CN113361928A - Crowdsourcing task recommendation method based on special-pattern attention network - Google Patents
Crowdsourcing task recommendation method based on special-pattern attention network Download PDFInfo
- Publication number
- CN113361928A CN113361928A CN202110631471.2A CN202110631471A CN113361928A CN 113361928 A CN113361928 A CN 113361928A CN 202110631471 A CN202110631471 A CN 202110631471A CN 113361928 A CN113361928 A CN 113361928A
- Authority
- CN
- China
- Prior art keywords
- crowdsourcing
- task
- node
- tasks
- feature
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 18
- 239000013598 vector Substances 0.000 claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000005070 sampling Methods 0.000 claims abstract description 8
- 238000004140 cleaning Methods 0.000 claims abstract description 7
- 238000012216 screening Methods 0.000 claims abstract description 7
- 230000004931 aggregating effect Effects 0.000 claims abstract description 4
- 238000012545 processing Methods 0.000 claims abstract description 4
- 230000003993 interaction Effects 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 6
- 230000002452 interceptive effect Effects 0.000 claims description 6
- 230000007246 mechanism Effects 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 4
- 230000002776 aggregation Effects 0.000 claims description 4
- 238000004220 aggregation Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000003062 neural network model Methods 0.000 description 3
- 239000002904 solvent Substances 0.000 description 3
- 238000000605 extraction Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000005295 random walk Methods 0.000 description 1
- 238000013522 software testing Methods 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06Q10/06316—Sequencing of tasks or work
-
- 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
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention discloses a crowdsourcing task recommendation method based on a heteromorphic graph attention network, which comprises the steps of collecting crowdsourcing data, sequentially performing feature cleaning, feature screening and feature completion processing to obtain attribute feature vectors of crowdsourcing participants and tasks, constructing a heterogeneous crowdsourcing network, performing sampling and pre-training according to a set element path, obtaining node feature representation of the participants and the tasks, performing model training, and obtaining attention among node pairs; transmitting and aggregating neighbor node information according to the attention coefficient, and updating node characteristic representations of personnel and tasks; finally, according to the learned node feature representation, calculating similarity scores of the personnel and the tasks, and recommending the tasks according to the score sequence; according to the invention, aiming at the existing task information and personnel information in crowdsourcing, a crowdsourcing heterogeneous network is established, and the attention network is used for updating the feature vector in the crowdsourcing network, so that the crowdsourcing task recommendation accuracy is improved, and the completion efficiency and the completion quality are further improved.
Description
Technical Field
The invention relates to the technical field of crowdsourcing task recommendation, in particular to a crowdsourcing task recommendation method based on an attention network of an abnormal picture.
Background
With the development of economic globalization and network science and technology, crowdsourcing as a distributed problem solving mechanism based on the internet accomplishes tasks that are difficult to accomplish by a computer alone by integrating the computer and unknown masses on the internet. In recent years, the public has gained much attention in the academic and industrial fields of the computer field due to the public popularity of many application backgrounds, such as mass image annotation, online commodity evaluation, software testing, and the like.
In the crowdsourcing mode, how to accurately recommend tasks to participants who are more likely to complete the tasks is one of the key problems to be solved. The accurate recommendation of the tasks directly influences the completion efficiency and the completion quality of the crowdsourcing tasks. However, the lack of information and the limitations of recommendation methods make it difficult to accurately recommend existing tasks to participating personnel. Therefore, aiming at the problem of crowd-sourced task recommendation, how to accurately recommend a task to a participant who is more likely to complete the task faces a great challenge.
When crowdsourcing task recommendation is performed, only the task and personnel attributes are considered, and often a recommendation result which is not accurate enough is obtained. Because the attribute data is an independent data structure, the interaction information of the participators and the historical tasks, the similarity between the personnel and the tasks and the similarity between the tasks are crowd-sourced, and the attribute data is also very important for accurate task recommendation. Therefore, modeling crowd-sourced data as a heterogeneous crowd-sourced network will provide rich interaction information for the recommendation task. By learning the node representation of the personnel and the tasks in the crowdsourcing network, accurate task recommendation can be achieved by calculating the node representation similarity. However, in consideration of the heterogeneous characteristics of nodes in the network, the conventional network representation learning method cannot be directly applied to crowd-sourced task recommendation, and a targeted method for establishing a model according to the characteristics of crowd-sourced data is required.
