E-commerce platform promotion task allocation method and device, terminal and storage medium

Abstract

The invention discloses an e-commerce platform promotion task distribution method and device, a terminal and a storage medium, and the method comprises the steps: obtaining an original promotion content data set and an original marketing account data set in a first promotion period, and judging the matching condition between each piece of promotion content and each marketing account; according to the matching condition, the expected promotion distribution income of each piece of promotion content and each marketing account is initialized; based on the expected promotion distribution income, solving and randomly rounding through a linear programming model to obtain a distribution scheme which meets a preset constraint condition and has the maximum expected income total amount; and entering the next promotion period and updating the original marketing account data set according to the distribution scheme, if the maximum promotion period is not reached, obtaining the distribution scheme of the next promotion period based on the updated marketing account data set, and otherwise, ending the task. According to the method, promotion task allocation is realized through efficient matching between the promotion content and the marketing account, so that the promotion effect and the resource utilization efficiency are improved.

Description

Technical Field

[0001] The present application relates to the field of Internet e-commerce promotion, and in particular to an e-commerce platform promotion task allocation method, device, terminal and storage medium. Background Art

[0002] In recent years, with the rapid development of Internet technology and the change of consumers' shopping habits, e-commerce platforms have gradually become an important channel for commodity sales. In order to improve the promotion effect of commodities, e-commerce platforms need to accurately match according to the characteristics of different marketing accounts and commodity characteristics, so as to achieve the efficient use of marketing resources. This process not only involves the data analysis of a large number of commodities and marketing accounts, but also needs to consider various complex factors, such as promotion costs, promotion effects, and the competitive relationship between commodities, etc., to maximize the benefits under budget constraints.

[0003] In the prior art, to solve the problem of task allocation on e-commerce platforms, the commonly used methods include manual experience judgment, simple matching based on historical data, and single-index allocation strategies with fixed rules. Manual experience judgment mainly relies on the experience of operation personnel, and manually selects appropriate marketing accounts for promotion; simple matching based on historical data statistically analyzes the historical promotion effects and selects the accounts with better effects for re-promotion; the allocation strategy with fixed rules is to allocate tasks according to preset rules, such as according to the number of fans of the account or historical transaction volume.

[0004] However, the above methods generally have the following defects: they cannot update the promotion ability of marketing accounts in real time, resulting in a deviation between the task allocation result and the actual promotion effect; they fail to fully consider the trade-off between promotion costs and expected effects, and it is difficult to maximize the benefits under budget constraints; at the same time, due to ignoring the competitive relationship of promotion resources between commodities, it may lead to poor promotion effects of some commodities or waste of resources. Therefore, there is an urgent need for an e-commerce platform task allocation technology that can combine real-time fine-grained traffic analysis to overcome the deficiencies of the prior art. Summary of the Invention

[0005] In order to complete the efficient matching between multi-commodity marketing content and marketing accounts and achieve the task allocation of maximizing benefits, the present application provides an e-commerce platform promotion task allocation method, device, terminal and storage medium.

[0006] In the first aspect, the present application provides an e-commerce platform promotion task allocation method, adopting the following technical means: An e-commerce platform promotion task allocation method includes the following steps: When in the first promotion cycle of the promotion allocation task, obtain the original promotion content data set and the original marketing account data set, and judge the matching situation between each piece of promotion content and each marketing account; Initialize the expected promotion allocation revenue of each promotion content and each marketing account in the current promotion period according to the matching situation; In the current promotion period, based on the expected promotion allocation revenue, solve through a linear programming model and perform random rounding to obtain an allocation plan for the current promotion period that satisfies the preset constraint conditions and maximizes the total expected revenue; Enter the next promotion period and update the original marketing account dataset according to the allocation plan. If the maximum promotion period is not reached, obtain the allocation plan for the next promotion period based on the updated marketing account dataset, otherwise end the promotion allocation task.

[0007] By adopting the above technical solution, match each promotion content with each marketing account, thereby initializing the expected promotion allocation revenue between the two in the current promotion period, providing basic data support for subsequent allocation; obtain the optimal allocation plan through a linear programming model, so that the final expected revenue meets the constraint conditions and reaches the maximum, effectively combining historical data with real-time feedback, realizing the efficient matching and dynamic update of promotion content and marketing accounts for the promotion task allocation, thereby improving the promotion effect and resource utilization efficiency.

[0008] Preferably, the obtaining the original promotion content dataset and the original marketing account dataset, and judging the matching situation between each promotion content and each marketing account specifically includes the following steps: Obtain the original promotion content dataset and the original marketing account dataset of the promotion allocation task, and generate category labels for the promotion content according to the product category and promotion content style in the original promotion content dataset of the promotion content , that is, ; wherein, is the product category of the promotion content , and is the promotion content style of the promotion content ; Combine the historical content style and historical promoted product category of each marketing account extracted from the original marketing account dataset, and judge the matching situation between any marketing account and the current promotion content in the current promotion period ; Specifically, match each marketing account with the product category and promotion content style of the current promotion content to obtain the following four matching results The first matching result is that the marketing account has never promoted the product of the current promotion content, and the style of the historical content of the marketing account is inconsistent with the style of the promotion content; The second matching result is that the marketing account has promoted the product of the current promotion content, and the style of the historical content of the marketing account is consistent with the style of the promotion content; The third matching result is that the marketing account has promoted the product of the current promotion content, but the style of the historical content of the marketing account is inconsistent with the style of the promotion content; The fourth matching result is that the marketing account has never promoted the product of the current promotion content, but the style of the historical content of the marketing account is consistent with the style of the promotion content.

