CN114638415A

CN114638415A - Real-time space crowdsourcing task allocation method based on Geohash index

Info

Publication number: CN114638415A
Application number: CN202210263794.5A
Authority: CN
Inventors: 李云辉; 古天龙
Original assignee: Guilin University of Electronic Technology
Current assignee: Guilin University of Electronic Technology
Priority date: 2022-03-17
Filing date: 2022-03-17
Publication date: 2022-06-17

Abstract

The invention relates to the technical field of computers, in particular to a real-time space crowdsourcing task allocation method based on a Geohash index.

Description

Real-time space crowdsourcing task allocation method based on Geohash index

Technical Field

Background

Spatial Crowdsourcing (SC) refers to various types of data collection and sharing using ubiquitous mobile devices and mobile networks, and the working mode of Spatial Crowdsourcing is to recruit workers using a network and ask the workers to a specific location to perform a task. Which tasks are assigned to which workers is a major research issue for spatial crowdsourcing, namely task assignment (task assignment). Most of the existing SC task assignment studies are mainly from the perspective of task requesters (i.e., users who issue tasks) to formulate task assignment schemes. Such a scheme aims at optimizing the objectives of the task requester, such as maximizing the number of tasks assigned, maximizing the quality of the returned results, minimizing the remuneration paid. There are also a few SC task allocation methods that address the goals of optimizing the worker, such as minimizing unfairness in task allocation, maximizing rewards earned, or minimizing worker travel distance. However, concern over the desire of one party alone may defeat the motivation of the other party to participate in the crowdsourcing task, thereby compromising the growth of the crowdsourcing market. On the one hand, most space crowdsourcing task allocation methods limit the allocation to completion before the task can be completed, and do not fully consider the expectation that the task requester wants the task to be completed as soon as possible. On the other hand, workers often strike a balance between the amount paid and the cost paid, deciding whether to participate in a particular crowdsourcing task.

Furthermore, location features are the most important features for spatial crowd sourcing, and existing spatial task allocation methods all take into account the acceptable distance between workers or tasks. The reality is that the crowdsourcing platform is very large in the number of tasks and workers that arrive instantaneously. Computing the distance between the task and the worker in a large data set is very time consuming.

Disclosure of Invention

The invention aims to provide a real-time spatial crowdsourcing task allocation method based on a Geohash index, and solves the technical problem that the conventional spatial crowdsourcing task allocation method does not consider the expected targets of a task requester and a worker at the same time and cannot work on a large data set well.

In order to achieve the above object, the present invention provides a real-time spatial crowdsourcing task allocation method based on a Geohash index, comprising the following steps:

dividing the existing tasks to be distributed into blocks according to geographical positions by using a Geohash algorithm, wherein each block is represented by a character string called Geohash coding;

determining a block to which a new worker belongs;

calculating and acquiring a coverage task set of the new worker, and scoring the tasks in the coverage task set;

ranking each task based on the score;

and comprehensively scoring the number of task acceptance intentions of the new worker and the sequencing result to perform task allocation.

Further, in the process of dividing all tasks into blocks by using a Geohash algorithm, all tasks are divided into blocks according to the geographic positions of the tasks and in consideration of the radius range of the tasks acceptable by workers, and each block is respectively provided with a unique character string identifier.

Further, in determining the block to which the new worker belongs, if the new worker is located near the boundary of the divided block, the block coverage to which the new worker belongs is extended to 8 blocks adjacent to the divided block.

Further, the calculation of the coverage task set is determined according to the following relationship:

distance between the worker and the task within the worker's coverage area;

the arrival time of the worker and the release time of the task;

whether the task has been assigned to enough workers.

Further, the task in the coverage task set is scored according to the balance of comprehensively considering the waiting time of the task and the consideration and the travel cost of the worker, so that the waiting time of the task is minimized, and the reward acquisition and the travel cost of the worker are maximized.

Further, a binary insertion sorting method is adopted to sort each task based on scores.

