CN113378051A - Crowd-sourced task recommendation method based on user-task association of graph neural network - Google Patents
Crowd-sourced task recommendation method based on user-task association of graph neural network Download PDFInfo
- Publication number
- CN113378051A CN113378051A CN202110664156.XA CN202110664156A CN113378051A CN 113378051 A CN113378051 A CN 113378051A CN 202110664156 A CN202110664156 A CN 202110664156A CN 113378051 A CN113378051 A CN 113378051A
- Authority
- CN
- China
- Prior art keywords
- task
- user
- association
- representation
- 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.)
- Granted
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 46
- 238000000034 method Methods 0.000 title claims abstract description 31
- 239000013598 vector Substances 0.000 claims abstract description 35
- 238000012549 training Methods 0.000 claims abstract description 30
- 230000003993 interaction Effects 0.000 claims abstract description 17
- 238000010586 diagram Methods 0.000 claims description 38
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000007430 reference method Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012358 sourcing Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
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
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a crowdsourcing task recommendation method based on graph neural network user-task association, which comprises the steps of collecting real interaction information between users, crowdsourcing task information borne by user history and preference label information of the users for tasks, which are generated by the users, through a crowdsourcing task platform, and respectively constructing a user association graph, a task association graph and a user-task association graph; respectively training by using a graph neural network so as to obtain preliminary vector representations of users and products; the method comprises the steps of utilizing a new graph neural network to keep crowdsourcing task information born by a user historically; and further obtaining a task recommendation list aiming at the user through the learned user and product representation. The invention obtains the task recommendation information more aiming at the user, and can better solve a part of recommendation cold start questions if the newly added user is connected with the historical user.
Description
Technical Field
The invention relates to the field of crowdsourcing task recommendation, which is mainly applied to a recommendation system, in particular to design of a graph neural network and construction of an association graph.
Background
With the vigorous development of the internet, individuals or organizations can acquire required services and ideas using a large number of network users, compared to professional knowledge acquired by a small number of experts. "crowd sourcing" (crowdsource) is a hybrid word produced by mixing the crowd (crowd) and outsource (outsource) word senses in 2006. This is a mode of synthesizing a final result by allocating work to many participants, and therefore, it is necessary to ensure the quality of crowdsourcing task completion, to recommend appropriate tasks to appropriate users, and how to recommend tasks suitable for them to different users, which faces a great challenge.
The graph neural network is a branch of graph representation learning, and aims to better adapt to downstream tasks by representing nodes in the graph with low-dimensional dense vectors and simultaneously preserving structural information formed between the nodes in the graph. Most of the traditional graph neural networks do not consider the types of nodes, that is, most of the graph neural networks are based on the assumption that the node types are the same, and in the recommendation system, two types of nodes, namely users and tasks, exist at the same time, so how to better adapt to the requirements of the node types in the recommendation system and obtain better user and task representations face huge challenges.
The method for recommending the user in the traditional recommendation system comprises the following steps: the historical task carrying information of the user and the historical carrying information of the user related to the user are observed, and then the tasks similar to the historical carrying tasks of the user and the tasks carried by the historical carrying tasks of the user related to the user are recommended. However, the measure of task similarity is rough, and the user-associated user involves a small range, so that more suitable tasks cannot be observed. The users and the tasks are represented through the graph neural network, and the relevance between the users and the tasks can be measured through uniform similarity calculation, so that the crowdsourcing tasks suitable for the users can be better and more widely screened.
Disclosure of Invention
The purpose of the invention is as follows: the invention provides a crowd-sourced task recommendation method based on user-task association of a graph neural network, which solves the technical problems that the measurement of task similarity in a traditional crowd-sourced task recommendation system is rough and the number of recommendable tasks which can be contacted by a user is small.
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 graph neural network user-task association obtains interaction information among users and information of the users for carrying the crowd-sourced tasks, which are generated by the users in enough crowd-sourced tasks, and association information among the tasks, including task types, subtasks, parent tasks and the like, and respectively constructs three graph structures of a user association graph, a task association graph and the user-task association graph; training the obtained user association diagram and task association diagram by using a graph neural network respectively so as to obtain preliminary vector representation of the users and products, and reserving relationship information between the users and between the products in the correlation representation; inputting the user representation and the product representation obtained in the data preprocessing stage into a new graph neural network as features by using a user-task association graph, and optimizing the user representation and the product representation on the basis, thereby retaining the crowdsourcing task information accepted by the user historically; through the learned user and product representation, the similarity between the user and the task can be calculated, and further a task recommendation list for the user is obtained. The method specifically comprises the following steps:
step 1, data acquisition: the method comprises the steps of collecting real interaction information between users generated by the users, crowdsourcing task information borne by user history and preference label information of the users for tasks through a crowdsourcing task platform. And meanwhile, crowdsourcing task information sent out by the platform is collected, wherein the crowdsourcing task information comprises category labels of tasks, split and combined information among the tasks and other attributes of the tasks.
