CN114547473A - Crowdsourcing task data recommendation method based on decision model and genetic matrix decomposition method - Google Patents

Crowdsourcing task data recommendation method based on decision model and genetic matrix decomposition method Download PDF

Info

Publication number
CN114547473A
CN114547473A CN202210254290.7A CN202210254290A CN114547473A CN 114547473 A CN114547473 A CN 114547473A CN 202210254290 A CN202210254290 A CN 202210254290A CN 114547473 A CN114547473 A CN 114547473A
Authority
CN
China
Prior art keywords
task
user
matrix
matching
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.)
Pending
Application number
CN202210254290.7A
Other languages
Chinese (zh)
Inventor
冯毅雄
高晓勰
洪兆溪
胡炳涛
张志峰
谭建荣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN202210254290.7A priority Critical patent/CN114547473A/en
Publication of CN114547473A publication Critical patent/CN114547473A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Biophysics (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Biology (AREA)
  • Molecular Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Physiology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Genetics & Genomics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a crowdsourcing task data recommendation method based on a decision model and a genetic matrix decomposition method. The method comprises the following steps: 1) preprocessing historical data in a crowdsourcing platform to obtain preprocessed historical data, and establishing a task characteristic matrix and a user characteristic matrix based on the preprocessed historical data; 2) performing information matching on the task characteristic matrix and the user characteristic matrix by using a user knowledge fusion decision-making method to obtain an initial capability matching matrix; 3) establishing a matching decision model according to the task characteristic matrix and the user characteristic matrix; 4) and solving the matching decision model by utilizing a genetic matrix decomposition algorithm according to the initial capability matching matrix to obtain the matching degree between the user and the plurality of tasks, and recommending the crowdsourcing task for the user by the crowdsourcing platform based on the matching degree between the user and the plurality of tasks. The method and the device improve the precision and efficiency of task recommendation on the crowdsourcing platform, and effectively solve the practical problems of task overload and low task matching efficiency in crowdsourcing scenes.

