A Crowdsourcing Multi-Task Allocation Method to Prevent Malicious Bidding

Abstract

The present invention relates to a crowdsourcing multi-task allocation method for preventing malicious bidding, which involves the incentive problem under asymmetric information and belongs to the cross-technical field of computer science and economics. First, according to the preferred task set of the worker group and the bidding for this task set, a linear ratio is defined as the allocation basis, which makes the allocation function monotonic; then, according to the task preference set given by each worker and the calculated ratio, task allocation is carried out to obtain the set of workers to whom tasks are allocated; finally, the selected workers are paid, and the payment is the critical value payment to obtain the final task allocation result. According to the current crowdsourcing environment and the needs of the worker group, the present invention proposes multi-task crowdsourcing allocation; and ensures individual rationality, that is, the workers participating in the allocation will not suffer losses in interests, which indirectly improves the willingness of workers to participate; solves the problem of malicious bidding by the worker group in the crowdsourcing platform, and improves the anti-manipulation of the multi-task allocation method.

Description

Technical Field

[0001] The present invention relates to a crowdsourcing multi-task allocation method for preventing malicious bidding, which involves the incentive problem under asymmetric information and is an interdisciplinary technical field of computer science and economics. Background Art

[0002] In recent years, crowdsourcing has received extensive attention in both the industrial and academic fields. Jeff Howe gave the definition of crowdsourcing in 2006: the practice of a company or organization outsourcing the work tasks previously performed by employees to a non-specific (and usually large) public network in a free and voluntary manner. The tasks of crowdsourcing are usually undertaken by individuals, but may also involve tasks that require multiple people to collaborate to complete. Currently, crowdsourcing has been fully applied in many fields, such as identifying a large number of pictures on the Internet, evaluating the quality of online goods, and assessing the quality of machine translation results, etc. However, at the same time, the development of crowdsourcing has also encountered huge challenges. The worker group in the crowdsourcing system is uncertain, and it is very likely that they will submit false bids for their own interests to obtain higher rewards. Therefore, it is crucial to design an effective and feasible task allocation mechanism. At the same time, with the development of technology, the number of tasks to be solved is increasing day by day, and the relevance between various tasks is becoming more and more obvious, such as the complementarity and substitutability between tasks, which are important factors worthy of consideration. And the skills mastered by the worker group are also increasing day by day, and they are more inclined to complete a series of tasks with strong relevance. It is necessary to allocate multiple tasks in a package according to the preferences of the workers. Therefore, in the current situation of the development of crowdsourcing, reasonably allocating numerous tasks to workers and making payments can not only improve the performance of the crowdsourcing system, but also provide a new model for the development of the crowdsourcing platform.

[0003] Qin et al. (<Tsinghua Science and Technology>, 2018, 645 - 659) modeled the task assignment problem in crowdsourcing as a reverse auction, where the task poster is the buyer, the workers are the sellers, and the crowdsourcing platform acts as the auctioneer. The crowdsourcing platform first screens the workers and task posters within a certain time node; then it assigns tasks to the screened workers and pays the selected workers, with the payment provided by the selected task posters; finally, the task posters and workers rate each other. This method solves the situation where workers provide false bids for personal interests, but only considers the case of simple task assignment, that is, each worker is assigned at most one task. Zhao et al. (<33rd AAAI Conference on Artificial Intelligence>, 2019, 2629 - 2636) proposed a preference-aware spatial task assignment system based on workers' time preferences, using historical data and two other context matrices as auxiliary tools to simulate workers' preferences, but there is still a gap with the true preferences of workers and cannot fully reflect the true thoughts of workers. Aloufi et al. (<IEEE International IOT, Electronics and Mechatronics Conference>, 2020, 619 - 623) evaluated the performance of a spatial crowdsourcing task assignment method. The results on real datasets prove that this method improves the task assignment rate while protecting the location privacy of the worker group, but it does not consider the malicious bidding behavior of the worker group for their own interests. Pan et al. (<EURASIP Journal on Wireless Communications and Networking>, 2021, 1 - 21) proposed a two-stage GH-AT algorithm based on the greedy algorithm and the improved Hungarian algorithm, with the goal of minimizing the travel cost to solve the problem of three-objective online task assignment in spatio-temporal crowdsourcing, but it does not consider the misreporting problem caused by personal factors of participants, that is, it does not have good anti-malicious bidding performance. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a crowdsourcing multi-task assignment method for preventing malicious bidding to solve the problem of dishonest bidding of the worker group in the crowdsourcing platform.

