Dynamic big data label generation method and device and computer equipment

Abstract

The invention provides a dynamic big data label generation method and device and computer equipment, and the method comprises the steps: obtaining a user data set of a preset dimension, then carrying out the fitting with time according to target dimension values and a difference value between the target dimension values, and obtaining a dynamic big data label according to a fitting function and a to-be-predicted time point; and a final dynamic big data label can be obtained. The method has the beneficial effects that merchants on an e-commerce platform can serve as a reference for the change trend of the data according to the dynamic big data labels, and can further determine the layout of the quantity and types of the products on shelf, so that the merchants are greatly helped to formulate commodity shelf and inventory strategies, and the experience of the merchants and shopping personnel is improved.

Description

Technical Field

[0001] The present invention relates to the field of big data, and in particular to a method, device and computer equipment for generating dynamic big data tags. Background Art

[0002] E-commerce platforms can dynamically label and analyze market data, allowing merchants on the e-commerce platforms to grasp market trends and changes in real time, and adjust product strategies and market layout in a timely manner to seize market opportunities.

[0003] At present, big data tags are mainly generated through real-time data to reflect the characteristics of the current data situation. However, this type of big data tag does not have the ability to predict the changing trend of data, resulting in the inability of merchants on the e-commerce platform to plan the number and types of products to be put on the shelves in the future. Summary of the invention

[0004] The main purpose of the present invention is to provide a method, device and computer equipment for generating dynamic big data tags, aiming to solve the problem that existing big data tags are unable to predict the changing trend of data.

[0005] The present invention provides a method for generating dynamic big data tags, comprising:

[0006] Obtain user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, and aggregate the user data values ​​corresponding to each preset dimension to form a user data set corresponding to each preset dimension;

[0007] Calculating a data fluctuation value in each of the user data sets; wherein the data fluctuation value is used to reflect the data fluctuation situation in the user data sets;

[0008] Selecting a user data set whose data fluctuation value is greater than or equal to a preset fluctuation value, and taking the user data set as a target user data set, and recording a dimension corresponding to the user data set as a target dimension;

[0009] The first weight values ​​of the target dimension values ​​of each target dimension are respectively set correspondingly, and the difference between the target dimension values ​​corresponding to the target dimension at two adjacent time points is calculated according to the time sequence, and the weighted sum is performed to obtain the average difference;

[0010] Based on the average difference and according to the softmax function, a second weight value of the target dimension value that is later in time sequence between two adjacent time points is set, and the second weight value of the first target dimension value is set to a constant;

[0011] Perform weighted summation on each target dimension value and the corresponding first weight value and second weight value to obtain a label value for each target dimension value;

[0012] The label value of each target dimension value and the corresponding time point are fitted through a preset fitting function to obtain a fitting function of each target dimension;

[0013] The time point to be predicted is obtained, and is input into the fitting function of each target dimension respectively to obtain the corresponding dynamic big data label of the time point to be predicted.

[0014] Furthermore, the step of fitting the label value of each target dimension value and the corresponding time point by a preset fitting function to obtain the fitting function of each target dimension includes:

[0015] By formula Calculate the rate of change of the label value corresponding to each target dimension value, where m represents the time point, Indicates the label value corresponding to the kth target dimension at time point l+m, Indicates the label value corresponding to the kth target dimension at time point lm, y kl' Indicates the rate of change of the label value corresponding to the kth target dimension value at time point l;

[0016] The minimum error calculation formula Limit the minimum error between the curve to be fitted and each label value of the same target dimension, and fit the fitting function to obtain the fitting function f for each target dimension k (x k )=a 0k +a 1k x k +…+a tk x k t , where a 0k , a 1k ..., a tk are constants, x kl represents the lth time point, δ kl Indicates time point x kl The error value of the kth target dimension, δ' kl Indicates time point x kl The error value of the change rate of the label value corresponding to the k-th target dimension value, f k (x kl ) indicates the time point is x kl When the kth target dimension corresponds to the fitting value of the limited fitting curve, y kl Indicates time point x kl The actual value of the kth target dimension, f k '(x kl ) is to limit the fitting curve to the time point x klThe derivative of , n represents the number of time points.

