CN108133330B - Social crowdsourcing task allocation method and system - Google Patents
Social crowdsourcing task allocation method and system Download PDFInfo
- Publication number
- CN108133330B CN108133330B CN201810032010.1A CN201810032010A CN108133330B CN 108133330 B CN108133330 B CN 108133330B CN 201810032010 A CN201810032010 A CN 201810032010A CN 108133330 B CN108133330 B CN 108133330B
- Authority
- CN
- China
- Prior art keywords
- task
- workers
- crowdsourcing
- worker
- tasks
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Abstract
The invention discloses a social crowdsourcing task allocation method and a system thereof, wherein according to crowdsourcing tasks published by task publishers, matching degree estimation algorithm calculation is carried out on workers and the crowdsourcing tasks to obtain a group of workers with the highest crowdsourcing task matching degree; a greedy algorithm is adopted to calculate task allocation, and a group of workers with the largest overall matching degree is selected as a final allocation result from a group of workers with the highest crowdsourcing task matching degree, so that each task has different workers to be allocated, workers are waited to pick up the task, and crowdsourcing task allocation is completed.
Description
Technical Field
The invention relates to the technical field of internet, in particular to a social crowdsourcing task allocation method and a social crowdsourcing task allocation system.
Background
Crowdsourcing is to use some mechanism to make the group participate in some things together to achieve some goal. Crowdsourcing is a distributed problem solving and production model. Crowdsourcing is to solve the problem difficult to understand by a machine through group intelligence, and to spread the problem to a worker group in a public bidding way. Crowdsourcing services in conjunction with social networks have become a hot issue for research in the area of crowdsourcing.
The traditional crowdsourcing system does not analyze the influence of a social network on task completion quality, and a matching degree estimation algorithm of the traditional crowdsourcing system cannot well solve the problem of cold start of the system. Some existing platforms are not designed according to the characteristics of task requirements, so that tasks and personnel cannot be effectively and reasonably arranged and optimized, and the characteristics of many existing crowdsourcing platforms cannot completely adapt to the requirements of specific customers.
The traditional crowdsourcing platform only provides a task searching function based on keywords, cannot provide task searching based on interest and preference of workers, cannot be recommended according to the characteristics of the workers, and is difficult to meet the personalized requirements of the workers.
Disclosure of Invention
In view of the above drawbacks or shortcomings, an object of the present invention is to provide a social crowdsourcing task allocation method and system.
In order to achieve the above purpose, the technical scheme of the invention is as follows:
a social crowdsourcing task allocation method comprises
1) According to the crowdsourcing tasks issued by the task issuers, carrying out matching degree estimation algorithm calculation on the workers and all the crowdsourcing tasks to obtain a group of workers with the highest crowdsourcing task matching degree;
2) calculating task allocation by adopting a greedy algorithm, and selecting a group of workers with the largest overall matching degree as a final allocation result from a group of workers with the highest crowdsourcing task matching degree, so that each task has different workers to be allocated;
3) and waiting for the worker members to pick up the tasks, and completing crowdsourcing task distribution.
The step 3) is followed by a step 4):
4) and when the task remains after the worker members in the optimal worker group pick up the task, repeating the processes of the steps 1) -3) in the remaining workers until all the tasks are picked up.
After the task is completed, recommending the task publisher and the worker to be friends by analyzing the social network diagram according to a friend recommendation algorithm, and writing the task completion condition into a history.
The estimating calculation of the matching degree of the workers and the crowdsourcing tasks in the step 1) to obtain a group of workers with the highest matching degree of the crowdsourcing tasks specifically comprises:
1.1, get worker task and complete diagram a, a ═ V, E, where V ═ wute T denotes a node set, the set of all available workers in the system is denoted as W, the crowdsourcing task set is T, E denotes an edge set, and element E denotes an edge setijExpress the worker wiFor task tjThe matching degree is represented by the ratio of the number of tasks successfully completed by the worker to the total number of tasks picked up;
obtaining a social network graph S, S ═ V, E, a node set V ═ W, a set of all available workers in the system is marked as W, and an edge set E is formed by an element EijShows wiAnd wjIs a symmetric friend relationship, and the weight value represents wiAnd wjDegree of similarity of (d), equivalent to eijThe similarity degree between different users is calculated by the interest label attribute information of the users, and indirect friends of the users are potential friends;
1.2, according to the social network diagram S, obtaining a high-quality l-dimensional feature matrix U, and assuming that there are m users in the system, there are S ═ UTZ, i.e. U ∈ Rl×mAnd Z ∈ Rl×mIs an implicit user and factor feature vector, each column UiAnd ZkImplicit feature vectors representing specific users and specific factors, respectively;
1.3, estimating the matching degree of the worker to the task according to the task completion graph A of the worker, the social network graph S and the implicit characteristic vectors of the user and specific factors.