[0009] By adopting the above technical solutions, according to the promotion content dataset and the marketing account dataset, category labels of the promotion content are generated, and combined with the historical content style and the historical promoted product category of the marketing account, the matching situation between each marketing account and the current promotion content is accurately judged; through the matching analysis of the product category and the promotion content style, the adaptability between the marketing account and the current promotion content is divided into four matching results, so as to achieve comprehensive coverage of the matching situation and improve the accuracy of task assignment.

[0010] Preferably, initializing the expected promotion allocation revenue of each promotion content and each marketing account in the current promotion period according to the matching situation specifically includes the following steps: Initializing the matching degree of the th promotion content and the th marketing account according to the matching situation, as well as the historical revenue of the promotion content on the marketing account ; Initializing the expected promotion allocation revenue of the promotion content assigned to the marketing account according to the matching degree and the historical revenue, that is ; wherein, is the expected promotion allocation revenue of the current promotion period , is the matching degree, and is the historical revenue.

[0011] By adopting the above technical solution, by initializing the matching degree and historical revenue between the promotion content and the marketing accounts, the accurate initialization of the expected promotion revenue distribution is finally achieved, laying a foundation for formulating an efficient promotion distribution strategy in the follow-up, contributing to maximizing the revenue under the budget constraint, improving the promotion efficiency of the e-commerce platform, and thus enhancing the overall promotion effect and resource utilization efficiency.

[0012] The original marketing account dataset also includes the transaction revenues of various categories of goods; For the first matching result, the matching degree and the historical revenue are ; ; For the second matching result, the matching degree and the historical revenue are ; ; For the third matching result, the matching degree and the historical revenue are ; ; For the fourth matching result, the matching degree and the historical revenue are ; ; Among them, is the historical content style of the marketing account in the th promotion period, is the historical promoted product category of the marketing account in the th promotion period, is the transaction revenue of the product with the category of .

[0013] By adopting the above technical solution, setting different matching degrees and historical revenues for each matching result makes the task allocation more scientific and reasonable, can better reflect the actual promotion effect, avoids the waste of resources caused by blind allocation, and provides a reliable basis for the initialization of the expected promotion revenue distribution in the follow-up by quantifying the matching degree and historical revenue, contributing to maximizing the revenue under the budget constraint and improving the promotion efficiency of the e-commerce platform.

[0014] Preferably, based on the expected promotion revenue distribution, by solving through a linear programming model and performing stochastic rounding, an allocation plan for the current promotion period that meets the preset constraint conditions and has the maximum total expected revenue is obtained, which specifically includes the following steps: Define slack variables , where , and obtain a constraint formula based on the slack variables and preset constraint conditions; Establish a linear programming model based on the slack variables and the expected promotion allocation revenue. The linear programming model is , where is the total expected revenue; Solve the linear programming model according to the constraint formula to obtain the slack variables that maximize the total expected revenue ; Perform stochastic rounding on the obtained slack variables to obtain a rounding result, and generate an allocation plan for the current promotion period based on the rounding result.

[0015] By adopting the above technical solution, a linear programming model is established based on the expected promotion allocation revenue and constraint conditions, and the slack variables that maximize the total expected revenue are obtained by solving the model. Then, stochastic rounding is performed on the slack variables to generate an optimal allocation plan. Stochastic rounding effectively balances the gap between the theoretical optimal solution and practical operability, enabling the finally generated allocation plan to maximize the promotion revenue within the constraints, thereby improving the task allocation efficiency and promotion effect of the e-commerce platform.

[0016] Preferably, the constraint conditions include: The first constraint condition is that the total cost of the allocation plan does not exceed the preset periodic budget, and the total cost includes the cost of allocating each piece of the promotion content to the corresponding marketing account; The second constraint condition is that within each promotion period, each piece of the promotion content is promoted at least once; The third constraint condition is that the total number of the promotion content allocated to each marketing account does not exceed the corresponding carrying limit of the marketing account.

[0017] By adopting the above technical solution, the constraint conditions are set in detail, realizing the refined management of the promotion task allocation on the e-commerce platform. Through the first constraint condition, the financial controllability of the promotion activity is ensured, avoiding the risk of overspending. Through the second constraint condition, the situation where some promotion content is ignored is avoided. Through the third constraint condition, the decline in effect caused by over-promotion is prevented, and at the same time, the long-term promotion ability of the marketing account is protected.

