CN110196763A - A kind of efficient multi-task planning method of time domain continuous type space crowdsourcing - Google Patents
A kind of efficient multi-task planning method of time domain continuous type space crowdsourcing Download PDFInfo
- Publication number
- CN110196763A CN110196763A CN201910383620.0A CN201910383620A CN110196763A CN 110196763 A CN110196763 A CN 110196763A CN 201910383620 A CN201910383620 A CN 201910383620A CN 110196763 A CN110196763 A CN 110196763A
- Authority
- CN
- China
- Prior art keywords
- task
- crowdsourcing
- distribution
- algorithm
- time domain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
- G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/5038—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
Abstract
The invention discloses a kind of efficient multi-task planning methods of time domain continuous type space crowdsourcing, comprising the following specific steps (1) baseline algorithm, (2) are based on independent task union of sets row algorithm, the parallel algorithm of (3) based on allocation unit.By above-mentioned, the efficient multi-task planning method of time domain continuous type space crowdsourcing of the invention, after receiving multiple crowdsourcing task requests at the same time, it needs to maximize general assignment quality under conditions of general assignment limited budget, the distribution to multiple crowdsourcing tasks is completed, efficient parallel frame is proposed to handle distribution conflict, realizes the concurrent and efficient distribution of more crowdsourcing tasks, with stronger exploitativeness, it is more conducive to practical application.
Description
Technical field
The present invention relates to database, Temporal-spatial data management, data analysis, data mining, crowdsourcing, space crowdsourcing fields, special
It is not to be related to a kind of efficient multi-task planning method of time domain continuous type space crowdsourcing.
Background technique
The space crowdsourcing of time domain continuous type is a new class of space crowdsourcing form, requires constantly to divide in task completion time
Designated position is reached with worker and executes crowdsourcing task, and then obtains a series of perception data of Continuous-times.
Smart machine is widely used so that a large amount of time domain continuous type space crowdsourcing task is published, while to multiple crowds
The demand that packet task carries out task distribution is growing.For single time domain continuous space crowdsourcing task, can by the task when
Between be divided into equal part time quantum in dimension, completed to be cooperated by multiple worker's timesharing;In multiple the type crowdsourcing tasks
Under allocation scenarios, it is possible that worker distributes collision problem: i.e. some worker's synchronization is assigned to two differences
The same time quantum of task will seriously affect the efficiency of distribution to distribution conflict occur.
Summary of the invention
The invention mainly solves the technical problem of providing a kind of efficient multi-task plannings of time domain continuous type space crowdsourcing
Method proposes efficient parallel frame for the multi-task planning problem of time domain continuous type space crowdsourcing to handle distribution punching
It is prominent, realize the concurrent and efficient distribution of more crowdsourcing tasks.
In order to solve the above technical problems, one technical scheme adopted by the invention is that: it is empty to provide a kind of time domain continuous type
Between crowdsourcing efficient multi-task planning method, comprising the following specific steps
(1) baseline algorithm
The location information of the crowdsourcing task-set and worker that are received in given a period of time, each time domain continuous type is empty
Between crowdsourcing task the subtasks of timeslices such as be divided into, each subtask and worker carry out one-to-one matching, and are appointed according to son
Business is set as the maximization of general assignment quality, constraint condition with its corresponding expense of the distance between worker calculating, optimization aim
The worker for being set as limited to the master budget and quantity of task expense and distribution gathers;
(2) it is based on independent task union of sets row algorithm
Independent division is carried out to crowdsourcing task-set, guarantees that there is no distribution conflicts between each independent task collection, concurrently
Complete distribution;
(3) based on the parallel algorithm of allocation unit
Parallel algorithm based on allocation unit, allocation unit are a task, and each task in task-set includes point
With conflict list, the timeslice of distribution conflict can occur for record, and host process reads task and distributes a process for each task,
Then, each task completes distribution using approximate algorithm.
In a preferred embodiment of the present invention, the function of the general assignment quality is the sum of each crowdsourcing Task Quality, it is assumed that
Single crowdsourcing Task Quality is described with entropy function, and entire crowdsourcing task-set T includes | T | a crowdsourcing task, each task t are divided into m
The subtask of the timeslices such as a, then general assignment quality is
WhereinIndicate the quality of each subtask.