Graph Neural Networks (GNNs) are a Neural network model that learns about network structure data representations. And the graph neural network updates the feature representation of each node by combining the linear transformation of the node features and the neighbor feature aggregation and the iterative learning of a downstream task. The graph attention network is a graph neural network combined with an attention mechanism, and can learn weights of feature aggregation of different neighbor nodes, so that more effective node feature representation learning is realized.
Disclosure of Invention
The purpose of the invention is as follows: most of the existing graph neural networks aim at homogeneous networks, and crowdsourcing networks have node heterogeneity of tasks and personnel, and the existing graph neural network model needs to be modified to extract heterogeneous semantic information. The invention provides a crowd-sourced task recommendation method based on an attention network of a heteromorphic image, which aims at the existing task information and personnel information in crowd-sourced mode to establish a crowd-sourced heterogeneous network. Network structure attribute extraction is carried out based on the meta-path, and feature vectors in the crowdsourcing network are updated by using the attention network, so that the crowdsourcing task recommendation accuracy is improved, and the completion efficiency and the completion quality of the crowdsourcing task are improved.
The technical scheme is as follows: in order to achieve the purpose, the invention adopts the technical scheme that:
a crowd-sourced task recommendation method based on an abnormal picture attention network comprises the following steps:
step S1, data acquisition;
collecting crowdsourcing data generated by crowdsourcing participants and crowdsourcing task objects in a plurality of real application scenes; the crowdsourcing data comprises crowdsourcing participant names, sexes, ages, advantages, historical task information, task ids, task descriptions, task affiliated modules and task types;
step S2, preprocessing data;
sequentially performing feature cleaning, feature screening and feature completion processing on the acquired crowdsourcing data to respectively obtain attribute feature vectors of crowdsourcing participants and crowdsourcing tasks; according to historical interaction information and characteristic information of crowdsourcing participants and crowdsourcing tasks, which are connected with each other, and a KNN algorithm, a crowdsourcing network is constructed, sampling and pre-training are carried out according to a set element path, and structural feature vectors of the crowdsourcing participants and the crowdsourcing tasks based on structural information are obtained respectively; finally, splicing the personnel structure characteristic vector and the task structure characteristic vector based on the structure information to obtain node characteristic representation of personnel and tasks;
step S3, training a model;
calculating the attention between the node pairs according to the node characteristic representation of the personnel and the task obtained in the step S2, transmitting and aggregating neighbor node information according to the calculated attention coefficient, and updating the node characteristic representation of the personnel and the task;
step S4, recommending tasks;
in the recursive update step S3, the node feature representation of the person and the task is calculated based on the learned node feature representation, and the task recommendation is performed in order of score by calculating the similarity score between the person and the task.
Further, in step S2, according to the historical interaction information of the crowdsourcing participants and the crowdsourcing task establishing connection, and the feature information of the crowdsourcing participants and the crowdsourcing task, a crowdsourcing network is constructed by combining the KNN algorithm, and specifically includes:
s2.1, performing feature cleaning, feature screening and feature completion on all data to obtain node attribute feature vector Hatt；
And S2.2, selecting the personnel and the tasks with the historical interactive connection according to the historical interactive information of the crowdsourcing participants and crowdsourcing task connection, and establishing personnel-task connection. Respectively calculating Euclidean distances of characteristic information representation of each crowdsourcing participant and other crowdsourcing participants by adopting a KNN algorithm to obtain a pairwise similarity index set SwCalculating Euclidean distance of the feature representation of each crowdsourcing task and other crowdsourcing tasks to obtain pairwise similarity index set St. For each crowd-sourced participant, S is selectedwThe first K same nodes with the maximum similarity index value are established to form person-person continuous edges(ii) a For each crowdsourcing task, selecting StEstablishing task-task connection edges for the first K same nodes with the maximum similarity index value, thereby obtaining a crowdsourcing heterogeneous network G;
step S2.3, according to the set meta-path set ρ ═ { ρ ═ ρ }1,ρ2,…,ρMSampling, and obtaining a node structure characteristic vector set H by using a metapath2vec algorithmst＝{Hst1,Hst2,…,HstM}; the Metapath2vec algorithm is a heterogeneous graph structural feature extraction method based on meta-paths, sampling is carried out through random walk of the predefined meta-paths, then node feature representation training is carried out through a shallow neural network, and semantic information and structural information in the heterogeneous network can be effectively reserved.
