CN111260161B - A method and device for delivering crowdsourcing tasks - Google Patents

Abstract

The invention discloses a method and equipment for issuing crowdsourcing tasks, which relates to the field of Internet data technology and is used to solve the problem of relatively unfair workers' compensation caused by existing crowdsourcing platforms pricing crowdsourcing tasks according to task categories, as well as the lack of consideration of workers' wages. To solve the problem of the price adjustment mechanism after the order is sent to the mid-term report platform, the method of the present invention includes: when determining the pricing of the triggering task, by preset member order price prediction model, according to the price influencing factors of the task, determine the price of the task The order price; publish the order carrying the task identifier and the order price of the task to the crowdsourcing platform; use the preset dynamic price adjustment rules to adjust the order according to the order receiving status of the task order on the crowdsourcing platform. The issuance price of the tasks in the order is adjusted.

Description

A method and device for delivering crowdsourcing tasks

Technical field

The invention relates to the field of Internet data technology, and in particular to a method and device for delivering crowdsourcing tasks.

Background technique

With the rapid development of Internet technology, the crowdsourcing model has been widely used in various aspects such as express delivery, taxi hailing, and freight transportation. Since in the crowdsourcing model, tasks are released to the public through the Internet and other channels, and the public chooses whether to accept the task, the order completion rate of the task is greatly affected by the quality of the task. At the same time, with the rapid development of the Internet, the market for broadband services and smart home decoration and maintenance services has become widespread. The traditional broadband installation business is "grid-based" management. Installation and maintenance workers are responsible for broadband installation and maintenance in fixed areas. Due to the lack of competition, the quality of regional installation and maintenance is greatly affected by workers, and the difficulty of broadband installation in each area is different. Nowadays, Some pricing strategies are based on fixed pricing based on business categories, resulting in workers’ pay levels not being differentiated based on job difficulty, and workers’ pay being relatively unfair across regions.

To sum up, the existing crowdsourcing platform’s pricing method for crowdsourcing tasks fixes the crowdsourcing tasks according to the task category, which leads to relatively unfair workers’ compensation, and the dynamic pricing methods for crowdsourcing tasks are mostly targeted at the workers of the above crowdsourcing tasks. The price prediction or adjustment before the order is issued does not take into account the price adjustment mechanism after the work order is issued.

Contents of the invention

The present invention provides a method and equipment for issuing crowdsourcing tasks to solve the problem that when existing crowdsourcing platforms price crowdsourcing tasks, the fixed pricing of crowdsourcing tasks according to task categories results in relatively unfair workers' wages, and the problem for the crowd Dynamic pricing methods for package tasks mostly focus on price prediction or adjustment before the work order is issued for the above-mentioned crowdsourcing tasks, and do not consider the price adjustment mechanism after the work order is issued.

In a first aspect, embodiments of the present invention provide a method for delivering crowdsourcing tasks. The method includes:

When determining the pricing of the triggered task, the order price of the task is determined based on the price influencing factors of the task through the preset member order price prediction model;

Publish the order carrying the task identifier and the order price of the task to the crowdsourcing platform;

According to the order receiving status of the order for the task on the crowdsourcing platform, the issuance price of the task in the order is adjusted using the preset dynamic price adjustment rules.

In a second aspect, an embodiment of the present invention provides a device for issuing crowdsourcing tasks. The device includes: a processor and a transceiver, wherein the processor is configured to use the transceiver:

When determining the pricing of the triggered task, the order price of the task is determined based on the price influencing factors of the task through the preset member order price prediction model;

Publish the order carrying the task identifier and the order price of the task to the crowdsourcing platform;

According to the order receiving status of the order for the task on the crowdsourcing platform, the issuance price of the task in the order is adjusted using the preset dynamic price adjustment rules.

In a third aspect, an embodiment of the present invention provides a device for issuing crowdsourcing tasks. The device includes: at least one processing unit and at least one storage unit, wherein the storage unit stores program code. When the program code When executed by the processing unit, the processing unit is caused to execute any step of the device described in the second aspect of the present invention.

In a fourth aspect, the present application also provides a computer storage medium on which a computer program is stored. When the program is executed by a processing unit, the steps of the method described in the first aspect are implemented.

The method and equipment for issuing crowdsourcing tasks provided by the present invention have the following beneficial effects compared with the existing technology:

1) In the present invention, the order price of the task is predicted by presetting the member order price prediction model. Compared with the existing technology that only prices the above tasks based on the business category, the predicted order price is more accurate. Ensure relative fairness in workers’ compensation;

2) Compared with the existing technology, in the present invention, after the order carrying the order price of the task is released to the crowdsourcing platform, a method of dynamically adjusting the order price of the task according to the order receiving situation is added, which improves the efficiency of the process. The completion rate of crowdsourcing platform orders.

Description of the drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings needed to describe the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. Those of ordinary skill in the art can also obtain other drawings based on these drawings without exerting any creative effort.

Figure 1A is a schematic diagram of a method for delivering crowdsourcing tasks provided by Embodiment 1 of the present invention;

Figure 1B is a schematic diagram of factors affecting the price of tasks provided by Embodiment 1 of the present invention;

Figure 1C is a schematic diagram of a specific process for establishing a preset member order price prediction model provided in Embodiment 1 of the present invention;

Figure 1D is a schematic flowchart of specific steps of a method for delivering crowdsourcing tasks provided in Embodiment 1 of the present invention;

Figure 2A is a schematic diagram of a first device for delivering crowdsourcing tasks provided in Embodiment 2 of the present invention;

Figure 2B is a schematic diagram of a second device for delivering crowdsourcing tasks provided in Embodiment 2 of the present invention;

FIG. 2C is a schematic diagram of a third device for delivering crowdsourcing tasks provided in Embodiment 2 of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be described in further detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of them. . Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

Here are some explanations of some words that appear in the text:

1. In the embodiment of the present invention, the term "and/or" describes the association relationship of associated objects, indicating that there can be three relationships, for example, A and/or B, which can mean: A exists alone, A and B exist simultaneously, alone There are three situations B. The character "/" generally indicates that the related objects are in an "or" relationship.

2. In the embodiment of the present invention, the term "task" refers to tasks that can be delivered to the crowdsourcing platform, such as broadband installation tasks, taxi-hailing tasks, etc.

The application scenarios described in the embodiments of the present invention are to more clearly illustrate the technical solutions of the embodiments of the present invention, and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. Those of ordinary skill in the art will know that with the emergence of new application scenarios It appears that the technical solutions provided by the embodiments of the present invention are equally applicable to similar technical problems. Among them, in the description of the present invention, unless otherwise stated, the meaning of "plurality" is two or more.

