Research on Mission Pricing of Crowdsourcing APP

In order to have a better study of pricing of crowdsourcing APP tasks, mission data was collected about an already completed project, including the location, pricing and fulfillment of each mission. Get the law of the task pricing of the completed projects by analyzing the location distribution of the task points and the members, and then analyze the reason of the task failure. Besides, combine the information data of the members to improve the original pricing model based on the K-means clustering. Focus on the task package pricing, design a better task pricing model to improve the success rate of the task.


Introduction
In the era of big data, the collection of data presents many complicated problems.It is hard to satisfy the diversified data needs of organizations for an organization with only individual organization.The self-service crowdsourcing platform based on mobile Internet such as "making pictures and making money" adopts the shared economy mode, that is, the crowdsourcing mode, which greatly saves the investigation cost and shortens the investigation period, so that it provides a large amount of data for the organization at a lower cost.
At present, the research on crowdsourcing task mainly focuses on the developmental status of crowdsourcing and its advantages.HOWE J (2006) proposed the concept of crowdsourcing.Yan Jie et al (2017) explored the definition, patterns, composition and workflow of crowdsourcing .Zhao Jiamin (2015) obtained the conclusion that the advantages of crowdsourcing are improving efficiency and the fully elastic production scale.Huang Mincong (2016) studied the development of crowdsourcing research platform and its operating mode, and proposed the ways to operating crowdsourcing platform.Liu Ya-ru (2016) thinks crowdsourcing delivery model is more excellent that the traditional distribution methods, which achieves the integration of social idle resources.However, with the development of crowdsourcing platforms, platforms have gradually exposed problems.Pang Jiangang (2015) put forward that Internet-based crowdsourced communities are good for exploiting the wisdom of the general public, but due to the asymmetry of information, there are some risks, so the risk management mechanism of crowdsourced communities should be improved.In response to the fraudulent activities in the crowdsourcing platform , Fei Youli et al. (2015) proposed that the credit evaluation system should be improved, the fraudulent screening mechanism should be optimized, and the relevant legal system should be established and improved.
Some researchers also study the distribution of crowdsourcing tasks from the dynamic crowdsourcing task.On the premise of considering the dynamic characteristics of distribution demand, Zhao Xinglong (2016) established a dynamic model of crowdsourcing distribution path to improve the public Package distribution staff word delivery revenue and customer service experience.Qian Guang (2017) introduced personalized service into crowdsourcing logistics model to explore the impact of personalization in total logistics on customer satisfaction, namely the completion of tasks and the failure of identification.Wu Jinghong and Lv Nengfang (2016) analyzed the key players such as the outsourcing parties and outsourcing parties in the crowdsourcing mode of express delivery position of the task is far away from more members, the corresponding will be increased.Another is the task pricing and the distribution of the task itself has a negative correlation between the concentration, that is, when the task distribution is more intensive, the all of tasks' price will be reduced.