[0017] Furthermore, the step of fitting the fitting function to obtain the fitting function of each target dimension includes:

[0018] Taking partial derivatives of the two right sides of the error minimum calculation formula, we get Convert to matrix and simplify to get the first matrix of formula ① And the second matrix of formula ② Among them, x 1 ~x n Indicates the time point, y 1 ~y n Indicates the actual value, y 1' ~y n' represents the fitted value;

[0019] According to the first matrix and the second matrix, find the parameter a 0k , a 1k ..., a tk And substitute them into the curve to be fitted respectively to obtain the fitting function of each target dimension.

[0020] Furthermore, the step of calculating the data fluctuation value in each of the user data sets includes:

[0021] Extracting the maximum value and the minimum value in the user data set;

[0022] According to the formula Calculate the discrete value of the dimension value corresponding to each user data set, where x ij represents the value of the i-th dimension in the j-th user data set, min(x ij ) represents the minimum value of the elements in the jth user data set, max(x ij ) The maximum value of the elements in the jth user data set, Y ij represents the discrete value corresponding to the i-th dimension value in the j-th user data set;

[0023] According to the formula Calculate the data fluctuation value of each user data set, where Among them, E j represents the data fluctuation value of the jth user data set, when p ij = 0, define p ij represents the probability value corresponding to the i-th dimension value of the j-th data fluctuation value, and n represents the number of the user data sets.

[0024] Furthermore, before the step of obtaining user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, the method further includes:

[0025] Detect whether there is an undetermined dimension in the data of the e-commerce platform;

[0026] If the undetermined dimension exists, detecting the correlation value between the undetermined dimension and each preset dimension;

[0027] If there is a correlation value that exceeds the preset correlation value, the undetermined dimension is determined as the preset dimension.

[0028] Furthermore, after the step of obtaining the time point to be predicted and inputting it into the fitting function of each target dimension respectively to obtain the corresponding dynamic big data label of the time point to be predicted, it also includes:

[0029] The types and quantities of products on the shelves are adjusted according to the dynamic big data tags.

[0030] Furthermore, before the step of obtaining user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, the method further includes:

[0031] Obtain data information of the dynamic big data label to be generated;

[0032] The preset dimension is determined according to the data information.

[0033] The present invention also provides a dynamic big data label generation device, comprising:

[0034] An acquisition module is used to acquire user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, and aggregate the user data values ​​corresponding to each preset dimension to form a user data set corresponding to each preset dimension;

[0035] A first calculation module, used to calculate the data fluctuation value in each of the user data sets; wherein the data fluctuation value is used to reflect the data fluctuation situation in the user data set;

[0036] A selection module, configured to select a user data set whose data fluctuation value is greater than or equal to a preset fluctuation value, and use the user data set as a target user data set, and record a dimension corresponding to the user data set as a target dimension;

[0037] A first setting module is used to set first weight values ​​of the target dimension values ​​of each target dimension respectively based on the target dimension values ​​of each target dimension, and to calculate the difference between the target dimension values ​​corresponding to the target dimension at two adjacent time points according to the time sequence, and to obtain an average difference by weighted summation;

[0038] A second setting module is used to set a second weight value of a target dimension value that is later in time sequence between two adjacent time points based on the average difference and according to a softmax function, and to set the second weight value of the first target dimension value to a constant;

[0039] A second calculation module is used to perform weighted summation on each target dimension value and the corresponding first weight value and second weight value to obtain a label value of each target dimension value;

[0040] A fitting module is used to fit the label value of each target dimension value and the corresponding time point through a preset fitting function to obtain a fitting function of each target dimension;

[0041] The input module is used to obtain the time point to be predicted and input it into the fitting function of each target dimension respectively to obtain the corresponding dynamic big data label of the time point to be predicted.

[0042] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0043] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0044] The beneficial effects of the present invention include obtaining a user data set of preset dimensions, and then fitting the target dimension values ​​and the difference between the target dimension values ​​with time. According to the fitting function and the time point to be predicted, the final dynamic big data label can be obtained, so that merchants on the e-commerce platform can use the dynamic big data label as a reference for the data change trend, and can further determine the layout of the number and type of products on the shelves, which greatly helps merchants to formulate strategies for product listing and inventory, and improves the experience of merchants and shoppers. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flowchart of a method for generating dynamic big data tags according to an embodiment of the present invention;

[0046] Figure 2 It is a schematic block diagram of the structure of a dynamic big data tag generation device according to one embodiment of the present invention;

[0047] Figure 3 A schematic block diagram of the structure of a computer device according to an embodiment of the present application.