[0018] Preferably, enter the next promotion cycle and update the original marketing account dataset according to the allocation scheme. If the maximum promotion cycle is not reached, obtain the allocation scheme for the next promotion cycle based on the updated marketing account dataset, otherwise end the promotion allocation task, which specifically includes the following steps: Increment the current promotion cycle by one to make the promotion allocation task enter the next promotion cycle, and determine whether the current promotion cycle reaches the preset maximum promotion cycle; If the maximum promotion cycle is reached, end the promotion allocation task and output the allocation schemes for each promotion cycle; If the maximum promotion cycle is not reached, execute the promotion allocation task according to the allocation scheme and obtain the promotion results of the previous promotion cycle. Based on the promotion results, obtain the updated historical content style, historical promoted product categories, and the transaction revenues of products in each category, and obtain the matching situation of the current promotion cycle. Based on the matching situation, initialize the expected promotion allocation revenues of each promotion content and each marketing account in the current promotion cycle to obtain the allocation scheme until the maximum promotion cycle is reached.

[0019] By adopting the above technical solution, when the maximum promotion cycle is not reached, the expected promotion allocation revenues of each promotion content and each marketing account are re-initialized by using the updated marketing account dataset, which can timely reflect the actual promotion effect changes of the marketing accounts, thereby improving the accuracy of the allocation scheme; generating a new allocation scheme based on the updated expected promotion allocation revenues ensures that the task allocation in each promotion cycle is optimized based on the latest data, maximizing the promotion revenue. When the preset maximum promotion cycle is reached, the promotion task is terminated, improving the controllability of the system, and finally realizing the dynamic update and optimization of the promotion task allocation on the e-commerce platform.

[0020] In a second aspect, the present application provides an e-commerce platform promotion task allocation device, adopting the following technical means: An e-commerce platform promotion task allocation device includes the following modules: A content matching module, configured to obtain the original promotion content dataset and the original marketing account dataset when in the first promotion cycle of the promotion allocation task, and determine the matching situation between each promotion content and each marketing account; An expected revenue initialization module, configured to initialize the expected promotion allocation revenues of each promotion content and each marketing account in the current promotion cycle according to the matching situation; The allocation plan generation module is configured to, within the current promotion period, based on the expected promotion allocation revenue, solve through a linear programming model and perform random rounding to obtain an allocation plan for the current promotion period that meets the preset constraint conditions and maximizes the total expected revenue; The promotion update module is configured to enter the next promotion period and update the original marketing account dataset according to the allocation plan. If the maximum promotion period is not reached, obtain the allocation plan for the next promotion period based on the updated marketing account dataset, otherwise end the promotion allocation task.

[0021] By adopting the above technical solutions, a complete promotion task allocation system is built, providing necessary software technical support for the efficient promotion of e-commerce platform content, significantly improving the efficiency and intelligence level of the matching between promotion content and marketing accounts, achieving the optimal allocation of resources, and meeting the requirements of technological progress.

[0022] In a third aspect, the present application provides an intelligent terminal, adopting the following technical solutions: An intelligent terminal includes a memory and a processor. At least one instruction, at least one program, a code set or an instruction set is stored in the memory, and the at least one instruction, at least one program, the code set or the instruction set is loaded and executed by the processor to implement the e-commerce platform promotion task allocation method as described above.

[0023] In a fourth aspect, the present application provides a computer-readable storage medium, adopting the following technical solutions: A computer-readable storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, at least one program, the code set or the instruction set is loaded and executed by a processor to implement the e-commerce platform promotion task allocation method as described above.

[0024] In summary, the present application includes at least one of the following beneficial technical effects: (1) By initializing and periodically updating the expected promotion allocation revenue of each promotion content and each marketing account, the present application can accurately reflect the actual promotion ability of marketing accounts, improving the accuracy and real-time performance of task allocation.

[0025] (2) The present application generates an allocation plan by solving a linear programming model and performing random rounding, maximizing the total expected revenue under the condition of meeting the constraint conditions, and achieving revenue maximization under budget constraints.

[0026] (3) The present application considers the competition relationship of promotion resources among products, and by dynamically adjusting the allocation strategy, avoids the problems of resource waste and poor promotion effect, and improves the overall promotion efficiency. Brief Description of the Drawings

[0027] Figure 1 is a flowchart of the method for allocating promotion tasks on an e-commerce platform according to an embodiment of the present application; Figure 2 is a flowchart of steps S1 and S2 of the method for allocating promotion tasks on an e-commerce platform according to an embodiment of the present application; Figure 3 is a flowchart of step S3 of the method for allocating promotion tasks on an e-commerce platform according to an embodiment of the present application; Figure 4 is a flowchart of step S4 of the method for allocating promotion tasks on an e-commerce platform according to an embodiment of the present application; Figure 5 is a structural diagram of the device for allocating promotion tasks on an e-commerce platform according to an embodiment of the present application. Detailed Description of the Embodiment

[0028] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings. The described embodiments are only possible technical implementations of the present invention, not all possible implementations. Those skilled in the art can fully combine the embodiments of the present invention to obtain other embodiments without creative labor, and these embodiments are also within the protection scope of the present invention.

[0029] The inventors of the present application found that, based on a given budget, the task allocation on an e-commerce platform needs to select the most suitable marketing account for the promotion content of each product to maximize the expected promotion effect of the product. However, the matching situation between different marketing accounts and promotion content is complex and changeable, and at the same time affected by cost, expected effect and promotion times limit. There is an urgent need for an efficient task allocation technology based on real-time fine-grained traffic analysis. For this reason, the present application mainly adopts the following method. By initializing and periodically updating the adaptation degree between promotion content and marketing accounts, promotion costs, and historical promotion data, and combining a linear programming model to generate an optimal allocation plan, the effect of maximizing revenue under budget constraints is achieved.