[0005] The technical solution of the present invention is as follows: A crowdsourcing multi-task allocation method for preventing malicious bidding. First, a linear ratio is defined as the basis for task allocation, making the allocation function monotonic with respect to the workers' bids and preferred task sets. Then, according to the task preference sets given by each worker and the calculated ratio, task allocation is performed to obtain the set of workers to whom tasks are allocated. Finally, payment is made to this set, and the obtained payment is the pivotal value payment, resulting in the final task allocation result.

[0006] The specific steps are as follows:

[0007] Step1: The task publisher publishes n heterogeneous tasks on the crowdsourcing platform, represented by the set T = {1, 2, …, n}. There are m heterogeneous workers on the crowdsourcing platform, represented by the set W = {1, 2, …, m}. Each worker has and only has one task preference set S i and the overall bid v for this set i .

[0008] Step2: Start task allocation and initialize the relevant sets. The set of tasks that have been allocated The set of workers to whom tasks have been allocated The set of payments obtained by the group of workers is P = {p1, p2, …, p m}.

[0009] Step3: For the set of workers W to be allocated, select the workers who meet the conditions to join W f , and T f = T f ∪S i .

[0010] Step4: Make payments to the workers in the set W f .

[0011] The specific content of Step3 is as follows:

[0012] Step3.1: Define a ratio as the basis for allocation, which makes the task allocation process monotonic with respect to both valuations and task sets. That is, if worker i is allocated a task with a bid v i and a preferred task set S i , then if v i ′ < v i , worker i will be allocated a task with a bid v i ′, and if worker i has a preferred task set S i ′, worker i will also be allocated a task.

[0013] Step3.2: Calculate the ratio r of the workers in the set W i, sort them in ascending order and use the monotonically increasing sequence \(R = r\) (1) , \(r\) (2) , …, \(r\) (m) to represent.

[0014] Step3.3: Select workers in turn according to the monotonically increasing sequence \(R\). Select worker \(i\). If its set of preferred tasks add it to the set \(W\) f , and \(T\) f = \(T\) f ∪ \(S\) i . Otherwise, do not add.

[0015] Step3.4: Repeat Step3.3 until \(|T\) f | = \(|T|\) or the set of workers \(W\) has been fully traversed, and end the task assignment.

[0016] Specifically, Step3.2 is as follows:

[0017] Step3.2.1: Calculate the ratio \(r\) of worker \(i\) in the set \(W\) i , if then add \(i\) to the sequence \(R\), otherwise go to Step3.2.2.

[0018] Step3.2.2: Compare \(r\) i with the element values in the sequence \(R\) in turn. If an element can be found such that \(r\) i is less than the value of this element, then insert it in the position before this element. Otherwise, go to Step3.2.3.

[0019] Step3.2.3: Insert \(r\) i at the end of the sequence \(R\).

[0020] Step3.2.4: Repeat Step 3.2.1 - Step 3.2.3 until the set \(W\) has been fully traversed and the sequence \(R\) is constructed.

[0021] Specifically, Step4 is as follows:

[0022] Step4.1: Define a task-related set, represented by \(T\) a to represent.

[0023] Step4.2: Initialize For worker \(i\) in the set \(W\) f , sequentially find worker \(j\) according to the monotonically increasing sequence \(R\). Worker \(j\) determines the critical value payment of worker \(i\).

[0024] Step4.3: If \(j\in W\) f and \(i\neq j\), then \(T\) a= T a ∪ S i , go to Step4.4.

[0025] Step4.4: If and then

[0026]

[0027] Step4.5: Repeat Step4.2 - Step4.4 until all workers in the set W f have received payment or the sequence R has been fully traversed, and the payment is completed.

[0028] The principle of the present invention is:

[0029] With the development of Internet technology, it has become increasingly convenient for workers to obtain information, and the skills they master are also becoming more and more abundant. At the same time, people's needs are becoming increasingly complex, and it has become a common phenomenon for a worker to receive and complete multiple tasks. Generally speaking, the tasks published on the crowdsourcing platform can be divided into two categories: one is relatively simple and independent simple tasks, such as online Q&A, image recognition, data annotation, online human translation, etc.; the other is relatively complex and related complex tasks, and the rewards that workers expect to receive for completing these tasks vary depending on the combination of tasks. When designing a task allocation method, the complementarity or substitutability between tasks is a factor worthy of consideration. Considering the above factors, the present invention proposes a crowdsourcing multi-task allocation that meets the needs of the multi-task crowdsourcing background and the worker group. Workers participate in the allocation process by submitting a set of preferred tasks and a total bid for the entire set. If finally assigned tasks, they will surely be assigned the entire set of preferred tasks.