[0048] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0049] 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.

[0050] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly. The connection can be a direct connection or an indirect connection.

[0051] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there may be three relationships. For example, A and B can represent three situations: A exists alone, A and B exist at the same time, and B exists alone.

[0052] In addition, in the present invention, descriptions such as "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0053] Reference Figure 1 The present invention proposes a method for generating dynamic big data tags, comprising:

[0054] S1: Obtain user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, and aggregate the user data values ​​corresponding to each preset dimension to form a user data set corresponding to each preset dimension;

[0055] S2: Calculate the data fluctuation value in each of the user data sets; wherein the data fluctuation value is used to reflect the data fluctuation situation in the user data sets;

[0056] S3: Selecting a user data set whose data fluctuation value is greater than or equal to a preset fluctuation value, and taking the user data set as a target user data set, and recording a dimension corresponding to the user data set as a target dimension;

[0057] S4: first weight values ​​of the target dimension values ​​of the target dimensions are respectively set accordingly, and the difference between the target dimension values ​​corresponding to the target dimensions at two adjacent time points is calculated according to the time sequence, and the weighted sum is performed to obtain the average difference;

[0058] S5: setting a second weight value of a target dimension value that is later in time sequence between two adjacent time points based on the average difference and according to a softmax function, and setting the second weight value of the first target dimension value to a constant;

[0059] S6: Perform weighted summation on each target dimension value and the corresponding first weight value and second weight value to obtain a label value of each target dimension value;

[0060] S7: Fitting the label value of each target dimension value and the corresponding time point through a preset fitting function to obtain a fitting function of each target dimension;

[0061] S8: Obtain the time point to be predicted, and input it into the fitting function of each target dimension respectively to obtain the corresponding dynamic big data label of the time point to be predicted.

[0062] As described in step S1 above, the user data values ​​of each preset dimension at multiple time points within a set time period are obtained from the e-commerce platform, and the user data values ​​corresponding to each preset dimension are aggregated to form a user data set corresponding to each preset dimension. The preset dimension is a predetermined dimension, which is determined according to the merchant corresponding to the e-commerce platform. For example, if the merchant is a clothing store, the required data includes user behavior data (browsing volume, click volume, purchase volume, etc.), user basic information (age, gender, region, etc.), transaction data (order amount, order quantity, return and exchange rate, etc.), time dimension (specific time point or time period) and other different dimensional data. Different merchants on the e-commerce platform have different preset dimensions. Specifically, the preset dimensions corresponding to different merchants can be pre-set, or they can be set manually in real time. Then, according to the set time period, the set time period can be the time period of the past year or the time period of the most recent month, preferably the time period close to the current time, such data is more meaningful for reference, and then the user data sets corresponding to each preset dimension are formed, that is, a user data set includes the dimension values ​​of this preset dimension at multiple time points within the set time period.

[0063] As described in the above steps S2-S3, the data fluctuation value in each of the user data sets is calculated; wherein the data fluctuation value is used to reflect the data fluctuation situation in the user data set, and the user data set whose data fluctuation value is greater than or equal to the preset fluctuation value is selected, and the user data set is used as the target user data set, and the dimension corresponding to the user data set is recorded as the target dimension. The data fluctuation value can specifically be variance, standard deviation, range, mean absolute deviation and coefficient of variation, etc. When the data fluctuation value is greater than or equal to the preset value, it means that the dimension has significant reference value, while for the dimension whose data fluctuation value is less than the preset value, it is not very helpful for the indicator. Therefore, the dimension whose data fluctuation value is less than the preset value can be eliminated, and the target user data set whose data fluctuation value is greater than or equal to the preset value and the corresponding target dimension are analyzed and processed.