[0030] The method for allocating promotion tasks on an e-commerce platform provided by an embodiment of the present application includes obtaining and judging the matching situation of a promotion content data set and a marketing account data set, initializing the expected promotion allocation revenue, solving through a linear programming model and randomly rounding to obtain an allocation plan that meets preset constraint conditions, and updating the marketing account data set according to the execution result until the maximum promotion period is reached. The whole process can achieve the maximization of revenue under budget constraints and improve the promotion efficiency. The following is a further detailed description of the present application.

[0031] The method for allocating promotion tasks on an e-commerce platform provided by an embodiment of the present application, as Figure 1 shown, includes the following steps: S1. When in the first promotion cycle of the promotion assignment task, obtain the original promotion content dataset and the original marketing account dataset, and judge the matching situation between each promotion content and each marketing account. As Figure 2 shown, the specific steps are as follows: S11. Obtain the current promotion cycle , and judge whether the current is in the first promotion cycle of the promotion assignment task, that is, judge whether is 1. If , then the current promotion cycle is in the first promotion cycle of the promotion assignment task.

[0032] If the current promotion cycle is not the first promotion cycle, execute step S42.

[0033] S12. When in the first promotion cycle of the promotion assignment task, obtain the original promotion content dataset.

[0034] In this embodiment, define the original promotion content dataset , and the original promotion content dataset includes the product name, promotion content serial number , product category, promotion content copywriting, promotion content style, and promotion content video.

[0035] Among them, the product category is a positive integer not exceeding 9 digits, and each fixed-digit number represents a first-level product classification; the promotion content style is generated according to the style library and is a positive integer not exceeding 4 digits.

[0036] The category of the promotion content is composed of the product category and the promotion content style.

[0037] For example, in a specific implementable manner, a tweet promoting baby bath products is: "Bathing a cute baby is really so healing!!" Its product category is "maternity and baby products", and the promotion content style is "daily life".

[0038] S13. Generate the category label of the promotion content based on the product category and promotion content style in the original promotion content dataset.

[0039] In this embodiment, extract the product category and the promotion content style from respectively, and generate the category label of the promotion content . The promotion content includes tweets and video ads, etc.

[0040] Specifically, the category label of the promotion content , and use the category label as the category code of the th promotion content.

[0041] S14. Obtain the original marketing account dataset and define the original marketing account dataset , where the original marketing account dataset includes account ID, historical content style of the account, historical promotion traffic conversion rate of the account for different categories of goods, historical promoted product categories of the account, transaction revenue of different categories of goods, and promotion content traffic of different categories of goods.

[0042] Specifically, the operations of obtaining the original promotion content dataset and the original marketing account dataset can include various methods. For example, relevant data can be automatically extracted from the e-commerce platform database through a data scraping tool or an API interface.

[0043] S15. Combine the historical content style and historical promoted product categories of each marketing account extracted from the original marketing account dataset, and judge the matching situation between any marketing account and the current promotion content within the current promotion period.

[0044] Specifically, match each marketing account with the product category and promotion content style of the current promotion content.

[0045] In this embodiment, record the historical content style of the th marketing account collected at the beginning of the th cycle as , the historical promoted product category as , the transaction revenue of different categories of goods as , where the transaction revenue of the th category of goods .

[0046] According to the category label of the th promotion content , the historical content style of the th cycle marketing account and the historical promoted product category of the th cycle marketing account , judge the matching situation and obtain the following four matching results, S151. First matching result: The marketing account has never promoted the product of the current promotion content, and the historical content style of the marketing account is inconsistent with the promotion content style.

[0047] S152. Second matching result: The marketing account has promoted the product of the current promotion content, and the historical content style of the marketing account is consistent with the promotion content style. ​​​

[0048] S153. The third matching result: The marketing account has promoted the product of the current promoted content, but the historical content style of the marketing account is inconsistent with the promoted content style.

[0049] S154. The fourth matching result: The marketing account has never promoted the product of the current promoted content, but the historical content style of the marketing account is consistent with the promoted content style.

[0050] Therefore, the above marketing accounts and the current promoted content can be summarized into the four situations in Table 1 below: Table 1 Matching result corresponding table of the marketing account and the current promoted content ; Among them, the first matching result corresponds to Situation 1, the second matching result corresponds to Situation 2, the third matching result corresponds to Situation 3, and the fourth matching result corresponds to Situation 4.

[0051] S2. Initialize the expected promotion distribution revenue of each promoted content and each marketing account during the current promotion period according to the matching situation, as Figure 2 shown, specifically including the following steps: S21. Initialize the matching degree of the th promoted content and the th marketing account, as well as the historical revenue of the promoted content on the marketing account .

[0052] In this embodiment, the original marketing account dataset also includes the transaction revenue of various categories of products.