In the step 2), a greedy algorithm is adopted to calculate the task allocation, and in a group of workers with the highest crowdsourcing task matching degree, a group of workers with the largest overall matching degree sum is selected as a final allocation result, so that each task has different workers to be allocated specifically including:
calculating the distribution result of the tasks by adopting a greedy algorithm, obtaining the maximum sum of matching degrees, and obtaining a task-worker matching degree estimation list l for the task ti belonging to T and issued by the crowdsourcing taska(ii) a By analyzing the social networking graph S, a potential friends list l of the publisher is obtained based on the interests and indirect friends of the userf(ii) a If the worker w and the publisher have an indirect friend relationship or the interest similarity of the worker w and the publisher is higher than a preset value, adding w into a potential friend list of the publisher; for theThe first k workers with high matching degree are selected as an optimal worker group list l of the tasks ti issued by the crowdsourcing taskso。
The calculation of the assignment result of the task by adopting a greedy algorithm comprises the following steps:
finding the task with the maximum matching degree of the current workers by adopting a greedy method, deleting the task and related workers in the dictionary of the task optimal worker group until all the tasks are calculated, and returning a greedy distribution result; after the task is completed, if the worker and the task publisher are not in the friend relationship, the worker and the task publisher are recommended to be friends, and generation of the friend relationship is promoted.
A social crowdsourcing task oriented distribution system comprises a selection module, a distribution module and a recommendation module:
the selecting module is used for carrying out matching degree estimation algorithm calculation on the workers and the crowdsourcing tasks according to the crowdsourcing tasks issued by the task issuer to obtain a group of workers with the highest crowdsourcing task matching degree;
the allocation module is used for calculating task allocation by adopting a greedy algorithm, and selecting a group of workers with the largest overall matching degree as a final allocation result from a group of workers with the highest crowdsourcing task matching degree, so that each task has different workers to be allocated;
and the picking module is used for waiting for the worker members to pick up the tasks and completing crowdsourcing task distribution.
And when the task is remained after the worker members in the optimal worker group pick up the task, repeating each module task until all tasks are picked up.
The system further comprises a processing module, wherein the processing module is used for recommending the task publisher and the worker to be friends mutually according to a friend recommendation algorithm by analyzing the social network diagram after the task is completed, and writing the task completion condition into a history record.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a social crowdsourcing task allocation method, provides a worker task matching degree estimation algorithm based on probability matrix decomposition and a corresponding social crowdsourcing task allocation algorithm, and realizes a social crowdsourcing service method under the online connection of a social network, so that task recommendation can be more efficient and accurate, the algorithm performance is improved, and the quality of crowdsourcing service is improved.
Furthermore, the invention provides a social crowdsourcing task allocation system, which is designed and realized by a social crowdsourcing service prototype system from online to offline, calculates potential friends of a publisher by analyzing indirect friend relationships and interests of crowdsourcing workers, recommends crowdsourcing tasks based on the social relationships of the workers in a task recommendation process, and promotes generation of social relationships through publication and collection of the tasks, so that the relationships between social networks and task allocation are studied more deeply.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a model diagram of the degree of match estimation of the present invention;
FIG. 3 is a schematic diagram of the system architecture of the present invention;
FIG. 4 is a graph comparing experimental task completion rates of the present invention;
FIG. 5 is a chart comparing the success rate of the experimental task recommendation of the present invention;
FIG. 6 is a comparison of the elapsed time of the experiment of the present invention;
fig. 7 is a comparison graph of the number of newly added experimental friends according to the present invention.