Step S2.4, the obtained attribute feature vector is spliced with the structure feature vector to obtain a corresponding node feature representation H ═ Hatt||Hst1||Hst2||…||HstM]。
Further, the specific steps of the model training in step S3 include:
s3.1, based on the input node characteristics H, performing linear transformation through a linear layer, multiplying by a weight matrix W, and obtaining transformed node characteristics H';
s3.2, calculating attention coefficients alpha pair by pair for two nodes connected with each group of edges of the crowdsourcing network; specifically, firstly, the feature vectors of two nodes i and j are spliced to obtain [ H'i||H’j]Then multiplying by a learnable attention parameter vector a, outputting the result through LeakyReLU, and finally carrying out attention coefficient normalization operation by using a softmax function to obtain alpha;
s3.3, correspondingly multiplying the calculated attention coefficient alpha with the transformed node characteristics H', calculating the aggregation of neighbor attention information for each node, and outputting through a sigmoid activation function to obtain an aggregated central node characteristic vector;
step S3.4, adopting a multi-head attention mechanism, repeatedly executing the steps S3.1-S3.3 to obtain a multi-head node representation Hagg＝{Hagg1,Hagg2,…,HaggTAnd finally, splicing the node representations calculated by each attention head to obtain a final node representation Z ═ Hagg1||Hagg2||…||HaggT]。
Further, the task recommending step in step S4 specifically includes:
s4.1, acquiring node representation Z of personnel and tasks in the crowdsourcing network;
step S4.2, for each person piAcquiring a task set t ═ t in the network and without connection relation with the person1,t2,…,tN}；
Step S4.3, calculator piWith each task tjThe cosine similarity of the task to be recommended is obtained, and the similarity score s of the task to be recommended corresponding to each person is obtained, wherein s is { s }1,s2,…,sN}；
Step S4.4, for each person piAnd sorting according to the similarity scores s of the tasks to be recommended, and selecting the L tasks with the highest sorting for recommendation.
Has the advantages that:
the invention provides a crowd-sourced task recommendation method based on an abnormal picture attention network, which can effectively realize accurate task recommendation in a crowd-sourced scene and improve the working efficiency of a crowd-sourced platform. The method for establishing the heterogeneous crowdsourcing network is adopted, so that the interaction information in the data can be fully utilized; and the advantages of the graph neural network model are fully utilized, and the node representation can be effectively learned for task recommendation.
Drawings
FIG. 1 is a flowchart of a crowd-sourced task recommendation method based on an attention network of an anomaly map according to the present invention;
FIG. 2 is a flow chart of a data preprocessing process according to an embodiment of the present invention.
FIG. 3 is a flowchart of model training based on a graph attention network according to the present invention.
Detailed Description
The present invention will be further described with reference to the accompanying drawings.
Fig. 1 shows a crowd-sourced task recommendation method based on an attention network of an anomaly map, which includes the following steps:
and step S1, data acquisition.
Crowdsourcing data generated by crowdsourcing participants and crowdsourcing task objects in a plurality of real application scenes is collected, wherein the crowdsourcing data comprises crowdsourcing participant names, sexes, ages, advantages, historical task information, task ids, task descriptions, task belonged modules and task types.
And step S2, preprocessing data.
Performing feature cleaning, feature screening, feature completion and other processing on the acquired data to obtain attribute feature vectors of personnel and tasks; meanwhile, a crowdsourcing network is constructed according to historical interaction information and a KNN algorithm, a person-task connecting edge is constructed according to the historical interaction information, attribute similarity is calculated according to the KNN, and the person-person connecting edge and the task-task connecting edge are constructed, so that the crowdsourcing network is constructed. Then sampling and pre-training are carried out according to a set meta path to obtain structural feature vectors of the personnel and tasks based on the structural information; and finally, splicing the two types of vectors to obtain the node characteristic representation of the personnel and the task. In particular, the amount of the solvent to be used,
s2.1, performing feature cleaning, feature screening and feature completion on all data to obtain node attribute feature vector Hatt；
And S2.2, selecting the personnel and the tasks with the historical interactive connection according to the historical interactive information of the crowdsourcing participants and crowdsourcing task connection, and establishing personnel-task connection. Respectively calculating Euclidean distances of characteristic information representation of each crowdsourcing participant and other crowdsourcing participants by adopting a KNN algorithm to obtain a pairwise similarity index set SwCalculating Euclidean distance of the feature representation of each crowdsourcing task and other crowdsourcing tasks to obtain pairwise similarity index set St. For each crowd-sourced participant, S is selectedwEstablishing personnel-personnel connection edges among the first K similar nodes with the maximum similarity index value; for each crowdsourcing task, selecting StEstablishing task-task connection edges for the first K same nodes with the maximum similarity index value, thereby obtaining a crowdsourcing heterogeneous network G;
step S2.3, according to the set meta-path set ρ ═ { ρ ═ ρ }1,ρ2,…,ρMSampling, and obtaining a node structure characteristic vector set H by using a metapath2vec algorithmst＝{Hst1,Hst2,…,HstM}；
Step S2.4, the obtained attribute feature vector is spliced with the structure feature vector to obtain a corresponding node feature representation H ═ Hatt||Hst1||Hst2||…||HstM]。
And step S3, training a model.