[0053] S211. For the first matching result, it can be known that the marketing account has never promoted the product of the content , and the affiliated style is also inconsistent with the promoted content , indicating that the marketing account is completely unfamiliar with the promoted content , that is . Since the category label , therefore can be used to represent the product category of the promoted content , and can be used to represent the promoted content style of the promoted content .

[0054] In order to make the marketing account Attempt to promote the content without incurring too high a cost of trial and error. Initialize the matching degree to a small value. Therefore, set the matching degree to 0.01, that is , is the matching degree.

[0055] At this time, the marketing account has no historical data on the income from promoting the content Therefore, infer the possible income from promoting the content by the ability to promote other products. It is considered that the possible income is the sum of the incomes of all types of products, that is , is the historical income.

[0056] The basis for taking the sum of the incomes of all types of products as the possible income of this account is that the sum of the incomes of all types of products of this account can reflect the overall promotion and sales ability of this account. To balance the cost of trial and error, a smaller weight (matching degree) is assigned to it. If there is still a large speculated income, it means that it can be tried for promotion.

[0057] Therefore, the matching degree and historical income are ; .

[0058] S212. For the second matching result, it can be seen that the marketing account has promoted the products of the content , and the account style is also consistent with the content , that is . Therefore, set the matching degree to 1.

[0059] At this time, since the marketing account has historical data on the income from promoting the content , then let , there is .

[0060] Therefore, the matching degree and historical income are ; .

[0061] S213. For the third matching result, it can be seen that the marketing account has promoted the products of the content , but the account style is inconsistent with the content , that is . Therefore, set the matching degree to 0.5.

[0062] At this time, since the marketing account has historical data on the income from promoting the content For the historical data of revenue, let There is .

[0063] Therefore, the matching degree and historical revenue are ; .

[0064] S214. For the fourth matching result, it can be known that the marketing account has never promoted the products of but the account style is consistent with the content , that is . Therefore, the matching degree is set to 0.5

[0065] At this time, the marketing account has no historical data of revenue from promoted content. Infer the possible revenue of the promoted content through the ability to promote other products, and consider that the possible revenue is the sum of the revenues of all types of products, that is .

[0066] Therefore, the matching degree and historical revenue are ;

[0067] .

[0067] Among them, is the historical content style of the marketing account in the th promotion period, is the historical promoted product category of the marketing account in the th promotion period, is the transaction revenue of the product with the category of .

[0068] Therefore, the initialization process of the above matching degree and historical revenue can be expressed as First, judge whether it belongs to . If so , let , and ; Then judge whether it belongs to . If so , let ; Otherwise, let ; If , let , then judge whether it belongs to , if so, , then let , otherwise let .

[0069] Thus, an initialization scheme for the matching degree and historical revenue is obtained.

[0070] S22. In the th promotion period, the expected revenue from allocating the promotion content to the marketing account is jointly determined by the matching degree and the historical revenue . Based on the matching degree and historical revenue, the expected promotion allocation revenue for allocating the promotion content to the marketing account is initialized, that is, ; wherein, is the expected promotion allocation revenue for the current promotion period .

[0071] By initializing the expected promotion allocation revenue for each promotion content and each marketing account in the above steps, the adaptation degree of different marketing accounts to different promotion contents can be accurately evaluated, providing a scientific basis for subsequent allocation.

[0072] S3. In the current promotion period, based on the expected promotion allocation revenue, through linear programming model solving and random rounding, an allocation scheme for the current promotion period that meets the preset constraint conditions and has the maximum total expected revenue is obtained, as shown in Figure 3 and specifically includes the following steps: S31. Define the slack variable , where . According to the slack variable and the preset constraint conditions, a constraint formula is obtained.

[0073] The constraint conditions include: S311. The first constraint condition is that the total cost of the allocation scheme does not exceed the preset period budget, and the total cost includes the cost of allocating each promotion content to the corresponding marketing account; According to the first constraint condition, the first constraint formula is obtained, that is, ; wherein, is the cost of allocating the promotion content to the marketing account , is the period budget, and can both be adjusted according to the requirements of the promotion task.

[0074] S312. The second constraint condition is that in each promotion period, each piece of promotion content is promoted at least once; According to the second constraint condition, the second constraint formula is obtained, that is, .

[0075] S313. The third constraint condition is that the total number of promotion contents assigned to each marketing account does not exceed the corresponding carrying capacity limit of the marketing account.

[0076] According to the third constraint condition, the third constraint formula is obtained, that is, ; Among them, is the carrying capacity limit of the marketing account .

[0077] S32. Based on the slack variables and the expected promotion distribution revenue, a linear programming model is established. The linear programming model is , Among them, is the total expected revenue.

[0078] S33. Solve the linear programming model according to the first constraint formula, the second constraint formula, and the third constraint formula to obtain the slack variable that maximizes the total expected revenue.

[0079] S34. Perform random rounding on the obtained slack variables to get the rounding result, and generate the allocation plan for the current promotion period according to the rounding result.

[0080] In this embodiment, perform random rounding on the slack variables to obtain , indicates whether to assign the promotion content to the marketing account in the promotion period . Use 0 to represent non-assignment and 1 to represent assignment; Round to 1 with a probability of to obtain , and output the with 1 as the allocation plan for the promotion period.