With the increasing development of Internet technology, crowdsourcing models have been widely used in various industries. Formulating dynamic pricing strategies for crowdsourcing models in different scenarios has positive and far-reaching practical significance for the development of crowdsourcing models. However, the research objects of existing crowdsourcing price adjustment and pricing methods are mostly tasks with relatively small technical differences in technical difficulty, such as express delivery and taxi-hailing. There is no specific method to quantify the difficulty of the task, and it is not suitable for tasks with higher technical requirements. The embodiment of the present invention provides a dynamic pricing method for crowdsourcing tasks that have high technical requirements, low time sensitivity, long time consumption, and can receive multiple orders at the same time, taking broadband installation as an example, to achieve differentiation based on technical difficulty. Pricing, improving the relative fairness of salaries, achieving the goal of regulating the willingness of installation and maintenance workers to grab orders based on dynamic price adjustments, increasing the enthusiasm of crowdsourcing personnel to grab orders "across the grid", and increasing the order completion rate in each region.

In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be described in further detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

In view of the above scenario, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings of the description.

Example 1:

As shown in Figure 1A, this embodiment provides a method for delivering crowdsourcing tasks, which specifically includes the following steps:

Step 101: When determining the pricing of the triggered task, determine the order price of the above task through the preset member order price prediction model and based on the price influencing factors of the above task;

The above-mentioned preset member order price prediction model is used to predict the predicted order price of tasks on the crowdsourcing platform by members of the crowdsourcing platform. During implementation, the above-mentioned predetermined price is established in advance based on the data of historical task transaction orders on the crowdsourcing platform. Set up a member order price prediction model;

In the implementation, through the preset member order price prediction model and based on the price influencing factors of the above tasks, a price prediction result set containing the above preset online members' predicted order prices for the above tasks is obtained, such as {p' ₁ , p ' ₂ ,…,p' _j ,…p' _w }, where p ₁ ', p ₂ ', p _j ', and p _w ' are member 1, member 2, member j and member j within the preset area, respectively. Member w’s predicted order price for the above-mentioned business; the above-mentioned preset online members are all online members within the preset area where the above-mentioned task is located;

The range of the preset area where the above task is located is not too limited. Those skilled in the art can set it according to actual needs. The range of the preset area where the above task is located can, but is not limited to, include: taking the position point where the above task is located as the center of the circle, and The area of a circle with a radius of 10Km is used as the preset area where the above task is located, or the city where the above task is located is used as the preset area where the above task is located, or the city community where the above task is located is used As the preset area where the above tasks are located.

From the above price prediction result set, according to the principle of price from low to high, the predicted order receiving price with the lowest price is selected and determined as the order issuing price for the above task.

As shown in Figure 1B, the price influencing factors of the above tasks may, but are not limited to, include at least one of the following:

Characteristics of a single task; the characteristics of a single task mentioned above refer to the characteristics of the factors affecting the price of the task of the single crowdsourcing task itself. The characteristics of a single task mentioned above may, but are not limited to, include at least one of the following: task reservation time, task type, Task difficulty level, task location.

The characteristics of the task area; the characteristics of the above-mentioned task area refer to the characteristics of the factors that affect the price of the crowdsourcing tasks in the task area divided by clustering of the same type of crowdsourcing tasks mentioned above. The characteristics of the above-mentioned task area can be but are not limited to Including at least one of the following: task price level in the task area, order density in the task area, and member density in the task area.

Characteristics of members who log in to the crowdsourcing platform; the above-mentioned characteristics of members who log in to the crowdsourcing platform refer to the characteristics of the above-mentioned online members who log in to the crowdsourcing platform that affect the price of the task. The characteristics of the above-mentioned members who log in to the crowdsourcing platform can be but are not limited to Including at least one of the following: the number of orders received by the member, the price of the order received by the member in history, the area of the order received by the member in history, and the time of order received by the member.

Step 102: Publish the order carrying the above-mentioned task identifier and the order price of the above-mentioned task to the crowdsourcing platform;

During implementation, after determining the order price of the above tasks based on the price influencing factors of the above tasks, it also includes:

When it is determined that the above task is accepted, the order price in the completed work order information of the above task is used as the task price, and the price influencing factors of the task in the completed work order information are saved to the training data of the above-mentioned preset member order price prediction model The set is used as the training data for training the above-mentioned preset member order price prediction model during the use of the platform.

Step 103: According to the order receiving status of the order for the above task on the crowdsourcing platform, use the preset dynamic price adjustment rules to adjust the order price of the task in the above order.

In implementation, if it is determined that the above tasks have not been accepted within the preset time period, dynamic price adjustment will be triggered.

Determine whether the orders for the above tasks have reached the preset number of push rounds. If they have reached the preset number of push rounds, they will be transferred to the system order dispatch process;

If the preset number of push rounds is not reached, it is determined whether the order price of the above task reaches the highest predicted price in the price prediction result set of the task in the preset area. If it reaches the highest predicted price of the task, it will be transferred to the system dispatch process. If If the maximum predicted price of the task is not reached, adjust the issuance price of the above task to obtain the new issuance price of the above task.

In the first preset platform stage when tasks are released to the crowdsourcing platform, the issuance price of the above tasks is increased in a step-by-step manner according to the preset step-up price increase system to obtain a new issuance price;

In the second preset platform stage when the task is released to the crowdsourcing platform, each time the order price of the above task is adjusted, the lowest predicted order price in the above price prediction result set is selected as the new issuance price of the above task. unit price.

The above-mentioned transfer order dispatching process into the system can be understood as the crowdsourcing platform pushing the system to dispatch the orders for the above tasks to the appropriate online members within the preset area according to certain preset rules; the above certain preset rules, It may include but is not limited to: based on distance priority, dispatch the above orders to the online members within the preset area who are closest to the above task location, or, based on the number of orders received by members, dispatch the above orders to the preset area. The online member with the smallest number of orders, or based on the member’s historical order price, the above orders will be sent to the online member with the lowest historical order price within the preset area, etc.

In this embodiment, in the early stage of the crowdsourcing platform (that is, the first preset platform stage for task release), the order price of the above task is adjusted according to Formula 1:

Formula 1: p″ _r =p' _r +p ₀

In the above formula 1, p _r ″ is the adjusted order price, p _r ′ is the current order price, and p ₀ is the above-mentioned preset step plus value. Those skilled in the art can set the above p ₀ according to the actual situation. value.

In the mature stage of the crowdsourcing platform (i.e., the second preset platform stage for task release), the order price of the above tasks is adjusted according to Formula 2:

Formula 2: p″ _r =min{{p' ₁ ,p' ₂ ,…,p' _j ,…p' _w }-min{p' _{1 ,…p' 2 ,} …,p' _j ,…p' _w }}

In the above formula 2, {p' ₁ , p' ₂ ,...,p' _j ,...p' _w } is the default member order price prediction model, and is the result set of member order price prediction for task r.

In the above step 101, the default member order price prediction model is determined in the following way:

In implementation, the above-mentioned preset member order price prediction model is determined in the following ways: based on the vector machine model, the initial price prediction model and corresponding model parameters are determined; the price influencing factors of historical tasks and task pricing are used as training for the price prediction model Data set, train the above price prediction model to obtain the above preset member order price prediction model.