Detailed Description
The present invention will now be described in detail with reference to the drawings, wherein the described embodiments are only some, but not all embodiments of the invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, belong to the scope of the present invention.
The present invention relates to a number of concepts:
crowdsourcing task: each crowdsourcing Task has attributes of start time, end time, pickup position, collection position, Task details, and the like, the completion status of the crowdsourcing Task has success and failure, and the crowdsourcing Task requires the user to complete at a specific time. In the system, a single Task is denoted as T, and all Task sets are denoted as T (T)1,t2,...tn)。
Crowdsourcing task publishers: the method comprises the steps that a user issuing a crowdsourcing task is represented by a request, and the user issuing the crowdsourcing task is the request and has a geographic position, time information, an authentication state and interest tag attributes.
Crowd-sourcing task workers: and the other users in the system receive the crowdsourcing task issued by the request and become the Worker of the task, wherein the Worker has the geographic position, the time information, the authentication state and the interest tag attribute. The set of all available workers in the system is denoted as W, a single Worker is denoted by W, and W is dynamic. The system provides that unauthenticated users do not have the right to pick up tasks, users who have not completed picking up tasks are inNew crowdsourced tasks cannot be picked up until the current task is completed. Since Worker often cannot be qualified for the task of crowdsourcing, vector P is usedw={p1 w,p2 w,p3 w,…,pn wDenotes the degree of matching of w to the crowdsourcing task set T, where piwIndicating how well the worker w matches the task ti.
Completing the task of the worker as follows: denoted by a, given a directed weighted graph, a ═ V, E, where V ═ wute denotes a set of nodes, E denotes a set of edges, and element E denotes an elementijExpress the worker wiFor task tjThe matching degree of (a) is represented by the ratio of the number of tasks successfully completed by the worker to the total number of tasks picked up, and is wiPoint of direction tjThe higher the matching degree of the worker to the task, the higher the probability that the worker completes the task.
A social network diagram: given an undirected weighted graph S ═ (V, E), a set of nodes V ═ W, and an edge set E where the element E isijShows wiAnd wjIs a symmetric friend relationship, and the weight value represents wiAnd wjDegree of similarity of (d), equivalent to ejiThe similarity between different users is calculated by attribute information such as interest tags of the users, and indirect friends of the users are potential friends.
As shown in fig. 1, the present invention provides a social crowdsourcing task allocation method, including:
1) according to the crowdsourcing tasks issued by the task issuers, carrying out matching degree estimation algorithm calculation on the workers and all crowdsourcing tasks to obtain a group of workers with the highest crowdsourcing task matching degree;
as shown in fig. 2, which is a graph model of matching degree estimation, the idea of social matrix decomposition is to obtain a high-quality l-dimensional feature matrix U on the basis of analyzing a social network graph, and if there are m users in a system, there are S ═ UTZ, i.e. U ∈ Rl×mAnd Z ∈ Rl×mIs an implicit user and factor feature vector, each column UiAnd ZkImplicit feature vectors representing specific users and specific factors, respectively, are defined by existing social relationships as followsThe condition distribution of (2).
Here N (x | μ, σ)2) Is a mean value of mu and a variance of sigma2Is determined as a function of the probability density of the gaussian distribution of (a),is an exponential function, if wiAnd wkIs a friend relationship between them, thenOtherwiseRegression function g (x) 1/(1+ exp (-x)) willNormalized to [0,1 ]]In the meantime. A zero mean spherical gaussian prior is applied to the user and factor eigenvectors to obtain the following equation (2).
Thus, equation (3) is obtained by simple bayesian inference.
Estimate at [0,1 ]]The matching degree of the worker to the task is estimated to be represented by the ratio of the number of the tasks successfully completed by the worker to the total number of the tasks picked up. A. theijRepresents wiFor tjAnd U ∈ R, andl×mand V ∈ Rl×nA feature matrix, a column vector U, representing workers and tasksiAnd VjRespectively representing specific workers and toolsThe implicit feature vector of the volume task defines the following conditional distribution according to the existing matching degree.
Assuming that there are m workers and n tasks in the system, the degree of match
And applying a spherical Gaussian prior with zero mean to the characteristic vectors of the workers and the tasks to obtain a formula (5).
Through simple bayesian inference, the following equation (6) is obtained.