[0081] In the above steps, defining slack variables and establishing a linear programming model in combination with the constraint formulas helps to accurately quantify the feasibility and revenue potential of promotion tasks. Solving based on the linear programming model and performing random rounding ensures that the generated allocation plan meets the preset constraint conditions while maximizing the total expected revenue, thus achieving the optimal allocation of resources.

[0082] S4. Enter the next promotion cycle and update the original marketing account dataset according to the allocation plan. If the maximum promotion cycle is not reached, obtain the allocation plan for the next promotion cycle based on the updated marketing account dataset; otherwise, end the promotion allocation task, as Figure 4 shown, and specifically includes the following steps: S41. Increment the current promotion cycle by one to make the promotion allocation task enter the next promotion cycle, that is, set t = t + 1; S42. Determine whether the current promotion cycle reaches the preset maximum promotion cycle T, where 1 ≤ t ≤ T.

[0083] S43. If the maximum promotion cycle is reached, that is, t > T, end the promotion allocation task and output the allocation plans for each promotion cycle.

[0084] S44. If the maximum promotion cycle is not reached, execute the promotion allocation task according to the allocation plan and obtain the promotion result of the previous promotion cycle, that is, the promotion result of the promotion cycle in step S3.

[0085] S45. Obtain the updated historical content style, historical promoted product categories, and the transaction revenues of products in each category based on the promotion result. S46. Obtain the matching situation for the current promotion cycle based on the content of the updated marketing accounts. The steps are the same as those in S14 above and will not be elaborated here.

[0086] Initialize the expected promotion allocation revenues of each promotion content and each marketing account for the current promotion cycle according to the matching situation.

[0087] In this embodiment, after obtaining the updated matching situation, update the matching degree and historical revenue in step S21 to obtain the initialized matching degree and historical revenue for the current promotion cycle.

[0088] Then recalculate the expected promotion allocation revenue for the current promotion cycle based on the updated matching degree and historical revenue.

[0089] S47. Obtain the allocation plan for the current promotion cycle based on the updated expected promotion allocation revenue.

[0090] The promotion allocation task terminates until the maximum promotion cycle is reached.

[0091] The above steps continuously optimize the allocation plan when the maximum promotion cycle is not reached, execute tasks according to the allocation plan of the previous promotion cycle, and update the marketing account dataset, dynamically reflecting the promotion effect, supporting the accuracy of the next round of allocation decisions, ensuring that the entire promotion process is always in the best state, and ultimately achieving the maximization of revenue under the budget constraint.

[0092] Based on the same inventive concept described above, an embodiment of the present application also discloses an e-commerce platform promotion task allocation device, as Figure 5 shown, which includes the following modules: The content matching module is configured to, when in the first promotion cycle of the promotion allocation task, obtain the original promotion content dataset and the original marketing account dataset, and judge the matching situation between each promotion content and each marketing account; The expected revenue initialization module is configured to initialize the expected promotion allocation revenue of each promotion content and each marketing account in the current promotion cycle according to the matching situation; The allocation scheme generation module is configured to, in the current promotion cycle, based on the expected promotion allocation revenue, solve through a linear programming model and perform random rounding to obtain an allocation scheme for the current promotion cycle that meets the preset constraint conditions and has the maximum total expected revenue; The promotion update module is configured to enter the next promotion cycle and update the original marketing account dataset according to the allocation scheme. If the maximum promotion cycle is not reached, obtain the allocation scheme for the next promotion cycle based on the updated marketing account dataset, otherwise end the promotion allocation task.

[0093] In a specific feasible implementation, the content matching module includes the following units: The category determination unit is configured to obtain the original promotion content dataset and the original marketing account dataset of the promotion allocation task, and generate a category label for the promotion content , that is, , , wherein, is the product category of the promotion content , and is the promotion content style of the promotion content ; The content matching unit is configured to combine the historical content style and the historical promoted product category of each marketing account extracted from the original marketing account dataset to judge the matching situation between any marketing account and the current promotion content in the current promotion cycle; Specifically, match each marketing account with the product category and promotion content style of the current promotion content to obtain the following four matching results: The first matching result is that the marketing account has never promoted the product of the current promotion content, and the historical content style of the marketing account is inconsistent with the promotion content style; The first matching result is that the marketing account has never promoted the product of the current promotion content, and the historical content style of the marketing account is inconsistent with the promotion content style; The second matching result is that the marketing account has promoted the product of the current promoted content, and the historical content style of the marketing account is consistent with the promoted content style; The third matching result is that the marketing account has promoted the product of the current promoted content, but the historical content style of the marketing account is inconsistent with the promoted content style; The fourth matching result is that the marketing account has never promoted the product of the current promoted content, but the historical content style of the marketing account is consistent with the promoted content style.

[0094] In a specific implementable solution, the expected revenue initialization module includes the following units: The matching degree and historical revenue initialization unit is configured to initialize the matching degree of the th promoted content and the th marketing account according to the matching situation, as well as the historical revenue of the promoted content in the marketing account ; The expected revenue initialization unit is configured to initialize the expected promotion allocation revenue of the promoted content allocated to the marketing account , that is, ; where is the expected promotion allocation revenue for the current promotion period , is the matching degree, is the historical revenue.