Step 2, data preprocessing: the method comprises the steps of constructing a user association diagram through interaction information between users, constructing a task association diagram through split and combination information between tasks and category labels of the tasks, and constructing the user association diagram, the task association diagram and the user-task association diagram through crowdsourcing task information borne by user history.
Step 3, data representation: and obtaining user vector initial representation of the user through graph neural network training according to the obtained user association graph. And obtaining task vector initial representation of the task through graph neural network training according to the obtained task association diagram.
And 4, optimizing data: and (3) according to a user-task association diagram in data preprocessing, taking the initial user vector representation and the initial task vector representation obtained in the step (3) as initial inputs, and training through a graph neural network to obtain final representations of the user and the task.
Step 5, task recommendation: and (4) according to the final representation of the user and the task obtained by the calculation in the step (4), calculating the vector similarity between the user and different tasks and sequencing to obtain a task recommendation sequence finally aiming at the user.
Preferably: the method for establishing the user-task association diagram in the step 2 comprises the following steps:
step 2a), constructing a user association graph Guser：
Guser＝{Vuser，Euser，Xuser}
Wherein, VuserRepresenting a set of users, Vuser＝{u1，u2，...，ui，...，uM}，uiIndicates the ith user, the number of M users, EuserRepresenting a set of user edges, XuserCharacteristic attributes of the user are represented.
Step 2b), constructing a task association graph Gtask：
Gtask＝{Vtask，Etask，Xtask}
Wherein, VtaskRepresenting a set of tasks, Vtask＝{t1，t2，...，tj，...，tN}，tjDenotes the jth task, N denotes the number of tasks, EtaskRepresenting tasksSet of edges, XtaskThe characteristic attributes of the task are represented.
Step 2c), constructing the obtained user-task association graph Guser-task：
Guser-task＝{Vuser-task，Euser-task，Xuser-task}
Wherein, Vuser-taskRepresenting a unified set of users and tasks, Vuser-task＝{Vuser∪Vtask}，Euser-taskRepresenting sets of user and task edges, Xuser-task＝{Xuser∪XtaskDenotes the characteristic attributes of the user and task.
Preferably: and 3, obtaining the user vector initial representation of the user through graph neural network training according to the obtained user association diagram. Obtaining a task vector initial representation method of the task through graph neural network training according to the obtained task association diagram:
and 3a), constructing sides by using interaction information among users by using a user association diagram, taking a preference label of the user as supervision information, taking a characteristic attribute of the user as characteristic input, and performing iterative training by using a diagram convolution network to obtain corresponding user vector initial representation, wherein the user vector initial representation can retain the interaction information among the users.
And 3b), constructing edges by using the task association graph and the association between the tasks, taking the category labels of the tasks as supervision information, taking the characteristic attributes of the tasks as input, and performing iterative training by using a graph convolution network to obtain corresponding task vector initial representation, wherein the task vector initial representation can keep the association information between the tasks.
Preferably: in the step 4, the method for training through the graph neural network to obtain the final representation of the user and the task comprises the following steps:
and 4a), taking the user representation and the task representation in the user-task association diagram as input, and training through a graph neural network to obtain a new user representation and a new task representation.
And 4b), returning the new user representation and the new task representation to the graph neural network model in the step 3 for judgment, so as to optimize model parameters and further optimize the user representation and the task representation.
And 4c), iterating the step 4a) and the step 4b), and finally achieving convergence, so that better user representation and task representation are obtained as final representations of the user and the task.
Preferably: in the step 5, according to the final representation of the user and the task obtained in the step 4, by calculating the similarity between the user and different task representations and adopting the vector similarity as a reference method, the relevance between the user and different tasks can be scored, so that a proper crowdsourcing task sequence is screened out.
Preferably: other attributes of the task itself include the party involved, the type of task.