Before training the above-mentioned preset member order price prediction model, first determine the input volume, output volume, sample set and test set of the above-mentioned preset member order price prediction model;

A) Determination of input amount:

First, by analyzing the order dispatching process of the crowdsourcing platform, we systematically sort out the factors that may affect the price of orders received by members on the crowdsourcing platform (factors affecting the price of the above tasks), and then communicate with members through each factor. Correlation analysis of the order price of the task, and based on the correlation analysis results, select the input variable as the above price prediction model;

The above-mentioned factors that affect the order price of a task may include, but are not limited to, at least one of the following:

Characteristics of individual tasks;

Characteristics of the mission area;

Characteristics of members who log in to the crowdsourcing platform.

In this embodiment, a summary is made from three aspects: task characteristics, regional characteristics, and member characteristics. The specific factors are shown in Figure 1B.

The above correlation analysis methods may include, but are not limited to, chart correlation analysis, covariance and covariance matrix, correlation coefficient, single regression and multiple regression, information entropy and mutual information, etc.;

To illustrate the process, the Pearson correlation coefficient is used here as an example of this step.

The formulas for correlation calculation are as follows: Formula 3 and Formula 4:

Formula 3:

Formula 4:

In the above formulas 3 and 4, Correl(X,Y) is the Pearson correlation coefficient of variables X and Y. The above X is the price influencing factor of the input task, the above Y is the order price of the output task, and n is the sample. The number of T statistic constructed to verify the significance of the correlation.

When calculating the correlation using the above formulas 3 and 4, compare the calculation results of the T value with the T test critical value table to determine whether the T statistic exceeds the critical value. If the T statistic does not exceed the critical value, then the variables X, Y is significantly relevant, otherwise it is not relevant.

B) Determination of output:

The predicted order price of the task by members of the crowdsourcing platform is used as the output variable of the above price prediction model.

C) Creation of sample set and test set:

According to the variable screening results of the above correlation analysis, the data is divided into dimensions according to the price influencing factors of the screened tasks, characteristic data samples are collected, and training sample sets {x _i ,y _i },y _i ∈(0,1 ) and test set {x' _i ,y' _i };

The xi in the above-mentioned training sample set and the xi _' in the test set are both factors affecting the price of the task. The yi in the above-mentioned training sample set and the y _{i '} in the test set are both _the order prices of the above-mentioned tasks;

The above training sample set is used for training the above price prediction model, including the selection of the selected kernel function and model parameters;

The above test set is used to verify the above price prediction model, including the prediction accuracy and prediction efficiency of the above price prediction model;

The data structure of the data in the above test set is consistent with the above training sample set in order to obtain a price prediction model with accurate predictions.

(1) Based on the vector machine model, determine the initial price prediction model and corresponding model parameters, including:

1) Based on the vector machine model, select a kernel function as the initial price prediction model;

Among them, for predicting the price of members’ orders, the algorithms that can be used include but are not limited to KNN algorithm, artificial neural network algorithm, support vector machine algorithm, etc.;

To illustrate the process, a support vector machine is used as an example of this step.

The basic idea of support vector machine regression is to map data x to a high-dimensional space F through a nonlinear mapping φ, and perform linear regression in this space, as shown in Formula 5 below:

Formula 5:

f(x)=(ω ^· φ(x))+b

φ.R ⁿ →F,ω∈F

In the above formula 5, b is the threshold value. The optimization goal of the support vector machine is to minimize the structural risk. R(ω) is introduced to quantify the structural risk of the model:

R(ω)=∑e(f(x _i )-y _i )+λ||ω|| ²

After minimizing R(ω), we get: ω=∑(α-α ^* )φ(x _i )

From this, the prediction model of the support vector machine is:

f(x)=∑(α-α ^* )(φ(x _i )φ(x))+b

Among them, let k( _xi ,x)=φ( _xi )φ(x) be the kernel function of the model.

2) According to the preset model parameter generation method, determine the kernel parameters corresponding to the above kernel function as the model parameters corresponding to the above initial price prediction model.

The above-mentioned kernel function k( _xi ,x) and kernel parameters (model parameters) include: polynomial function, RBF function, Sigmoid function, etc.;

The kernel function can be selected through various methods such as empirical selection method, cross-validation method, and repeated sampling method, but is not limited to it;

The above-mentioned kernel parameters can be selected but are not limited to the following methods: cross-validation method, exhaustive method, gradient descent method, grid search method, genetic algorithm, particle swarm algorithm, etc.;

In this embodiment, the gradient descent method is taken as an example. First, a set of initial values of kernel parameters of the selected kernel function are determined based on the empirical selection method, and then the negative gradient direction is used to determine the new search direction for the next iteration, so that The objective function to be optimized gradually decreases, and the specific formula is as follows: Formula 1:

Formula 6: x _i+1 = x _i -a*g _i

In the above formula 6, a is the learning rate, which can be set by technicians to a smaller constant value according to the actual situation; g _i is the gradient; x _i+1 and x _i are the above-mentioned kernel parameters, where x _i is the current kernel parameter, x _i+1 is the kernel parameter used in the next round.

(2) Use the price influencing factors of historical tasks and task pricing as training data sets for the price prediction model, including:

1. Obtain relevant information about all historical tasks on the above-mentioned crowdsourcing platform. The relevant information about the above-mentioned historical tasks includes the price influencing factors of historical tasks and the order price:

In implementation, the attribute characteristics of the task (i.e., the above-mentioned price influencing factors) can be divided according to the characteristics of the crowdsourcing task, which can include but is not limited to the type of task and the length of task service;

The above-mentioned historical task-related information includes the work order information data of the historical task. During implementation, the work order information data of the above-mentioned historical task can be stored in a preset database;

2. Based on the above historical task-related information, determine the benchmark pricing for each type of task in the corresponding task area:

1) Classify all historical tasks according to the task types of the above historical tasks.

In implementation, the service market of historical tasks can be used as an indicator to measure difficulty, and the work order information data of the above historical tasks can be classified according to task categories.

2) Conduct data mining analysis on the task service duration of all historical tasks of the same type, and obtain the task difficulty levels of all historical tasks of the same type mentioned above.

You can, but are not limited to, use K-means cluster analysis method/artificial neural network to conduct data mining analysis on the task service time of all historical tasks of the same type, cluster the above historical tasks of the same type, and classify them according to the clustering results. The difficulty of the above historical tasks of the same type is divided into several levels.

3) Based on the task difficulty levels and order prices of all historical tasks of the same type mentioned above, determine the benchmark pricing of the above historical tasks of the same type in each corresponding task area.

Factors affecting the price of the above historical tasks include the location of the task;

According to the task difficulty level and task location of all the above-mentioned historical tasks of the same type, perform task area clustering on all the above-mentioned historical tasks of the same type, and obtain the task areas of the above-mentioned historical tasks of the same type;

Perform statistical analysis on the order prices of tasks in each task area corresponding to each type of historical tasks, and determine the benchmark price of each type of historical tasks in each corresponding task area.