[0064] As described in the above steps S4-S5, the first weight values ​​of the target dimension values ​​of each target dimension are respectively set correspondingly, and the difference between the target dimension values ​​corresponding to the target dimension at two adjacent time points is calculated according to the time sequence, and the weighted sum is obtained to obtain the average difference. Among them, the method of calculating the first weight can be through the formula α i =softmax(s i ) is calculated, where s i represents the target dimension value, α i Represents the first weight value. Then, calculating the average difference can reflect the change of the data between the two time points. Therefore, based on the average difference and according to the softmax function, the second weight value of the target dimension value with the later time sequence in the two adjacent time points is set, and the second weight value of the first target dimension value is set to a constant. Specifically, according to the formula β i =softmax(δ i ) is calculated, where β i represents the second weight value, δ i Represents the average difference, so that the first weight value and the second weight value of each target dimension value can be determined.

[0065] As described in the above step S6, each target dimension value and the corresponding first weight value and second weight value are weightedly summed to obtain the label value of each target dimension value, wherein the weighted summation method is to multiply the target dimension value and the corresponding first weight value and second weight value by the corresponding coefficients respectively, and then sum them up to obtain the label value of each target dimension value.

[0066] As described in step S7 above, the label value of each target dimension value and the corresponding time point are fitted by a preset fitting function to obtain the fitting function of each target dimension. The fitting method can be a linear function fitting or a quadratic function fitting, preferably a polynomial function fitting, and the specific fitting method will be described in detail later, which will not be repeated here, so as to obtain the final fitting function.

[0067] As described in step S8 above, the time point to be predicted is obtained, and is input into the fitting function of each target dimension respectively, so as to obtain the corresponding dynamic big data label of the time point to be predicted. The time point to be predicted is obtained, and then is input into the fitting function of the target dimension respectively, so as to obtain the numerical value of each target dimension, and then obtain the dynamic big data label. According to the fitting function and the time point to be predicted, the final dynamic big data label can be obtained, so that merchants on the e-commerce platform can use the dynamic big data label as a reference for the change trend of the data, and can further determine the layout of the number and type of products on the shelves, which greatly helps merchants to formulate strategies for product listing and inventory, and improves the experience of merchants and shoppers.

[0068] In one embodiment, the step S7 of fitting the label value of each target dimension value and the corresponding time point by a preset fitting function to obtain the fitting function of each target dimension includes:

[0069] S701: By formula Calculate the rate of change of the label value corresponding to each target dimension value, where m represents the time point, Indicates the label value corresponding to the kth target dimension at time point l+m, Indicates the label value corresponding to the kth target dimension at time point lm, y kl' Indicates the rate of change of the label value corresponding to the kth target dimension value at time point l;

[0070] S702: Calculate the minimum error value using the formula Limit the minimum error between the curve to be fitted and each label value of the same target dimension, and fit the fitting function to obtain the fitting function of each target dimension Among them, a 0k , a 1k ..., a tk are constants, x kl represents the lth time point, δ kl Indicates time point x kl The error value of the kth target dimension, δ' kl Indicates time point x kl The error value of the change rate of the label value corresponding to the k-th target dimension value, fk (x kl ) indicates the time point is x kl When the kth target dimension corresponds to the fitting value of the limited fitting curve, y kl Indicates time point x kl The actual value of the kth target dimension, f k '(x kl ) is to limit the fitting curve to the time point x kl The derivative of , n represents the number of time points.

[0071] As described in the above steps S701-S702, a fitting calculation method is provided. This fitting method not only takes into account the size of each label value, but also takes into account the slope (i.e., the rate of change) of the label value. By combining the numerical value size and the rate of change, the minimum error between the curve to be fitted and each label value of the same target dimension is limited, and the fitting curve is fitted to obtain the fitting function corresponding to each target dimension.

[0072] In one embodiment, the step S702 of fitting the fitting function to obtain the fitting function of each target dimension includes:

[0073] S7021: Take partial derivatives of the right sides of the two equations of the minimum error calculation formula to obtain Convert to matrix and simplify to get the first matrix of formula ① And the second matrix of formula ② Among them, x 1 ~x n Indicates the time point, y 1 ~y n Indicates the actual value, y 1' ~y n' represents the fitted value;

[0074] S7022: Calculate parameter a according to the first matrix and the second matrix 0k , a 1k ..., a tk And substitute them into the curve to be fitted respectively to obtain the fitting function of each target dimension.

[0075] As described in steps S7021-S7022 above, taking partial derivatives of the right sides of the two equations of the error minimum calculation formula, we obtain Convert to matrix and simplify to get the first matrix of formula ① And the second matrix of formula ② Calculate the parameter a based on the first matrix and the second matrix 0k , a 1k ..., a tkAnd they are substituted into the curve to be fitted respectively to obtain the fitting function of each target dimension, and the size and change rate of the label value are involved in the calculation, so that the fitting function can fit the data more accurately.

[0076] In one embodiment, the step S2 of calculating the data fluctuation value in each of the user data sets includes:

[0077] S201: extracting the maximum value and the minimum value in the user data set;

[0078] S202: According to the formula Calculate the discrete value of the dimension value corresponding to each user data set, where x ij represents the value of the i-th dimension in the j-th user data set, min(x ij ) represents the minimum value of the elements in the jth user data set, max(x ij ) The maximum value of the elements in the jth user data set, Y ij represents the discrete value corresponding to the i-th dimension value in the j-th user data set;

[0079] S203: According to the formula Calculate the data fluctuation value of each user data set, where Among them, E j represents the data fluctuation value of the jth user data set, when p ij = 0, define p ij represents the probability value corresponding to the i-th dimension value of the j-th data fluctuation value, and n represents the number of the user data sets.

[0080] As described in the above steps S201-S203, an algorithm for calculating data fluctuation values ​​is provided. Specifically, the maximum and minimum values ​​in the user data set are first obtained, and the data fluctuation of the entire user data set is reflected according to the maximum and minimum values. Then, the discrete values ​​corresponding to the values ​​of each target dimension in each user data set are calculated and normalized, so as to prevent the final result from being biased due to excessive data. Then, the probability p of the standard deviation corresponding to the values ​​of each target dimension is calculated. ij , calculate the data fluctuation value of the jth user data set. The data fluctuation value calculated according to the above calculation formula fully considers the fluctuation of the values ​​of each target dimension in the same dimension, and also fully considers the impact of an extremely large or small value on the overall data fluctuation value, so that the calculated data fluctuation value is more referenceable.

[0081] In one embodiment, before the step S1 of acquiring user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, the method further includes:

[0082] S001: Detect whether there is an undetermined dimension in the data of the e-commerce platform;

[0083] S002: If the undetermined dimension exists, detecting the correlation value between the undetermined dimension and each preset dimension;

[0084] S003: If there is a correlation value that exceeds the preset correlation value, the undetermined dimension is determined as the preset dimension.

[0085] As described in the above steps S001-S003, in some embodiments, when setting the preset dimension, the consideration may not be very comprehensive. Therefore, each dimension in the data can be traversed to detect whether there is still a pending dimension, and then the correlation value between the pending dimension and each preset dimension is calculated, wherein the method of calculating the correlation value is not limited. For example, the pending dimension and the preset dimension can be converted into vectors, and then the similarity calculation is performed, and the similarity value is used as the correlation value. Alternatively, the text information under the pending dimension and the text information under the preset dimension can be obtained, and then the text vectors of the two text information are identified according to the semantic recognition technology, and the similarity value is calculated as the correlation value. Alternatively, the data information under the pending dimension and the data information under the preset dimension can be obtained, and then converted into corresponding data vectors through a preset data conversion method, and then the similarity between the data vectors is calculated and used as the correlation value.

[0086] In one embodiment, after the step S8 of obtaining the time point to be predicted and inputting it into the fitting function of each target dimension respectively to obtain the corresponding dynamic big data label of the time point to be predicted, the method further includes:

[0087] S901: Adjust the types and quantities of products on the shelves according to the dynamic big data tags.

[0088] As described in the above step S901, after obtaining the dynamic big data tag, it can be forwarded directly to the merchant. The merchant adjusts the types and quantities of listed products according to the dynamic big data tag. In some embodiments, the e-commerce platform can also calculate the types and quantities of the merchant's listed products and then send them to the merchant as a reference.

[0089] In one embodiment, before the step S1 of acquiring user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, the method further includes:

[0090] S011: Obtain data information of the dynamic big data tag to be generated;

[0091] S012: Determine the preset dimension according to the data information.

[0092] As described in the above steps S011-S012, the preset dimensions are determined, and the data information of the dynamic big data tags to be generated is obtained. The data information can be obtained from the merchants, namely the business scope, service items, transaction volume per unit time, etc. of the merchants. Based on the information, the preset dimensions that the user needs to consider can be determined, thereby realizing personalized business guidance services for each merchant.

[0093] Reference Figure 2 , a dynamic big data label generation device, comprising:

[0094] The acquisition module 10 is used to acquire user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, and aggregate the user data values ​​corresponding to each preset dimension to form a user data set corresponding to each preset dimension;

[0095] A first calculation module 20 is used to calculate the data fluctuation value in each of the user data sets; wherein the data fluctuation value is used to reflect the data fluctuation situation in the user data set;

[0096] A selection module 30 is used to select a user data set whose data fluctuation value is greater than or equal to a preset fluctuation value, and use the user data set as a target user data set, and record a dimension corresponding to the user data set as a target dimension;

[0097] A first setting module 40 is used to set first weight values ​​of the target dimension values ​​of each target dimension respectively based on the target dimension values ​​of each target dimension, and to calculate the difference between the target dimension values ​​corresponding to the target dimension at two adjacent time points according to the time sequence, and to perform weighted summation to obtain an average difference;

[0098] A second setting module 50 is used to set a second weight value of a target dimension value that is later in time sequence between two adjacent time points based on the average difference and according to a softmax function, and to set the second weight value of the first target dimension value to a constant;

[0099] A second calculation module 60 is used to perform weighted summation on each target dimension value and the corresponding first weight value and second weight value to obtain a label value of each target dimension value;

[0100] A fitting module 70 is used to fit the label value of each target dimension value and the corresponding time point through a preset fitting function to obtain a fitting function of each target dimension;

[0101] The input module 80 is used to obtain the time point to be predicted and input it into the fitting function of each target dimension respectively to obtain the corresponding dynamic big data label of the time point to be predicted.

[0102] In one embodiment, the fitting module 70 includes:

[0103] The rate of change calculation submodule is used to calculate the rate of change through the formula Calculate the rate of change of the label value corresponding to each target dimension value, where m represents the time point, Indicates the label value corresponding to the kth target dimension at time point l+m, Indicates the label value corresponding to the kth target dimension at time point lm, y kl' Indicates the rate of change of the label value corresponding to the kth target dimension value at time point l;

[0104] The minimum error calculation submodule is used to calculate the minimum error formula Limit the minimum error between the curve to be fitted and each label value of the same target dimension, and fit the fitting function to obtain the fitting function f for each target dimension k (x k )=a 0k +a 1k x k +…+a tk x k t , where a 0k , a 1k ..., a tk are constants, x kl represents the lth time point, δ kl Indicates time point x kl The error value of the kth target dimension, δ' kl Indicates time point x kl The error value of the change rate of the label value corresponding to the k-th target dimension value, f k (x kl ) indicates the time point is x kl When the kth target dimension corresponds to the fitting value of the limited fitting curve, y kl Indicates time point x kl The actual value of the kth target dimension, f k '(x kl ) is to limit the fitting curve to the time point x kl The derivative of , n represents the number of time points.

[0105] In one embodiment, the error minimum value calculation submodule includes:

[0106] A calculation unit is used to take partial derivatives of the right sides of the two equations of the error minimum calculation formula to obtain Convert to matrix and simplify to get the first matrix of formula ① And the second matrix of formula ② Among them, x 1 ~x n Indicates the time point, y 1 ~y n Indicates the actual value, y 1' ~y n' represents the fitted value;

[0107] Substitution unit, used to find the parameter a according to the first matrix and the second matrix 0k , a 1k ..., a tk And substitute them into the curve to be fitted respectively to obtain the fitting function of each target dimension.

[0108] In one embodiment, the first calculation module 20 includes:

[0109] A maximum value extraction submodule, used to extract the maximum value and the minimum value in the user data set;

[0110] Discrete value calculation submodule is used to calculate the discrete value according to the formula Calculate the discrete value of the dimension value corresponding to each user data set, where x ij represents the value of the i-th dimension in the j-th user data set, min(x ij ) represents the minimum value of the elements in the jth user data set, max(x ij ) The maximum value of the elements in the jth user data set, Y ij represents the discrete value corresponding to the i-th dimension value in the j-th user data set;

[0111] The data fluctuation value calculation submodule is used to calculate the data fluctuation value according to the formula Calculate the data fluctuation value of each user data set, where Among them, E j represents the data fluctuation value of the jth user data set, when p ij = 0, define p ij represents the probability value corresponding to the i-th dimension value of the j-th data fluctuation value, and n represents the number of the user data sets.

[0112] In one embodiment, the dynamic big data tag generation device further includes:

[0113] A pending dimension detection module is used to detect whether there is a pending dimension in the data of the e-commerce platform;

[0114] A correlation value detection module, used for detecting the correlation value between the undetermined dimension and each preset dimension if the undetermined dimension exists;

[0115] The determination module is used to determine the undetermined dimension as a preset dimension if there is a correlation value exceeding the preset correlation value.

[0116] In one embodiment, the dynamic big data tag generation device further includes:

[0117] The adjustment module is used to adjust the types and quantities of products on the shelves according to the dynamic big data tags.

[0118] In one embodiment, the dynamic big data tag generation device further includes:

[0119] A data information acquisition module is used to acquire data information of the dynamic big data tag to be generated;

[0120] A preset dimension determination module is used to determine the preset dimension according to the data information.

[0121] The beneficial effects of the present invention include obtaining a user data set of preset dimensions, and then fitting the target dimension values ​​and the difference between the target dimension values ​​with time. According to the fitting function and the time point to be predicted, the final dynamic big data label can be obtained, so that merchants on the e-commerce platform can use the dynamic big data label as a reference for the data change trend, and can further determine the layout of the number and type of products on the shelves, which greatly helps merchants to formulate strategies for product listing and inventory, and improves the experience of merchants and shoppers.

[0122] Reference Figure 3 In an embodiment of the present application, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store various user data values, etc. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the dynamic big data label generation method described in any of the above embodiments can be implemented.

[0123] Those skilled in the art will understand that Figure 3 The structure shown in is merely a block diagram of a portion of the structure related to the present application solution and does not constitute a limitation on the computer device to which the present application solution is applied.

[0124] An embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the dynamic big data label generation method described in any of the above embodiments can be implemented.

[0125] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0126] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, device, article or method including the element.

[0127] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0128] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0129] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included in the scope of the claims of the present invention.

Claims

1. A method for generating dynamic big data tags, characterized in that: include: Obtain user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, and aggregate the user data values ​​corresponding to each preset dimension to form a user data set corresponding to each preset dimension; Calculating a data fluctuation value in each of the user data sets; wherein the data fluctuation value is used to reflect the data fluctuation situation in the user data sets; Selecting a user data set whose data fluctuation value is greater than or equal to a preset fluctuation value, and taking the user data set as a target user data set, and recording a dimension corresponding to the user data set as a target dimension; The first weight values ​​of the target dimension values ​​of each target dimension are respectively set correspondingly, and the difference between the target dimension values ​​corresponding to the target dimension at two adjacent time points is calculated according to the time sequence, and the weighted sum is performed to obtain the average difference; Based on the average difference and according to the softmax function, a second weight value of the target dimension value that is later in time sequence between two adjacent time points is set, and the second weight value of the first target dimension value is set to a constant; Perform weighted summation on each target dimension value and the corresponding first weight value and second weight value to obtain a label value for each target dimension value; The label value of each target dimension value and the corresponding time point are fitted through a preset fitting function to obtain a fitting function of each target dimension; The time point to be predicted is obtained, and is input into the fitting function of each target dimension respectively to obtain the corresponding dynamic big data label of the time point to be predicted.

2. The method for generating dynamic big data tags according to claim 1, characterized in that: The step of fitting the label value of each target dimension value and the corresponding time point by a preset fitting function to obtain the fitting function of each target dimension includes: By formula Calculate the rate of change of the label value corresponding to each target dimension value, where m represents the time point, Indicates the label value corresponding to the kth target dimension at time point l+m, Indicates the label value corresponding to the kth target dimension at time point lm, y kl’ Indicates the rate of change of the label value corresponding to the kth target dimension value at time point l; The minimum error calculation formula Limit the minimum error between the curve to be fitted and each label value of the same target dimension, and fit the fitting function to obtain the fitting function f for each target dimension k (x k )=a 0k +a 1k x k +…+a tk x k t , where a 0k , a 1k ..., a tk are constants, x kl represents the lth time point, δ kl Indicates time point x kl The error value of the kth target dimension, δ' kl Indicates time point x kl The error value of the change rate of the label value corresponding to the k-th target dimension value, f k (x kl Indicates time point x kl When the kth target dimension corresponds to the fitting value of the limited fitting curve, y kl Indicates time point x kl The actual value of the kth target dimension, f k’ (x kl ) is to limit the fitting curve to the time point x kl The derivative of , n represents the number of time points.