And calculating the attention between the node pairs according to the node characteristic representation of the personnel and the task obtained in the step S2, transmitting and aggregating the neighbor node information according to the calculated attention coefficient, and updating the node characteristic representation of the personnel and the task. In particular, the amount of the solvent to be used,
s3.1, based on the input node characteristics H, performing linear transformation through a linear layer, multiplying by a weight matrix W, and obtaining transformed node characteristics H';
s3.2, calculating attention coefficients alpha pair by pair for two nodes connected with each group of edges of the crowdsourcing network; specifically, firstly, the feature vectors of two nodes i and j are spliced to obtain [ H'i||H’j]Then multiplying by a learnable attention parameter vector a, outputting the result through LeakyReLU, and finally carrying out attention coefficient normalization operation by using a softmax function to obtain alpha;
step S3.3, correspondingly multiplying the calculated attention coefficient α by the transformed node feature H ', and adding vectors obtained by multiplying the attentions of the neighboring nodes to each node i to obtain an aggregated central node feature vector H' i, i.e., H 'i ═ Σ α ij H' j, j ∈ Ni;
step 3.4, representing H ' of all nodes, and taking the H ' as the input of a sigmoid activation function to obtain a node feature vector Hagg after nonlinear activation, namely Hagg is sigmoid (H '); wherein sigmoid (x) is 1/(1+ e-x);
step S3.5, adopting a multi-head attention mechanism, and repeatedly executing the steps S3.1-S3.3 to obtain a multi-head nodeRepresents Hagg＝{Hagg1,Hagg2,…,HaggTAnd finally, splicing the node representations calculated by each attention head to obtain a final node representation Z ═ Hagg1||Hagg2||…||HaggT]。
Step S4, recommending tasks;
in the recursive update step S3, the node feature representation of the person and the task is calculated based on the learned node feature representation, and the task recommendation is performed in order of score by calculating the similarity score between the person and the task. In particular, the amount of the solvent to be used,
s4.1, acquiring node representation Z of personnel and tasks in the crowdsourcing network;
step S4.2, for each person piAcquiring a task set t ═ t in the network and without connection relation with the person1,t2,…,tN}；
Step S4.3, calculator piWith each task tjThe cosine similarity of the task to be recommended is obtained, and the similarity score s of the task to be recommended corresponding to each person is obtained, wherein s is { s }1,s2,…,sN}；
Step S4.4, for each person piAnd sorting according to the similarity scores s of the tasks to be recommended, and selecting the L tasks with the highest sorting for recommendation.
The above description is only of the preferred embodiments of the present invention, and it should be noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.
Claims (4)
1. A crowd-sourced task recommendation method based on an attention network of an abnormal picture is characterized by comprising the following steps:
step S1, data acquisition;
collecting crowdsourcing data generated by crowdsourcing participants and crowdsourcing tasks in a plurality of real application scenes; the crowdsourcing data comprises crowdsourcing participant names, sexes, ages, advantages, historical task information, task ids, task descriptions, task affiliated modules and task types;
step S2, preprocessing data;
sequentially performing feature cleaning, feature screening and feature completion processing on the acquired crowdsourcing data to respectively obtain attribute feature vectors of crowdsourcing participants and tasks; according to historical interaction information of connection between crowdsourcing participants and crowdsourcing tasks and feature information of the crowdsourcing participants and the crowdsourcing tasks, constructing a crowdsourcing network based on a KNN algorithm, sampling and pre-training according to a set element path, and respectively obtaining structural feature vectors of the crowdsourcing participants and the crowdsourcing tasks based on structural information; finally, splicing the personnel structure characteristic vector and the task structure characteristic vector based on the structure information to obtain node characteristic representation of personnel and tasks;
step S3, training a model;
calculating the attention between the node pairs according to the node characteristic representation of the personnel and the task obtained in the step S2, transmitting and aggregating neighbor node information according to the calculated attention coefficient, and updating the node characteristic representation of the personnel and the task;
step S4, recommending tasks;
in the recursive update step S3, the node feature representation of the person and the task is calculated based on the learned node feature representation, and the task recommendation is performed in order of score by calculating the similarity score between the person and the task.
2. The method for recommending crowdsourcing tasks based on the heteromorphic image attention network as claimed in claim 1, wherein the step S2 of constructing the crowdsourcing network based on the KNN algorithm specifically comprises:
s2.1, performing feature cleaning, feature screening and feature completion on all data to obtain node attribute feature vector Hatt；
And S2.2, selecting the personnel and the tasks with the historical interactive connection according to the historical interactive information of the crowdsourcing participants and crowdsourcing task connection, and establishing personnel-task connection. Respectively calculating Euclidean distances of characteristic information representation of each crowdsourcing participant and other crowdsourcing participants by adopting a KNN algorithmTo obtain a pairwise similarity index set SwCalculating Euclidean distance of the feature representation of each crowdsourcing task and other crowdsourcing tasks to obtain pairwise similarity index set St(ii) a For each crowd-sourced participant, S is selectedwEstablishing personnel-personnel connection edges among the first K similar nodes with the maximum similarity index value; for each crowdsourcing task, selecting StEstablishing task-task connection edges for the first K same nodes with the maximum similarity index value, thereby obtaining a crowdsourcing heterogeneous network G;
step S2.3, according to the set meta-path set ρ ═ { ρ ═ ρ }1,ρ2,…,ρMSampling, and obtaining a node structure characteristic vector set H by using a metapath2vec algorithmst＝{Hst1,Hst2,…,HstM}；
Step S2.4, the obtained attribute feature vector is spliced with the structure feature vector to obtain a corresponding node feature representation H ═ Hatt||Hst1||Hst2||…||HstM]。
3. The method for recommending crowdsourcing tasks based on the heteromorphic image attention network as claimed in claim 1, wherein the specific steps of model training in step S3 include:
s3.1, based on the input node characteristics H, performing linear transformation through a linear layer, multiplying by a weight matrix W, and obtaining transformed node characteristics H';
s3.2, calculating attention coefficients alpha pair by pair for two nodes connected with each group of edges of the crowdsourcing network; specifically, firstly, the feature vectors of two nodes i and j are spliced to obtain [ H'i||H’j]Then multiplying by a learnable attention parameter vector a, outputting the result through LeakyReLU, and finally carrying out attention coefficient normalization operation by using a softmax function to obtain alpha;
s3.3, correspondingly multiplying the calculated attention coefficient alpha with the transformed node characteristics H', calculating the aggregation of neighbor attention information for each node, and outputting through a sigmoid activation function to obtain an aggregated central node characteristic vector;
step S3.4, adopting a multi-head attention mechanism, repeatedly executing the steps S3.1-S3.3 to obtain a multi-head node representation Hagg＝{Hagg1,Hagg2,…,HaggTAnd finally, splicing the node representations calculated by each attention head to obtain a final node representation Z ═ Hagg1||Hagg2||…||HaggT]。
4. The crowd-sourced task recommendation method based on the heteromorphic image attention network as claimed in claim 1, wherein the task recommendation step in step S4 specifically includes:
s4.1, acquiring node representation Z of personnel and tasks in the crowdsourcing network;
step S4.2, for each person piAcquiring a task set t ═ t in the network and without connection relation with the person1,t2,…,tN}；
Step S4.3, calculator piWith each task tjThe cosine similarity of the task to be recommended is obtained, and the similarity score s of the task to be recommended corresponding to each person is obtained, wherein s is { s }1,s2,…,sN}；
Step S4.4, for each person piAnd sorting according to the similarity scores s of the tasks to be recommended, and selecting the L tasks with the highest sorting for recommendation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110631471.2A CN113361928B (en) | 2021-06-07 | 2021-06-07 | Crowd-sourced task recommendation method based on heterogram attention network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110631471.2A CN113361928B (en) | 2021-06-07 | 2021-06-07 | Crowd-sourced task recommendation method based on heterogram attention network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113361928A true CN113361928A (en) | 2021-09-07 |
| CN113361928B CN113361928B (en) | 2023-08-25 |
Family
ID=77532742
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110631471.2A Active CN113361928B (en) | 2021-06-07 | 2021-06-07 | Crowd-sourced task recommendation method based on heterogram attention network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113361928B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114004667A (en) * | 2021-09-17 | 2022-02-01 | 重庆大学 | Knowledge crowdsourcing cold start task modeling and recommending method |