In the implementation, task difficulty is added to the above task characteristics. Among all historical tasks of the same type, the work orders of the historical tasks are clustered by task area based on the task location and task difficulty level. The tasks are clustered according to the task area clustering results. Partition and store the partition results in the database.

The above method for clustering task areas can include, but is not limited to, the hybrid algorithm of DBSCAN and K-means;

The specific implementation method of performing task area clustering on historical task work orders based on task location and task difficulty level is as follows: taking the community as the basic unit and numbering the clustering result task areas.

In the implementation, based on the task area of the task difficulty partition and the distribution density of the work order area of the historical tasks, the benchmark price of each type of historical tasks in the corresponding task area is calculated, while ensuring that the work orders of the above historical tasks are Under the premise that the total return budget remains unchanged, the above-mentioned benchmark price of each task area is mapped to the average service time of the tasks in each area, and the benchmark price of each task area is differentially calculated; in this embodiment, it is assumed that The number of task areas is divided into n. The average monthly number of work orders in the i-th task area is m _i . The average service time of tasks in the i-th area is The fixed price before the task adopts the dynamic pricing strategy is p ₀ , then the benchmark price of the task in the i-th task area is p _i , and its calculation formula is as shown in Equation 7:

Formula 7:

Arranged in order from high to low by the benchmark price of tasks in the task area, the benchmark price of tasks in area {1,2,…,n} is {p ₁ , p ₂ ,…, p _n }. In this embodiment, in the initial stage of the transformation of the broadband service from the "network format" indiscriminate order dispatch mode to the crowdsourcing order dispatch model, the benchmark price of the task in the task area is used as the initial order issuance price, combined with dynamic price adjustment after the order is issued. The module dynamically adjusts the work order price. In the middle and later stages of order dispatch under the crowdsourcing model, the task price will be realized by a combination of two modules: dynamic price adjustment before order issuance and dynamic price adjustment after order issuance.

3. Use the price influencing factors of the above historical tasks and the corresponding benchmark pricing as the training data set for the price prediction model.

In this embodiment, based on the task difficulty level of historical tasks, the distance between the task and the member, the regional density of the task, the number of work orders received by members and other task attributes, effective data sets are screened as a training set to crowdsource the price of orders received by members. Carry out prediction model training and prediction.

(3) Use the price influencing factors of historical tasks and task pricing as training data sets for the price prediction model, and train the above-mentioned price prediction model to obtain the above-mentioned preset member order price prediction model;

1) Use the price influencing factors of historical tasks and the order price of the task as the input of the price prediction model, and obtain the member’s predicted order price for each historical task output by the price prediction model;

2) Use the correlation analysis method to train the above-mentioned current price prediction model based on the order prices and predicted order prices of the above-mentioned historical tasks.

In the implementation, the correlation analysis method is used to determine that the correlation between the predicted order price and task pricing of any historical task reaches the first preset threshold, or when the number of times the above-mentioned current price prediction model is trained reaches the preset number of times, It is determined that the current price prediction model mentioned above is the default member order price prediction model.

The above correlation analysis method may, but is not limited to, include at least one of the following:

Chart correlation analysis method, covariance and covariance matrix method, correlation coefficient method, single regression method, multiple regression method, information entropy method, and mutual information method.

When the correlation analysis results between the order price and the predicted order price of the above historical tasks meet the preset conditions, it is determined that the training is completed, and the trained price prediction model and corresponding model parameters are used as the preset member order price prediction model;

When the correlation analysis results between the order price and the predicted order price of the above historical tasks do not meet the preset conditions, determine whether the sample data in the training data set have been trained by the current price prediction model. If the sample data in the training data set When there is sample data that has not been trained by the current price prediction model, use the untrained sample data to continue training the current price prediction model;

If the sample data in the training data set are all outdated by the current price prediction model, the model parameters of the above price prediction model can be adjusted according to the preset learning rate, and the sample data in the above training data set can be reused to train the above price prediction model;

When the adjustment of the model parameters of the above price prediction model reaches the preset stop parameter adjustment condition, a kernel function and the corresponding kernel parameters can be reselected as the price prediction model and model parameters, and the sample data in the above training data set can be reused for training and reselection. price prediction model; the above-mentioned preset stop parameter adjustment conditions may include, but are not limited to, including model parameter adjustment reaching the preset number of adjustments.

As an optional implementation, after the preset member order price prediction model is obtained for the first time, when the model optimization conditions are met, the updated training data set can be reused to train and optimize the above preset member order price prediction model. , in order to achieve better model effects and make the model more accurate in predicting the price of orders received by members for crowdsourcing tasks;

The above-mentioned conditions for satisfying the model optimization are not too limited. Those skilled in the art can set them according to actual needs. For example, when the interval is a fixed period of time, it is determined that the above-mentioned model optimization conditions are met, or when the above-mentioned preset member after training is completed, the When the number of times the single price prediction model is used reaches the preset number of times, it is determined that the above model optimization conditions are met.

As shown in Figure 1C, the following is a specific process for establishing a preset member order price prediction model:

Step 1101, collect historical task related information;

Step 1102, divide historical tasks of the same type according to task difficulty levels;

Step 1103, perform task area clustering on historical tasks of the same type based on task difficulty and task location;

Step 1104: For historical tasks of the same type, task areas are divided according to the task area clustering results;

Step 1105: Statistically analyze the order prices of historical tasks within each task area of the same type of historical tasks;

Step 1106, differentially calculate the benchmark price for each task area, and input the calculation results into the training data set in step 1108;

Step 1107, determine the input and output of the price prediction model;

Step 1108, create a training data set (sample set and test set);

Step 1109: Based on the vector machine model, select the price prediction model and corresponding model parameters;

Step 1110, normalize the training data;

Step 1111: Use the training data set data to train the price prediction model to obtain a preset member order price prediction model.

As shown in Figure 1D, the following is a specific step-by-step process of a method for issuing crowdsourcing tasks:

Step 1) Obtain the price influencing factors of task r;

Step 2) Obtain all online member information within the preset area where task r is located;

Step 3) Input the price influencing factors of task r into the preset member order price prediction model to obtain the price prediction result set within the preset area where task r is located;

Step 4) Select the lowest predicted price from the price prediction result set as the issuing price of task r;

Step 5) Push the order carrying the order price of task r to members within the preset area;

Step 6) Determine whether task r has been ordered within the preset time period. If the order has been accepted, go to step 7; otherwise, go to step 8;

Step 7) Save the transaction order data of task r into the training data set;

Step 8) Determine whether the number of order pushes for task r reaches the preset number of push rounds. If so, go to step 9; otherwise, go to step 10;

Step 9) Stop price adjustment and use the system order dispatch process to dispatch task r;

Step 10) Determine whether the order price of the above task reaches the highest predicted price in the price prediction result set of the preset area. If so, go to step 9, otherwise go to step 11;

Step 11) Determine whether the crowdsourcing platform is in the early stage of the platform. If so, go to step 12; otherwise, go to step 13;