Compared with the prior art, the invention has the following beneficial effects:
1. the method and the device respectively represent the users and the tasks based on the graph neural network, and simultaneously keep information among the users, among the tasks, historical tasks of the users and the like, so that task recommendation information more specific to the users is obtained, and if connection between the newly added users and the historical users is obtained, part of recommended cold start problems can be better solved.
2. According to the invention, the user and the task are represented through the graph neural network, and the relevance between the user and the task can be measured through relatively uniform similarity calculation, so that the crowdsourcing task suitable for the user can be better and more widely screened.
Drawings
FIG. 1 is an overall flow chart of the method of the present invention;
FIG. 2 is a flow chart of data representation and data optimization performed by the neural network of the present invention.
Detailed Description
The present invention is further illustrated by the following description in conjunction with the accompanying drawings and the specific embodiments, it is to be understood that these examples are given solely for the purpose of illustration and are not intended as a definition of the limits of the invention, since various equivalent modifications will occur to those skilled in the art upon reading the present invention and fall within the limits of the appended claims.
A crowd-sourced task recommendation method based on graph neural network user-task association is disclosed, as shown in FIG. 1, and comprises the following steps:
step 1, data acquisition: the method comprises the steps of collecting and obtaining enough real interaction information between users, crowdsourcing task information born by user history and preference label information of the users for tasks, wherein the interaction information is generated by the users, and the crowdsourcing task information is generated by the users. And meanwhile, crowdsourcing task information sent out by the platform is collected, wherein the crowdsourcing task information comprises category labels of tasks, split and combined information among the tasks and other attributes of the tasks. Other attributes of the task itself include the party involved, the type of task, etc.
Step 2, data preprocessing: the method comprises the steps of constructing a user association diagram through interaction information between users, constructing a task association diagram through split and combination information between tasks and category labels of the tasks, and constructing the user association diagram, the task association diagram and the user-task association diagram through crowdsourcing task information borne by user history.
Firstly, collecting data required by constructing a related structure diagram, wherein the data comprises interaction information between users and task information born by user history; and the information forming the association among the tasks comprises the task category, the splitting and merging of the tasks, the owner and all platforms of the tasks, and the like. And on the basis of collecting the relevant data, the construction of the relevant structure chart is completed.
Step 2a), constructing a user association graph Guser：
Guser＝{Vuser，Euser，Xuser}
Wherein, VuserRepresenting a set of users, each user becoming a node in a user association graph, Vuser＝{u1，u2，...，ui，...，uM}，uiIndicates the ith user, the number of M users, EuserRepresenting a set of user edges, being used if there is an interaction between users, e.g. there is a commute between friendsForm an edge, X, between householdsuserCharacteristic attributes of the user are represented.
Step 2b), constructing a task association graph Gtask：
Gtask＝{Vtask，Etask，Xtask}
Wherein, VtaskRepresenting a set of tasks, each task becoming a node in a task dependency graph, Vtask＝{t1，t2，...，tj，...，tN}，tjDenotes the jth task, N denotes the number of tasks, EtaskRepresenting sets of task edges, and forming edges, X, between tasks if there is an interaction between the tasks, e.g. from the same parent task or from the same contracting partytaskThe characteristic attributes of the task are represented.
Step 2c), constructing the obtained user-task association graph Guser-task：
Guser-task＝{Vuser-task，Euser-task，Xuser-task}
Wherein, Vuser-taskRepresenting a unified set of users and tasks, Vuser-task＝{Vuser∪Vtask}，Euser-taskRepresenting a set of user and task edges, and forming an edge, X, between the user and the task if the user historically takes over the associated taskuser-task＝{Xuser∪XtaskDenotes the characteristic attributes of the user and task.
Step 3, data representation: as shown in fig. 2, the initial representation of the user vector of the user is obtained by training through a neural network (here, GCN is used as a method) according to the obtained user association diagram. And obtaining task vector initial representation of the task through graph neural network training according to the obtained task association diagram.
And training related graph structures and information through a graph neural network to finally obtain related representations of users and tasks. The interactive information among the users is constructed, the preference labels of the users are used as supervision information, the characteristic attributes of the users are used as characteristic input, and then training is carried out; in the task part, the association between tasks is constructed into edges, the category label of the task itself is used as supervision information, the characteristic attribute of the task itself is used as input, and training is carried out, specifically as follows:
step 3a), using the user association graph GuserConstructing the interactive information among users, using the preference label of the user as the supervision information, and the characteristic attribute X of the useruserAs characteristic input, iterative training is carried out by using Graph convolution network (Graph relational Networks) to obtain corresponding user vector initial representation
The representation can preserve information of interaction between users,
representing a real number.