The beneficial effects of the present invention are: the efficient multi-task planning method of time domain continuous type space crowdsourcing of the invention,
After receiving multiple crowdsourcing task requests simultaneously, need to maximize general assignment quality under conditions of general assignment limited budget, it is complete
The distribution of pairs of multiple crowdsourcing tasks proposes efficient parallel frame to handle distribution conflict, realize more crowdsourcing tasks and
Hair and efficiently distribution, have stronger exploitativeness, are more conducive to practical application.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing, in which:
Fig. 1 is the experiment effect figure that the runing time under different task quantity compares;
Fig. 2 is the experimental result picture that the runing time under different task position distribution compares.
Specific embodiment
The technical scheme in the embodiments of the invention will be clearly and completely described below, it is clear that described implementation
Example is only a part of the embodiments of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, this field is common
Technical staff's all other embodiment obtained without making creative work belongs to the model that the present invention protects
It encloses.
The embodiment of the present invention includes:
(1) baseline algorithm
The location information of the crowdsourcing task-set and worker that are received in given a period of time, each time domain continuous type is empty
Between crowdsourcing task the subtasks of timeslices such as be divided into, each subtask and worker carry out one-to-one matching, and are appointed according to son
Business and its corresponding expense of the distance between worker calculating.Optimization aim is set as the maximization of general assignment quality, constraint condition
The worker for being set as limited to the master budget and quantity of task expense and distribution gathers.
The multi-task planning problem of time domain continuous type space crowdsourcing specification can ask in polynomial time to broad sense distribution
Topic, to prove that the problem belongs to np hard problem, allocation algorithm time complexity is very high, needs that more efficient scheme is taken
Ingredient is matched.General assignment mass function can be proved to be a non-negative and dull submodular function, therefore can use heuristic close
Come effectively to carry out task distribution like algorithm.Basic ideas are that the every wheel of approximate algorithm avidly selects the maximum son of heuristic value to appoint
Business is to complete to distribute.
Wherein, the function of the general assignment quality is the sum of each crowdsourcing Task Quality, it is assumed that single crowdsourcing Task Quality
Described with entropy function, entire crowdsourcing task-set T includes | T | a crowdsourcing task, each task t are divided into the timeslices such as m
Subtask, then general assignment quality isWherein
Indicate the quality of each subtask.
The Baseline Methods for handling the multi-task planning problem of time domain continuous type space crowdsourcing are to be considered as all crowdsourcing tasks
One entirety is allocated according to single crowdsourcing task allocation plan.Since the every wheel of algorithm is needed in all time domain continuous types skies
Between crowdsourcing task all timeslices in select the highest subtask of heuristic value, therefore the calculating cost of this method is very
Height, efficiency is very low in practical application, especially task are distributed in real time.
The main problem of baseline algorithm is to need to be traversed for larger crowdsourcing set of tasks, and time complexity is with crowd
The increase of packet task quantity is exponentially increased.An actually crowdsourcing task is only and other a small amount of crowdsourcing tasks have distribution punching
It is prominent.It is clear that if all there is no distribution conflict between all crowdsourcing tasks, optimal scheduling scheme is only needed each
The optimal distributing scheme of crowdsourcing task is simply added.Therefore, in order to save calculating cost, it is necessary to utilize distribution punching as far as possible
It is prominent, crowdsourcing task is assigned among the set not conflicted mutually, the crowdsourcing set of tasks is found in each mutually independent set
In optimal distributing scheme.According to the independence of crowdsourcing task, the present invention proposes to be based on independent task union of sets row algorithm.