Step 12) Add the value to the issuance price of task r according to the preset step-by-step price increase to obtain a new issuance price, and enter step 5;

Step 13: Adjust the price according to the price adjustment rules in the mature stage of the platform to obtain a new order price. Select the predicted order price in the above price prediction result set, subtract the minimum value of the predicted order price with the lowest price, and use it as the new price for the above task. the order price and enter step 5;

The method proposed in this embodiment to propose differentiated pricing for services with high technical requirements lays a foundation for task pricing to move from a unified price to differentiation, and provides data support for dynamic price adjustment. Different dynamic price adjustment modules are proposed for different stages of platform development. In the early stage of the platform, the main goal is to collect data and increase the order rate. Later, the main goal is to optimize the price adjustment model and maintain a high order rate level. There are multiple feedback channels between each module, which is conducive to optimizing the price adjustment model. The price adjustment mechanism after the order is issued allows the dynamic price adjustment model to be in line with market demand at all times. Members' demand feedback on task prices is used to optimize the prediction model, improve the order completion rate or maintain Higher level of order completion rate.

The research objects of existing crowdsourcing price adjustment and pricing methods are mostly tasks with relatively small differences in technical difficulty such as express delivery and taxi-hailing. There is no specific method to quantify the difficulty of the task and is not suitable for tasks with higher technical requirements. This implementation In the example method, based on the task characteristics of mid-term reporting tasks, tasks of the same type are classified according to task difficulty levels and task areas are divided, laying the foundation for differentiated pricing in the later period;

The default premise in the dynamic pricing methods of existing crowdsourcing platforms is that task pricing has shown differentiation in line with market rules, and there is enough effective data to support the establishment of dynamic pricing models. The transition from fixed prices to differentiated pricing is not considered. The conversion problem is the collection of effective data required for modeling. In this embodiment, at different stages of the crowdsourcing platform, different methods are used to collect data for price prediction model modeling and model optimization, and the dynamic pricing method is implemented by Transform from fixed pricing to differentiated pricing.

Existing dynamic pricing methods are mostly aimed at price prediction or adjustment before the work order of the crowdsourcing task is released, and do not consider the price adjustment mechanism after the work order is issued. In this embodiment, after the work order of the crowdsourcing task is released, Based on the member's response to the work order, the member's situation, and the stage stated on the platform, the order price of the crowdsourcing task is dynamically adjusted; and the optimization goal of the existing dynamic price adjustment method for the crowdsourcing task is mostly to increase the order completion rate. Most of the crowdsourcing tasks are time-sensitive crowdsourcing tasks, and the method provided in this embodiment not only considers time-sensitive crowdsourcing tasks, but also considers low time sensitivity, long-lasting tasks, and crowdsourcing personnel's acceptance. Factors such as order volume and work order density in the task area optimize the effect of dynamic price adjustment for crowdsourcing tasks.

It should be noted that the method of delivering a broadband task on a crowdsourcing platform listed in the embodiment of the present invention is just an example, and any method of delivering a task on a crowdsourcing platform is applicable to the implementation of the present invention. example.

Example 2:

Based on the same inventive concept, the embodiment of the present invention also provides a first device for issuing crowdsourcing tasks. As shown in Figure 2A, the first device includes: a processor 201 and a transceiver 202, wherein the above processing The transceiver is used to:

When determining the pricing of the triggered task, the order price of the above task is determined based on the price influencing factors of the above task through the preset member order price prediction model;

Publish the order carrying the above task identifier and the order price of the above task to the crowdsourcing platform;

According to the order reception status of the above-mentioned task orders on the crowdsourcing platform, the preset dynamic price adjustment rules are used to adjust the issuance price of the tasks in the above-mentioned orders.

In implementation, the above-mentioned processor determines the above-mentioned preset member order price prediction model in the following manner:

Based on the vector machine model, determine the initial price prediction model and corresponding model parameters;

The price influencing factors of historical tasks and task pricing are used as training data sets for the price prediction model, and the above price prediction model is trained to obtain the above preset member order price prediction model.

The above processor is specifically used to: select a kernel function as the initial price prediction model based on the vector machine model;

According to the preset model parameter generation method, the kernel parameters corresponding to the above kernel function are determined as the model parameters corresponding to the above initial price prediction model.

Factors affecting the price of the above tasks include at least one of the following:

Characteristics of individual tasks;

Characteristics of the mission area;

Characteristics of members who log in to the crowdsourcing platform.

The characteristics of the above-mentioned single task include at least one of the following:

Task reservation time;

Type of task;

Task difficulty level;

Mission location.

The characteristics of the above-mentioned task area include at least one of the following:

Task price levels in the task area;

Order density in the task area;

Membership density in the mission area.

The above-mentioned characteristics of members who log in to the crowdsourcing platform include at least one of the following:

The number of orders received by members;

Member’s historical order price;

Member history order receiving area;

Member’s order receiving time;

Member location.

The above-mentioned processor is specifically used to: obtain all historical task-related information on the above-mentioned crowdsourcing platform. The above-mentioned historical task-related information includes the price influencing factors of historical tasks and the order price;

Based on the above historical task-related information, determine the benchmark pricing for each type of task in the corresponding task area;

Use the price influencing factors of the above historical tasks and the corresponding benchmark pricing as the training data set for the price prediction model.

The above processor is specifically used to: classify all historical tasks according to the task types of the above historical tasks;

Conduct data mining analysis on the task service duration of all historical tasks of the same type, and obtain the task difficulty levels of all historical tasks of the same type mentioned above.

Based on the task difficulty levels and order prices of all the above-mentioned historical tasks of the same type, determine the benchmark pricing of the above-mentioned historical tasks of the same type in each corresponding task area.

The above-mentioned processor is specifically used to: use K-means cluster analysis method/artificial neural network to perform data mining analysis on the task service time of all historical tasks of the same type.

The above-mentioned processor is specifically used to: perform task area clustering on all the above-mentioned historical tasks of the same type based on the task difficulty levels and task locations of all the above-mentioned historical tasks of the same type, and obtain the task areas of the above-mentioned historical tasks of the same type;

Perform statistical analysis on the order prices of tasks in each task area corresponding to each type of historical tasks, and determine the benchmark price of each type of historical tasks in each corresponding task area.

The above-mentioned processor is specifically used to: use the price influencing factors of historical tasks and task pricing as input quantities of the price prediction model, and obtain the member's predicted order price for each historical task output by the price prediction model;

Use the correlation analysis method to train the above current price prediction model based on the task pricing and predicted order price of the above historical tasks.

The above-mentioned processor is specifically used to: use the correlation analysis method to determine when the correlation between the predicted order price and task pricing of any historical task reaches the first preset threshold, or when the number of times of training the above-mentioned current price prediction model reaches the preset times, the above current price prediction model is determined to be the default member order price prediction model.

The above correlation analysis method includes at least one of the following:

Chart correlation analysis method;

Covariance and covariance matrix methods;

Correlation coefficient method;

Univariate regression method;

Multiple regression method;

Information entropy method;

Mutual information method.