Step 3b), using the task correlation graph GtaskThe correlation between tasks is constructed into edges, the category label of the task is used as supervision information, and the characteristic attribute X of the task is used as supervision informationtaskAs input, iterative training is performed by using a graph convolution network to obtain a corresponding task vector initial representation
The representation can preserve the association information between tasks.
And 4, optimizing data: and (3) according to a user-task association diagram in data preprocessing, taking the initial user vector representation and the initial task vector representation obtained in the step (3) as initial inputs, and training through a graph neural network to obtain final representations of the user and the task.
And (3) taking the user representation and the task representation obtained in the step (3) as input through the constructed user-task association graph and the graph neural network, and performing iterative training by using a graph convolution network, so that task information borne by the user history is reserved, and further the related representation information of the user and the task is optimized, wherein the method specifically comprises the following steps:
step 4a), associating the user with the task Guser-taskUser meter inShown as RuserTask represents RtaskAs input, training is performed through a graph neural network to obtain a new user representation R'userAnd New task represents R'task。
Step 4b), representing the new user as R'userAnd New task represents R'taskAnd returning to the graph neural network model in the step 3 for judgment, so as to optimize model parameters and further optimize user representation and task representation.
And 4c), iterating the step 4a) and the step 4b), and finally achieving convergence, so that better user representation and task representation are obtained as final representations of the user and the task.
Step 5, task recommendation: and (4) according to the final representation of the user and the task obtained by the calculation in the step (4), calculating the vector similarity between the user and different tasks and sequencing to obtain a task recommendation sequence finally aiming at the user.
According to the final representation of the user and the task obtained in the step 4, similarity between the user and different task representations is calculated, and vector similarity is used as a reference method, so that relevance between the user and different tasks can be scored, and a proper crowdsourcing task sequence is screened out.
In actual deployment, because newly added users and tasks may exist, the users and the tasks can be simply initialized by simply judging the connection relation between the users or the tasks and the existing nodes in the graph, and meanwhile, the representations of the users and the tasks are regularly updated, so that the model is ensured to have a good effect.
In summary, the invention provides a crowd-sourced task recommendation method based on user-task association of a graph neural network, and the graph neural network is used for representing users and tasks, so that the relevance between the users and the tasks can be measured through relatively uniform similarity calculation, and therefore, the crowd-sourced tasks suitable for the users can be better and more widely screened.
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 (6)
1. A crowd-sourced task recommendation method based on user-task association of a graph neural network is characterized by comprising the following steps:
step 1, data acquisition: the method comprises the steps that real interaction information between users, crowdsourcing task information borne by user history and preference label information of the users for tasks, which are generated by the users, are collected through a crowdsourcing task platform; meanwhile, crowdsourcing task information sent out by the platform is collected, wherein the crowdsourcing task information comprises category labels of tasks, splitting and merging information among the tasks and other attributes of the tasks;
step 2, data preprocessing: constructing a user association diagram through interaction information between users, constructing a task association diagram through split and merge information between tasks and category labels of the tasks, and constructing a user association diagram, a task association diagram and a user-task association diagram through crowdsourcing task information borne by user history;
step 3, data representation: obtaining user vector initial representation of the user through graph neural network training according to the obtained user association diagram; obtaining task vector initial representation of the task through graph neural network training according to the obtained task association diagram;
and 4, optimizing data: according to a user-task association diagram in data preprocessing, taking the user vector initial representation and the task vector initial representation obtained in the step 3 as initial inputs, and training through a graph neural network to obtain final representations of the user and the task;
step 5, task recommendation: and (4) according to the final representation of the user and the task obtained by the calculation in the step (4), calculating the vector similarity between the user and different tasks and sequencing to obtain a task recommendation sequence finally aiming at the user.