Further, in the process of comprehensively scoring the number of task acceptance intentions of the new worker and the sorting result for task allocation, according to the number of task acceptance intentions of the new worker, the same number of tasks are selected from high to low according to the task scoring result and are allocated to the new worker.

The invention provides a real-time space crowdsourcing task allocation method based on a Geohash index, which comprises the steps of firstly determining a task area covered by each incoming worker by utilizing a Geohash algorithm, then converting double-target optimization of the workers on minimizing a trip distance and maximizing acquired reward into single-target optimization by utilizing a linear weighting and evaluating function, grading the tasks in a covered task set, and performing task allocation by integrating the task acceptance intention quantity and the sequencing result of the workers, thereby solving the technical problems that the current space crowdsourcing task allocation method lacks the technical problems of simultaneously considering the expected targets of a task requester and the workers and cannot work on a large data set well.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.

Fig. 1 is a schematic flow chart of a real-time spatial crowdsourcing task allocation method based on a Geohash index according to the present invention.

FIG. 2 is a schematic diagram of geographic location blocks of an embodiment of the present invention.

FIG. 3 is a comparison graph of average worker travel distances in an embodiment of the invention.

FIG. 4 is a comparison graph of average rewards earned by workers in a particular embodiment of the invention.

Fig. 5 is a graph comparing average delays for tasks using various methods in an embodiment of the invention.

FIG. 6 is a graph comparing CPU cost using various methods in an embodiment of the present invention.

Detailed Description

Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.

Referring to fig. 1, the present invention provides a real-time spatial crowdsourcing task allocation method based on Geohash index, which includes the following steps:

s1: dividing the existing tasks to be distributed into blocks according to geographical positions by using a Geohash algorithm;

s2: determining a block to which a new worker belongs;

s3: calculating and acquiring a coverage task set of the new worker, and scoring the tasks in the coverage task set;

s4: ranking each task based on the score;

s5: and comprehensively scoring the number of task acceptance intentions of the new worker and the sequencing result to perform task allocation.

The real-time spatial crowdsourcing task allocation method based on the Geohash index is further elaborated by combining specific embodiments as follows:

first for the case when a task is assigned to a single worker:

step 1, selecting a proper Geohash coding length, and establishing an index of an existing task by using a Geohash algorithm, as shown in FIG. 2, the Geohash coding length is 4, and the task t is₁The Geohash index of (1) is a character string of '9 qh 1', task t₂The Geohash index of (a) is the string "9 qh 0".

Step 2, for the newly arrived worker, determining the task coverage area where the newly arrived worker is, as shown in fig. 1, the worker w₁Is the character string "9 qh 0", and thus the block denoted by "9 qh 0" and its adjacent 8 blocks are w₁Task coverage area CG (w)₁)。

Step 3, aiming at coverage area CG (w)₁) Task within, according to task and worker w₁Distance between, time of task release, time of worker arrivalWorker w₁The radius range of the acceptable tasks, the number of assignments of the tasks, the number of tasks acceptable to the worker, the tasks covered by the worker are determined, and the tasks are scored.

In particular if task t_j∈CG(w₁) Is released by the time r_jIs less than or equal to w₁Time of arrival a₁,t_jAnd w₁The distance between

Less than or equal to R, t_jThe number of allocations required maxW_j>0, worker w₁Acceptable number of tasks maxT₁>0, then task t_jIs by w₁An overlay task, and a candidate task. Suppose using C (w)₁) Denotes w₁For any t, then_j∈C(w₁) Having t of_j∈CW(w₁) And satisfies the conditions (1) to (4):

r_j≤a_i (1)

maxW_j>0 (3)

maxT₁>0 (4)

for any t_j∈C(w₁) Evaluate "worker-task" pairs (w)₁,t_j). This is a multi-objective optimization problem, since the objective of task assignment needs to consider the trade-off between worker return and travel cost, and also consider the delay of waiting for assignment after the task is released. In multi-objective optimization, on the one hand, multiple objectives may be contradictory. For example, to maximize reward and minimize latency. Therefore, it is necessary to convert a plurality of targets into the same-direction targets. On the other hand, in multi-objective optimization, the values of the targets may be very different, and the influence of the large-value target on the scoring result is larger, so that the specification is required before scoringAnd (4) transforming each target. The invention utilizes linear weighting and quantification of the balance between reward and travel cost of workers, utilizes multiplication function quantification and considers the multi-objective optimization problem of the balance of workers and task delay at the same time, as shown in formulas (5) to (8).

