belongs to, the corresponding interval naming value is directly selected to assign a value to this real and valid information, and the real and valid information after the assignment is completed is transmitted to the portrait generation end. If the interval to which it belongs is not queried, no assignment is performed.

7. The system for generating user portraits of a retail platform based on cloud computing according to claim 1, characterized in that: The process of generating the user portrait of the target user at the portrait generating end is as follows: Determine the multiple major categories divided by the retail platform, and then determine the group of major categories with the largest number of subordinate categories based on the subordinate categories of real and effective information; Then, determine several corresponding subcategories in this large category, and then confirm the subcategory based on the specific IP ownership of real and effective information; The values ​​of several real and effective information in each different small category are summed up, and the specific value obtained by the summation is calibrated as the ranking value of the corresponding small category; Based on the numerical values ​​of the sorting values, several small categories are sorted from large to small, and the top three small categories are extracted and three groups of user portraits are generated. The three groups of user portraits generated are transmitted to the push center.

8. The system for generating user portraits of a retail platform based on cloud computing according to claim 7, characterized in that: The push center includes a time period parameter acquisition unit, a parameter analysis unit and a push unit; The time period parameter acquisition unit acquires different time periods when the target user uses the retail platform, and transmits the acquired different time periods to the parameter analysis unit; The parameter analysis unit performs a comprehensive ratio analysis on the different time periods obtained and the preset pushable time periods, selects the maximum ratio, and thus determines the best push time period.

9. The system for generating user portraits of a retail platform based on cloud computing according to claim 8, characterized in that: The parameter analysis unit determines the best push time by: According to the rest time in the morning, noon and evening, three groups of push time periods are determined. The push time periods are all planned in advance and operated by the operators themselves; Based on the different time periods obtained, the total usage time of the target users corresponding to each group of different pushable time periods is confirmed, and then the usage ratio value is determined, where: usage ratio value = total usage time ÷ total time of pushable time periods; From the three groups of usage ratio values, determine the maximum ratio, and mark the push time period to which the maximum ratio belongs as the optimal push time period; The push unit randomly pushes the three groups of user portraits to the corresponding target user terminals during the optimal push time period and when the target users are using the retail platform.

10. A method for generating a user portrait of a retail platform based on cloud computing, the method being applied to the user portrait generation system according to any one of claims 1 to 9, characterized in that: include: Determine target users based on the specific purchase volume of registered users; Extract historical browsing information generated by target users from the retail platform and ensure that the historical browsing information meets the standards; Perform a preliminary screening of historical browsing information, remove invalid historical browsing information by confirming the average browsing time, and retain valid historical browsing information; Cluster the viewing speed of valid historical browsing information, remove valid historical browsing information with large clustering deviation, and retain real and valid information; Based on the total browsing time and click count of the real and effective information, determine the information feature value, and based on the information feature value, determine the specific value of the real and effective information; Based on the specific assignment and classification of different real and valid information, the specific ownership of the real and valid information is confirmed and a user portrait of the corresponding target user is generated.

Images