Wherein, N (x | mu, sigma)2) Is a mean value of mu and a variance of sigma2Is determined as a function of the probability density of the gaussian distribution of (a),is an exponential function, if wiAnd wkIs a friend relationship between them, thenOtherwiseRegression function g (x) 1/(1+ exp (-x)) willNormalized to [0,1 ]]A isijRepresents wiFor tjAnd U ∈ R, andl×mand V ∈ Rl×nRepresenting workers and tasksThe characteristic matrix, column vector UiAnd VjImplicit feature vectors representing specific workers and specific tasks, respectively.
2) Calculating task allocation by adopting a greedy algorithm, and selecting a group of workers with the largest overall matching degree as a final allocation result from a group of workers with the highest crowdsourcing task matching degree, so that each task has different workers to be allocated;
calculating the distribution result of the tasks by adopting a greedy algorithm, obtaining the maximum sum of matching degrees, and obtaining a task-worker matching degree estimation list l for the task ti belonging to T and issued by the crowdsourcing taska(ii) a By analyzing the social networking graph S, a potential friends list l of the publisher is obtained based on the interests and indirect friends of the userf(ii) a If the worker w and the publisher have an indirect friend relationship or the interest similarity of the worker w and the publisher is higher than a preset value, adding w into a potential friend list of the publisher; for theThe first k workers with high matching degree are selected as an optimal worker group list l of the tasks ti issued by the crowdsourcing taskso。
The calculation of the assignment result of the task by adopting a greedy algorithm comprises the following steps:
finding the task with the maximum matching degree of the current workers by adopting a greedy method, deleting the task and related workers in the dictionary of the task optimal worker group until all the tasks are calculated, and returning a greedy distribution result; after the task is completed, if the worker and the task publisher are not in a friend relationship, the worker and the task publisher are recommended to be friends, the generation of the friend relationship is promoted, and the task completion quality in the system is effectively improved.
4) And when the task remains after the worker members in the optimal worker group pick up the task, repeating the processes of the steps 1) -3) in the remaining workers until all the tasks are picked up.
It should be noted that, after the task is completed, the task publisher and the worker are recommended to each other as friends by analyzing the social network diagram according to a friend recommendation algorithm, and the task completion condition is written into the history.
As shown in fig. 3, the present invention further provides a social crowdsourcing task allocation system, which includes a selection module 1, an allocation module 2, and a getting module 3:
the system comprises a selecting module 1, a task publisher and a task matching module, wherein the selecting module is used for calculating a matching degree estimation algorithm of workers and crowdsourcing tasks according to the crowdsourcing tasks published by the task publisher to obtain a group of workers with the highest matching degree of the crowdsourcing tasks;
the allocation module 2 is used for calculating task allocation by adopting a greedy algorithm, and selecting a group of workers with the largest overall matching degree as a final allocation result from a group of workers with the highest crowdsourcing task matching degree, so that each task has different workers to be allocated;
and the getting module 3 is used for waiting for the worker members to get the tasks and completing crowdsourcing task distribution.
And when the task is remained after the worker members in the optimal worker group pick up the task, repeating each module task until all tasks are picked up.
The system further comprises a processing module 4, wherein the processing module is used for recommending the task publisher and the worker to be friends mutually according to a friend recommendation algorithm by analyzing the social network diagram after the task is completed, and writing the task completion condition into a history.
3) And recommending the crowdsourcing task to the worker members in the optimal worker group, waiting for the worker members to pick up the task, and finishing the crowdsourcing task distribution.
The experimental results are as follows:
the data set of the experiment is generated by three groups of different random number seeds, and the statistical result is the average value of the random number seeds, the experimental program is completed by Python language and runs on an Intel (R) core (TM) i7-6500@2.50GHz 2.59GHz CPU windows 10 system.
The simulation data set comprises W users, m tasks are generated every day in d days, a task publisher is randomly generated in the W users, the positions of the users and the tasks in the system are all in a d x d rectangle, and the key data generation method comprises the following steps:
1) the distribution process of the crowdsourcing task comprises the following steps: after the tasks are released, the tasks are recommended to relevant users at the next integral point after the release, and each integral point has more than one released tasks. If the time-location information of the recommended user and the task are similar, then there will be a probability pRecommendingAnd after the recommendation, the task which is not successfully picked up is picked up by all members in the organization, and users similar to the task time-position information in the members can pick up the task according to the probability pReception deviceGetting task, the successful users will get with probability pComplete the processFriends and potential friends may be more aggressive in picking up and completing tasks than non-friend or non-potential friend workers.