Equations (6) - (8) convert all optimization objectives into a maximization problem and normalize the values to (0, 1) at the same time]. Wherein the content of the first and second substances,

represents the worker w_iAnd task t_jThe distance between the two or more of the two or more,

represents the worker w_iCompletion of task t_jThe remuneration to be obtained is,

representing a task t_jIs assigned to worker w_iTime delay of previous wait, weight coefficient w₀And w₁Indicating the preference of the worker for reward and distance traveled, w₀+w₁＝1,∑_v∈{0,1}

w

_v1. The invention assumes that the task allocation satisfies the flexible adjustment of the distance of travel between the remuneration and the payment of the worker. For example, by setting w₀＝w₁The method can lead the worker to preferentially select the task with small trip distance from a plurality of candidate tasks with the same reward; on the contrary, among a plurality of candidate tasks with the same travel distance, a task with a large reward is preferentially selected. And w may be set if the worker is more likely to see the reward for completing the task₀<w₁. The present invention best attempts to satisfy the trade-offs and task concerns of workersIn order to reduce the task latency requirement, equation 5 is shown. Therefore, the formula in the formula (5) is used

To worker w_iAnd its candidate task t_j∈C(w₁) Is scored. When the subscript i is 1, i.e. the worker w₁。

Step 4, adopting a binary insertion sorting method to pair C (w)₁) And (6) sorting.

Step 5, selecting the front maxT₁Assignment of tasks to workers w₁. At the same time, the work to be allocated remains the number of needs to be allocated minus 1.

In reality, a large number of workers emerge on a crowdsourcing platform in a short time. I.e., often in reality, need to be allocated concurrently to a large number of workers. Thus, the following experiment simulates the case of concurrent assignment of tasks.

Large worker Allocation experiments

1. A data set.

The present invention uses check-in data from Gowalla in the United states of 10 2010. The data set includes a user ID, a venue location, a check-in time, and a location randomly selected from a plurality of check-in records of the user as the user's location. A total of 151849 tasks and 18203 workers were extracted. The reward for the task, the number of people required by the worker and the number of tasks to be distributed are generated with an even distribution. And setting the radius R of the task acceptable to the worker as 30km, w₀＝w₁＝0.5。

The workers were divided into 50 groups and tasks were assigned to the 50 groups of workers concurrently using multiple processes.

2. Evaluation index

(1) Average distance of travel

(2) Average acquired reward

(3) Average time delay

(4) Cost of the CPU: complete all tasks-the cost of CPU time allocated by the worker.

3. Comparison algorithm

(1) TAW: from the perspective of the worker alone, maximizing reward is considered while minimizing travel distance.

(2) DSTA-A considers the trade-off of workers in terms of obtaining reward and distance between trips, and simultaneously considers reducing the task waiting time delay.

(3) The DSTA-GH is a real-time space crowdsourcing task allocation algorithm based on the Geohash index. The algorithm introduces a Geohash algorithm on the basis of DSTA-A, and aims to improve the task allocation efficiency so that the task allocation algorithm can work on a large data set well.

4. Results of the experiment

The results of concurrent task allocation for 50 groups of workers are shown in FIGS. 3-5. Compared with the TAM, the space crowdsourcing task allocation algorithm which simultaneously considers the task delay and the balance of the worker on the reward and the travel cost, namely DSTA-A and DSTA-GH, greatly reduces the task waiting time (figure 5) and the travel cost of the worker (figure 3), and meanwhile, the reward obtained by the worker is also increased remarkably (figure 4).

The total CPU cost for all assignments between 50 group workers and all tasks is shown in FIG. 6. DSTA-A takes slightly more time than TAW, since it takes the extra delay of the task into account. But the DSTA-GH was shortened by about 99.97% compared to DSTA-a, indicating that DSTA-GH performed well on large scale data sets.

Compared with the prior art, the invention has the following advantages:

1. the task area of its valve cover is determined for each incoming worker using the Geohash algorithm. Since the Geohash is encoded by using dichotomy and Base32, the longitude range-180, 180 and the latitude range-90, 90 are divided into a plurality of sub-regions and Geohash blocks. Each sub-region is indexed with a string. For the current worker, the division of the previous task is already determined, and only the task area covered by the current worker, the sub-area to which the worker belongs and the adjacent 8 sub-areas need to be determined. The advantages are that: (1) the Geohash algorithm is simple, and the task area covered by the current worker can be quickly determined; (2) the Geohash algorithm filters out tasks which cannot be within the acceptable radius range of the workers, and reduces the search space, so that the efficiency of task retrieval is improved; (3) once the Geohash encoding length is determined, the extent of each sub-region does not change with the arrival of new tasks and workers, and therefore the Geohash data index does not need to be maintained at additional cost.

2. And converting the double-target optimization of the worker on minimizing the travel distance and maximizing the acquired reward into the single-target optimization by utilizing a linear weighting and evaluation function. The weighting coefficients are all set to be 0.5, namely, in a plurality of tasks with the same remuneration, a worker preferentially selects a task with a small travel distance; on the contrary, among a plurality of tasks with the same travel distance, the worker preferentially selects the task with much reward. The linear weighting and evaluation functions convert the multi-objective optimization problem into a single-objective optimization problem, simplify calculation, and can characterize the balance between the reward and the travel cost of workers.

3. And converting the optimization target of the worker and the double-target optimization problem of the task requester about minimizing the task delay into a single-target optimization problem by using a multiplication evaluation function. A multiplicative merit function is a function that strictly increases in respect of any one of its factors, and the value of the function is maximal only when all of its factors take a maximum value. Thus, the multiplicative evaluation function accurately describes the problem of dual target optimization with respect to the worker and the task requester, and is computationally simple.

While the invention has been described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. A real-time space crowdsourcing task allocation method based on a Geohash index is characterized by comprising the following steps:

dividing the existing tasks to be distributed into blocks according to geographical positions by using a Geohash algorithm;

determining a block to which a new worker belongs;

ranking each task based on the score;

2. The Geohash index-based real-time spatial crowd-sourced task allocation method according to claim 1,

in the process of dividing all tasks into blocks by using a Geohash algorithm, all tasks are divided into blocks according to the geographic positions of the tasks in consideration of the radius range of the tasks acceptable by workers, and each block is respectively provided with a unique character string identifier.

3. The Geohash index-based real-time spatial crowd-sourced task allocation method according to claim 1,

in determining the block to which the new worker belongs, if the new worker is located near the boundary of the divided block, the block coverage to which the new worker belongs is extended to 8 blocks adjacent to the divided block.

4. The Geohash index-based real-time spatial crowd-sourced task allocation method according to claim 1,

the calculation of the covering task set is determined according to the following relation:

distance between the worker and the task within the worker's coverage area;

the arrival time of the worker and the release time of the task;

whether the task has been assigned to enough workers.

5. The Geohash index-based real-time spatial crowd-sourced task allocation method according to claim 1,

and scoring the tasks in the coverage task set according to the balance of comprehensively considering task waiting time and worker consideration and trip cost, minimizing the task waiting time, and maximizing the worker obtaining remuneration and minimizing the trip cost.

6. The Geohash index-based real-time spatial crowd-sourced task allocation method according to claim 1,

and ranking the scores of each task by adopting a binary insertion ranking method.

7. The Geohash index-based real-time spatial crowd-sourced task allocation method according to claim 1,

and in the process of comprehensively scoring the task acceptance intention quantity and the sequencing result of the new worker for task allocation, selecting the same quantity of tasks from high to low according to the task acceptance intention quantity of the new worker and combining the task scoring result to allocate the tasks to the new worker.