2. The crowd-sourced task recommendation method based on graph neural network user-task association of claim 1, wherein: the method for establishing the user-task association diagram in the step 2 comprises the following steps:
step 2a), constructing a user association graph Guser：
Guser＝{Vuser，Euser，Xuser}
Wherein, VuserRepresenting a set of users, Vuser＝{u1，u2，...，ui，...，uM}，uiIndicates the ith user, the number of M users, EuserRepresenting a set of user edges, XuserThe characteristic attribute of the user is represented;
step 2b), constructing a task association graph Gtask：
Gtask＝{Vtask，Etask，Xtask}
Wherein, VtaskRepresenting a set of tasks, Vtask＝{t1，t2，...，tj，...，tN}，tjDenotes the jth task, N denotes the number of tasks, EtaskRepresenting a set of task edges, XtaskThe characteristic attributes of the task are represented;
step 2c), constructing the obtained user-task association graph Guser-task：
Guser-task＝{Vuser-task，Euser-task，Xuser-task}
Wherein, Vuser-taskRepresenting a unified set of users and tasks, Vuser-task＝{Vuser∪Vtask}，Euser-taskRepresenting sets of user and task edges, Xuser-task＝{Xuser∪XtaskDenotes the characteristic attributes of the user and task.
3. The crowd-sourced task recommendation method based on graph neural network user-task association of claim 2, wherein: in step 3, obtaining user vector initial representation of the user through graph neural network training according to the obtained user association diagram; obtaining a task vector initial representation method of the task through graph neural network training according to the obtained task association diagram:
step 3a), constructing edges by using interaction information among users by using a user association diagram, taking a preference label of each user as supervision information, taking a characteristic attribute of each user as characteristic input, and performing iterative training by using a diagram convolution network to obtain corresponding user vector initial representation, wherein the user vector initial representation can keep the interaction information among the users;
and 3b), constructing edges by using the task association graph and the association between the tasks, taking the category labels of the tasks as supervision information, taking the characteristic attributes of the tasks as input, and performing iterative training by using a graph convolution network to obtain corresponding task vector initial representation, wherein the task vector initial representation can keep the association information between the tasks.
4. The graph neural network-based crowd-sourced task recommendation method for user-task association according to claim 3, wherein: in the step 4, the method for training through the graph neural network to obtain the final representation of the user and the task comprises the following steps:
step 4a), taking the user representation and the task representation in the user-task association diagram as input, and training through a graph neural network to obtain a new user representation and a new task representation;
step 4b), returning the new user representation and the new task representation to the graph neural network model in the step 3 for judgment, so as to optimize model parameters and further optimize the user representation and the task representation;
and 4c), iterating the step 4a) and the step 4b), and finally achieving convergence, so that better user representation and task representation are obtained as final representations of the user and the task.
5. The graph neural network-based crowd-sourced task recommendation method for user-task association according to claim 4, wherein: in the step 5, according to the final representation of the user and the task obtained in the step 4, by calculating the similarity between the user and different task representations and adopting the vector similarity as a reference method, the relevance between the user and different tasks can be scored, so that a proper crowdsourcing task sequence is screened out.
6. The graph neural network-based crowd-sourced task recommendation method for user-task association according to claim 5, wherein: other attributes of the task itself include the party involved, the type of task.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110664156.XA CN113378051B (en) | 2021-06-16 | 2021-06-16 | User-task association crowdsourcing task recommendation method based on graph neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110664156.XA CN113378051B (en) | 2021-06-16 | 2021-06-16 | User-task association crowdsourcing task recommendation method based on graph neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113378051A true CN113378051A (en) | 2021-09-10 |
| CN113378051B CN113378051B (en) | 2024-03-22 |
Family
ID=77574501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110664156.XA Active CN113378051B (en) | 2021-06-16 | 2021-06-16 | User-task association crowdsourcing task recommendation method based on graph neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113378051B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120303710A1 (en) * | 2011-05-27 | 2012-11-29 | Gracenote, Inc. | Systems and methods for generating and employing a social media graph |