Establishment of Pricing Model
(1) Establishment of pricing model In this paper, the static crowdsourcing task is considered, that is, the number of platform members does not change over time.The model includes multiple tasks and multiple platform members for one task.
Each member i a contains two attributes: one is the cost of completing the task i c , and another one is the number of tasks i w that the platform members can complete, namely the task reservation quota.However, in practice, the task publishers don't know these two attributes of the members, and their purpose is to be able to receive the tasks from the platform members under the lowest cost.Therefore, this paper judges the quality of the completion of the task based on the credit value of the member, so as to maximize the task publishers' profits.
Therefore, before determining the task pricing, the paper first explore the quality of the completed work by analyzing the creditworthiness of the members.At the pricing stage, according to the cost and the number of tasks that can be accomplished by the member, we use the pricing mechanism to design a platform member.
Finally, according to the first part of the quality results are sorted in descending order, priority to select high-quality members to complete the task.The specific algorithm is as follows: Crowdsourcing platform pricing mechanism design algorithm： Step 1：Enter the platform member attribute data   ( , ),( , ), ( , ) , j c relates to the distance between members and task and member's credit value, j w represents maximum limit for the task; Step 2: Initialize, set 1 i  , for each task i b , repeat the following steps: Step 3：Set 1 j  , sort the member data j a by   Step 4：If the current member j a has selected the number of tasks j j w w  , then 1  ,otherwise, repeat this step; Step 5：End step 2 cycle; Step 6：The algorithm is over.
When the unit price of task is i p c  , the number of tasks assigned to platform members i a can be expressed as i w .
There are three main reasons for designing this pricing strategy: First, the cost is sorted according to the distance between platform member and task location, which can avoid pricing too high or too low; Secondly, under the condition of maximizing the profit of the task publisher, select the tasks are priced low enough to control the cost of the task; Thirdly, for the platform members, the task price is in line with the expected return of most members, so it will have a high acceptance, and will not make the task unattended, leading to the failure of the task.
(2) Taking Shenzhen as an Example to Solve Task Pricing Model Taking Shenzhen as an example, substitute the task information data and membership information of Shenzhen into the pricing model based on mechanism design theory to implement this pricing scheme for different tasks in this project, and the crowdsourcing task are priced.The part of the results shown in  of the e task ondly, price of the task package.Among them select the maximum value of the task can offset the members alone to complete each task required in the package, and the average value of the task pricing is representative in the package.On the other hand, in order to meet the maximization of the revenue of the task publisher and the platform member, the ratio can be 0.618, the golden section, so that the task publisher has a minimum cost, and membership can receive the most effective task.
Task package pricing expression is as follows:   max 0.618 ( 1) 1, 2, Among them, P represents the cost of a task package, n represents the number of tasks packaged in a task package, max i p represents the maximum value of the pricing of tasks i p received by a member within a task package, p represents the mean value of the pricing of tasks received by members within a task package.
Based on the mechanism design theory of the task pricing model, using MATLAB software to calculate the task data in Shenzhen, get the tasks' numbers and price after being packaged as shown in Table 3. Comparing with 3.2 results, found that the packaged task in Shenzhen City, the total cost was 7952.33 yuan, and the total cost of packaging before the task was 11,152.33yuan.We can see that a more concentrated package of tasks can reduce the cost of the task publisher of publishing the task of pricing after being integrated to a great extent.
On the other hand, members with higher creditworthiness can choose the task package closest to them within the scope of their pre-defined task quota, and accomplish multiple tasks within the task package at one time, resulting in higher economic benefits.Compared with the selection of members after each task is released separately, the distribution of focused tasks after being packaged meets the target of reducing the member's paying cost and obtaining a considerable profit.
At the same time, by observing the formula of the task package pricing, the more number k of K-means clustering, the smaller of cost of the task publisher, but each member will be bound by the task quota when receiving the task.Therefore, when the number of clustering is too small, the number of tasks in each task package will be too much, which will result in many members can't receive the task package, resulting in the increase of the failure rate of the task.Therefore, it is necessary to determine the reasonable K-means clustering number k .

Practical Significance
Through the research of this paper, it is found that the task attribute and the membership attribute lead to different competitive conditions in the crowdsourcing task market, which causes the task publisher to adopt different bidding strategies.
Specific with the actual situation, the task of pricing may also have the following conditions: the higher price of crowdsourcing platform similar tasks recently, the higher the task pricing; the shorter the task requirements of the completion of the task, the higher the pricing, etc.All of these results are determined by the incentive mechanism theory.On the crowdsourcing task platform, the task of publishers and members are often expected to be able to get their own income maximization.Platform members will choose some of their more favorable tasks to complete.Therefore, high pricing of tasks is a big attraction for members.And task publishers balance other factors based on pricing, such as ensuring completion of tasks, quality of accomplishment, and so on.
However, platform members may not be able to accomplish the task efficiently and effective after receiving the task.
To ensure the quality of the task completion, the crowdsourcing platform should improve the business risk management.The platform not only needs to verify the real name of the platform member "package", but also improve the credit rating system.Tu Shuli (2015) mentioned that in the delivery of crowdsourcing delivery model, in order to avoid the loss and damage of items in the distribution process, the platform will take full insurance insurers, cargo loss compensation protection and other measures.Therefore, in order to ensure the successful completion of the task, the platform can imitate the delivery mode of crowdsourcing management, to collect margin members, the failure of the task of compensation protection and other measures.In the meantime, some certain task recipients will get quick answers but not right, which can optimize their time efficiency and earn more money.To avoid this phenomenon, crowdsourcing platforms should also take steps to identify counter-frauds.

Essay Innovation
In the process of solving the problem, this paper integrates the mechanism design theory and task pricingattribute information theory, and applies these theories in the new field of mobile Internet -the task pricing of self-service labor crowdsourcing platform.The paper explores the way for the development of crowdsourcing model in new field, and points out the influencing factors of task bidding, which is conducive to the adoption of shared economy model crowdsourcing website platform to better serve the task of publishers, but also providing some reference for the tasks of the task of pricing publisher.

Table 2 :
Table 2. Pricing of Tasks in Shenzhen Based on Mechanism Design Theory

Table 3 .
Shenzhen the tasks' numbers and price after being package