| CN116701706A (en) * | 2023-07-29 | 2023-09-05 | 腾讯科技（深圳）有限公司 | Data processing method, device, equipment and medium based on artificial intelligence |
| CN114004667B (en) * | 2021-09-17 | 2024-05-14 | 重庆大学 | Knowledge crowdsourcing cold start task modeling and recommending method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111027838A (en) * | 2019-12-04 | 2020-04-17 | 杨剑峰 | Crowdsourcing task pushing method, device, equipment and storage medium thereof |
| CN112508256A (en) * | 2020-12-01 | 2021-03-16 | 安徽大学 | User demand active prediction method and system based on crowdsourcing |
-
2021
- 2021-06-07 CN CN202110631471.2A patent/CN113361928B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111027838A (en) * | 2019-12-04 | 2020-04-17 | 杨剑峰 | Crowdsourcing task pushing method, device, equipment and storage medium thereof |
| CN112508256A (en) * | 2020-12-01 | 2021-03-16 | 安徽大学 | User demand active prediction method and system based on crowdsourcing |
Non-Patent Citations (2)
| Title |
|---|
| CHONG JUN WANG 等: "Mechanism Design for Cross-Market Task Crowdsourcing", IEEE/ACM INTERNATIONAL SYMPOSIUM ON QUALITY OF SERVICE, pages 1 - 10 * |
| 李晓峰;李东;王妍玮;: "面向医疗系统的多样性数据深层神经网络推荐算法", 沈阳大学学报(自然科学版), no. 03, pages 55 - 61 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114004667A (en) * | 2021-09-17 | 2022-02-01 | 重庆大学 | Knowledge crowdsourcing cold start task modeling and recommending method |
| CN114004667B (en) * | 2021-09-17 | 2024-05-14 | 重庆大学 | Knowledge crowdsourcing cold start task modeling and recommending method |
| CN116701706A (en) * | 2023-07-29 | 2023-09-05 | 腾讯科技（深圳）有限公司 | Data processing method, device, equipment and medium based on artificial intelligence |
| CN116701706B (en) * | 2023-07-29 | 2023-09-29 | 腾讯科技（深圳）有限公司 | Data processing method, device, equipment and medium based on artificial intelligence |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113361928B (en) | 2023-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | SaliencyGAN: Deep learning semisupervised salient object detection in the fog of IoT | |
| CN109582864B (en) | Course recommendation method and system based on big data science and dynamic weight adjustment | |
| CN108920544A (en) | A kind of personalized position recommended method of knowledge based map | |
| CN109871504B (en) | Course recommendation system based on heterogeneous information network and deep learning | |
| CN112016601B (en) | Network model construction method based on knowledge graph enhanced small sample visual classification | |
| CN112967112B (en) | Electronic commerce recommendation method for self-attention mechanism and graph neural network | |
| CN110880019A (en) | Method for adaptively training target domain classification model through unsupervised domain | |
| CN115995018A (en) | Long tail distribution visual classification method based on sample perception distillation | |
| CN114332984B (en) | Training data processing method, device and storage medium | |
| CN107391577B (en) | Work label recommendation method and system based on expression vector | |
| US20220237917A1 (en) | Video comparison method and apparatus, computer device, and storage medium | |
| CN109086463B (en) | Question-answering community label recommendation method based on regional convolutional neural network | |
| CN109446414A (en) | A kind of software information website fast tag recommended method based on neural network classification | |
| CN106649886A (en) | Method for searching for images by utilizing depth monitoring hash of triple label | |
| Zhang et al. | Multimodal analogical reasoning over knowledge graphs | |
| CN112364747A (en) | Target detection method under limited sample | |
| CN113868537B (en) | Recommendation method based on multi-behavior session graph fusion | |
| CN113361928B (en) | Crowd-sourced task recommendation method based on heterogram attention network | |
| CN116304336A (en) | Course recommendation method integrating knowledge graph and graph neural network | |
| CN115631008B (en) | Commodity recommendation method, device, equipment and medium | |
| CN109299291A (en) | A kind of Ask-Answer Community label recommendation method based on convolutional neural networks | |
| CN114240539B (en) | Commodity recommendation method based on Tucker decomposition and knowledge graph | |
| CN115601745A (en) | Multi-view three-dimensional object identification method facing application end | |
| CN115019342A (en) | Endangered animal target detection method based on class relation reasoning | |
| CN113011911A (en) | Data prediction method, device, medium and electronic equipment based on artificial intelligence |
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 |