[0030] Therefore, by achieving incentive compatibility under the single-minded model (each worker has one and only one set of task preferences and a total bid for the entire set, and if assigned tasks, will be assigned the entire set of his task preferences), it is possible to encourage the worker group to honestly provide their bids, if and only if the following two conditions are met: ① The allocation function is monotonic with respect to the information of the worker group; ② The payment received by the worker assigned to the task is his critical value payment, that is, the critical value between being assigned the task and not being assigned the task.

[0031] The worker group is the core force of the crowdsourcing platform. Without them, the crowdsourcing platform cannot function properly. Crowdsourcing has been widely used as a new model for data collection and problem solving. It uses the "power of the crowd" to meet real challenges. When a task is released through the crowdsourcing platform, it must be accepted and completed by the workers, and then the completed results are submitted back to the crowdsourcing platform. In order to make workers have a basic willingness to participate, the present invention introduces and realizes individual rationality-each worker who participates in the crowdsourcing task allocation will obtain non-negative benefits, that is, for the worker who is assigned the task, the payment he receives will not be lower than his bid. Therefore, the process of worker group participation in task allocation will not cause them to lose benefits, and their enthusiasm for participation is improved from the side. The relevant theoretical proof is as follows:

[0032] Assume that worker i’s task preference set is S i , the bid is v i , the payment received is p i , the utility is u i .

[0033] (1) Worker i is assigned a task

[0034]

[0035] (2) Worker i is not assigned a task

[0036] u i =p i -v i =0-0=0.

[0037] Therefore, the present invention satisfies individual rationality.

[0038] For multi-task allocation, many methods focus on solving the problems of task completion quality and low cost. These methods all assume that the worker group honestly provides their abilities and accurately estimates the compensation corresponding to the labor they provide, which does not conform to the actual situation. The worker group is complex and they may behave dishonestly for their own interests. The present invention models it as a reverse auction model, and incorporates the idea of the incentive mechanism into the two links of allocation and payment in the auction model, achieving incentive compatibility. Whether the worker group's bid for the task preference set increases or decreases, it will not be higher than the benefits obtained by honest bidding. It has good anti-malicious bidding performance and provides a solid and strong foundation for further research. The relevant theoretical proofs are as follows:

[0039] Assume that worker i’s task preference set is S i , the honest bidding is v i , the payment received is p i , the utility is u i ; False bidding is v i ′, the payment obtained is pi ′, with utility u i ′.

[0040] (1) When worker i bids honestly and is not assigned a task, p i = 0, u i = 0

[0041] ① High-bidding strategy, i.e., v i < v′ i

[0042] Then i.e., r i < r′ i , selecting workers who may be assigned tasks in ascending order, then worker i will still not be assigned a task, so u i = u′ i = 0.

[0043] ② Low-bidding strategy, i.e., v i > v′ i

[0044] Then i.e., r i > r′ i , selecting workers who may be assigned tasks in ascending order, then there are the following two cases:

[0045] A: Worker i will still not be assigned a task, so u i = u′ i = 0.

[0046] B: Worker i is assigned a task, then there must be a worker j such that his payment is the critical value payment, and there is

[0047] Then The utility of worker i is negative.

[0048] (2) When worker i bids honestly and is assigned a task, p i > 0, u i ≥ 0

[0049] ① High-bidding strategy, i.e., v i < v′ i

[0050] Then i.e., r i < r′ i , selecting workers who may be assigned tasks in ascending order, then there are the following two cases:

[0051] A: Worker i will not be assigned a task, so u′ i = 0.

[0052] B: If worker i is assigned a task, then there must be a same worker j such that his payment is the pivotal value payment, and there is then

[0053] ② Low-bidding strategy, i.e., v i > v' i

[0054] If worker i is assigned a task, then there must be a same worker j such that his payment is the pivotal value payment, and there is then

[0055] In summary, the present invention satisfies incentive compatibility.

[0056] The beneficial effects of the present invention are as follows: According to the current crowdsourcing environment and the needs of the worker group, the present invention proposes multi-task crowdsourcing allocation; and ensures individual rationality, that is, the workers participating in the allocation will not have losses in interests, which improves the participation willingness of the workers indirectly; at the same time, an incentive mechanism is introduced to solve the problem of malicious bidding of the worker group in the crowdsourcing platform and improve the anti-manipulation of the multi-task allocation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 is the step flow chart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] The present invention will be further described below in conjunction with the drawings and specific embodiments.

[0059] Embodiment 1: As Figure 1 shown, a crowdsourcing multi-task allocation method for preventing malicious bidding first defines a linear ratio as the basis for task allocation, making the allocation function monotonic with respect to the workers' bids and preferred task sets; then tasks are allocated according to the task preference sets given by each worker and the calculated ratio to obtain the set of workers assigned tasks; finally, payments are made to this set, and the obtained payments are pivotal value payments to obtain the final task allocation result.

[0060] The specific steps of the method are as follows:

[0061] Step1: There are 20 workers and 10 tasks. The worker set is W = {w1, w2,..., w 20}, and the task set is T = {t1, t2,..., t 10}. The preferred task sets and bidding information of the worker group are shown in Table 1.

[0062] Table 1

[0063] Worker ID Set of Preferred Tasks Bid <![CDATA[w1]]> <![CDATA[{t3}]]> 2.82 <![CDATA[w2]]> <![CDATA[{t3}]]> 5.30 <![CDATA[w3]]> <![CDATA[{t9,t2,t4}]]> 5.86 <![CDATA[w4]]> <![CDATA[{t7,t5,t1}]]> 3.51 <![CDATA[w5]]> <![CDATA[{t7,t2,t1}]]> 4.09 <![CDATA[w6]]> <![CDATA[{t4,t5,t8}]]> 2.27 <![CDATA[w7]]> <![CDATA[{t3,t9}]]> 3.71 <![CDATA[w8]]> <![CDATA[{t6,t8}]]> 3.72 <![CDATA[w9]]> <![CDATA[{t2,t9}]]> 3.52 <![CDATA[w 10 > <![CDATA[{t1}]]> 2.31 <![CDATA[w 11 > <![CDATA[{t2,t6,t9}]]> 2.42 <![CDATA[w 12 > <![CDATA[{t3}]]> 4.09 <![CDATA[w 13 > <![CDATA[{t8}]]> 3.49 <![CDATA[w 14 > <![CDATA[{t6}]]> 2.93 <![CDATA[w 15 > <![CDATA[{t3,t9,t5}]]> 2.46 <![CDATA[w 16 > <![CDATA[{t4,t8,t9}]]> 3.82 <![CDATA[w 17 > <![CDATA[{t 10 ,t9}]]> 3.80 <![CDATA[w 18 > <![CDATA[{t9}]]> 1.96 <![CDATA[w 19 > <![CDATA[{t4}]]> 3.57 <![CDATA[w 20 > <![CDATA[{t1,t 10 ,t7}]]> 5.44

[0064] Step 2: Start task allocation and initialize relevant sets, including the set of tasks that have been allocated The set of workers to whom tasks have been allocated The set of payments obtained by the group of workers is P = {p1, p2,..., p 20}.

[0065] Step 3: For the set of workers W to be allocated, select workers who meet the conditions and add them to W f , and T f = T f ∪ S i .

[0066] Step 3.1: Define a ratio as the basis for allocation, which makes the task allocation process monotonic for both valuations and task sets. That is, if worker i is allocated a task with a bid v i and a preferred task set S i , then if v i ′ < v i , worker i will be allocated a task with a bid v i ′; if worker i is also allocated a task with a preferred task set S i ′.

[0067] Step 3.2: Calculate the ratio r i of the workers in set W, sort them in ascending order, and represent them with a monotonically increasing sequence R = (r (1) , r (2) , …, r (m) ).

[0068] Step 3.2.1: Calculate the ratio r i of worker i in set W. If , then add i to sequence R; otherwise, go to Step 3.2.2.

[0069] Step 3.2.2: Compare the value of r i with the elements in sequence R one by one. If an element can be found such that r i is less than the value of that element, then insert it into the position before that element; otherwise, go to Step 3.2.3.

[0070] Step 3.2.3: Insert r i at the end of sequence R.

[0071] Step 3.2.4 Repeat Step 3.2.1 - Step 3.2.3 until set W has been fully traversed and sequence R is constructed.

[0072] The construction process of sequence R is shown in Table 2.

[0073] Table 2

[0074]

[0075]

[0076] Step3.3: Select workers in sequence according to the monotonically increasing sequence R. If worker i is selected and its set of preferred tasks Add it to the set W f , and T f = T f ∪ S i ; otherwise, do not add.

[0077] Step3.4: Repeat Step3.3 until | f ||T| or the set of workers W has been fully traversed, and the task assignment ends.

[0078] The task assignment process and related sequence information are shown in Table 3.

[0079] Table 3

[0080]

[0081]

[0082] Step4: Pay the workers in the set W f = {w6, w 11 , w 10 , w1}.

[0083] Step4.1: Define a task-related set, denoted by T a .

[0084] Step4.2: Initialize For worker i in the set W f , sequentially search for worker j according to the monotonically increasing sequence R. Worker j determines the critical value payment of worker i.

[0085] Step4.3: If j ∈ W f and i ≠ j, then T a = T a ∪ S i , and go to Step4.4.

[0086] Step4.4: If and then

[0087]

[0088] Step 4.5: Repeat Step 4.2 - Step 4.4 until all workers in set W f have received payment or sequence R has been fully traversed, and the payment is completed.

[0089] The payment process for worker w6 is shown in Table 4.

[0090] Table 4

[0091]

[0092] For worker w 11 the payment process is shown in Table 5.

[0093] Table 5

[0094]

[0095] For worker w 10 the payment process is shown in Table 6.

[0096] Table 6

[0097]

[0098] The payment process for worker w1 is shown in Table 7.

[0099] Table 7

[0100]

[0101]

[0102] The allocation result is shown in Table 8.

[0103] Table 8

[0104] Worker ID Set of Preferred Tasks Bid Payment <![CDATA[w6]]> <![CDATA[{t4,t5,t8}]]> 2.27 3.51 <![CDATA[w 11 > <![CDATA[{t2,t6,t9}]]> 2.42 3.39 <![CDATA[w 10 > <![CDATA[{t1}]]> 2.31 3.14 <![CDATA[w1]]> <![CDATA[{t3}]]> 2.82 4.09

[0105] According to Embodiment 1, the present invention realizes crowdsourcing multi - task allocation, and the payments obtained by the workers assigned tasks are all higher than their bids, satisfying individual rationality. At the same time, the present invention also realizes incentive compatibility, which is illustrated by the following two cases.

[0106] (1) Workers who bid honestly are assigned tasks, and those who bid falsely are not assigned tasks.

[0107] In the case where the worker group bids honestly, from Table 8, it can be seen that worker w 10 is assigned a task. If worker w 10Out of self - interest, he wants to increase his bid to get a higher payment. That is, he honestly bids 2.31, but participates in the task assignment with a false bid of 3.8. The other workers still bid honestly, and the task assignment is carried out. According to the assignment method described in the present invention, the assignment result is shown in Table 9.

[0108] Table 9

[0109] Worker ID Set of Preferred Tasks Bid Payment <![CDATA[w6]]> <![CDATA[{t4,t5,t8}]]> 2.27 3.51 <![CDATA[w 11 > <![CDATA[{t2,t6,t9}]]> 2.42 3.39 <![CDATA[w1]]> <![CDATA[{t3}]]> 2.82 4.09 <![CDATA[w 20 > <![CDATA[{t1,t 10 ,t7}]]> 5.44 6.58

[0110] From Table 9, it can be seen that when worker w 10 provides a false bid, he is no longer assigned tasks and does not receive payment. Therefore, worker w 10 will only honestly provide his bid.

[0111] (2) A worker's honest bid is not assigned tasks, while a false bid is assigned tasks

[0112] In the case where the worker group bids honestly, from Table 8, it can be seen that worker w 12 is not assigned tasks. If worker w 12 out of self - interest, wants to lower his bid so as to be assigned tasks to obtain payment. That is, he honestly bids 4.09, but participates in the task assignment with a false bid of 2.22. The other workers still bid honestly, and the task assignment is carried out. According to the assignment method described in the present invention, the assignment result is shown in Table 10.

[0113] Table 10

[0114] Worker ID Set of Preferred Tasks True Bid False Bid Payment <![CDATA[w6]]> <![CDATA[{t4,t5,t8}]]> 2.27 2.27 3.51 <![CDATA[w 11 > <![CDATA[{t2,t6,t9}]]> 2.42 2.42 3.39 <![CDATA[w 12 > <![CDATA[{t3}]]> 4.09 2.22 2.82 <![CDATA[w 10 > <![CDATA[{t1}]]> 2.31 2.31 3.14

[0115] From Table 10, it can be seen that when worker w 12 provides a false bid, he is assigned tasks and also receives payment, but the payment he receives is lower than his true bid, making his utility negative, which harms the interests of worker w 12 . Therefore, worker w 12 will only honestly provide his bid.

[0116] Based on the above two situations, it can be seen that the worker group, out of self - interest, will only honestly provide their bids, which makes the present invention satisfy incentive compatibility.

[0117] The specific implementation manners of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the above - mentioned implementation manners. Within the scope of knowledge possessed by those of ordinary skill in the art, various changes can be made without departing from the gist of the present invention.

Claims

1. A crowdsourcing multi-task allocation method for preventing malicious bidding, characterized in that: First, define a linear ratio as the basis for task allocation, making the allocation function monotonic with respect to the workers' bids and preferred task sets. Then, based on the task preference sets given by each worker and the calculated ratio, perform task allocation to obtain the set of workers to whom tasks are assigned. Finally, make payments to this set, with the obtained payment being the pivotal value payment, to obtain the final task allocation result. The specific steps are as follows: Step1: The task publisher publishes n heterogeneous tasks on the crowdsourcing platform, represented by the set T = {1, 2, …, n}. There are m heterogeneous workers on the crowdsourcing platform, represented by the set W = {1, 2, …, m}. Each worker has exactly one task preference set S i and the overall bid v for this set i ; Step 2: Start task assignment and initialize relevant sets, including the set of tasks that have been assigned The set of workers who have been assigned tasks The set of payments obtained by the group of workers is P = {p1, p2, …, p m}; Step 3: For the set of workers \(W\) to be assigned, select the workers that meet the conditions and add them to \(W\). f , and \(T\) f = \(T\) f \(\cup S\) i ; Step4: Pay the workers in set W f ; Specifically, Step 3 is as follows: Step 3.1: Define a ratio Let \(i = 1, 2, \ldots, m\) be used as the allocation basis, that is, if worker \(i\) bids \(v\) i and the preferred task set \(S\) i obtains the task allocation, then if \(v\) i ' < \(v\) i , worker \(i\) bids \(v\) i ' will be assigned the task. If worker \(i\) uses the preferred task set \(S\) i ' will also be assigned the task; Step 3.2: Calculate the ratio r of the workers in set W i , sort them in ascending order, and use the monotonically increasing sequence R = r (1) , r (2) , …, r (m) to represent; Step 3.3: Select workers in sequence according to the monotonically increasing sequence R. If worker i is selected and its set of preferred tasks add it to the set W f , and T f = T f ∪ S i ; otherwise, do not add it. Step3.4: Step3.3 until |T f | = |T| or the worker set W has been fully traversed, and the task assignment ends.

2. The crowdsourcing multi-task allocation method for preventing malicious bidding according to claim 1, wherein Specifically, Step 3.2 is as follows: Step 3.2.1: Calculate the ratio r of worker i in set W i , if then add i to sequence R, otherwise go to Step 3.2.2; Step3.2.2: Compare r i with the element values in sequence R one by one. If an element can be found such that r i is less than the value of this element, then insert it into the position before this element; otherwise, go to Step3.2.

3. Step3.2.3: Insert r i at the end of sequence R; Step 3.2.4: Repeat Step 3.2.1 - Step 3.2.3 until the set W has been fully traversed and the sequence R is completed.

3. The crowdsourcing multi-task allocation method for preventing malicious bidding according to claim 1, characterized in that Specifically, Step 4 is as follows: Step4.1: Define a task-related set, denoted by T a as shown Step4.2: Initialization For worker i in set W f search for worker j in ascending order according to sequence R successively, and worker j determines the critical value payment of worker i; Step4.3: If j ∈ W f and i ≠ j, then T a = T a ∪ S i , go to Step4.4; Step4.4: If and then Step4.5: Repeat Step4.2 - Step4.4 until all workers in set W f have been paid or sequence R has been fully traversed, and the payment is completed.

Images