Description

Crowdsourcing task data recommendation method based on decision model and genetic matrix decomposition method
Technical Field
The invention relates to a crowdsourcing task recommendation method, in particular to a crowdsourcing task data recommendation method based on a decision model and a genetic matrix decomposition method.
Background
The development of the sharing economy promotes the rapid development of the crowdsourcing mode, the number and the types of tasks on a crowdsourcing design platform are increased rapidly, and the problem of information overload is revealed. Facing a lot of information, it is difficult for users to quickly select a task suitable for themselves. In order to reduce the search cost, most users select tasks which are recently released or ranked on the first two pages, and the higher search cost may reduce the participation of the users and is not favorable for ensuring the task completion quality. The task selection method is mainly used for helping a user to select tasks related to the user, and the two mainstream task selection methods are task search and task recommendation. Because the task searching is to search for the task from the user, the user is difficult to find the task meeting the personalized requirement under the condition of information overload; and the task recommendation is to identify the personalized features of the user by using historical information in the crowdsourcing design platform from the perspective of the crowdsourcing design platform and to carry out personalized push on the user. Therefore, research task recommendation is an effective way for solving the problem of crowdsourcing information overload, and has important significance.
Disclosure of Invention
In order to solve the problems existing in the background technology, the invention provides a crowdsourcing task data recommendation method based on a decision model and a genetic matrix decomposition method, which realizes the configuration of tasks and users suitable for crowdsourcing scenes and provides decision support for crowdsourcing task allocation.
The invention is realized by the following technical scheme:
the invention comprises the following steps:
1) preprocessing historical data in a crowdsourcing platform to obtain preprocessed historical data, and establishing a task characteristic matrix and a user characteristic matrix based on the preprocessed historical data;
2) performing information matching on the task characteristic matrix and the user characteristic matrix by using a user knowledge fusion decision-making method to obtain an initial capability matching matrix;
3) establishing a matching decision model according to the task characteristic matrix and the user characteristic matrix;
4) and solving the matching decision model by utilizing a genetic matrix decomposition algorithm according to the initial capability matching matrix to obtain the matching degree between the user and the plurality of tasks, and recommending the crowdsourcing task for the user by the crowdsourcing platform based on the matching degree between the user and the plurality of tasks.
In the step 1), standardized information extraction is respectively carried out on the attribute information of the task and the user in the historical data by using a natural language NLP method, the standardized information of the task and the user is respectively obtained, and the preprocessed historical data is formed by the standardized information of the task and the user.
The task characteristic matrix, the user characteristic matrix and the initial capability matching matrix satisfy the following relations:
A＝W·H
wherein, W represents a user feature matrix, the dimension of the user feature matrix W is m × k, m represents the total number of users, k represents the maximum task number distributed by each user,
Figure BDA0003547964250000021
for the ith row element in the user profile matrix,
Figure BDA0003547964250000022
attribute information indicating the ith user, i 1, 2. H represents a task feature matrix, the dimension of the task feature matrix H is k x n, n represents the total number of tasks,
Figure BDA0003547964250000023
for the jth row element in the task feature matrix,
Figure BDA0003547964250000024
attribute information indicating the j-th task, j 1, 2. A is an initial capability matching matrix, and the dimension of the initial capability matching matrix A is m x n, aijMatching the element of the ith row and the jth column in the matrix A for the initial capacity
Figure BDA0003547964250000025
aijIndicating a capability match value for user i with task j.
The target equation of the matching decision model is as follows:
Figure BDA0003547964250000026
wherein f (W, H) represents a feature matrix matching difference value, A represents an initial capability matching matrix, W represents a user feature matrix, H represents a task feature matrix,
Figure BDA0003547964250000027
a square root operation representing the sum of squares of the absolute values; min represents the minimization operation;
the constraint conditions are as follows:
k is more than or equal to 1 and less than or equal to K, K represents the number of tasks actually distributed to the user, and K represents the total number of tasks;
one task is allocated to each user;
T＜T0where T represents the actual time that the user completed the task, T0Indicating the time required to complete the task.
In the step 4), the crowdsourcing platform gives a matching recommendation list of the user and the plurality of tasks based on the matching degree between the user and the plurality of tasks, so that crowdsourcing task recommendation is realized.
The invention has the beneficial effects that:
the invention provides a crowdsourcing task data recommendation method based on a decision model and a genetic matrix decomposition method. The method has the advantages that the effect of mutation operators in the calculation results is researched in the implementation case, the local search capability of the algorithm is enhanced, the important effect is played on the aspect of keeping population diversity, and a powerful basis is provided for parameter setting in the solving process. Overall, the task recommendation method saves time and effort to select a suitable task from a large number of tasks of the crowd-sourcing platform to distribute to the respective users.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic diagram of performance analysis during the solution of an objective equation.
Detailed Description
The invention is described in further detail below with reference to the following figures and specific embodiments:
as shown in fig. 1, the present invention comprises the steps of:
1) preprocessing historical data in a crowdsourcing platform to obtain preprocessed historical data, and establishing a task characteristic matrix and a user characteristic matrix based on the preprocessed historical data; the history data includes attribute information of the task and the user. And respectively constructing a task characteristic matrix and a user characteristic matrix based on the standardized information of the task and the user.
In the step 1), standardized information extraction is respectively carried out on the attribute information of the task and the user in the historical data by using a natural language NLP method, the standardized information of the task and the user is respectively obtained, and the preprocessed historical data is formed by the standardized information of the task and the user.
2) Performing information matching on the task characteristic matrix and the user characteristic matrix by using a user knowledge fusion decision-making method to obtain an initial capability matching matrix;
the task characteristic matrix, the user characteristic matrix and the initial capability matching matrix satisfy the following relations:
A＝W·H
wherein, W represents a user feature matrix, the dimension of the user feature matrix W is m × k, m represents the total number of users, k represents the maximum task number distributed by each user,
Figure BDA0003547964250000031
for the ith row element in the user profile matrix,
Figure BDA0003547964250000032
attribute information indicating the ith user, i 1, 2. H represents a task feature matrix, the dimension of the task feature matrix H is k x n, n represents the total number of tasks,
Figure BDA0003547964250000033
for the jth row element in the task feature matrix,
Figure BDA0003547964250000034
attribute information indicating the j-th task, j 1, 2. A is an initial capability matching matrix, and the dimension of the initial capability matching matrix A is m x n, aijMatching the element of the ith row and the jth column in the matrix A for the initial capacity
Figure BDA0003547964250000035
aijIndicating a capability match value for user i with task j.
3) Establishing a matching decision model according to the task characteristic matrix and the user characteristic matrix;
the objective equation for the matching decision model is:
Figure BDA0003547964250000036
wherein f (W, H) represents a feature matrix matching difference value, A represents an initial capability matching matrix, W represents a user feature matrix, H represents a task feature matrix,
Figure BDA0003547964250000037
a square root operation representing the sum of squares of the absolute values of the matrix elements in parentheses; min represents the minimization operation;
the constraint conditions are as follows:
k is more than or equal to 1 and less than or equal to K, K represents the number of tasks actually distributed to the user, and K represents the total number of tasks;
one task is allocated to each user;
T＜T0t represents user completion taskActual time of transaction, T0Indicating the time required to complete the task.
4) And solving the matching decision model by utilizing a genetic matrix decomposition algorithm according to the initial capability matching matrix to obtain the matching degree between the user and the plurality of tasks, and recommending the crowdsourcing task for the user by the crowdsourcing platform based on the matching degree between the user and the plurality of tasks.
In the step 4), the crowdsourcing platform gives a matching recommendation list of the user and the plurality of tasks based on the matching degree between the user and the plurality of tasks, so that crowdsourcing task recommendation is realized.
In the simulation experiment, the number of tasks is set to be 500, the number of users is set to be 200, and the maximum number of tasks allocated to each user is 10. Mutation operators in genetic matrix factorization algorithms play an important role in the calculation results. The method not only enhances the local search capability of the algorithm, but also plays an important role in keeping population diversity. However, the accuracy of the algorithm is affected because the mutation probability of the mutation operator is too large or too small. The invention explores P through the implementation of the casem-T and Pm-relationships of RMSE to verify the performance of the scheme.
As shown in FIG. 2, when the probability value P is mutatedmWhen the value is gradually increased from 0.05, the convergence value of the algorithm of the invention is gradually increased, which means that the convergence speed of the algorithm is continuously increased. But when mutating the probability value PmBeyond 0.25, the convergence value begins to decrease, indicating that the convergence performance of the algorithm is degraded. Probability value of mutation PmIncreasing from 0.05, the RMSE value decreases gradually, indicating that the task assignment accuracy of the algorithm increases gradually, since a lower RMSE value indicates a higher accuracy of the algorithm assigning tasks. But when the mutation probability value P of the algorithmmWhen the increase is continued beyond 0.25, the RMSE gradually rises, indicating that the accuracy of the algorithm to assign tasks gradually decreases. The experimental result shows that the mutation probability value P corresponding to the maximum task allocation precision of the algorithmmIs 0.25.