The above-mentioned processor is specifically used to: obtain a price prediction result set including the above-mentioned preset online member's predicted order price for the above-mentioned task based on the price influencing factors of the above-mentioned task through the preset member order price prediction model, the above-mentioned preset online member Members are all online members within the preset area where the above tasks are located;

From the above price prediction result set, according to the principle of price from low to high, the predicted order receiving price with the lowest price is selected and determined as the order issuing price for the above task.

The above-mentioned processor is also used to: when determining that the above-mentioned task is accepted, the order-receiving price in the transaction work order information of the above-mentioned task is used as the task price, and the price influencing factors of the task in the transaction work order information are saved to the above-mentioned preset member interface. Training dataset for single price prediction model.

The above processor is specifically used to trigger dynamic price adjustment when it is determined that the above task has not been accepted within a preset time period.

The above processor is specifically used to: determine whether the order for the above task reaches the preset number of releases, and if it reaches the preset number of push rounds, transfer to the system order dispatch process;

If the preset number of push rounds is not reached, determine whether the order price of the above task reaches the highest predicted price in the price prediction result set within the preset area. If the highest predicted price is reached, the system will transfer to the order dispatch process. If it does not reach the highest predicted price, The highest predicted price is used to adjust the issuance price of the above task to obtain the new issuance price of the above task.

The above-mentioned processor is specifically used to: in the first preset platform stage, add value to the issuance price of the above task according to the preset step-by-step step-by-step price increase to obtain a new issuance price;

In the second preset platform stage, every time the order price of the above task is adjusted, the minimum value of the predicted order price in the above price prediction result set is subtracted from the value of the lowest predicted order price, as The new billing price for the above tasks.

As shown in Figure 2B, this embodiment also provides a second device for issuing crowdsourcing tasks. The second device includes: at least one processing unit 203 and at least one storage unit 204, wherein the above storage unit stores program code. When the above program code is executed by the above processing unit, the above processing unit is caused to execute any content of the first device described in this embodiment. Since the second device is the device in the method in this embodiment, and the problem-solving principle of this device is similar to that of the method, the implementation of the device can be referred to the implementation of the method, and repeated details will not be repeated.

As shown in Figure 2C, this embodiment also provides a third device for issuing crowdsourcing tasks. The third device includes:

The initial order price determination unit 2001 is used to determine the order price of the above task through the preset member order price prediction model and based on the price influencing factors of the above task when determining the pricing of the trigger task;

The order publishing unit 2002 is used to publish the order carrying the above-mentioned task identifier and the order price of the above-mentioned task to the crowdsourcing platform;

The dynamic price adjustment unit 2003 is used to adjust the issuance price of the tasks in the above-mentioned orders by using preset dynamic price adjustment rules based on the order receiving status of the above-mentioned task orders on the crowdsourcing platform.

In the implementation, it also includes a preset member order price prediction model building unit 2004, which is used to determine the initial price prediction model and corresponding model parameters based on the vector machine model; use the price influencing factors of historical tasks and task pricing as price predictions The training data set of the model is used to train the above price prediction model to obtain the above preset member order price prediction model.

The above-mentioned preset member order price prediction model construction unit is used to select a kernel function as the initial price prediction model based on the vector machine model;

According to the preset model parameter generation method, the kernel parameters corresponding to the above kernel function are determined as the model parameters corresponding to the above initial price prediction model.

Factors affecting the price of the above tasks include at least one of the following:

Characteristics of individual tasks;

Characteristics of the mission area;

Characteristics of members who log in to the crowdsourcing platform.

The characteristics of the above-mentioned single task include at least one of the following:

Task reservation time;

Type of task;

Task difficulty level;

Mission location.

The characteristics of the above-mentioned task area include at least one of the following:

Task price levels in the task area;

Order density in the task area;

Membership density in the mission area.

The above-mentioned characteristics of members who log in to the crowdsourcing platform include at least one of the following:

The number of orders received by members;

Member’s historical order price;

Member history order receiving area;

Member’s order receiving time;

Member location.

The above-mentioned preset member order price prediction model construction unit is used to obtain all historical task-related information on the above-mentioned crowdsourcing platform. The above-mentioned historical task-related information includes the price influencing factors of historical tasks and the order price;

Based on the above historical task-related information, determine the benchmark pricing for each type of task in the corresponding task area;

Use the price influencing factors of the above historical tasks and the corresponding benchmark pricing as the training data set for the price prediction model.

The above-mentioned preset member order price prediction model construction unit is used to classify all historical tasks according to the task types of the above-mentioned historical tasks;

Conduct data mining analysis on the task service duration of all historical tasks of the same type, and obtain the task difficulty levels of all historical tasks of the same type mentioned above.

Based on the task difficulty levels and order prices of all the above-mentioned historical tasks of the same type, determine the benchmark pricing of the above-mentioned historical tasks of the same type in each corresponding task area.

The above-mentioned preset member order price prediction model construction unit is used to use K-means cluster analysis method/artificial neural network to perform data mining analysis on the task service duration of all historical tasks of the same type.

The above-mentioned preset member order price prediction model construction unit is used to perform task area clustering on all the historical tasks of the above-mentioned same type based on the task difficulty level and task location of all the above-mentioned historical tasks of the same type, and obtain the above-mentioned historical tasks of the same type. The task area of the task;

Perform statistical analysis on the order prices of tasks in each task area corresponding to each type of historical tasks, and determine the benchmark price of each type of historical tasks in each corresponding task area.

The above-mentioned preset member order price prediction model construction unit is used to use the price influencing factors of historical tasks and task pricing as input quantities of the price prediction model, and obtain the member's predicted order price for each historical task output by the price prediction model;

Use the correlation analysis method to train the above current price prediction model based on the task pricing and predicted order price of the above historical tasks.

The above-mentioned preset member order price prediction model construction unit is used to use the correlation analysis method to determine the correlation between the predicted order price of any historical task and the task pricing reaches the first preset threshold, or to determine the above-mentioned current price prediction model. When the number of times of training reaches the preset number, the above-mentioned current price prediction model is determined to be the preset member order price prediction model.

The above correlation analysis method includes at least one of the following:

Chart correlation analysis method;

Covariance and covariance matrix methods;

Correlation coefficient method;

Univariate regression method;

Multiple regression method;

Information entropy method;

Mutual information method.

The above-mentioned initial order price determination unit is used to obtain a price prediction result set including the above-mentioned preset online member's predicted order price for the above-mentioned task based on the price influencing factors of the above-mentioned task through the preset member order-taking price prediction model. The above-mentioned The default online members are all online members within the preset area where the above tasks are located;

From the above price prediction result set, according to the principle of price from low to high, the predicted order receiving price with the lowest price is selected and determined as the order issuing price for the above task.

The above-mentioned order publishing unit is also used to determine that when the above-mentioned task is accepted, the order-receiving price in the transaction work order information of the above-mentioned task is used as the task price, and the price influencing factors of the task in the transaction work order information are saved to the above-mentioned default member Training data set for order price prediction model.