3. The method for generating dynamic big data tags according to claim 2, characterized in that: The step of fitting the fitting function to obtain the fitting function of each target dimension includes: Taking partial derivatives of the two right sides of the error minimum calculation formula, we get Convert to matrix and simplify to get the first matrix of formula ① And the second matrix of formula ② Among them, x1~x n Indicates time point, y1~y n Indicates the actual value, y 1’ ~y n’ represents the fitted value; According to the first matrix and the second matrix, find the parameter a 0k , a 1k ..., a tk And substitute them into the curve to be fitted respectively to obtain the fitting function of each target dimension.

4. The method for generating dynamic big data tags according to claim 1, characterized in that: The step of calculating the data fluctuation value in each of the user data sets includes: Extracting the maximum value and the minimum value in the user data set; According to the formula Calculate the discrete value of the dimension value corresponding to each user data set, where x ij represents the value of the i-th dimension in the j-th user data set, min(x ij ) represents the minimum value of the elements in the jth user data set, max(x ij ) The maximum value of the elements in the jth user data set, Y ij represents the discrete value corresponding to the i-th dimension value in the j-th user data set; According to the formula Calculate the data fluctuation value of each user data set, where Among them, E j represents the data fluctuation value of the jth user data set, when p ij = 0, define p ij represents the probability value corresponding to the i-th dimension value of the j-th data fluctuation value, and n represents the number of the user data sets.

5. The method for generating dynamic big data tags according to claim 1, characterized in that: Before the step of obtaining user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, the method further includes: Detect whether there is an undetermined dimension in the data of the e-commerce platform; If the undetermined dimension exists, detecting the correlation value between the undetermined dimension and each preset dimension; If there is a correlation value that exceeds the preset correlation value, the undetermined dimension is determined as the preset dimension.

6. The method for generating dynamic big data tags according to claim 1, characterized in that: After the step of obtaining the time point to be predicted and inputting it into the fitting function of each target dimension respectively to obtain the dynamic big data label corresponding to the time point to be predicted, the method further includes: The types and quantities of products on the shelves are adjusted according to the dynamic big data tags.

7. The method for generating dynamic big data tags according to claim 1, characterized in that: Before the step of obtaining user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, the method further includes: Obtain data information of the dynamic big data label to be generated; The preset dimension is determined according to the data information.

8. A dynamic big data label generation device, characterized in that: include: An acquisition module is used to acquire user data values ​​of each preset dimension at multiple time points within a set time period from the e-commerce platform, and aggregate the user data values ​​corresponding to each preset dimension to form a user data set corresponding to each preset dimension; A first calculation module, used to calculate the data fluctuation value in each of the user data sets; wherein the data fluctuation value is used to reflect the data fluctuation situation in the user data set; A selection module, configured to select a user data set whose data fluctuation value is greater than or equal to a preset fluctuation value, and use the user data set as a target user data set, and record a dimension corresponding to the user data set as a target dimension; A first setting module is used to set first weight values ​​of the target dimension values ​​of each target dimension respectively based on the target dimension values ​​of each target dimension, and to calculate the difference between the target dimension values ​​corresponding to the target dimension at two adjacent time points according to the time sequence, and to obtain an average difference by weighted summation; A second setting module is used to set a second weight value of a target dimension value that is later in time sequence between two adjacent time points based on the average difference and according to a softmax function, and to set the second weight value of the first target dimension value to a constant; A second calculation module is used to perform weighted summation on each target dimension value and the corresponding first weight value and second weight value to obtain a label value of each target dimension value; A fitting module is used to fit the label value of each target dimension value and the corresponding time point through a preset fitting function to obtain a fitting function of each target dimension; The input module is used to obtain the time point to be predicted and input it into the fitting function of each target dimension respectively to obtain the corresponding dynamic big data label of the time point to be predicted.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images