Claims (5)

1. A crowd-sourced task data recommendation method based on a decision model and a genetic matrix decomposition method is characterized by comprising the following steps:
1) preprocessing historical data in a crowdsourcing platform to obtain preprocessed historical data, and establishing a task characteristic matrix and a user characteristic matrix based on the preprocessed historical data;
2) performing information matching on the task characteristic matrix and the user characteristic matrix by using a user knowledge fusion decision-making method to obtain an initial capability matching matrix;
3) establishing a matching decision model according to the task characteristic matrix and the user characteristic matrix;
4) and solving the matching decision model by utilizing a genetic matrix decomposition algorithm according to the initial capability matching matrix to obtain the matching degree between the user and the plurality of tasks, and recommending the crowdsourcing task for the user by the crowdsourcing platform based on the matching degree between the user and the plurality of tasks.
2. The method for recommending crowdsourcing task data based on decision model and genetic matrix factorization (NLP) of claim 1, wherein in step 1), standardized information extraction is performed on attribute information of tasks and users in historical data by using a Natural Language (NLP) method, standardized information of the tasks and the users is obtained, and preprocessed historical data is formed by the standardized information of the tasks and the users.
3. The method for recommending crowdsourcing task data based on decision model and genetic matrix decomposition method according to claim 1, wherein the task feature matrix, the user feature matrix and the initial capability matching matrix satisfy the following relations:
A＝W·H
wherein, W represents a user feature matrix, the dimension of the user feature matrix W is m × k, m represents the total number of users, k represents the maximum task number distributed by each user,
Figure FDA0003547964240000011
for the ith row element in the user profile matrix,
Figure FDA0003547964240000012
attribute information indicating the ith user, i 1, 2. H represents a task feature matrix, the dimension of the task feature matrix H is k x n, n represents the total number of tasks,
Figure FDA0003547964240000013
for the jth row element in the task feature matrix,
Figure FDA0003547964240000014
attribute information indicating the j-th task, j 1, 2. A is an initial capability matching matrix, and the dimension of the initial capability matching matrix A is m x n, aijMatching the element of the ith row and the jth column in the matrix A for the initial capacity
Figure FDA0003547964240000015
aijIndicating a capability match value for user i with task j.
4. The method of claim 1, wherein the objective equation for matching the decision model is as follows:
Figure FDA0003547964240000016
wherein f (W, H) represents a feature matrix matching difference value, A represents an initial capability matching matrix, W represents a user feature matrix, H represents a task feature matrix,
Figure FDA0003547964240000021
a square root operation representing the sum of squares of the absolute values; min represents the minimization operation;
the constraint conditions are as follows:
k is more than or equal to 1 and less than or equal to K, K represents the number of tasks actually distributed to the user, and K represents the total number of tasks;
one task is allocated to each user;
T＜T0where T represents the actual time that the user completed the task, T0Indicating the time required to complete the task.
5. The method as claimed in claim 1, wherein in the step 4), the crowdsourcing platform gives a matching recommendation list of the user and the plurality of tasks based on a matching degree between the user and the plurality of tasks, so as to implement crowdsourcing task recommendation.
CN202210254290.7A 2022-03-15 2022-03-15 Crowdsourcing task data recommendation method based on decision model and genetic matrix decomposition method Pending CN114547473A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210254290.7A CN114547473A (en) 2022-03-15 2022-03-15 Crowdsourcing task data recommendation method based on decision model and genetic matrix decomposition method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210254290.7A CN114547473A (en) 2022-03-15 2022-03-15 Crowdsourcing task data recommendation method based on decision model and genetic matrix decomposition method