2) Task: the start times are randomly distributed among 8 to 16; the ending time of the task is randomly distributed from the starting time to 22 hours; the release and pick-up positions of the tasks are randomly distributed among the rectangles of d x d.
3) And (2) Worker: randomly generating a time-location pair per day for each user, and tagging pages from all interest tagging nets for each user1Obtain interest labels of users) randomly selects two of the users as interests of the users and establishes a friend relationship graph between the users by using a Barabasi-Albert model.
4) And friend relationship generation: after the task is successfully completed, calculating the interest similarity of the Worker and the Requester, and calculating the probability p of becoming a friend according to the potential friend relationship of the two partiesMaking friends。
The experimental part generates 100 users in the above way, the number of vertex connections on the social network diagram is 5, 20 new tasks are generated each day, and the task completion condition with the task quantity of 1000 in 50 days is simulated.
The proposed social-oriented task allocation algorithm (shortly called SoTS) is similar to the traditional similar-based task allocation algorithm (shortly called SoTS)ICF) method, and experimental setting of lambdaS=λU=λV=λZThe number of the recessive feature vectors is set to be 5, and the matching degree of all the workers in the two days before the experiment is set to be 0.9 because no history of the completion condition of the workers exists in the system in the initial stage of the experiment.
Selecting SIMS of 0.7 and k of 3 for experiment to obtain
The experimental statistics are shown in fig. 4-5.
Fig. 4 compares the change of the task completion rates of the SoTS and the ICF with time, the abscissa represents time, the unit is week, and the ordinate represents the task completion rate, and it can be seen that the task completion rates of the SoTS and the ICF algorithms fluctuate around 0.9, and the task completion rate of the SoTS algorithm is improved by 6.17% compared with the ICF.
FIG. 5 compares the change of task recommendation success rates of SoTS and ICF with time, wherein the abscissa represents time in weeks and the ordinate represents task completion rate, and the SoTS algorithm recommendation success rate is improved by 3.67% compared with the ICF.
FIG. 6 compares SoTS and ICF pick-up task times, with the abscissa representing time in weeks and the ordinate representing the sum of the pick-up task times in hours for all tasks during a week, the pick-up task time decreasing from an average of 387.8 minutes per week to 373.6 minutes, indicating that SoTS can allocate tasks more quickly than ICF.
Fig. 7 compares the numbers of newly added friends of SoTS and ICF, where the abscissa represents time, the unit is week, the ordinate represents the number of newly added friends, which represents the number of new friends in the system in one week, and the number of newly added friends is increased from 7.4 to 10.5 per week on average, which shows that SoTS significantly promotes the formation of friend relationships in the crowdsourcing platform compared to ICF.
It will be apparent to those skilled in the art that the foregoing specific embodiments are merely preferred embodiments of the invention, and thus, modifications and variations may be made in the invention to which the invention pertains, which will still embody the principles of the invention and which will still achieve the objects of the invention, within the scope of the invention as defined by the appended claims.