[0095] In a specific implementable solution, the matching degree and historical revenue initialization unit includes the following sub-units: The first initialization sub-unit is configured to the original marketing account dataset also includes the transaction revenue of various categories of products, For the first matching result, the matching degree and the historical revenue are ; ; The second initialization sub-unit is configured to for the second matching result, the matching degree and the historical revenue are ; ; The third initialization sub-unit is configured to for the third matching result, the matching degree and the historical revenue are ; ; The fourth initialization subunit is configured such that for the fourth matching result, the matching degree and the historical revenue are ; ; wherein is the historical content style of the marketing account in the th promotion period, and is the historical promoted product category of the marketing account in the th promotion period, and is the transaction revenue of the product with the category of .

[0096] In a specific feasible implementation, the allocation scheme generation module includes the following units: The conditional constraint unit is configured to define a slack variable , wherein , and obtain a constraint formula based on the slack variable and a preset constraint condition; The linear solution unit is configured to establish a linear programming model based on the slack variable and the expected promotion allocation revenue. The linear programming model is , wherein is the total expected revenue. Solve the linear programming model according to the constraint formula to obtain the slack variable that maximizes the total expected revenue; The random rounding unit is configured to perform random rounding on the solved slack variable to obtain a rounding result, and generate an allocation scheme for the current promotion period based on the rounding result.

[0097] In a specific feasible implementation, the conditional constraint unit includes the following subunits: The conditional constraint subunit is configured such that the constraint conditions include: The first constraint condition is that the total cost of the allocation scheme does not exceed the preset period budget, and the total cost includes the cost of allocating each piece of promotion content to the corresponding marketing account; The second constraint condition is that in each promotion period, each piece of promotion content is promoted at least once; The third constraint condition is that the total number of promotion contents allocated to each marketing account does not exceed the corresponding marketing account's carrying capacity.

[0098] In a specific feasible implementation, the promotion update module includes the following units: The maximum period judgment unit is configured to increment the current promotion period by one, causing the promotion allocation task to enter the next promotion period, and judge whether the current promotion period has reached the preset maximum promotion period; If the maximum promotion period is reached, end the promotion allocation task and output the allocation scheme for each promotion period; A periodic update unit, configured to, if the maximum promotion period is not reached, execute the promotion allocation task according to the allocation scheme and obtain the promotion results of the previous promotion period, obtain the updated historical content style, historical promoted product categories, and the transaction revenues of products in each category based on the promotion results, obtain the matching situation of the current promotion period, initialize the expected promotion allocation revenues of each promotion content and each marketing account for the current promotion period according to the matching situation, and obtain the allocation scheme until the maximum promotion period is reached.

[0099] Based on the same inventive concept as above, an embodiment of the present application also discloses a computer-readable storage medium, in which at least one instruction, at least one program, a code set or an instruction set is stored, and the at least one instruction, at least one program, the code set or the instruction set can be loaded and executed by a processor to implement the e-commerce platform promotion task allocation method provided in the above method embodiment.

[0100] Also based on the same inventive concept as above, an embodiment of the present application also discloses a computer-readable storage medium, in which at least one instruction, at least one program, a code set or an instruction set is stored, and the at least one instruction, at least one program, the code set or the instruction set is loaded and executed by a processor to implement the e-commerce platform promotion task allocation method as described above.

[0101] Those of ordinary skill in the art can understand that all or part of the steps for implementing the above embodiments can be completed by hardware, or can be completed by a program instructing relevant hardware. The program can be stored in the computer-readable storage medium. The computer-readable storage medium includes, for example: various media that can store program codes such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs.

[0102] The above are only optional embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for allocating promotion tasks on an e-commerce platform, characterized in that, Including the following steps: When in the first promotion cycle of the promotion assignment task, obtain the original promotion content dataset and the original marketing account dataset, and judge the matching situation between each promotion content and each marketing account; Initialize the expected promotion assignment revenue of each promotion content and each marketing account in the current promotion cycle according to the matching situation; In the current promotion cycle, based on the expected promotion assignment revenue, solve through a linear programming model and perform random rounding to obtain an assignment plan for the current promotion cycle that meets the preset constraint conditions and has the maximum total expected revenue; Enter the next promotion cycle and update the original marketing account dataset according to the assignment plan. If the maximum promotion cycle is not reached, obtain the assignment plan for the next promotion cycle based on the updated marketing account dataset, otherwise end the promotion assignment task.

2. The method for allocating promotion tasks of an e-commerce platform according to claim 1, wherein The obtaining of the original promotion content dataset and the original marketing account dataset, and the judgment of the matching situation between each promotion content and each marketing account specifically include the following steps: Obtain the original promotion content dataset and the original marketing account dataset for the promotion assignment task, and generate promotion content based on the product categories and promotion content styles in the original promotion content dataset of the category label , that is ; Among them, is the product category of the promotion content , is the promotion content style of the promotion content . Based on the historical content style and historical promoted product categories of each marketing account extracted from the original marketing account dataset, determine, during the current promotion period for any of the marketing accounts the matching situation with the current promoted content ; Specifically, match each marketing account with the product category and the promotion content style of the current promotion content to obtain the following four matching results: The first matching result: The marketing account has never promoted the product of the current promotion content, and the historical content style of the marketing account is inconsistent with the promotion content style; The second matching result: The marketing account has promoted the product of the current promotion content, and the historical content style of the marketing account is consistent with the promotion content style; The third matching result: The marketing account has promoted the product of the current promotion content, but the historical content style of the marketing account is inconsistent with the promotion content style; The fourth matching result: The marketing account has never promoted the product of the current promotion content, but the historical content style of the marketing account is consistent with the promotion content style.