Publications (1)

Publication Number Publication Date
CN114547473A true CN114547473A (en) 2022-05-27

Family

ID=81664269

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210254290.7A Pending CN114547473A (en) 2022-03-15 2022-03-15 Crowdsourcing task data recommendation method based on decision model and genetic matrix decomposition method

Country Status (1)

Country Link
CN (1) CN114547473A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115018322A (en) * 2022-06-07 2022-09-06 山东京德智汇科技有限公司 Intelligent crowdsourcing task allocation method and system
CN115018322B (en) * 2022-06-07 2024-07-02 山东冀联人力资源服务有限公司 Intelligent crowdsourcing task distribution method and system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115018322A (en) * 2022-06-07 2022-09-06 山东京德智汇科技有限公司 Intelligent crowdsourcing task allocation method and system
CN115018322B (en) * 2022-06-07 2024-07-02 山东冀联人力资源服务有限公司 Intelligent crowdsourcing task distribution method and system

Similar Documents

Publication Publication Date Title
Lin et al. Smart manufacturing scheduling with edge computing using multiclass deep Q network
Deb Introduction to evolutionary multiobjective optimization
US7043461B2 (en) Process and system for developing a predictive model
CN110782015A (en) Training method and device for network structure optimizer of neural network and storage medium
CN108154430A (en) A kind of credit scoring construction method based on machine learning and big data technology
CN114912705A (en) Optimization method for heterogeneous model fusion in federated learning
CN110222838B (en) Document sorting method and device, electronic equipment and storage medium
CN112101525A (en) Method, device and system for designing neural network through NAS
CN112308298B (en) Multi-scenario performance index prediction method and system for semiconductor production line
CN113283179A (en) Short-term load prediction method based on multi-target LSTM integrated network
CN113032367A (en) Dynamic load scene-oriented cross-layer configuration parameter collaborative tuning method and system for big data system
CN107329887A (en) A kind of data processing method and device based on commending system
CN115481727A (en) Intention recognition neural network generation and optimization method based on evolutionary computation
CN110110447B (en) Method for predicting thickness of strip steel of mixed frog leaping feedback extreme learning machine
CN112270058B (en) Optical network multichannel transmission quality prediction method based on echo state network
CN117421989A (en) Agent model-assisted parallel collaboration method for high-dimensional expensive optimization problem
CN113139639A (en) MOMBI-based smart city application-oriented multi-target calculation migration method and device
CN114547473A (en) Crowdsourcing task data recommendation method based on decision model and genetic matrix decomposition method
CN115562760B (en) Deep learning model layered unloading method based on edge computing node classification table
CN113094973B (en) Civil aircraft demand optimization method based on multi-objective optimization algorithm
CN113780146B (en) Hyperspectral image classification method and system based on lightweight neural architecture search
Starkey et al. A comparison of particle swarm optimization and genetic algorithms for a multi-objective Type-2 fuzzy logic based system for the optimal allocation of mobile field engineers
CN115470927A (en) Automatic extraction method of surrogate model, terminal and storage medium
CN108256694A (en) Based on Fuzzy time sequence forecasting system, the method and device for repeating genetic algorithm
Xue et al. RARTS: an efficient first-order relaxed architecture search method

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