| US20170337287A1 (en) * | 2003-06-25 | 2017-11-23 | Susan (Zann) Gill | Intelligent integrating system for crowdsourcing and collaborative intelligence in human- and device- adaptive query-response networks |
| CN110737778A (en) * | 2019-09-04 | 2020-01-31 | 北京邮电大学 | Knowledge graph and Transformer based patent recommendation method |
| CN111506835A (en) * | 2020-04-17 | 2020-08-07 | 北京理工大学 | Data feature extraction method fusing user time features and individual features |
| CN111881350A (en) * | 2020-07-23 | 2020-11-03 | 清华大学 | Recommendation method and system based on mixed graph structured modeling |
| CN112508256A (en) * | 2020-12-01 | 2021-03-16 | 安徽大学 | User demand active prediction method and system based on crowdsourcing |
| CN112559878A (en) * | 2020-12-24 | 2021-03-26 | 山西大学 | Sequence recommendation system and recommendation method based on graph neural network |
| CN112733018A (en) * | 2020-12-31 | 2021-04-30 | 哈尔滨工程大学 | Session recommendation method based on graph neural network GNN and multi-task learning |
| CN112749323A (en) * | 2019-10-31 | 2021-05-04 | 北京沃东天骏信息技术有限公司 | Method and device for constructing user portrait |
-
2021
- 2021-06-16 CN CN202110664156.XA patent/CN113378051B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170337287A1 (en) * | 2003-06-25 | 2017-11-23 | Susan (Zann) Gill | Intelligent integrating system for crowdsourcing and collaborative intelligence in human- and device- adaptive query-response networks |
| US20120303710A1 (en) * | 2011-05-27 | 2012-11-29 | Gracenote, Inc. | Systems and methods for generating and employing a social media graph |
| CN110737778A (en) * | 2019-09-04 | 2020-01-31 | 北京邮电大学 | Knowledge graph and Transformer based patent recommendation method |
| CN112749323A (en) * | 2019-10-31 | 2021-05-04 | 北京沃东天骏信息技术有限公司 | Method and device for constructing user portrait |
| CN111506835A (en) * | 2020-04-17 | 2020-08-07 | 北京理工大学 | Data feature extraction method fusing user time features and individual features |
| CN111881350A (en) * | 2020-07-23 | 2020-11-03 | 清华大学 | Recommendation method and system based on mixed graph structured modeling |
| CN112508256A (en) * | 2020-12-01 | 2021-03-16 | 安徽大学 | User demand active prediction method and system based on crowdsourcing |
| CN112559878A (en) * | 2020-12-24 | 2021-03-26 | 山西大学 | Sequence recommendation system and recommendation method based on graph neural network |
| CN112733018A (en) * | 2020-12-31 | 2021-04-30 | 哈尔滨工程大学 | Session recommendation method based on graph neural network GNN and multi-task learning |
Non-Patent Citations (2)
| Title |
|---|
| WENQI FAN ET AL.: "Graph Neural Networks for Social Recommendation", 《THE WORLD WIDE WEB CONFERENCE》, pages 417 - 426 * |
| 王健宗 等: "图神经网络综述", 《计算机工程》, pages 1 - 12 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113378051B (en) | 2024-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Competing bandits in matching markets | |
| Zeng et al. | Intuitionistic fuzzy social network hybrid MCDM model for an assessment of digital reforms of manufacturing industry in China | |
| CN108133330B (en) | Social crowdsourcing task allocation method and system | |
| CN111881350B (en) | Recommendation method and system based on mixed graph structured modeling | |
| Feng et al. | An adjustable approach to fuzzy soft set based decision making | |
| CN110597992B (en) | Semantic reasoning method and device based on knowledge graph and electronic equipment | |
| CN111125537A (en) | Session recommendation method based on graph representation | |
| CN111967972A (en) | Financial product recommendation method and device | |
| WO2020211240A1 (en) | Joint construction method and apparatus for prediction model, and computer device | |
| CN112184391A (en) | Recommendation model training method, medium, electronic device and recommendation model | |
| CN115775026B (en) | Federal learning method based on tissue similarity | |
| Alfantoukh et al. | Multi-stakeholder consensus decision-making framework based on trust: a generic framework | |
| Xu et al. | A hesitant fuzzy programming method for hybrid MADM with incomplete attribute weight information | |
| CN107346333B (en) | Online social network friend recommendation method and system based on link prediction | |
| CN114444693A (en) | Man-machine interaction topic transfer method based on collaborative knowledge graph | |
| Liu et al. | Wheel hub customization with an interactive artificial immune algorithm | |
| CN111126607B (en) | Data processing method, device and system for model training | |
| CN113378051A (en) | Crowd-sourced task recommendation method based on user-task association of graph neural network | |
| Panula-Ontto | AXIOM method for cross-impact modeling and analysis | |
| Chli et al. | Using the max-sum algorithm for supply chain emergence in dynamic multiunit environments | |
| CN113377656B (en) | Public testing recommendation method based on graph neural network | |
| CN114547480A (en) | Deep learning recommendation method and system based on multi-platform fusion | |
| CN107122852A (en) | A kind of microblog users interest Forecasting Methodology based on PMF | |
| CN114357283A (en) | Recommendation method based on comment and self-attention mechanism | |
| CN110599358B (en) | Cross-social network user identity association method based on probability factor graph model |
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 |