The above-mentioned order issuance unit is used to determine that when the above-mentioned task has not been accepted within a preset time period, dynamic price adjustment will be triggered.

The above dynamic price adjustment unit is used to determine whether the order for the above task reaches the preset number of releases. If it reaches the preset number of push rounds, it will be transferred to the system order dispatch process;

If the preset number of push rounds is not reached, determine whether the order price of the above task reaches the highest predicted price in the price prediction result set within the preset area. If the highest predicted price is reached, the system will transfer to the order dispatch process. If it does not reach the highest predicted price, The highest predicted price is used to adjust the issuance price of the above task to obtain the new issuance price of the above task.

The above-mentioned dynamic price adjustment unit is used in the first preset platform stage to add value to the issuance price of the above task according to the preset step-by-step step-by-step price increase to obtain a new issuance price;

In the second preset platform stage, every time the order price of the above task is adjusted, the minimum value of the predicted order price in the above price prediction result set is subtracted from the value of the lowest predicted order price, as The new billing price for the above tasks.

Embodiment three:

Embodiments of the present invention also provide a computer-readable non-volatile storage medium, including program code. When the program code is run on a computing terminal, the program code is used to cause the computing terminal to execute the above implementation of the present invention. Steps of the method in Example 1.

The present application is described above with reference to block diagrams and/or flowcharts illustrating methods, apparatus (systems) and/or computer program products according to embodiments of the application. It will be understood that one block of the block diagrams and/or flowchart illustrations, and combinations of blocks of the block diagrams and/or flowchart illustrations, can be implemented by computer program instructions. These computer program instructions may be provided to a general-purpose computer, a processor of a special-purpose computer, and/or other programmable data processing apparatus to produce a machine, such that the instructions, executed via the computer processor and/or other programmable data processing apparatus, create a machine for A method that implements the functions/actions specified in the block diagram and/or flowchart blocks.

Correspondingly, the present application can also be implemented using hardware and/or software (including firmware, resident software, microcode, etc.). Furthermore, the present application may take the form of a computer program product on a computer-usable or computer-readable storage medium having computer-usable or computer-readable program code embodied in the medium for use by an instruction execution system or Used in conjunction with the instruction execution system. In the context of this application, a computer-usable or computer-readable medium may be any medium that can contain, store, communicate, transmit, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. device or equipment use.

Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the invention. In this way, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims (32)