Claims (8)
1. A social crowdsourcing task allocation method is characterized by comprising
1) According to the crowdsourcing tasks issued by the task issuers, carrying out matching degree estimation algorithm calculation on the workers and all crowdsourcing tasks to obtain a group of workers with the highest crowdsourcing task matching degree;
2) calculating task allocation by adopting a greedy algorithm, and selecting a group of workers with the largest overall matching degree as a final allocation result from a group of workers with the highest crowdsourcing task matching degree, so that each task has different workers to be allocated;
3) waiting for the worker members to get the tasks and completing the crowdsourcing tasks;
the performing matching degree estimation calculation on the workers and the crowdsourcing task in the step 1) to obtain a group of worker calculation with the highest crowdsourcing task matching degree further comprises:
1.1, get worker task and complete diagram a, a ═ V, E, where V ═ wute T denotes a node set, the set of all available workers in the system is denoted as W, the crowdsourcing task set is T, E denotes an edge set, and element E denotes an edge setijExpress the worker wiFor task tjThe matching degree is represented by the ratio of the number of tasks successfully completed by the worker to the total number of tasks picked up;
obtaining a social network graph S, S ═ V, E, a node set V ═ W, a set of all available workers in the system is marked as W, and an edge set E is formed by an element EijShows wiAnd wjIs a symmetric friend relationship, and the weight value represents wiAnd wjDegree of similarity of (d), equivalent to eijThe similarity degree between different users is calculated by the interest label attribute information of the users, and indirect friends of the users are potential friends;
1.2, according to the social network diagram S, obtaining a high-quality l-dimensional feature matrix U, and assuming that there are m users in the system, there are S ═ UTZ, i.e. U ∈ Rl×mAnd Z ∈ Rl×mIs an implicit user and factor feature vector, each column UiAnd ZkImplicit feature vectors representing specific users and specific factors, respectively;
1.3, estimating the matching degree of the worker to the task according to the task completion graph A of the worker, the social network graph S and the implicit characteristic vectors of the user and specific factors.
2. The social crowdsourcing task oriented distribution method according to claim 1, further comprising, after the step 3), a step 4):
4) and when the task remains after the worker members in the optimal worker group pick up the task, repeating the processes of the steps 1) -3) in the remaining workers until all the tasks are picked up.
3. The social-oriented crowdsourcing task distribution method according to claim 1 or 2, wherein after the task is completed, the task publisher and the worker are recommended to each other as friends by analyzing a social network diagram according to a friend recommendation algorithm, and task completion conditions are written into a history.
4. The social-oriented crowdsourcing task allocation method according to claim 1, wherein the step 2) adopts a greedy algorithm to calculate task allocation, and selects a group of workers with the largest overall matching degree as a final allocation result from a group of workers with the highest crowdsourcing task matching degree, so that each task has different workers to be allocated specifically comprises:
calculating the distribution result of the tasks by adopting a greedy algorithm, obtaining the maximum matching degree sum, and obtaining a task-worker matching degree estimation list l for the task ti belonging to T and issued by the crowdsourcing taska(ii) a By analyzing the social networking graph S, a potential friends list l of the publisher is obtained based on the interests and indirect friends of the userf(ii) a If the worker w and the publisher have an indirect friend relationship or the interest similarity of the worker w and the publisher is higher than a preset value, adding w into a potential friend list of the publisher; for theSelecting the product with high matching degreeThe first k workers, the optimal worker group list l of tasks ti issued as crowd-sourced taskso。
5. The social-oriented crowdsourcing task allocation method according to claim 4, wherein computing the allocation result of the task by adopting a greedy algorithm comprises:
finding the task with the maximum matching degree of the current workers by adopting a greedy method, deleting the task and related workers in the dictionary of the task optimal worker group until all the tasks are calculated, and returning a greedy distribution result; after the task is completed, if the worker and the task publisher are not in the friend relationship, the worker and the task publisher are recommended to be friends, and generation of the friend relationship is promoted.
6. The utility model provides a crowd-sourced task distribution system towards social, which characterized in that includes selects module, distribution module and draws the module:
the selecting module is used for carrying out matching degree estimation algorithm calculation on the workers and the crowdsourcing tasks according to the crowdsourcing tasks issued by the task issuer to obtain a group of workers with the highest crowdsourcing task matching degree;
the allocation module is used for calculating task allocation by adopting a greedy algorithm, and selecting a group of workers with the largest overall matching degree as a final allocation result from a group of workers with the highest crowdsourcing task matching degree, so that each task has different workers to be allocated;
the receiving module is used for waiting for the members of the workers to receive the tasks and completing the crowdsourcing tasks;
the selecting module is specifically configured to perform matching degree estimation calculation on the workers and the crowdsourcing tasks, and obtain a group of worker calculation with the highest matching degree of the crowdsourcing tasks, further including:
1.1, get worker task and complete diagram a, a ═ V, E, where V ═ wute T denotes a node set, the set of all available workers in the system is denoted as W, the crowdsourcing task set is T, E denotes an edge set, and element E denotes an edge setijExpress the worker wiFor task tjThe matching degree is represented by the ratio of the number of tasks successfully completed by the worker to the total number of tasks picked up;
obtaining a social network graph S, S ═ V, E, a node set V ═ W, a set of all available workers in the system is marked as W, and an edge set E is formed by an element EijShows wiAnd wjIs a symmetric friend relationship, and the weight value represents wiAnd wjDegree of similarity of (d), equivalent to eijThe similarity degree between different users is calculated by the interest label attribute information of the users, and indirect friends of the users are potential friends;
1.2, according to the social network diagram S, obtaining a high-quality l-dimensional feature matrix U, and assuming that there are m users in the system, there are S ═ UTZ, i.e. U ∈ Rl×mAnd Z ∈ Rl×mIs an implicit user and factor feature vector, each column UiAnd ZkImplicit feature vectors representing specific users and specific factors, respectively;
1.3, estimating the matching degree of the worker to the task according to the task completion graph A of the worker, the social network graph S and the implicit characteristic vectors of the user and specific factors.