3. The e-commerce platform promotion task allocation method according to claim 2, wherein The initialization of the expected promotion assignment revenue of each promotion content and each marketing account in the current promotion cycle according to the matching situation specifically includes the following steps: Initialize the matching degree of the th piece of the promotion content and the th marketing account according to the matching situation, and the historical revenue of the promotion content on the marketing account; Initialize the expected promotion allocation revenue for the promoted content according to the matching degree and the historical revenue assigned to the marketing account , that is ; Among them, is the expected promotion allocation revenue for the current promotion period , is the matching degree, and is the historical revenue.

4. The e-commerce platform promotion task assignment method according to claim 3, characterized in that: The original marketing account dataset also includes the transaction revenue of various categories of products; For the first matching result, the matching degree and the historical revenue are: ； ； For the second matching result, the matching degree and the historical revenue are: ； ； For the third matching result, the matching degree and the historical revenue are: ； ； For the fourth matching result, the matching degree and the historical revenue are: ； ； Among them, is the historical content style of the marketing account in the th promotion period, is the historical promoted product category of the marketing account in the th promotion period, is the transaction revenue of the product with the category of .

5. The e-commerce platform promotion task allocation method according to claim 3, wherein The obtaining of the assignment plan for the current promotion cycle that meets the preset constraint conditions and has the maximum total expected revenue by solving through a linear programming model and performing random rounding based on the expected promotion assignment revenue specifically includes the following steps: Define slack variables , where , and obtain a constraint formula based on the slack variables and preset constraint conditions; Establish a linear programming model based on the slack variable and the expected promotion assignment revenue. The linear programming model is: ， Among them, is the total expected return; Solving the linear programming model according to the constraint formula to obtain the slack variable that maximizes the total expected return ; Perform random rounding on the obtained slack variable to obtain a rounding result, and generate an assignment plan for the current promotion cycle based on the rounding result.

6. The method for allocating promotion tasks of an e-commerce platform according to claim 5, wherein The constraint conditions include: The first constraint condition is that the total cost of the allocation scheme does not exceed a preset cycle budget, and the total cost includes the cost of allocating each piece of the promoted content to the corresponding marketing account; The second constraint condition is that within each promotion cycle, each piece of the promoted content is promoted at least once; The third constraint condition is that the total number of the promoted content allocated to each marketing account does not exceed the corresponding carrying limit of the marketing account.

7. The method for allocating promotion tasks on an e-commerce platform according to claim 2, wherein Enter the next promotion cycle and update the original marketing account data set according to the allocation scheme. If the maximum promotion cycle is not reached, obtain the allocation scheme for the next promotion cycle based on the updated marketing account data set. Otherwise, end the promotion allocation task. Specifically, it includes the following steps: Increment the current promotion cycle by one to make the promotion allocation task enter the next promotion cycle, and determine whether the current promotion cycle reaches the preset maximum promotion cycle; If the maximum promotion cycle is reached, end the promotion allocation task and output the allocation schemes for each promotion cycle; If the maximum promotion cycle is not reached, execute the promotion allocation task according to the allocation scheme and obtain the promotion result of the previous promotion cycle. Obtain the updated historical content style, historical promoted product categories, and the transaction revenues of products in each category based on the promotion result, and obtain the matching situation of the current promotion cycle. Initialize the expected promotion allocation revenues of each piece of the promoted content and each marketing account in the current promotion cycle according to the matching situation to obtain the allocation scheme until the maximum promotion cycle is reached.

8. An e-commerce platform promotion task allocation device, characterized in that It includes the following modules: The content matching module is configured to, when in the first promotion cycle of the promotion allocation task, obtain the original promoted content data set and the original marketing account data set, and determine the matching situation between each piece of the promoted content and each marketing account; The expected revenue initialization module is configured to initialize the expected promotion allocation revenues of each piece of the promoted content and each marketing account in the current promotion cycle according to the matching situation; The allocation scheme generation module is configured to, in the current promotion cycle, based on the expected promotion allocation revenues, solve through a linear programming model and perform random rounding to obtain the allocation scheme for the current promotion cycle that meets the preset constraint conditions and has the maximum total expected revenue; The promotion update module is configured to enter the next promotion cycle and update the original marketing account data set according to the allocation scheme. If the maximum promotion cycle is not reached, obtain the allocation scheme for the next promotion cycle based on the updated marketing account data set. Otherwise, end the promotion allocation task.

9. An intelligent terminal, characterized in that, It includes a memory and a processor. At least one instruction, at least one program, a code set, or an instruction set is stored in the memory, and the at least one instruction, at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the e-commerce platform promotion task allocation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the e-commerce platform promotion task allocation method according to any one of claims 1 to 7.

Images