1. A method for issuing crowdsourcing tasks, characterized in that the method includes: According to the different development stages of the crowdsourcing platform, the platform stage of the crowdsourcing platform is divided into the first default platform stage with the development goal of increasing the order rate, and the second default platform stage with the development goal of maintaining the order rate. platform stage; When determining the pricing of a triggering task, a price prediction result set containing a preset online member's predicted order price for the task is obtained through a preset member order price prediction model and based on the price influencing factors of the task. Online members are all online members within the preset area where the task is located; From the price prediction result set, according to the principle of price from low to high, select the predicted order receiving price with the lowest price and determine it as the order issuing price of the task; Publish the order carrying the task ID of the task and the order price of the task to the crowdsourcing platform; When it is determined that the task has not been ordered within the preset time period, it is determined whether the order of the task has reached the preset number of releases. If it has reached the preset number of push rounds, it will be transferred to the system order dispatch process; If the preset number of push rounds is not reached, it is determined whether the order price of the task reaches the highest predicted price in the price prediction result set within the preset area. If the highest predicted price is reached, the order dispatch process is transferred to the system; If the highest predicted price is not reached, when it is confirmed that the crowdsourcing platform is in the first preset platform stage, the issuance price of the task will be increased according to the preset step by step price increase to obtain a new issuance price; When it is confirmed that the crowdsourcing platform is in the second preset platform stage, each time the order price of the task is adjusted, the lowest predicted order price in the price prediction result set is selected as the new price of the task. Issuing price. 2. The method of claim 1, wherein the preset member order price prediction model is determined as follows: Based on the vector machine model, determine the initial price prediction model and corresponding model parameters; The price influencing factors of historical tasks and task pricing are used as training data sets for the price prediction model, and the price prediction model is trained to obtain the preset member order price prediction model. 3. The method of claim 2, characterized in that, based on the vector machine model, determining the initial price prediction model and corresponding model parameters includes: Based on the vector machine model, select a kernel function as the initial price prediction model; According to the preset model parameter generation method, the kernel parameters corresponding to the kernel function are determined as the model parameters corresponding to the initial price prediction model. 4. The method according to claim 1 or 2, characterized in that the price influencing factors of the task include at least one of the following: Characteristics of individual tasks; Characteristics of the mission area; Characteristics of members who log in to the crowdsourcing platform. 5. The method of claim 4, wherein the characteristics of the single task include at least one of the following: Task reservation time; Type of task; Task difficulty level; Mission location. 6. The method of claim 4, wherein the characteristics of the task area include at least one of the following: Task price levels in the task area; Order density in the task area; Membership density in the mission area. 7. The method of claim 4, wherein the characteristics of members who log in to the crowdsourcing platform include at least one of the following: The number of orders received by members; Member’s historical order price; Member history order receiving area; Member’s order receiving time; Member location. 8. The method of claim 2, characterized in that using price influencing factors of historical tasks and task pricing as a training data set for the price prediction model, including: Obtain relevant information about all historical tasks on the crowdsourcing platform. The relevant information about historical tasks includes the price influencing factors of historical tasks and the order price; Determine the benchmark pricing for each type of task in each corresponding task area based on the historical task-related information; The price influencing factors of the historical tasks and the corresponding benchmark pricing are used as the training data set of the price prediction model. 9. The method of claim 8, wherein the price influencing factors of the historical tasks include task type and task service duration, and determining the benchmark pricing of each type of task in each corresponding task area includes: Classify all historical tasks according to the task type of the historical task; Conduct data mining analysis on the task service duration of all historical tasks of the same type, and obtain the task difficulty levels of all historical tasks of the same type; Based on the task difficulty levels and order prices of all historical tasks of the same type, the base pricing of the historical tasks of the same type in each corresponding task area is determined. 10. The method according to claim 9, characterized in that performing data mining analysis on the task service duration of all historical tasks of the same type includes: Use K-means cluster analysis method/artificial neural network to conduct data mining analysis on the task service time of all historical tasks of the same type. 11. The method of claim 9, wherein the price influencing factors of the historical tasks include task locations, and the same type is determined based on the task difficulty levels and order prices of all historical tasks of the same type. Baseline pricing for historical tasks in the corresponding task area, including: According to the task difficulty level and task location of all historical tasks of the same type, perform task area clustering on all historical tasks of the same type to obtain the task area of the historical tasks of the same type; Perform statistical analysis on the order prices of tasks in each task area corresponding to each type of historical tasks, and determine the benchmark price of each type of historical tasks in each corresponding task area. 12. The method of claim 2, wherein the price influencing factors of historical tasks and task pricing are used as training data sets for the price prediction model, and the price prediction model is trained to obtain the preset member order acceptance. Price prediction models, including: Use the price influencing factors of historical tasks and task pricing as inputs to the price prediction model, and obtain the member’s predicted order price for each historical task output by the price prediction model; The price prediction model is trained using a correlation analysis method based on the task pricing of the historical tasks and the predicted order price. 13. The method of claim 12, wherein the price prediction model is trained using a correlation analysis method based on the task pricing of the historical tasks and the predicted order price, including: Use the correlation analysis method to determine that the correlation between the predicted order price and task pricing of any historical task reaches the first preset threshold, or when the number of times the price prediction model is trained reaches the preset number, the price is determined The prediction model is the preset member order price prediction model. 14. The method of claim 12, wherein the correlation analysis method includes at least one of the following: Chart correlation analysis method; Covariance and covariance matrix methods; Correlation coefficient method; Univariate regression method; Multiple regression method; Information entropy method; Mutual information method. 15. The method according to claim 1, characterized in that, after determining the issuance price of the task according to the price influencing factors of the task, it further includes: When it is determined that the task is accepted, the order price in the completed work order information of the task is used as the task price, and the price influencing factors of the task in the completed work order information are saved to the preset member order price prediction model training data set. 16. A device for issuing crowdsourcing tasks, characterized in that the device includes: a processor and a transceiver, wherein the processor is configured to use the transceiver: According to the different development stages of the crowdsourcing platform, the platform stage of the crowdsourcing platform is divided into the first default platform stage with the development goal of increasing the order rate, and the second default platform stage with the development goal of maintaining the order rate. platform stage; When determining the pricing of a triggering task, a price prediction result set containing a preset online member's predicted order price for the task is obtained through a preset member order price prediction model and based on the price influencing factors of the task. Online members are all online members within the preset area where the task is located; From the price prediction result set, according to the principle of price from low to high, select the predicted order receiving price with the lowest price and determine it as the order issuing price of the task; Publish the order carrying the task ID of the task and the order price of the task to the crowdsourcing platform; When it is determined that the task has not been ordered within the preset time period, it is determined whether the order of the task has reached the preset number of releases. If it has reached the preset number of push rounds, it will be transferred to the system order dispatch process; If the preset number of push rounds is not reached, it is determined whether the order price of the task reaches the highest predicted price in the price prediction result set within the preset area. If the highest predicted price is reached, the order dispatch process is transferred to the system; If the maximum predicted price is not reached, then When it is confirmed that the crowdsourcing platform is in the first preset platform stage, the issuance price of the task is added according to the preset ladder value, and a stepwise price increase is performed to obtain a new issuance price; When it is confirmed that the crowdsourcing platform is in the second preset platform stage, each time the order price of the task is adjusted, the lowest predicted order price in the price prediction result set is selected as the new price of the task. Issuing price. 17. The device of claim 16, wherein the processor determines the preset member order price prediction model in the following manner: Based on the vector machine model, determine the initial price prediction model and corresponding model parameters; The price influencing factors of historical tasks and task pricing are used as training data sets for the price prediction model, and the price prediction model is trained to obtain the preset member order price prediction model. 18. The device of claim 17, wherein the processor is specifically configured to: Based on the vector machine model, select a kernel function as the initial price prediction model; According to the preset model parameter generation method, the kernel parameters corresponding to the kernel function are determined as the model parameters corresponding to the initial price prediction model. 19. The device according to claim 16 or 17, wherein the price influencing factors of the task include at least one of the following: Characteristics of individual tasks; Characteristics of the mission area; Characteristics of members who log in to the crowdsourcing platform. 20. The device of claim 19, wherein the characteristics of the single task include at least one of the following: Task reservation time; Type of task; Task difficulty level; Mission location. 21. The device of claim 19, wherein the characteristics of the task area include at least one of the following: Task price levels in the task area; Order density in the task area; Membership density in the mission area. 22. The device of claim 19, wherein the characteristics of members who log in to the crowdsourcing platform include at least one of the following: The number of orders received by members; Member’s historical order price; Member history order receiving area; Member’s order receiving time; Member location. 23. The device of claim 17, wherein the processor is specifically configured to: Obtain relevant information about all historical tasks on the crowdsourcing platform. The relevant information about historical tasks includes the price influencing factors of historical tasks and the order price; Determine the benchmark pricing for each type of task in each corresponding task area based on the historical task-related information; The price influencing factors of the historical tasks and the corresponding benchmark pricing are used as the training data set of the price prediction model. 24. The device of claim 23, wherein the processor is specifically configured to: Classify all historical tasks according to the task type of the historical task; Conduct data mining analysis on the task service duration of all historical tasks of the same type, and obtain the task difficulty levels of all historical tasks of the same type; Based on the task difficulty levels and order prices of all historical tasks of the same type, the base pricing of the historical tasks of the same type in each corresponding task area is determined. 25. The device of claim 24, wherein the processor is specifically configured to perform data mining on the task service duration of all historical tasks of the same type using K-means cluster analysis method/artificial neural network. analyze. 26. The device of claim 24, wherein the processor is specifically configured to: perform processing on all historical tasks of the same type based on task difficulty levels and task locations of all historical tasks of the same type. Task area clustering to obtain task areas for historical tasks of the same type; Perform statistical analysis on the order prices of tasks in each task area corresponding to each type of historical tasks, and determine the benchmark price of each type of historical tasks in each corresponding task area. 27. The device of claim 17, wherein the processor is specifically configured to: use the price influencing factors of historical tasks and task pricing as input quantities of the price prediction model, and obtain the member output of the price prediction model for each member. The predicted order price of each historical task; The price prediction model is trained using a correlation analysis method based on the task pricing of the historical tasks and the predicted order price. 28. The device of claim 27, wherein the processor is specifically configured to: use a correlation analysis method to determine that the correlation between the predicted order price and task pricing of any historical task reaches a first preset threshold. when, or when the number of times the price prediction model is trained reaches a preset number of times, the price prediction model is determined to be the preset member order price prediction model. 29. The device of claim 27, wherein the correlation analysis method includes at least one of the following: Chart correlation analysis method; Covariance and covariance matrix methods; Correlation coefficient method; Univariate regression method; Multiple regression method; Information entropy method; Mutual information method. 30. The device of claim 16, wherein the processor is further configured to: When it is determined that the task is accepted, the order price in the completed work order information of the task is used as the task price, and the price influencing factors of the task in the completed work order information are saved to the preset member order price prediction model training data set. 31. A device for issuing crowdsourcing tasks, characterized in that the device includes: at least one processing unit and at least one storage unit, wherein the storage unit stores program code. When the program code is processed by When the unit is executed, the processing unit is caused to execute the steps of the device described in any one of claims 16 to 30. 32. A computer-storable medium with a computer program stored thereon, characterized in that when the program is executed by a processor, the steps of the method according to any one of claims 1 to 15 are implemented.