7. The social-oriented crowdsourcing task distribution system of claim 6, further comprising repeating each module task until all tasks are picked, when there are more tasks remaining after picking up the tasks by the worker members in the optimal worker group.
8. The social-oriented crowdsourcing task distribution system of claim 7, further comprising a processing module, configured to, after the task is completed, recommend the task publisher and the worker to each other as a friend by analyzing the social network graph according to a friend recommendation algorithm, and write task completion into a history.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810032010.1A CN108133330B (en) | 2018-01-12 | 2018-01-12 | Social crowdsourcing task allocation method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810032010.1A CN108133330B (en) | 2018-01-12 | 2018-01-12 | Social crowdsourcing task allocation method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108133330A CN108133330A (en) | 2018-06-08 |
CN108133330B true CN108133330B (en) | 2022-04-08 |
Family
ID=62400511
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810032010.1A Active CN108133330B (en) | 2018-01-12 | 2018-01-12 | Social crowdsourcing task allocation method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108133330B (en) |
Families Citing this family (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108846577B (en) * | 2018-06-14 | 2021-11-05 | 河南科技大学 | Group task allocation method based on context analysis |
CN111161555A (en) * | 2018-11-07 | 2020-05-15 | 北京嘀嘀无限科技发展有限公司 | Information collection method and system |
CN109460978B (en) * | 2018-11-15 | 2021-12-03 | 泉州师范学院 | Social relationship cognition-based crowd-sourcing service node selection method |
CN110378663A (en) * | 2019-04-19 | 2019-10-25 | 西北工业大学 | A kind of complicated movement crowdsourcing method for allocating tasks based on Greedy strategy |
CN110097277B (en) * | 2019-04-28 | 2021-10-08 | 北京交通大学 | Time window-based crowdsourcing task dynamic allocation method |
CN110196763A (en) * | 2019-05-09 | 2019-09-03 | 中国科学技术大学苏州研究院 | A kind of efficient multi-task planning method of time domain continuous type space crowdsourcing |
CN110276520B (en) * | 2019-05-15 | 2023-01-10 | 创新先进技术有限公司 | Project case screening method and device |
CN110298546A (en) * | 2019-05-23 | 2019-10-01 | 北京邮电大学 | A kind of method for allocating tasks and device |
CN110503331A (en) * | 2019-08-20 | 2019-11-26 | 西北工业大学 | A kind of complicated movement crowdsourcing task analytic approach towards multi-constraint condition |
CN110764898B (en) * | 2019-09-02 | 2023-03-03 | 平安科技(深圳)有限公司 | Task allocation method and device, readable storage medium and terminal equipment |
CN111291973B (en) * | 2020-01-17 | 2023-09-29 | 麦科思(无锡)数据科技有限公司 | Space crowdsourcing task allocation method based on alliance |
CN111582641B (en) * | 2020-04-02 | 2023-04-28 | 长沙理工大学 | Task allocation method and system for crowdsourcing service of Internet of vehicles |
CN111738576B (en) * | 2020-06-10 | 2023-08-01 | 中北大学 | Fair crowdsourcing workflow distribution method |
CN112396279B (en) * | 2020-06-17 | 2022-08-26 | 南京邮电大学 | Robust crowdsourcing data analysis method based on trust model |
CN112363822B (en) * | 2020-09-22 | 2024-02-13 | 长沙理工大学 | Collaborative task allocation method for crowdsourcing platform |
CN112148986B (en) * | 2020-10-09 | 2022-09-30 | 安徽大学 | Top-N service re-recommendation method and system based on crowdsourcing |
CN112819210B (en) * | 2021-01-20 | 2023-01-13 | 杭州电子科技大学 | Online single-point task allocation method capable of being rejected by workers in space crowdsourcing |
CN113222391A (en) * | 2021-05-06 | 2021-08-06 | 南京财经大学 | Global scoring method for registrants of crowdsourcing business platform |
