Task issuing method and device for active driver, electronic equipment and medium

Abstract

The invention provides a task issuing method and device for active drivers, electronic equipment and a medium, and the method comprises the steps: predicting the probability that each active driver is paired with different odd numbers under different order average reward amounts through employing a trained multi-classification model; obtaining a pairing odd number expectation of each active driver under different order average reward amounts; according to the pairing order number expectation, the subsidy cost corresponding to different order average reward amounts and the subsidy budget, obtaining the order average reward amount of each active driver; and according to the average reward amount of each active driver and the paired odd number expectation under the average reward amount of each active driver, obtaining a task of each active driver and issuing the task. According to the method, an appropriate task is provided for each active driver through reward differentiation and task differentiation, so that the order receiving number of the active drivers is maximized as much as possible under the effective subsidy budget, and the optimal task incentive effect is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of online car-hailing, and in particular to a method and device for issuing tasks to active drivers, electronic equipment, and media. Background Art

[0002] In the era of mobile Internet, modern travel for users has changed the traditional way of taking taxis. By utilizing the characteristics of mobile Internet and integrating online and offline, the traditional way of drivers waiting for passengers has been changed. Drivers can accept orders based on the passengers' wishes according to their destinations, saving the communication costs between drivers and passengers, reducing the empty driving rate, and maximizing the savings of resources and time for both drivers and passengers.

[0003] In the prior art, the platform often uses incentives to increase the enthusiasm of drivers to accept orders. These incentives generally include the following: First, rewards for completing orders, that is, after the driver completes the order, the corresponding reward is issued according to the route and time of the order; second, when the passenger sends an order request, according to certain rules, the order request is added with reward information, and then the order request and reward information are pushed to the driver's client. The incentive measures of these schemes are fixed and not customized for drivers, resulting in poor incentive effects for the task. Summary of the invention

[0004] In view of the above defects in the prior art, the present invention provides a method and device for issuing tasks for active drivers, an electronic device, and a medium to solve the technical problems of single task rewards and poor task incentive effects.

[0005] To achieve the above-mentioned purpose and other related purposes, the present invention provides a method for publishing tasks for active drivers, comprising: using a trained multi-classification model to predict the probability of each active driver pairing with different numbers of orders under different average reward amounts per order; obtaining the expected number of pairing orders for each active driver under different average reward amounts per order based on the probability of each active driver pairing with different numbers of orders under different average reward amounts per order; obtaining the average reward amount for each active driver based on the expected number of pairing orders under different average reward amounts, the subsidy costs corresponding to the different average reward amounts, and the subsidy budget; obtaining and publishing the task for each active driver based on the average reward amount for each active driver and the expected number of pairing orders under the average reward amount.

[0006] In one embodiment of the present invention, the input features of the multi-classification model include characteristics of active drivers and average reward amount per order, and the characteristics of active drivers include attribute information, historical data, and tracking point information of active drivers.

[0007] In one embodiment of the present invention, the label of the multi-classification model is an odd-number interval, and the historical average daily matching odd numbers of all active drivers are aggregated to obtain the odd-number range corresponding to each of the odd-number intervals; using the trained multi-classification model, the probability of each of the active drivers matching with different odd numbers under different average reward amounts is predicted, including: using the trained multi-classification model, predicting the probability of each of the active drivers hitting different odd-number intervals under different average reward amounts.

[0008] In one embodiment of the present invention, based on the probability of each active driver pairing different numbers of orders under different average order reward amounts, the expected number of pairing orders for each active driver under different average order reward amounts is obtained, including: based on the probability of each active driver hitting different number intervals under different average order reward amounts, obtaining the probability that each active driver pairs at least K orders under different average order reward amounts; based on the probability that each active driver pairs at least K orders under different average order reward amounts, calculating the lower bound of the expected number of pairing orders for each active driver under different average order reward amounts, so as to obtain the expected number of pairing orders for each active driver under different average order reward amounts.

[0009] In one embodiment of the present invention, the average reward amount per order for each active driver is obtained based on the expected number of pairing orders for each active driver at different average reward amounts per order, the subsidy costs corresponding to different average reward amounts per order, and the subsidy budget, including: obtaining the input parameters of the optimization problem based on the expected number of pairing orders for each active driver at different average reward amounts per order and the subsidy costs corresponding to different average reward amounts per order; obtaining the constraints of the optimization problem based on the subsidy budget and the fact that each active driver has only one average reward amount per order; constructing and solving the optimization problem based on the input parameters and constraints of the optimization problem and taking maximizing the number of pairing orders for active drivers as the objective function, to obtain the Boolean variables of each active driver at different average reward amounts per order; obtaining the average reward amount per order for each active driver based on the Boolean variables of each active driver at different average reward amounts per order.

[0010] In one embodiment of the present invention, after the step of obtaining the expected number of pairing orders for each active driver at different average reward amounts based on the probability of pairing different numbers of orders for each active driver at different average reward amounts, the step also includes: classifying all active drivers to obtain several categories of active drivers; obtaining the constraints of the optimization problem based on the subsidy budget and the fact that each active driver has only one average reward amount, including: obtaining the constraints of the optimization problem based on the subsidy budget, the fact that each active driver has only one average reward amount, and the active drivers belonging to the same category have the same average reward amount.

[0011] In one embodiment of the present invention, all active drivers are classified to obtain several categories of active drivers, including: obtaining the expected number of natural pairing odds of the active drivers based on the expected number of pairing odds when the average reward amount per odd is 0; performing grid division according to the vehicle model of the active drivers and the expected number of natural pairing odds to obtain multiple grids; and classifying each of the active drivers into a corresponding grid, and the active drivers belonging to the same grid are regarded as the same category of active drivers.

[0012] In one embodiment of the present invention, the task of each active driver is obtained and published according to the average reward amount per order of each active driver and the expected number of pairing orders under the average reward amount per order, including: obtaining the threshold and reward amount of the first-level sub-tasks according to the average reward amount per order of each active driver and the expected number of pairing orders under the average reward amount per order; obtaining the reward amount of the subsequent sub-tasks according to the reward amount of the parent task and the reward amount of the first-level sub-task, and obtaining the threshold of the subsequent sub-tasks according to the preset threshold of the subsequent task; obtaining the parent task of each active driver according to the threshold and reward amount of the first-level sub-task and the threshold and reward amount of the subsequent sub-task and publishing it.

[0013] In one embodiment of the present invention, the subsequent subtasks only include second-level subtasks; the reward amount of the subsequent subtask is obtained based on the reward amount of the parent task and the reward amount of the first-level subtask, and the threshold of the subsequent subtask is obtained based on a preset subsequent task threshold, including: obtaining the reward amount of the second-level subtask based on the reward amount of the parent task and the reward amount of the first-level subtask; obtaining the threshold of the second-level subtask based on the historical average daily pairing number of active drivers.

[0014] In one embodiment of the present invention, the reward amount of the second-level sub-task is obtained according to the reward amount of the parent task and the reward amount of the first-level sub-task, including: obtaining the reward amount of the second-level sub-task according to the reward amount of the parent task, the reward amount of the first-level sub-task, whether the active driver has a task dispatch record at the preset time, whether the active driver has a receipt record at the preset time, and a preset random increase amount.

[0015] To achieve the above-mentioned purpose and other related purposes, the present invention also provides a task publishing device for active drivers, comprising: a first calculation unit, used to predict the probability of each active driver pairing with different numbers of orders under different average reward amounts per order by using a trained multi-classification model; a second calculation unit, used to obtain the expected number of pairing orders for each active driver under different average reward amounts per order based on the probability of each active driver pairing with different numbers of orders under different average reward amounts per order; a third calculation unit, used to obtain the average reward amount per order for each active driver based on the expected number of pairing orders for each active driver under different average reward amounts per order, the subsidy costs corresponding to different average reward amounts, and the subsidy budget; a task processing unit, used to obtain and publish the task for each active driver based on the average reward amount per order for each active driver and the expected number of pairing orders under the average reward amount per order.

[0016] To achieve the above-mentioned purpose and other related purposes, the present invention also provides an electronic device, including a processor, a memory and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute a computer program stored in the memory to implement a method provided in any one of the above embodiments.

[0017] To achieve the above-mentioned object and other related objects, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is used to enable a computer to execute the method provided in any one of the above-mentioned embodiments.

[0018] Beneficial effects of the present invention: The present invention proposes a method and device for issuing tasks for active drivers, an electronic device, and a medium. The method predicts the probability of an active driver pairing with different numbers of orders under different average order reward amounts through a multi-classification model, thereby obtaining an expected number of paired orders, and calculates the average order reward amount for each active driver based on the expected number of paired orders. Through reward differentiation and task differentiation, a suitable task is provided for each active driver, thereby maximizing the number of orders accepted by the active driver as much as possible under an effective subsidy budget, thereby achieving the best task incentive effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 A flowchart of a first task publishing method provided by an embodiment of the present invention;

[0021] Figure 2 A multi-classification model training flow chart provided in one embodiment of the present invention;

[0022] Figure 3 A flowchart of the calculation of the expected number of pairs provided by an embodiment of the present invention;

[0023] Figure 4 A flowchart for calculating the average reward amount for active drivers according to an embodiment of the present invention;

[0024] Figure 5 A flowchart of a second task publishing method provided by an embodiment of the present invention;

[0025] Figure 6 Flow chart of active driver classification provided by one embodiment of the present invention

[0026] Figure 7 A multi-layer task configuration flow chart provided for an embodiment of the present invention;

[0027] Figure 8 A flowchart of the second layer subtask configuration provided by an embodiment of the present invention;

[0028] Fig. 9 A flowchart for obtaining the threshold and reward amount of the first-level subtask provided by an embodiment of the present invention;

[0029] Fig.10 A flowchart for obtaining the threshold and reward amount of the second-level subtask provided in one embodiment of the present invention;

[0030] Fig.11 A schematic diagram of a task issuing device provided by an embodiment of the present invention;

[0031] Fig.12 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention.

[0032] Explanation of the reference numerals: 101, first computing unit; 102, second computing unit; 103, third computing unit; 104, task processing unit; 201, processor; 202, memory. DETAILED DESCRIPTION

[0033] The following describes the embodiments of the present invention through specific specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. It should be noted that the following embodiments and the features in the embodiments can be combined with each other without conflict. In addition to the specific methods, equipment, and materials used in the embodiments, according to the technical personnel in this technical field's grasp of the prior art and the records of the present invention, any methods, equipment, and materials of the prior art that are similar or equivalent to the methods, equipment, and materials described in the embodiments of the present invention can also be used to implement the present invention.

[0034] It should be understood that the terms used in the embodiments of the present invention are for describing specific embodiments rather than for limiting the scope of protection of the present invention. Unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those generally understood by those skilled in the art.

[0035] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In some of the embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0036] The flowchart and block diagram in the accompanying drawings illustrate the system architecture, functions and operations that may be implemented according to the methods and computer program products of various embodiments disclosed in the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment, or a part of a code, and the module, the program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0037] See also Figure 1 , Figure 1 A method for issuing tasks for active drivers provided in one embodiment of the present invention includes steps S100, S200, S400 and S500.

[0038] Step S100, using the trained multi-classification model, predict the probability of each active driver being paired with different numbers of orders under different average order reward amounts. In this step, the probability of pairing different numbers of orders is predicted by the trained multi-classification model, so that the expected number of paired orders can be conveniently calculated based on this probability. The multi-classification model is a type of model used to deal with multi-classification problems in machine learning. In multi-classification problems, the goal is to assign the input data to one of three or more categories based on its characteristics. In this step, the category of the multi-classification model corresponds to the number of orders, such as 0 orders, 1 order, 2 orders, etc. The multi-classification model can be, for example, multinomial logistic regression, decision tree, random forest, support vector machine, etc.

[0039] In a specific embodiment of the present invention, the input features of the multi-classification model include the features of active drivers and the average reward amount per order. The features of active drivers include the attribute information, historical data, and tracking information of active drivers. Among them, the attribute information of active drivers can be, for example, the driver's registered city, driver's gender, vehicle type, etc.; the driver's historical data can be, for example, the number of orders received and the number of completed orders within a certain number of days; the driver's tracking information can be, for example, the number of sliding orders and refreshing orders within a certain number of days.

[0040] See also Figure 2 In a specific embodiment of the present invention, the label of the multi-classification model is an odd-number interval, and the historical daily average matching odd numbers of all active drivers are aggregated to obtain the odd-number range corresponding to each odd-number interval. For example, the market data of the most recent year can be counted, and the daily average matching odd numbers of all active drivers can be aggregated to ensure the balance of the number of drivers in each odd-number interval. For example, 0 orders are the first odd-number interval, 1 to 2 orders are the second odd-number interval, and 3 orders and above are the third odd-number interval, so that multiple odd-number intervals can be obtained. At this time, step S100 includes: using the trained multi-classification model to predict the probability of each active driver hitting different odd-number intervals under different average reward amounts. In this embodiment, each order is no longer regarded as a category of a multi-classification model, but the odd-number interval is used as a classification category, which can effectively reduce the number of classification categories and reduce the complexity of the multi-classification model.

[0041] In the specific implementation, the multi-classification model is initialized first, and then the multi-classification model is trained according to the historical active driver task data to obtain a trained multi-classification model. The model only needs to be trained once. Then, the characteristics of each active driver and a certain average reward amount are input into the trained multi-classification model, and the probability of the active driver hitting different odd intervals under the average reward amount can be obtained. It should be noted that the multi-classification model can only predict the probability of a certain active driver hitting different odd intervals under a certain average reward amount at a time. If you want to obtain the probability of each active driver hitting different odd intervals under different average reward amounts, multiple predictions are required. Assuming that there are N active drivers and the average reward amount has M levels, for example, the reward amount is 0, 1, 2, ..., M-1 yuan, then after performing N×M predictions in step S100, N×M probabilities of hitting different odd intervals can be obtained. The probability of driver i hitting the kth odd interval under the average reward amount j can be recorded as P ijk , where i=1,2,…,N and j=0,1,2,…,M-1.

[0042] It should be noted that the average reward amount does not necessarily have to be continuous or at equal intervals. It can also be 0, 5, 10, 15 yuan, or 0, 1, 5, 10 yuan. The specific average reward amount can be set according to the actual activity situation.

[0043] Step S200: According to the probability of each active driver matching different numbers of orders under different average order reward amounts, the expected number of matching orders for each active driver under different average order reward amounts is obtained. Since the goal of task release is to increase the number of matching orders as much as possible within the budget, this step needs to solve the expected number of matching orders for each active driver under different average order reward amounts.

[0044] See also Figure 3 In a specific embodiment of the present invention, step S200 includes step S201 and step S202.

[0045] Step S201: According to the probability of each active driver hitting different odd number intervals under different average order reward amounts, the probability of each active driver matching at least K orders under different average order reward amounts is obtained. Taking the above three odd number intervals as an example, let the probability of driver i hitting the three odd number intervals under the average order reward amount j be p, p, and p, respectively. ij1 、p ij2 、p ij3 , then the probability of at least 0 pairings is p ij1 +p ij2 +p ij3 , the probability of at least one pairing is p ij2 +p ij3, the probability of at least 3 pairs is p ij3 .

[0046] Step S202: Calculate the lower bound of the expected number of paired orders for each active driver at different average reward amounts based on the probability that each active driver will match at least K orders under different average reward amounts, so as to obtain the expected number of paired orders for each active driver at different average reward amounts. Specifically, the probability q of each active driver matching at least K orders under different average reward amounts is calculated. ij , according to formula c ij =max(K*q ij ) Calculate the lower bound c of the pairing odd-number expectation of driver i under the average reward amount j ij Here, the “infimum” refers to the minimum possible value of the expected value of the number of paired pairs of driver i under a given average reward amount j. That is, no matter how the driver works, his expected value of the number of paired pairs will not be lower than this infimum. ij That is, the paired singular expectation of this step.

[0047] Step S400: Obtain the average reward amount per order for each active driver based on the expected number of paired orders under different average reward amounts per order, the subsidy cost corresponding to different average reward amounts per order, and the subsidy budget. In this step, the specific amount of the subsidy cost corresponding to different average reward amounts per order has nothing to do with the driver, and is generally only related to the average reward amount per order. For a certain taxi-hailing platform, it is a known value and can be recorded as w j The subsidy budget corresponds to the total amount of the proposed incentives, which can be recorded as C. This step is based on c ij 、w j And C assigns a different average reward amount per order to each active driver to achieve the best incentive effect.

[0048] See also Figure 4 In a specific embodiment of the present invention, step S400 includes steps S401 to S404.

[0049] Step S401: According to the expected number of paired orders for each active driver under different average reward amounts and the subsidy costs corresponding to different average reward amounts, the input parameters of the optimization problem are obtained. ij 、w j As input parameters of the optimization problem, they are all known quantities.

[0050] Step S402: According to the subsidy budget, each active driver has only one average reward amount per order, and the constraints of the optimization problem are obtained. In order to better define the optimization problem and solve it, the decision variables of the optimization problem can be defined as: the average reward amount per order of driver i is a Boolean variable j, denoted as x ij , where xij Either 0 or 1, this is what the optimization problem needs to solve. With the definition of decision variables, these two constraints can be expressed by mathematical formulas. The sum of subsidies for all active drivers must be less than or equal to the subsidy budget, that is, ∑ ij (c ij *x ij x j )≤C, each active driver has only one reward amount, that is, ∑ j x ij =1.

[0051] Step S403: Based on the input parameters and constraints of the optimization problem, the optimization problem is constructed and solved with the objective function of maximizing the number of active driver pairings, and the Boolean variables of each active driver under different average reward amounts are obtained. The objective function is to maximize the number of pairings, which is expressed in mathematical language as maximizing the expected sum of the number of pairings of drivers, which can be expressed as: max[∑ ij (c ij *x ij )]. For any active driver, after solving the optimization problem, M Boolean variables can be obtained, and only one Boolean variable corresponding to the average reward amount is 1, and the other Boolean variables corresponding to the average reward amount are all 0. For example, for the first active driver, the solution of the decision variable is: x 10 =0,x 11 =0,x 12 =1,x 13 =0, ... The result is a Boolean variable for the first active driver at different reward amounts.

[0052] Step S404: Obtain the average reward amount for each active driver based on the Boolean variables of each active driver under different average reward amounts. 12 is 1, and the others are 0, indicating that the task reward amount for the first active driver corresponds to the third-tier reward, for example, 2 yuan.

[0053] See also Figure 5In a specific embodiment of the present invention, in the above steps S401 to S404, when the optimization problem is defined and solved, it is necessary to solve the Boolean variables of each active driver under different reward amounts, that is, it is necessary to solve N×M decision variables. A large number of decision variables will affect the speed of solving. In this embodiment, in order to speed up the solution, step S200 also includes step S300: classify all active drivers to obtain several categories of active drivers; and when constructing constraints, let the average reward amount for active drivers belonging to the same category be the same. For example, there are 10,000 active drivers and 5 types of average reward amounts. Through classification, the 10,000 active drivers are divided into 16 categories, then the decision variables change from the initial 10,000×5 to 16×5, which greatly reduces the amount of calculation.

[0054] See also Figure 6 In a specific embodiment of the present invention, step S300 includes steps S301 to S303.

[0055] Step S301: Obtain the expected number of natural pairings of active drivers based on the expected number of pairings when the average reward amount per order is 0. According to the expected number of pairings of each active driver at different average reward amounts in step S200, the expected number of pairings when the average reward amount per order is 0 is found, and the expected number of natural pairings of each active driver can be obtained. The expected number of natural pairings reflects the expected number of pairings of the active driver when there is no reward, which can be used as a consideration for classifying active drivers.

[0056] Step S302: Grid division is performed according to the vehicle type and the natural matching odd number expectation of the active driver to obtain a plurality of grids. The vehicle type of the active driver may be, for example, economy type, ordinary type, luxury type, or other vehicle type classifications. In this step, the vehicle type and the natural matching odd number expectation are used as the basis for classification, so that the active drivers can be easily classified (also referred to as grouping). In addition, other classification considerations may also be added.

[0057] Step S303, classify each active driver into a corresponding grid, and active drivers belonging to the same grid are regarded as the same type of active drivers. Different active drivers have different sensitivities to the average reward amount, so an average reward amount can be set for each active driver, but in fact, there are a large number of active drivers, and many driver groups are similar. By dividing the active drivers into grids based on features such as vehicle type and natural pairing odd number expectation, the active driver group can be divided into subsets with similar features. The purpose of this is to make the driver's behavior and response to rewards more consistent in each subset, so that the solution to the optimization problem is more effective and accurate.

[0058] For example, in a specific embodiment of the present invention, the natural pairing odd number expectations are divided into 4 buckets (bucketing is a method of data binning or segmentation, which is a common data preprocessing technology used to divide the values ​​of continuous variables into a finite number of discrete intervals or "buckets". The purpose of this is to simplify the complexity of the model and improve the interpretability of the model), and taking the above-mentioned economy, ordinary, and luxury models as examples, all active drivers can be divided into 3×4 types, a total of 12 combinations (or categories).

[0059] Step S500: According to the average reward amount per order of each active driver and the expected number of paired orders under the average reward amount per order, the task of each active driver is obtained and released. By solving the above optimization problem, the final average reward amount per order j of driver i and the corresponding expected number of paired orders can be obtained. The expected number of paired orders can be used as the task threshold, the product of the average reward amount per order and the expected number of paired orders as the task reward amount. The task reward amount and threshold of each driver (each type of driver) are customized, so the task release effect is more reasonable and efficient.

[0060] When issuing tasks, a single task reward may not attract the interest of active drivers, or active drivers may continue to sleep after completing tasks just for the reward. In this case, a more complex task system can be introduced. For example, a ladder task can be used. The main form of a ladder task is: the outer layer is the parent task, and the inner layer is composed of multiple subtasks. The subtasks need to set thresholds and reward amounts respectively. There is a sequence between subtasks. After completing the previous subtask, you can enter the next subtask. The reward amounts between subtasks are issued independently. This will achieve a better incentive effect.

[0061] See also Figure 7 In a specific embodiment of the present invention, step S500 includes steps S510 to S530.

[0062] Step S510: According to the average reward amount per order of each active driver and the expected number of paired orders under the average reward amount per order, the threshold and reward amount of the first-level subtask are obtained. The expected number of paired orders can be used as the threshold of the first-level subtask, and the product of the average reward amount per order and the expected number of paired orders can be used as the reward amount of the first-level subtask.

[0063] Step S520: According to the parent task reward amount and the first-level subtask reward amount, the subsequent subtask reward amount is obtained, and according to the preset subsequent task threshold, the subsequent subtask threshold is obtained. The subsequent subtask can have multiple layers and be set as needed. The reward amount of the subsequent subtask can be a fixed value, or an overall parent task reward amount can be preset for each active driver, and then the reward amount of the subsequent subtask is calculated based on the parent task reward amount and the first-level subtask reward amount.

[0064] See also Figure 8 In a specific embodiment of the present invention, the subsequent subtasks only include the second-level subtasks; in this case, step S520 may include: S521, according to the parent task reward amount and the first-level subtask reward amount, obtain the second-level subtask reward amount; S522, according to the historical average daily matching number of active drivers, obtain the second-level subtask threshold. For example, the second-level subtask reward amount can be directly equal to the parent task reward amount minus the first-level subtask reward amount, and the second-level subtask threshold is equal to the historical average daily matching number of active drivers plus 1.

[0065] In a specific embodiment of the present invention, step S521 includes: obtaining the reward amount of the second-level sub-task based on the reward amount of the parent task, the reward amount of the first-level sub-task, whether the active driver has a task dispatch record at the preset time, whether the active driver has a receipt record at the preset time, and a preset random increase amount.

[0066] For example, the reward amount of the parent task minus the reward amount of the first-level subtask can be used as the default value of the reward amount of the second-level subtask.

[0067] If the active driver has no dispatch record on T-1 day (the day before the task is released) (the task has not been dispatched to the active driver), then when the task is released on T day, the actual payment value of the reward amount of the second-level subtask will be equal to the default value of the reward amount of the second-level subtask.

[0068] If the active driver has a distribution record on T-1 day, but no claim record, it means that the active driver has seen the task but has not claimed it. In this case, the reward amount can be increased, and the actual distribution value of the second-level subtask reward amount is equal to the default value of the second-level subtask reward amount plus a to b yuan (0 < a < b). This can further increase the enthusiasm of active drivers to claim tasks.

[0069] If the active driver has a distribution record and a receipt record on T-1, it means that the active driver has seen and received the task. At this time, the task can continue to be distributed, but the reward amount needs to be reduced. The actual distribution value of the reward amount of the second-level subtask is equal to the default value of the reward amount of the second-level subtask minus a to b yuan (0<a<b) at random.

[0070] It can be understood that in addition to the reward amount of the second-level sub-tasks being adjustable according to certain rules, the reward amount of the first-level sub-tasks can also be fine-tuned according to certain rules, and the threshold of the sub-tasks can also be fine-tuned according to certain rules to increase the enthusiasm of active drivers to receive task rewards.

[0071] Step S530: According to the threshold and reward amount of the first-level subtasks, the threshold and reward amount of the subsequent subtasks, the parent task of each active driver is obtained and published. The threshold and reward amount of all subtasks are combined to obtain the parent task.

[0072] It should be noted that when there are subsequent subtasks, the reward amount of the first-level subtask is calculated based on the proportion when calculating the optimization problem. Assuming that the total subsidy budget is C, and the reward amount of the first-level subtask accounts for 70% of the total subsidy budget, then the formula ∑ ij (c ij *x ij *w j )≤C should be adjusted to ∑ ij (c ij *x ij *w j )≤0.7C. In addition, when fine-tuning the reward amount of a subtask, the fine-tuning amount also needs to be considered to avoid exceeding the total budget.

[0073] It should be noted that the steps of the above methods are divided only for the purpose of clear description. When implementing, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they contain the same logical relationship, they are all within the protection scope of this patent. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, but not changing the core design of the algorithm and process, are all within the protection scope of this patent. For example, Fig. 9 It shows a flowchart of obtaining the threshold and reward amount of the first-level subtask in another execution order. Fig.10 It shows a flowchart of obtaining the threshold and reward amount of the second-level subtask in another execution order.

[0074] See also Fig.11 , Fig.11A task publishing device for active drivers provided in one embodiment of the present invention includes a first computing unit 101, a second computing unit 102, a third computing unit 103 and a task processing unit 104, wherein the first computing unit 101 is used to use a trained multi-classification model to predict the probability of each active driver pairing with different numbers of orders under different average order reward amounts; the second computing unit 102 is used to obtain the expected number of pairing orders for each active driver under different average order reward amounts based on the probability of each active driver pairing with different numbers of orders under different average order reward amounts; the third computing unit 103 is used to obtain the average order reward amount for each active driver based on the expected number of pairing orders for each active driver under different average order reward amounts, the subsidy cost corresponding to different average order reward amounts, and the subsidy budget; the task processing unit 104 is used to obtain and publish the task of each active driver based on the average order reward amount of each active driver and the expected number of pairing orders under the average order reward amount.

[0075] It should be noted that the above-mentioned functional modules can be fully or partially integrated into one physical entity during actual implementation, or they can be physically separated. And these modules can all be implemented in the form of software called by processing elements; they can also be all implemented in the form of hardware; some modules can also be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware. In addition, these modules can be fully or partially integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, some or all of the steps of the above method, or the above functional modules can be completed by hardware integrated logic circuits in the processor element or instructions in software form.

[0076] It should be noted that the task issuing device of this embodiment is a device corresponding to the above-mentioned task issuing method, and the functional modules in the task issuing device may correspond to the corresponding steps in the task issuing method. The task issuing device of this embodiment can be implemented in conjunction with the task issuing method, that is, in the absence of conflict, the relevant technical details mentioned in the task issuing method of the above-mentioned embodiment can also be applied to the task issuing device of this embodiment.

[0077] See also Fig.12 , Fig.12 An electronic device provided for one embodiment of the present invention includes a processor 201, a memory 202 and a communication bus; the communication bus is used to connect the processor 201 and the memory 202; the processor 201 is used to execute a computer program stored in the memory 202 to implement the above-mentioned active driver task publishing method.

[0078] The above-mentioned processor can be, for example, a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components; the above-mentioned memory may include random access memory (RAM), and may also include non-volatile memory (Non-volatile Memory), such as at least one disk storage.

[0079] The above-mentioned electronic device is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0080] The electronic device mentioned above can be any electronic product that can perform human-computer interaction with a user, such as a personal computer, a tablet computer, a smart phone, a personal digital assistant (PDA), a game console, an interactive network television (IPTV), a smart wearable device, etc.

[0081] The electronic device may also include a network device and / or a user device, wherein the network device includes, but is not limited to, a single network server, a server group consisting of multiple network servers, or a cloud consisting of a large number of hosts or network servers based on cloud computing.

[0082] The network where the above electronic devices are located includes but is not limited to the Internet, wide area network, metropolitan area network, local area network, virtual private network (VPN), etc.

[0083] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and the computer program is used to enable a computer to execute the above-mentioned active driver task issuance method.

[0084] In general, the present invention utilizes algorithmic capabilities to identify active drivers' willingness to accept orders, generates personalized driver tasks, and maximizes the number of paired orders; and utilizes algorithmic capabilities to control and predict budgets, prevents the occurrence of events such as over-budget, and improves controllability and stability; and also utilizes algorithmic capabilities to automatically publish tasks based on online traffic, thereby reducing labor costs.

[0085] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A method for issuing tasks to active drivers, characterized in that: include: Using the trained multi-classification model, predict the probability of each active driver being paired with different numbers of orders under different average order reward amounts; According to the probability of each active driver pairing different numbers of orders under different average order reward amounts, obtaining the expected number of pairing orders for each active driver under different average order reward amounts; According to the expected number of paired orders of each active driver under different average order reward amounts, the subsidy costs corresponding to different average order reward amounts, and the subsidy budget, the average order reward amount of each active driver is obtained; According to the average reward amount per order of each active driver and the expected number of paired orders under the average reward amount per order, the tasks of each active driver are obtained and published.

2. The method for issuing tasks to active drivers according to claim 1, characterized in that: The input features of the multi-classification model include the characteristics of active drivers and the average reward amount per order, and the characteristics of active drivers include attribute information, historical data and tracking point information of active drivers.

3. The task issuing method for active drivers according to claim 1, characterized in that: The label of the multi-classification model is an odd-number interval, and the historical daily average paired odd numbers of all active drivers are aggregated to obtain an odd-number range corresponding to each odd-number interval; Using the trained multi-classification model, the probability of each active driver being paired with different numbers of orders under different average order reward amounts is predicted, including: The trained multi-classification model is used to predict the probability of each active driver hitting different order number intervals under different average order reward amounts.

4. The method for issuing tasks to active drivers according to claim 3, characterized in that: According to the probability of each active driver pairing different numbers of orders under different average order reward amounts, the expected number of pairing orders for each active driver under different average order reward amounts is obtained, including: According to the probability of each active driver hitting different single number intervals under different average order reward amounts, the probability of each active driver matching at least K orders under different average order reward amounts is obtained; According to the probability that each active driver matches at least K orders under different average reward amounts per order, the lower bound of the expected number of matching orders for each active driver under different average reward amounts per order is calculated to obtain the expected number of matching orders for each active driver under different average reward amounts per order.

5. The method for issuing tasks to active drivers according to claim 1, characterized in that: According to the expected number of paired orders of each active driver under different average order reward amounts, the subsidy costs and subsidy budgets corresponding to different average order reward amounts, the average order reward amount of each active driver is obtained, including: According to the expected number of paired orders of each active driver under different average order reward amounts and the subsidy costs corresponding to different average order reward amounts, the input parameters of the optimization problem are obtained; According to the subsidy budget and each active driver having only one average reward amount, the constraint condition of the optimization problem is obtained; According to the input parameters and constraints of the optimization problem, and taking maximizing the number of active driver pairings as the objective function, an optimization problem is constructed and solved to obtain Boolean variables of each active driver under different average reward amounts; The average reward amount per order of each active driver is obtained according to the Boolean variables of each active driver under different average reward amounts per order.

6. The method for issuing tasks to active drivers according to claim 5, characterized in that: After the step of obtaining the expected number of paired orders for each active driver at different average reward amounts according to the probability of each active driver pairing different numbers of orders at different average reward amounts, the following step further comprises: Classify all active drivers to obtain several categories of active drivers; According to the subsidy budget and each active driver having only one average reward amount, the constraints of the optimization problem are obtained, including: According to the subsidy budget, each of the active drivers has only one average reward amount per order, and the average reward amount per order for active drivers belonging to the same category is the same, so as to obtain the constraint conditions of the optimization problem.

7. The method for issuing tasks to active drivers according to claim 6, characterized in that: All active drivers are classified into several categories, including: According to the expected number of pairings of active drivers when the average reward amount per pair is 0, the expected number of natural pairings of active drivers is obtained; Performing grid division according to the vehicle type of the active driver and the natural pairing singular expectation to obtain a plurality of grids; Each of the active drivers is classified into a corresponding grid, and active drivers belonging to the same grid are regarded as the same type of active drivers.

8. The method for issuing tasks to active drivers according to claim 1, characterized in that: According to the average reward amount per order of each active driver and the expected number of paired orders under the average reward amount per order, the tasks of each active driver are obtained and published, including: According to the average reward amount per order of each active driver and the expected number of paired orders under the average reward amount per order, the threshold and reward amount of the first-level subtask are obtained; According to the reward amount of the parent task and the reward amount of the first-level subtask, the reward amount of the subsequent subtask is obtained, and according to the preset threshold of the subsequent task, the threshold of the subsequent subtask is obtained; According to the threshold and reward amount of the first-level subtasks and the threshold and reward amount of the subsequent subtasks, the parent task of each active driver is obtained and published.

9. The method for issuing tasks to active drivers according to claim 8, characterized in that: The subsequent subtasks only include second-level subtasks; According to the reward amount of the parent task and the reward amount of the first-level subtask, the reward amount of the subsequent subtask is obtained, and according to the preset threshold of the subsequent task, the threshold of the subsequent subtask is obtained, including: According to the reward amount of the parent task and the reward amount of the first-level subtask, the reward amount of the second-level subtask is obtained; The threshold for the second-level subtask is obtained based on the historical average daily pairing number of active drivers.

10. The method for issuing tasks to active drivers according to claim 9, characterized in that: According to the reward amount of the parent task and the reward amount of the first-level subtask, the reward amount of the second-level subtask is obtained, including: The reward amount of the second-level subtask is obtained based on the reward amount of the parent task, the reward amount of the first-level subtask, whether the active driver has a task dispatch record at the preset time, whether the active driver has a collection record at the preset time, and the preset random increase amount.

11. A task issuing device for active drivers, characterized in that: include: The first computing unit is used to predict the probability of each active driver pairing different numbers of orders under different average order reward amounts by using the trained multi-classification model; The second calculation unit is used to obtain the expected number of paired orders for each active driver under different average reward amounts according to the probability of each active driver pairing different numbers of orders under different average reward amounts; A third calculation unit is used to obtain the average reward amount per order for each active driver according to the expected number of paired orders under different average reward amounts per order for each active driver, the subsidy cost corresponding to the different average reward amounts per order, and the subsidy budget; The task processing unit is used to obtain and publish the task of each active driver according to the average reward amount per order of each active driver and the expected number of paired orders under the average reward amount per order.

12. An electronic device, characterized in that: The system comprises a processor, a memory and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute a computer program stored in the memory to implement the method according to any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer program is used to make a computer execute the method according to any one of claims 1 to 10.

Images