(2) it is based on independent task union of sets row algorithm
Parallel algorithm main thought based on independent task collection is to carry out independent division to crowdsourcing task-set, is guaranteed each only
There is no distribution conflict between vertical task-set, can concurrently complete to distribute.Because the number of tasks that each independent task is concentrated is remote
Much smaller than total task number, therefore the allocative efficiency compared with Baseline Methods of the parallel algorithm based on independent task collection is higher.But
When crowdsourcing task-set carries out independent divide, it may appear that the task quantity of a certain independent task collection is larger.Using parallel algorithm into
When row distribution, which inevitably reduces the runing time of entire parallel frame, reduces the effect of parallel frame
Rate.Therefore, the present invention proposes that the parallel algorithm based on allocation unit, allocation unit are a task.
(3) based on the parallel algorithm of allocation unit
The working principle of parallel frame is as follows: each task in task-set includes distribution conflict list, and record can be sent out
Timeslice estranged with conflict.Host process reads task and distributes a process for each task.Then, each task is using close
It completes to distribute like algorithm.In process operational process, such as meet following condition, process will be blocked:
Condition 1: if the subtask newly completed belongs to the timeslice in conflict list, blocking corresponding process, waits
It wakes up.
After 2: one tasks of condition are often assigned several timeslices, corresponding process is blocked, and is exchanged with host process
Information.
When condition 1 occurs, then it represents that distribution conflict may occur at this time, the process then blocked disappears to host process transmission
Breath, tell host process it be blocked, and pass heuristic value back host process.The purpose of condition 2 is to feed back to heuristic value
Host process, and then by host process according to the heuristic priority for being worth the process of determination.When host process obtains the heuristic of all processes
After value, whether the heuristic value for checking the blocking process as caused by condition 1 is greater than other all processes, only when heuristic value
When greater than all processes, which can be just waken up.When the totle drilling cost of the allocated task have been over it is given pre-
It calculates, then process will terminate, system executes rollback frame, sorts according to the size of heuristic value, and successively selection executes, until
Reach task master budget, the distribution portion beyond master budget can be then deleted in task-set.
Since the every wheel of algorithm needs that the highest subtask of heuristic value is selected to distribute.Therefore, this programme is also high to having
The process of heuristic value is provided with high priority, to reduce the calculating of low heuristic value, saves CPU computing resource.
Embodiment 1:
(1) pre- according to task issuing time, acquisition task-set information, including task location, task completion time and task
Working time and the location information of calculation and worker.
In order to assess the performance of crowdsourcing Task Quality Study on Problems in time domain continuity space of the present invention, it is based on practical Beijing
Taxi car data (reference can be made to " https: //www.microsoft.com/en-us/research/publication/t-
Drive-drivingdirections-based-on-taxi-trajectories/ "), to simulate worker's data.Each work
Author, which has a working range, can reach task location to execute task guaranteeing worker in a timeslice, work
The working range of person is random.
Due to time domain continuous type space crowdsourcing field lack benchmarks data, so present case using simulated data sets into
Row experiment.In order to which without loss of generality, normal distribution, Gaussian Profile, Zipf distribution will be respectively adopted to simulate in crowdsourcing task location
Actual scene.Crowdsourcing task issuing time range is one day, and the deadline of crowdsourcing task generates at random.
(2) available nearest worker is asked to the subtask of each timeslice in crowdsourcing task, according to each crowdsourcing task
Nearest worker's situation obtain distribution conflict list.
(3) it is one new process of each crowdsourcing task creation, according to the process of above scheme, realizes and multiple crowdsourcings are appointed
It is distributed while business.
The evaluation index that parallel frame uses is Riming time of algorithm, and different task quantity, different task position point are discussed
Plant, by the parallel algorithm based on allocation unit with based on independent task collection parallel algorithm, compared with baseline algorithm, display is based on
The high efficiency of the parallel algorithm of allocation unit.Specific experiment effect picture is as depicted in figs. 1 and 2.
It can be obtained by above-mentioned experiment, when crowdsourcing number of tasks amount increases, the parallel algorithm based on independent task collection is substantially better than
Baseline algorithm, the parallel algorithm execution efficiency based on allocation unit are better than the parallel algorithm based on independent task collection.Meanwhile not
With under task location distribution, the runing time of the parallel algorithm based on allocation unit is also much smaller than based on the parallel of independent task collection
Algorithm, the high efficiency of the parallel algorithm based on allocation unit are all effective in the distribution of a variety of different task locations.It is comprehensive real
Test it can be found that while to a large amount of crowdsourcing tasks distribution be in terms of run time it is feasible, can satisfy actual requirement, together
When, it may have it is widely applied scene.
Compared with the prior art the efficient multi-task planning method of time domain continuous type provided by the invention space crowdsourcing, is shown
It is as follows to write advantage:
(1) the multi-task planning problem for considering the space crowdsourcing of time domain continuous type, proposes baseline algorithm and divides task
Match, which can solve a variety of actual demands;
(2) it finds the distribution collision problem between crowdsourcing task, proposes to be based on independent task collection by the independence of task-set
Parallel algorithm, realize more efficient distribution;
(3) it proposes the parallel algorithm based on allocation unit, distribution conflict, tool is handled using the communication between task process
There is stronger exploitativeness, is more conducive to practical application.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright description is applied directly or indirectly in other relevant technology necks
Domain is included within the scope of the present invention.
Claims (2)
1. a kind of efficient multi-task planning method of time domain continuous type space crowdsourcing, which is characterized in that comprising the following specific steps
(1) baseline algorithm
The location information of the crowdsourcing task-set and worker that are received in given a period of time, by each time domain continuous type space crowd
Packet task such as is divided at the subtask of timeslices, and each subtask and worker carry out one-to-one matching, and according to subtask with
The distance between worker calculates its corresponding expense, and optimization aim is set as the maximization of general assignment quality, constraint condition setting
For the master budget and quantity to task expense and it is distributed limited worker's set;
(2) it is based on independent task union of sets row algorithm
Independent division is carried out to crowdsourcing task-set, guarantees concurrently to complete between each independent task collection there is no distribution conflict
Distribution;
(3) based on the parallel algorithm of allocation unit
Parallel algorithm based on allocation unit, allocation unit are a task, and each task in task-set includes distribution punching
The timeslice of distribution conflict can occur for prominent list, record, and host process reads task and distributes a process for each task, so
Afterwards, each task completes distribution using approximate algorithm.
2. the efficient multi-task planning method of time domain continuous type according to claim 1 space crowdsourcing, which is characterized in that institute
The function for the general assignment quality stated is the sum of each crowdsourcing Task Quality, it is assumed that single crowdsourcing Task Quality is described with entropy function,
Entirely crowdsourcing task-set T includes | T | a crowdsourcing task, each task t are divided into the subtask of the timeslices such as m, then total appoint
Business quality beWhereinIndicate each subtask
Quality.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910383620.0A CN110196763A (en) | 2019-05-09 | 2019-05-09 | A kind of efficient multi-task planning method of time domain continuous type space crowdsourcing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910383620.0A CN110196763A (en) | 2019-05-09 | 2019-05-09 | A kind of efficient multi-task planning method of time domain continuous type space crowdsourcing |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110196763A true CN110196763A (en) | 2019-09-03 |
Family
ID=67752610
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910383620.0A Pending CN110196763A (en) | 2019-05-09 | 2019-05-09 | A kind of efficient multi-task planning method of time domain continuous type space crowdsourcing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110196763A (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103164745A (en) * | 2011-12-13 | 2013-06-19 | 中国人民解放军第二炮兵工程学院 | Maintenance supply chain integration mechanism based on ant colony algorithm and multi-agent technology |
US20140343984A1 (en) * | 2013-03-14 | 2014-11-20 | University Of Southern California | Spatial crowdsourcing with trustworthy query answering |
CN107257381A (en) * | 2017-07-03 | 2017-10-17 | 深圳大学 | The task distribution system model and implementation method of a kind of secret protection space mass-rent |
CN108133330A (en) * | 2018-01-12 | 2018-06-08 | 东北大学 | One kind is towards social crowdsourcing method for allocating tasks and its system |
CN108241930A (en) * | 2017-12-29 | 2018-07-03 | 儒安科技有限公司 | A kind of method for allocating tasks of mobile crowdsourcing platform |
CN108876012A (en) * | 2018-05-28 | 2018-11-23 | 哈尔滨工程大学 | A kind of space crowdsourcing method for allocating tasks |
CN108920261A (en) * | 2018-05-23 | 2018-11-30 | 中国航天系统科学与工程研究院 | A kind of two-stage self-adapting dispatching method suitable for large-scale parallel data processing task |
-
2019
- 2019-05-09 CN CN201910383620.0A patent/CN110196763A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103164745A (en) * | 2011-12-13 | 2013-06-19 | 中国人民解放军第二炮兵工程学院 | Maintenance supply chain integration mechanism based on ant colony algorithm and multi-agent technology |
US20140343984A1 (en) * | 2013-03-14 | 2014-11-20 | University Of Southern California | Spatial crowdsourcing with trustworthy query answering |
CN107257381A (en) * | 2017-07-03 | 2017-10-17 | 深圳大学 | The task distribution system model and implementation method of a kind of secret protection space mass-rent |
CN108241930A (en) * | 2017-12-29 | 2018-07-03 | 儒安科技有限公司 | A kind of method for allocating tasks of mobile crowdsourcing platform |
CN108133330A (en) * | 2018-01-12 | 2018-06-08 | 东北大学 | One kind is towards social crowdsourcing method for allocating tasks and its system |
CN108920261A (en) * | 2018-05-23 | 2018-11-30 | 中国航天系统科学与工程研究院 | A kind of two-stage self-adapting dispatching method suitable for large-scale parallel data processing task |
CN108876012A (en) * | 2018-05-28 | 2018-11-23 | 哈尔滨工程大学 | A kind of space crowdsourcing method for allocating tasks |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102387173B (en) | MapReduce system and method and device for scheduling tasks thereof | |
Li et al. | Parallel-machine scheduling with machine-dependent maintenance periodic recycles | |
EP3454210B1 (en) | Prescriptive analytics based activation timetable stack for cloud computing resource scheduling | |
CN104915253B (en) | A kind of method and job processor of job scheduling | |
CN101527086A (en) | Method for implementing flight time slot allocation | |
CN105893158A (en) | Big data hybrid scheduling model on private cloud condition | |
CN101652750A (en) | Data processing device, distributed processing system, data processing method, and data processing program | |
Hu et al. | FlowTime: Dynamic scheduling of deadline-aware workflows and ad-hoc jobs | |
Reddy et al. | A study on optimisation of resources for multiple projects by using primavera | |
CN107239342A (en) | A kind of storage cluster task management method and device | |
CN108268546A (en) | A kind of method and device for optimizing database | |
CN104933495A (en) | Work task assessment system on Android-based mobile terminal | |
CN113886034A (en) | Task scheduling method, system, electronic device and storage medium | |
CN114281509B (en) | Spacecraft multiuser collaborative task planning system based on scene description | |
Klevanskiy et al. | Multi-project scheduling: multicriteria time-cost trade-off problem | |
CN108958942A (en) | A kind of distributed system distribution multitask method, scheduler and computer equipment | |
Luan et al. | SCHED²: Scheduling Deep Learning Training via Deep Reinforcement Learning | |
CN103325012A (en) | Parallel computing dynamic task distribution method applicable to grid security correction | |
CN110196763A (en) | A kind of efficient multi-task planning method of time domain continuous type space crowdsourcing | |
Aleksieienko et al. | Information support of the management process of the economic entity | |
CN113537952A (en) | Multi-team collaborative release management method, system, device and medium | |
CN109936471A (en) | A kind of resource allocation methods and device of more clusters | |
Chen et al. | Design and implementation of a scientific research funds analysis model based on boston matrix | |
Dong et al. | Software multi-project resource scheduling: A comparative analysis | |
Ho et al. | Deadline flow scheduling in datacenters with time-varying bandwidth allocations |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB03 | Change of inventor or designer information |
Inventor after: Xie Xike Inventor after: Wang Ting Inventor before: Xie Xike Inventor before: Zou Kai Inventor before: Li Qi |
|
CB03 | Change of inventor or designer information | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190903 |
|
RJ01 | Rejection of invention patent application after publication |