CN116415839B (en) * | 2023-02-10 | 2023-11-28 | 华南师范大学 | Crowd-sourced task allocation method and system based on interpretable machine learning |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102905368B (en) * | 2012-10-18 | 2015-06-10 | 无锡儒安科技有限公司 | Mobile auxiliary indoor positioning method and system based on smart phone platform |
US20150317607A1 (en) * | 2014-05-05 | 2015-11-05 | Groove Technologies, Inc. | System and Method for Matching Job Candidates to Job Opportunities |
US9688281B2 (en) * | 2015-03-23 | 2017-06-27 | Toyota Jidosha Kabushiki Kaisha | Proactive autocomplete of a user's in-vehicle operations |
US20160350694A1 (en) * | 2015-05-29 | 2016-12-01 | International Business Machines Corporation | Implementing workflow based on social network nodes |
CN105719010A (en) * | 2015-07-24 | 2016-06-29 | 北京小度信息科技有限公司 | Method and device for processing distribution tasks |
CN106209874B (en) * | 2016-07-18 | 2019-06-28 | 沈阳师范大学 | A kind of method for allocating tasks for intelligent perception distribution system |
CN106327090A (en) * | 2016-08-29 | 2017-01-11 | 安徽慧达通信网络科技股份有限公司 | Real task allocation method applied to preference crowd-sourcing system |
-
2018
- 2018-01-12 CN CN201810032010.1A patent/CN108133330B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN108133330A (en) | 2018-06-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108133330B (en) | Social crowdsourcing task allocation method and system | |
Cai et al. | Real-time bidding by reinforcement learning in display advertising | |
Tang et al. | Personalized recommendation via parameter-free contextual bandits | |
US9654593B2 (en) | Discovering signature of electronic social networks | |
US20190272553A1 (en) | Predictive Modeling with Entity Representations Computed from Neural Network Models Simultaneously Trained on Multiple Tasks | |
WO2016091173A1 (en) | User maintenance system and method | |
CN107688605B (en) | Cross-platform data matching process, device, computer equipment and storage medium | |
CN111966904B (en) | Information recommendation method and related device based on multi-user portrait model | |
CN109615411A (en) | Advertisement placement method and device, electronic equipment based on algorithm model | |
CN112256961B (en) | User portrait generation method, device, equipment and medium | |
Salek et al. | Hotspotting—a probabilistic graphical model for image object localization through crowdsourcing | |
CN111008335B (en) | Information processing method, device, equipment and storage medium | |
Jia et al. | Collaborative restricted Boltzmann machine for social event recommendation | |
CN112148986B (en) | Top-N service re-recommendation method and system based on crowdsourcing | |
WO2016165058A1 (en) | Social prediction | |
CN110020022A (en) | Data processing method, device, equipment and readable storage medium storing program for executing | |
WO2023036184A1 (en) | Methods and systems for quantifying client contribution in federated learning | |
CN113641811B (en) | Session recommendation method, system, equipment and storage medium for promoting purchasing behavior | |
CN113034168A (en) | Content item delivery method and device, computer equipment and storage medium | |
Liu et al. | Bid-aware active learning in real-time bidding for display advertising | |
CN113821574A (en) | User behavior classification method and device and storage medium | |
CN113222073A (en) | Method and device for training transfer learning model and recommendation model | |
CN112115354A (en) | Information processing method, information processing apparatus, server, and storage medium | |
CN110263959B (en) | Click rate estimation method, click rate estimation device, machine equipment and computer readable storage medium | |
CN114528973A (en) | Method for generating business processing model, business processing method and device |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |