CN112464042B - Task label generating method and related device for convolution network according to relationship graph - Google Patents
Task label generating method and related device for convolution network according to relationship graph Download PDFInfo
- Publication number
- CN112464042B CN112464042B CN202011342170.XA CN202011342170A CN112464042B CN 112464042 B CN112464042 B CN 112464042B CN 202011342170 A CN202011342170 A CN 202011342170A CN 112464042 B CN112464042 B CN 112464042B
- Authority
- CN
- China
- Prior art keywords
- task
- user
- data
- graph data
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- 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
-
- 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
- G06F16/901—Indexing; Data structures therefor; Storage structures
- G06F16/9024—Graphs; Linked lists
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present disclosure relates to the field of artificial intelligence, and in particular, to a method and an apparatus for generating task labels in a graph-based convolution network. Generating user task graph data according to the user data and the task data; inputting user task graph data into a relation graph convolution network, and obtaining at least one user characteristic and at least one task characteristic of the relation graph convolution network aiming at the user task graph data output; constructing a user task matrix according to the user characteristics and the task characteristics; and generating a target task label of a user corresponding to the user node according to probability distribution of the user node on at least one task label contained in the user task matrix. By the method, manual task assignment to the user by an administrator is avoided, task assignment efficiency is improved, the generated task labels are more relevant to the user, and task assignment matching degree is improved.
Description
Technical Field
The present disclosure relates to the field of artificial intelligence, and in particular, to a method and an apparatus for generating task labels in a graph-based convolution network.
Background
Deep learning is the inherent regularity and presentation hierarchy of learning sample data, and the information obtained during such learning is helpful in interpreting data such as text, images and sounds. An important process for generating sample data is labeling the data, i.e., tagging the data, which can be used as a lead experience for deep learning.
Currently, data annotation is typically performed by an annotation system or platform. The administrator distributes labeling tasks to a large number of users for labeling through the labeling system, and the same labeling task is typically distributed to multiple users and the answers are aggregated as labeling results.
However, in the above solution, the task distribution process generally selects labeling personnel or users for the target task by subjective judgment and experience of an administrator, which has low distribution efficiency and may cause poor matching degree between the task and the labeling personnel, and reduces labeling efficiency.
Disclosure of Invention
According to the technical problems, the task label generating method and the related device of the convolution network according to the relation diagram are provided, so that task assignment to a user manually by an administrator is avoided, task assignment efficiency is improved, the generated task label has stronger correlation with the user, and matching degree of task assignment is improved.
Other features and advantages of the present application will be apparent from the following detailed description, or may be learned in part by the practice of the application.
According to an aspect of the embodiment of the present application, there is provided a task tag generation method of a convolutional network according to a relationship diagram, including:
generating user task graph data according to the user data and the task data, wherein the user task graph data comprises at least one user node and at least one task node, the user node corresponds to user information contained in the user data, and the task node corresponds to task information contained in the task data;
inputting user task graph data into a relation graph convolution network, and obtaining at least one user characteristic and at least one task characteristic of the relation graph convolution network, wherein the user characteristic corresponds to a user node contained in the user task graph data, and the task characteristic corresponds to a task node contained in the user task graph data;
constructing a user task matrix according to the user characteristics and the task characteristics, wherein the user task matrix contains probability distribution of user nodes on at least one task label, and the task labels are in one-to-one correspondence with the task nodes;
And generating a target task label of a user corresponding to the user node according to probability distribution of the user node on at least one task label contained in the user task matrix.
In some embodiments of the present application, according to the above technical solution, before generating the user task graph data according to the user data and the task data, the method further includes:
generating historical user task graph data according to historical task data and historical user data, wherein the historical user data corresponds to a plurality of known task labels, and the known task labels correspond to the historical task information one by one;
and training the relation graph convolutional network according to the historical user task graph data and the known task labels.
In some embodiments of the present application, according to the above technical solution, generating user task graph data according to user data and task data includes:
generating user graph data according to the user data, wherein the user graph data comprises at least one user node, and generating task graph data according to the task data, and the task graph data comprises at least one task node;
and combining the user graph data and the task graph data to obtain the user task graph data.
In some embodiments of the present application, according to the above technical solution, generating user graph data according to user data includes:
acquiring user information contained in the user data and task record information respectively corresponding to each user information;
determining the same task completed among the users corresponding to the user information according to the task record information corresponding to the user information;
and generating user nodes in the user graph data according to the user information, and generating edges between the user nodes according to the same tasks completed between the users corresponding to the user information so as to obtain the user graph data.
In some embodiments of the present application, according to the above technical solution, generating task graph data according to task data includes:
task information contained in the task data and task history data corresponding to each piece of user information are acquired;
according to task history data respectively corresponding to the user information, determining the task information executed by the same user;
and generating task nodes in the task graph data according to the task information, and generating edges between the task nodes according to the task information executed by the same user to obtain the task graph data.
In some embodiments of the present application, according to the above technical solution, the merging processing is performed on the user task graph data and the task graph data to obtain user task graph data, including:
acquiring task record information corresponding to user nodes in the user graph data and task history data corresponding to task nodes in the task graph data;
according to the task record information and the task history data, determining an association relation between the user node and the task node, wherein the association relation indicates a user corresponding to the user node to finish a task corresponding to the task node;
and constructing edges between the user nodes and the task nodes between the user graph data and the task graph data according to the association relation between the user nodes and the task nodes so as to obtain the user task graph data.
In some embodiments of the present application, according to the above technical solution, generating, according to probability distribution of user nodes on at least one task label, a target task label of a user corresponding to the user nodes, where the target task label includes:
determining the maximum probability corresponding to the user node according to the user task matrix;
and if the maximum probability is greater than a preset probability distribution threshold, taking the task label corresponding to the maximum probability as a target task label of the user corresponding to the user node.
According to another aspect of the embodiments of the present application, there is provided a task tag generating device, including:
the user task graph data module is used for generating user task graph data according to the user data and the task data, wherein the user task graph data comprises at least one user node and at least one task node, the user node corresponds to user information contained in the user data, and the task node corresponds to task information contained in the task data;
the feature representation output module is used for inputting the user task graph data into the relation graph convolution network, obtaining at least one user feature of the relation graph convolution network, which is output aiming at the user task graph data, and at least one task feature, wherein the user feature corresponds to a user node contained in the user task graph data, and the task feature corresponds to a task node contained in the user task graph data;
the user task matrix construction module is used for constructing a user task matrix according to the user characteristics and the task characteristics, wherein the user task matrix contains probability distribution of user nodes on at least one task label, and the task labels are in one-to-one correspondence with the task nodes;
and the task label generating module is used for generating a target task label of a user corresponding to the user node according to probability distribution of the user node on at least one task label contained in the user task matrix.
According to another aspect of the embodiments of the present application, there is provided a task tag generation device for convolving a network according to a relationship diagram, the device including: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the task tag generation method of the graph-based convolution network as in the above technical solution via execution of the executable instructions.
According to another aspect of the embodiments of the present application, there is provided a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a task label generation method according to a relationship graph convolution network as in the above technical solution.
In the embodiment of the application, the correlation between the user and the user, the correlation between the task and the correlation between the user and the task are learned through the relation graph convolution network, and the task label with strong correlation with the user can be generated, so that the task is prevented from being manually assigned to the user by an administrator, the task assigning efficiency is improved, the generated task label has stronger correlation with the user, and the matching degree of task assignment is improved.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and together with the description, serve to explain the principles of the application. It is apparent that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained from these drawings without inventive effort for a person of ordinary skill in the art. In the drawings:
fig. 1 is a schematic diagram of an exemplary system architecture of the technical solution of the present application in an application scenario.
Fig. 2 is a flowchart of a task tag generation method according to a relationship graph rolling network according to an embodiment of the present application.
Fig. 3 is an exemplary schematic diagram of user graph data in an embodiment of the present application.
Fig. 4 is an example schematic diagram of task graph data in an embodiment of the present application.
Fig. 5 is an exemplary schematic diagram of user task graph data in an embodiment of the present application.
Fig. 6 is a block diagram of the task tag generation device in the embodiment of the present application.
Fig. 7 is a schematic diagram of a computer system suitable for use in implementing embodiments of the present application.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. However, the exemplary embodiments may be embodied in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.
Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the present application. One skilled in the relevant art will recognize, however, that the aspects of the application can be practiced without one or more of the specific details, or with other methods, components, devices, steps, etc. In other instances, well-known methods, devices, implementations, or operations are not shown or described in detail to avoid obscuring aspects of the application.
The block diagrams depicted in the figures are merely functional entities and do not necessarily correspond to physically separate entities. That is, the functional entities may be implemented in software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor devices and/or microcontroller devices.
The flow diagrams depicted in the figures are exemplary only, and do not necessarily include all of the elements and operations/steps, nor must they be performed in the order described. For example, some operations/steps may be decomposed, and some operations/steps may be combined or partially combined, so that the order of actual execution may be changed according to actual situations.
Fig. 1 is a schematic diagram of an exemplary system architecture of the technical solution of the present application in an application scenario. As shown in fig. 1, in an exemplary embodiment, the implementation environment may include: a terminal 101 and a server 102. The terminal 101 and the server 102 are provided with labeling systems. The labeling system is a system for labeling data, and effective labeling data generated by the labeling system provides source data for a model with strong generalization capability for deep learning training. Common labeling tasks in the labeling system are text labeling, voice labeling, translation labeling, image labeling and the like. In the labeling system, a server sends corresponding labeling tasks to each client, and a user sends the labeling tasks back to the server after completing the labeling tasks through the client. In some embodiments, the labeling system may score the user according to the task completion status of the user, such as the speed and number of task completion and the labeling result review status.
The terminal 101 and the server 102 establish a wired or wireless communication connection, and further realize data transmission with the server 102 through the communication connection.
The terminal 101 has a client running therein, which can provide a user interaction interface. User interaction operations can be triggered on the user interaction interface to conduct data annotation. The client may be, for example, an image annotation client, through which the user receives the assignment during image annotation.
It should be noted that, in the present embodiment, the terminal 101 may be a mobile phone, a tablet computer, a notebook computer, a desktop computer, or any other electronic device that may be used by a client to run, which is not limited herein. The client may be an application client or a web client, which is not limited in this regard.
The server 102 is configured to provide data support for user interaction triggered in the client, so that the client can operate normally in the terminal 101. Still taking the image labeling as an example, the server 102 may send the image to be labeled and the corresponding candidate task label to the client according to the image labeling operation triggered in the client, or receive the image labeling result from the client, so as to assign the subsequent labeling task content to the client.
The server 102 may be a single server device or a server group including a plurality of server devices, and is not limited herein.
The technical scheme provided by the application is described in detail below in connection with the specific embodiments.
Fig. 2 is a flowchart of a task tag generation method according to a relationship graph rolling network according to an embodiment of the present application. The method may be performed by the server 103 in the application scenario shown in fig. 1. As shown in fig. 2, in one embodiment, the method may include the steps of:
Step S210, generating user task graph data according to the user data and the task data, wherein the user task graph data comprises at least one user node and at least one task node, the user node corresponds to user information contained in the user data, and the task node corresponds to task information contained in the task data.
The main expression form of the user task graph data is an undirected weighted graph containing user nodes and task nodes, and the undirected weighted graph expresses the association relation and the association degree among users, tasks and between users and tasks related by the annotation system.
The user data includes user information and completion status. The user information includes user attributes such as age, history, profession, expertise, foreign language level, and the like. The task completion situation comprises the situation of tasks completed by a user, and mainly comprises information such as the number, the type and the time of the tasks. In the user task graph data, user information of users is contained in user nodes, edges between the user nodes represent association relations between the users relative to the completed tasks, and the larger the weight of the edges is, the closer the association relations are.
The task data includes task information and task records. The task information mainly includes a task type and task content. The task types include, for example, text annotation tasks, voice annotation tasks, translation annotation tasks, image annotation tasks, and the like. Task content is the task object of the corresponding task, for example, for a translation labeling task, task content is the original text and the translated text, introduction information about the translation labeling task itself, and the like. In the user task graph data, task information of tasks is included in task nodes, edges between task nodes represent association relations between tasks relative to users who complete the tasks, and the larger the weight of the edges is, the closer the association relations are.
The side between the user node and the task node represents the association relation between the user and the task, for example, the task completed by the user belongs to the association relation between the user and the task, and the association relation is tighter as the weight of the side is larger.
According to the user data and the task data, the association relations among users, between tasks and between users and tasks can be determined, and according to the association relations, the user task graph data can be generated.
S220, inputting the user task graph data into a relation graph convolution network, and obtaining at least one user characteristic and at least one task characteristic of the relation graph convolution network, wherein the user characteristic corresponds to a user node contained in the user task graph data, and the task characteristic corresponds to a task node contained in the user task graph data.
The relational graph rolling network (Relational Graph Convolutional Network, RGCN) is an extension on the large-scale relational data based on the graph rolling network, which is a graph rolling network that aggregates local neighbor information. In RGCN, a propagation model is defined for computing forward updates of nodes or entities in a relational graph:
Wherein, the liquid crystal display device comprises a liquid crystal display device,index, c, representing the neighbor set of node i at relationship r i,r Is a regularized constant which can be learned or extracted to be chosen as +.>Explicit summation of neighbors can be avoided using sparse matrix multiplication. The conversion of relationships to features is introduced in the RGCN, which depends on the type and direction of the edges. The conversion may be a linear message conversion and other more flexible functions may be used, such as a multi-layer neural network, while at the same time increasing the amount of computation required.
In the application, the relation graph rolling network calculates corresponding characteristic representation of each user and each task according to the user information and the task information in the user task graph data and the relation between the user and the task. The feature representation is typically a one-dimensional vector and may be a sparse feature vector, e.g., a one-hot vector. The user characteristic representation and the task characteristic representation are representations of user data and task data. Subsequently, embedding of the user (embedding) and embedding of the task (embedding) are learned through the RGCN.
The calculation process of the RGCN for each node in the user task graph may be, for example: and taking any node as a central node, carrying out convolution on the central node once, and updating the representation of the central node by aggregating the information of the neighbor nodes. The aggregation of the neighbor nodes is classified according to the types of the edges, corresponding conversion is carried out according to the different types of the edges, and the collected information is added through a regularization and finally a function is activated. Wherein the information of each vertex updates the shared parameters, and the parallel computation also includes self-connection, that is to say node self-expression.
S230, constructing a user task matrix according to the user characteristics and the task characteristics, wherein the user task matrix contains probability distribution of user nodes on at least one task label, and the task labels are in one-to-one correspondence with the task nodes.
In particular, constructing the user task matrix may employ a matrix multiplication approach. If there are 3 users and 3 tasks, a 3x3 user task matrix is constructed, with each element in the matrix representing a probability that the corresponding user is more adept at the corresponding task. The user task matrix is normalized by softmax, e.g., the probability distribution of each user over all tasks has a value between 0 and 1 and a sum of 1.
S240, generating a target task label of a user corresponding to the user node according to probability distribution of the user node on at least one task label contained in the user task matrix.
The probability distribution of each user to all task labels can be determined according to the user task matrix, and the task labels corresponding to each user can be generated according to a preset task distribution principle. For example, if a predetermined number of tasks are allocated to each user, the task labels may be arranged in descending order according to the probability amount and the task labels may be taken as a result according to the predetermined number, or all task labels with a probability greater than a predetermined threshold may be taken as a result, or, for each user, a task corresponding to the maximum probability for that user may be selected as a task label.
After determining the task tag of the user, the server may serve the task corresponding to the task tag to the user.
In the embodiment of the application, the correlation between the user and the user, the correlation between the task and the correlation between the user and the task are learned through the relation graph convolution network, and the task label with strong correlation with the user can be generated, so that the task is prevented from being manually assigned to the user by an administrator, the task assigning efficiency is improved, the generated task label has stronger correlation with the user, and the matching degree of task assignment is improved.
In some embodiments of the present application, on the basis of the above embodiments, before generating the user task graph data according to the user data and the task data in the step s210, the method may further include the steps of:
step S201, generating historical user task graph data according to historical task data and historical user data, wherein the historical user data corresponds to a plurality of known task labels, and the known task labels correspond to historical task information one by one;
and S202, training the relation graph convolutional network according to the historical user task graph data and the known task labels.
The corresponding relation between the historical user data and the task labels can be determined according to task records in the historical task data. For example, a task tag with the largest number of occurrences among tasks completed by a certain user included in the history user data is counted as the corresponding task tag. The content contained in the historical user task data is not repeated for the same number of times as the user task graph. Typically, the historical user data and the historical task data may directly employ historical task record information stored in the annotation system.
Based on historical user task graph data and known task labels, the relationship graph convolution network may be trained. Specifically, in the training process, historical user task graph data and known task labels are input into a relationship graph convolution network to obtain user characteristic representations and task characteristic representations. According to the obtained result and the corresponding relation between the historical user data and the known task labels, the solving of the loss function can be determined. The loss function can adopt various classification loss functions such as cross entropy and the like. The solution of the loss function is minimized by adjusting the parameters to determine the values of the various parameters in the relationship graph rolling network.
In the embodiment of the application, the historical user task graph data is constructed, and the relationship graph convolution network is trained by utilizing the historical user task graph data, so that a specific model conforming to the task condition and the user condition of the labeling system can be obtained, the accuracy of the output result of the relationship graph convolution network is improved, and the accuracy of task assignment is improved.
In some embodiments of the present application, based on the above embodiments, the generating user task graph data according to the user data and the task data includes:
s211, generating user graph data according to the user data, wherein the user graph data comprises at least one user node, and generating task graph data according to the task data, and the task graph data comprises at least one task node;
and S212, combining the user graph data and the task graph data to obtain the user task graph data.
Specifically, the user data graph is an undirected weighted graph with user information as nodes and the cooperative relationship between users relative to tasks as edges. The user graph data includes at least one user node. The task graph is an undirected weighted graph taking task information as a node and taking the association relation between tasks relative to users as an edge. The task graph data includes at least one task node.
After the user graph data and the task graph data are generated, the user graph data and the task graph data are combined, and the user task graph data can be obtained. Specifically, under the condition that the relation between nodes and edges in the user graph data and the task graph data is kept unchanged, edges are added between task nodes and user nodes according to the condition that a user completes a task, so as to be used for representing the association relation between the user and the task. Traversing each user node in the user graph data, determining the association relation between each user node and each task node in the task graph data, and establishing a corresponding side according to the association relation to obtain the user task graph data.
In the embodiment of the application, the user task graph is obtained by combining the user graph data and the task graph data, and the relationship between the user and the task is added under the condition of retaining the original data of the user graph data and the task graph data, so that the maintained data integrity is beneficial to improving the accuracy of the graph rolling network.
In some embodiments of the present application, on the basis of the above embodiments, generating user map data from the user data in step S211 includes:
s2111, acquiring user information contained in user data and task record information respectively corresponding to each piece of user information;
S2112, determining the same task completed among users corresponding to each user information according to task record information corresponding to each user information;
s2113, generating user nodes in the user graph data according to the user information, and generating edges between the user nodes according to the same tasks completed between the users corresponding to the user information so as to obtain the user graph data.
In one embodiment, the task record information mainly includes task types of the tasks completed by the user. The same task indicates a task of the same task type. Depending on the task type, the number of tasks that two users accomplish for the same type may be determined. In this embodiment, the edges between users in the user graph data represent the number of tasks of the same type that both users accomplish.
Specifically, referring to fig. 3, fig. 3 is an exemplary schematic diagram of user graph data in an embodiment of the present application. As shown in fig. 3, in this example, there is an edge between user a and user B, i.e., indicating that user a and user B have completed the same type of task, and the weight value of the edge indicates the number of tasks of the same type. For example, if user A has completed 30 phonetic annotation tasks and user B has completed 15 phonetic annotation tasks, the edge weight value is 15. For another example, if user a has completed 7 voice annotation tasks and 8 image annotation tasks, and user B has also completed 7 voice annotation tasks and 8 image annotation tasks, then the weight value of the edge is also 15, i.e., the weight value of the edge includes the number of tasks of the same type.
In another embodiment, the task record information includes task identifications of respective tasks completed by the user. The same task indicates that the task identifies the same task, i.e., the exact same task. Accordingly, edges between users in the user graph data represent the same number of tasks that the two users have completed. For example, if user B completes 4 picture annotation tasks, user C completes 5 picture annotation tasks and the pictures in 3 tasks are the same as the annotation tasks completed by user B, then there is an edge between user B and user C and the weight value of the edge is 3.
In this embodiment, user graph data is established according to the user information and the same task completed between the users corresponding to the user information, so that a cooperative relationship between a plurality of users and the task is fully reflected, and distribution rationality when the plurality of users are distributed in the same task is facilitated.
In some embodiments of the present application, on the basis of the above embodiments, generating task graph data from task data in step S211 includes:
s2114, task information contained in the task data and task history data corresponding to each piece of user information are obtained;
S2115, determining each piece of task information executed by the same user according to task history data respectively corresponding to each piece of user information;
step S2116, task nodes in the task graph data are generated according to the task information, and edges between the task nodes are generated according to the task information executed by the same user, so that the task graph data are obtained.
In one embodiment, the task history data mainly includes user information for completing the task and task completion time. In determining the information of each task executed by the same user, a time threshold may be set, and when the time difference between two tasks completed by the same user is within the time threshold, the two tasks are considered to have an association with respect to the user. In this embodiment, the edges between tasks in the task graph data represent the number of users who completed both tasks.
In particular, referring to fig. 4, fig. 4 is an exemplary schematic diagram of task graph data in an embodiment of the present application. As shown in fig. 4, in this example, there is an edge between task 1 and task 2, that is, indicating that task 1 and task 2 are completed by the same user, and the weight value of the edge indicates the number of users who completed task 1 and task 2. For example, if 40 users completed task 1 and 30 of them completed task 2, the weight value of the edge is 30.
In this embodiment, user graph data is established according to task information and each task information executed by the same user, so that association relations between a plurality of tasks and the user are fully reflected, and the user's model and skill are distributed when the tasks are distributed, so that the user completes task efficiency.
In some embodiments of the present application, based on the above embodiments, step s212 performs a merging process on the user task graph data and the task graph data to obtain user task graph data, where the merging process includes:
s2121, acquiring task record information corresponding to user nodes in user graph data and task history data corresponding to task nodes in task graph data;
s2122, determining an association relation between the user node and the task node according to the task record information and the task history data, wherein the association relation indicates a user corresponding to the user node to complete a task corresponding to the task node;
s2123, constructing edges between the user nodes and the task nodes between the user graph data and the task graph data according to the association relation between the user nodes and the task nodes so as to obtain the user task graph data.
Specifically, according to the task record information and the task history data, the specific task completed by each user and the number of times each specific task is completed can be specifically determined. According to the determined condition that each user completes the task, the association relation between the user node and the task node can be determined, and for the two nodes with the association relation, the edge between the user node and the task node can be constructed on the basis of the user graph data and the task graph data. The edge represents the case where the user completes the corresponding task, and the weight of the edge represents the number of times the user completes the task.
For convenience of description, please refer to fig. 5 in conjunction with fig. 3 and fig. 4, fig. 5 is an exemplary schematic diagram of user task graph data in an embodiment of the present application. Assuming that task 1 represents a translation annotation task, if user A completes 10 different translation annotation tasks, then accordingly, there is a side between user A and task 1 en route to the user task, and its weight is 10.
In another embodiment, task 1 represents one specific translation annotation task, such as annotating a translation of an article, and task 2 represents another specific translation annotation task. User a completes task 1 10 times and user B completes task 2 30 times, there is an edge in the user relationship graph between user a and task 1 with a weight of 10, and there is an edge between user B and task 2 with a weight of 30.
In the embodiment, the user task graph is generated based on the user graph data and the task graph data under the condition that the user completes the task, so that the relationship between the user and the task is fully considered in the calculation process of the relationship graph convolution network, the association degree between the task and the user is improved, and further, the feature representation of the user and the task is calculated more accurately.
In some embodiments of the present application, based on the above embodiments, the step s240 generates a target task label of a user corresponding to the user node according to probability distribution of the user node on at least one task label included in the user task matrix, including:
s241, determining the maximum probability corresponding to the user node according to the user task matrix;
and S242, if the maximum probability is greater than a preset probability distribution threshold, taking the task label corresponding to the maximum probability as a target task label of the user corresponding to the user node.
Specifically, the user task matrix includes the assignment probability of each task to each user. If the user task matrix includes 3 users and 3 tasks, for one of the users, there is a 3-dimensional vector, and each eigenvalue in the vector represents the probability that the corresponding task is assigned to that user. For example, the 3 tasks are text labeling, speech labeling, and image labeling, respectively, and for user X, the three-dimensional vector may be {0.7,0.2,0.1}.
In the vector for the specific user, the maximum probability can be determined, and if the maximum probability is greater than the probability distribution threshold, the association relationship between the user and the corresponding task is more compact, and the task label can be used as the task label of the user. For example, for the user X, the maximum probability is 0.7, and if the probability distribution threshold is 0.5, it is determined that the text label is the task label of the user.
In this embodiment, the task labels of the users are determined by comparing the maximum probability of the user nodes with the probability distribution threshold, which is beneficial to avoiding the task types of the users from being forcibly limited when the tendency of the users is not obvious, and is beneficial to improving the accuracy of task assignment.
It should be noted that although the steps of the methods in the present application are depicted in the accompanying drawings in a particular order, this does not require or imply that the steps must be performed in that particular order, or that all illustrated steps be performed, to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step to perform, and/or one step decomposed into multiple steps to perform, etc.
The following describes an implementation of the apparatus of the present application, which may be used to perform the task tag generation method according to the relationship graph rolling network in the above embodiments of the present application. Fig. 6 is a block diagram of the task tag generation device in the embodiment of the present application. As shown in fig. 6, the task tag generation device 300 may mainly include:
a user task graph data module 310, configured to generate user task graph data according to user data and task data, where the user task graph data includes at least one user node and at least one task node, the user node corresponds to user information contained in the user data, and the task node corresponds to task information contained in the task data;
The feature representation output module 320 is configured to input the user task graph data into the relationship graph convolutional network, obtain at least one user feature of the relationship graph convolutional network output for the user task graph data, and at least one task feature, where the user feature corresponds to a user node contained in the user task graph data, and the task feature corresponds to a task node contained in the user task graph data;
the user task matrix construction module 330 is configured to construct a user task matrix according to the user characteristics and the task characteristics, where the user task matrix contains probability distribution of user nodes on at least one task label, and the task labels are in one-to-one correspondence with the task nodes;
the task label generating module 340 is configured to generate a target task label of a user corresponding to the user node according to probability distribution of the user node on at least one task label included in the user task matrix.
In some embodiments of the present application, according to the above technical solution, the task tag generating device 300 further includes:
the historical user task graph data generation module is used for generating historical user task graph data according to historical task data and historical user data, wherein the historical user data corresponds to a plurality of known task labels, and the known task labels correspond to the historical task information one by one;
And the relation diagram convolution network training module is used for training the relation diagram convolution network according to the historical user task diagram data and the known task labels.
In some embodiments of the present application, according to the above technical solution, the user task graph data module 310 may include:
a user graph data generating unit, configured to generate user graph data according to user data, where the user graph data includes at least one user node, and generate task graph data according to task data, where the task graph data includes at least one task node;
and the merging processing unit is used for merging the user graph data and the task graph data to obtain the user task graph data.
In some embodiments of the present application, according to the above technical solution, the user graph data generating unit may include:
the user information acquisition subunit is used for acquiring user information contained in the user data and task record information corresponding to each user information respectively;
the same task determining subunit is used for determining the same task completed among the users corresponding to each user information according to the task record information corresponding to each user information;
and the user node generating subunit is used for generating user nodes in the user graph data according to the user information and generating edges between the user nodes according to the same tasks completed between the users corresponding to the user information so as to obtain the user graph data.
In some embodiments of the present application, according to the above technical solution, the user task graph data module 310 may include:
the task information acquisition subunit is used for acquiring task information contained in the task data and task history data corresponding to each piece of user information respectively;
the same user determining subunit is used for determining each task information executed by the same user according to the task history data respectively corresponding to each user information;
and the task node generating subunit is used for generating task nodes in the task graph data according to the task information and generating edges among the task nodes according to the task information executed by the same user so as to obtain the task graph data.
In some embodiments of the present application, according to the above technical solution, the merging processing unit may include:
the task record information acquisition subunit is used for acquiring task record information corresponding to the user nodes in the user graph data and acquiring task history data corresponding to the task nodes in the task graph data;
the association relation determining subunit is used for determining the association relation between the user node and the task node according to the task record information and the task history data, and the association relation indicates the user corresponding to the user node to finish the task corresponding to the task node;
And the user task graph data construction subunit is used for constructing edges between the user nodes and the task nodes between the user graph data and the task graph data according to the association relation between the user nodes and the task nodes so as to obtain the user task graph data.
In some embodiments of the present application, according to the above technical solution, the task tag generation module 340 may include:
the maximum probability determining unit is used for determining the maximum probability corresponding to the user node according to the user task matrix;
and the task label determining unit is used for taking the task label corresponding to the maximum probability as the target task label of the user corresponding to the user node if the maximum probability is larger than a preset probability distribution threshold value.
It should be noted that, the apparatus provided in the foregoing embodiments and the method provided in the foregoing embodiments belong to the same concept, and a specific manner in which each module performs an operation has been described in detail in the method embodiment, which is not described herein again.
Fig. 7 is a schematic diagram of a computer system suitable for use in implementing embodiments of the present application.
It should be noted that, the computer system 400 of the electronic device shown in fig. 7 is only an example, and should not impose any limitation on the functions and the application scope of the embodiments of the present application.
As shown in fig. 7, the computer system 400 includes a central processing unit (Central Processing Unit, CPU) 401, which can perform various appropriate actions and processes according to a program stored in a Read-Only Memory (ROM) 402 or a program loaded from a storage section 408 into a random access Memory (Random Access Memory, RAM) 403. In the RAM 403, various programs and data required for the system operation are also stored. The CPU 401, ROM 402, and RAM 403 are connected to each other by a bus 404. An Input/Output (I/O) interface 405 is also connected to bus 404.
The following components are connected to the I/O interface 405: an input section 406 including a keyboard, a mouse, and the like; an output portion 407 including a Cathode Ray Tube (CRT), a liquid crystal display (Liquid Crystal Display, LCD), and the like, a speaker, and the like; a storage section 408 including a hard disk or the like; and a communication section 409 including a network interface card such as a LAN (Local Area Network ) card, a modem, or the like. The communication section 409 performs communication processing via a network such as the internet. The drive 410 is also connected to the I/O interface 405 as needed. A removable medium 411 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is installed on the drive 410 as needed, so that a computer program read therefrom is installed into the storage section 408 as needed.
In particular, according to embodiments of the present application, the processes described in the various method flowcharts may be implemented as computer software programs. For example, embodiments of the present application include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flowcharts. In such an embodiment, the computer program may be downloaded and installed from a network via the communication portion 409 and/or installed from the removable medium 411. When executed by a Central Processing Unit (CPU) 401, performs the various functions defined in the system of the present application.
It should be noted that, the computer readable medium shown in the embodiments of the present application may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-Only Memory (ROM), an erasable programmable read-Only Memory (Erasable Programmable Read Only Memory, EPROM), flash Memory, an optical fiber, a portable compact disc read-Only Memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present application, however, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, with computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
It should be noted that although in the above detailed description several modules or units of a device for action execution are mentioned, such a division is not mandatory. Indeed, the features and functions of two or more modules or units described above may be embodied in one module or unit, in accordance with embodiments of the present application. Conversely, the features and functions of one module or unit described above may be further divided into a plurality of modules or units to be embodied.
From the above description of embodiments, those skilled in the art will readily appreciate that the example embodiments described herein may be implemented in software, or may be implemented in software in combination with the necessary hardware. Thus, the technical solution according to the embodiments of the present application may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (may be a CD-ROM, a usb disk, a mobile hard disk, etc.) or on a network, and includes several instructions to cause a computing device (may be a personal computer, a server, a touch terminal, or a network device, etc.) to perform the method according to the embodiments of the present application.
Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the application pertains.
It is to be understood that the present application is not limited to the precise arrangements and instrumentalities shown in the drawings, which have been described above, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the application is limited only by the appended claims.
Claims (10)
1. A task label generating method for a convolution network according to a relation chart is characterized by comprising the following steps:
generating user task graph data according to user data and task data, wherein the user task graph data comprises at least one user node and at least one task node, the user node corresponds to user information contained in the user data, and the task node corresponds to task information contained in the task data;
inputting the user task graph data into a relation graph convolution network, and obtaining at least one user characteristic and at least one task characteristic, which are output by the relation graph convolution network aiming at the user task graph data, wherein the user characteristic corresponds to a user node contained in the user task graph data, and the task characteristic corresponds to a task node contained in the user task graph data;
constructing a user task matrix according to the user characteristics and the task characteristics, wherein the user task matrix contains probability distribution of the user nodes on at least one task label, and the task labels are in one-to-one correspondence with the task nodes;
and generating a target task label of a user corresponding to the user node according to probability distribution of the user node on at least one task label contained in the user task matrix.
2. The method of claim 1, wherein prior to said generating user task graph data from user data and task data, the method further comprises:
generating historical user task graph data according to historical task data and historical user data, wherein the historical user data corresponds to a plurality of known task labels, and the known task labels correspond to the historical task data one by one;
and training the relation graph convolutional network according to the historical user task graph data and the known task labels.
3. The method of claim 1, wherein generating user task graph data from user data and task data comprises:
generating user graph data according to the user data, wherein the user graph data comprises the at least one user node, and generating task graph data according to the task data, and the task graph data comprises the at least one task node;
and combining the user graph data and the task graph data to obtain the user task graph data.
4. A method according to claim 3, wherein said generating user graph data from said user data comprises:
Acquiring user information contained in the user data and task record information corresponding to each piece of user information;
determining the same task completed among the users corresponding to the user information according to the task record information corresponding to the user information;
and generating user nodes in the user graph data according to the user information, and generating edges between the user nodes according to the same tasks completed between the users corresponding to the user information so as to obtain the user graph data.
5. A method according to claim 3, wherein said generating task graph data from said task data comprises:
acquiring task information contained in the task data and task history data corresponding to each piece of user information;
according to task history data respectively corresponding to the user information, determining the task information executed by the same user;
and generating task nodes in the task graph data according to the task information, and generating edges between the task nodes according to the task information executed by the same user to obtain the task graph data.
6. A method according to claim 3, wherein the merging the user task graph data and the task graph data to obtain the user task graph data includes:
acquiring task record information corresponding to user nodes in the user graph data and task history data corresponding to task nodes in the task graph data;
determining an association relation between the user node and the task node according to the task record information and the task history data, wherein the association relation indicates that a user corresponding to the user node finishes a task corresponding to the task node;
and constructing edges between the user nodes and the task nodes between the user graph data and the task graph data according to the association relation between the user nodes and the task nodes so as to obtain the user task graph data.
7. The method according to claim 1, wherein the generating the target task label of the user corresponding to the user node according to the probability distribution of the user node on at least one task label included in the user task matrix includes:
Determining the maximum probability corresponding to the user node according to the user task matrix;
and if the maximum probability is greater than a preset probability distribution threshold, taking the task label corresponding to the maximum probability as a target task label of the user corresponding to the user node.
8. A task tag generation device, comprising:
the system comprises a user task graph data module, a task data processing module and a task data processing module, wherein the user task graph data module is used for generating user task graph data according to user data and task data, the user task graph data comprises at least one user node and at least one task node, the user node corresponds to user information contained in the user data, and the task node corresponds to task information contained in the task data;
the characteristic representation output module is used for inputting the user task graph data into a relation graph convolution network, obtaining at least one user characteristic output by the relation graph convolution network aiming at the user task graph data and at least one task characteristic, wherein the user characteristic corresponds to a user node contained in the user task graph data, and the task characteristic corresponds to a task node contained in the user task graph data;
The user task matrix construction module is used for constructing a user task matrix according to the user characteristics and the task characteristics, wherein the user task matrix contains probability distribution of the user nodes on at least one task label, and the task labels are in one-to-one correspondence with the task nodes;
and the task label generating module is used for generating a target task label of a user corresponding to the user node according to probability distribution of the user node on at least one task label contained in the user task matrix.
9. A task tag generation device characterized by comprising:
a memory storing computer readable instructions;
a processor reading computer readable instructions stored in a memory to perform the method of any one of claims 1-7.
10. A computer readable storage medium having stored thereon computer readable instructions which, when executed by a processor of a computer, cause the computer to perform the method of any of claims 1-7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011342170.XA CN112464042B (en) | 2020-11-25 | 2020-11-25 | Task label generating method and related device for convolution network according to relationship graph |
| PCT/CN2021/084223 WO2021213156A1 (en) | 2020-11-25 | 2021-03-31 | Method and related apparatus for generating task label on basis of relationship graph convolutional network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011342170.XA CN112464042B (en) | 2020-11-25 | 2020-11-25 | Task label generating method and related device for convolution network according to relationship graph |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112464042A CN112464042A (en) | 2021-03-09 |
| CN112464042B true CN112464042B (en) | 2023-07-28 |
Family
ID=74808128
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011342170.XA Active CN112464042B (en) | 2020-11-25 | 2020-11-25 | Task label generating method and related device for convolution network according to relationship graph |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112464042B (en) |
| WO (1) | WO2021213156A1 (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112464042B (en) * | 2020-11-25 | 2023-07-28 | 平安科技（深圳）有限公司 | Task label generating method and related device for convolution network according to relationship graph |
| CN113887575B (en) * | 2021-09-13 | 2024-04-05 | 华南理工大学 | Graph dataset enhancement method based on self-adaptive graph rolling network |
| CN114510989B (en) * | 2021-12-23 | 2022-10-25 | 中国科学院软件研究所 | Normative evaluation method, device and equipment for image data set |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110363282A (en) * | 2019-06-06 | 2019-10-22 | 中国科学院信息工程研究所 | A kind of network node label Active Learning Method and system based on figure convolutional network |
| CN111199474A (en) * | 2019-12-16 | 2020-05-26 | 北京淇瑀信息科技有限公司 | Risk prediction method and device based on network diagram data of two parties and electronic equipment |
| CN111553215A (en) * | 2020-04-20 | 2020-08-18 | 深圳云天励飞技术有限公司 | Personnel association method and device, and graph convolution network training method and device |
| CN111932308A (en) * | 2020-08-13 | 2020-11-13 | 中国工商银行股份有限公司 | Data recommendation method, device and equipment |
| CN111931076A (en) * | 2020-09-22 | 2020-11-13 | 平安国际智慧城市科技股份有限公司 | Method and device for carrying out relationship recommendation based on authorized directed graph and computer equipment |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170200101A1 (en) * | 2016-01-07 | 2017-07-13 | Tata Consultancy Services Limited | Systems and methods for optimized task allocation |
| US10599686B1 (en) * | 2018-09-27 | 2020-03-24 | Babylon Partners Limited | Method and system for extracting information from graphs |
| CN111639696B (en) * | 2020-05-26 | 2023-08-01 | 北京邮电大学 | User classification method and device |
| CN112464042B (en) * | 2020-11-25 | 2023-07-28 | 平安科技（深圳）有限公司 | Task label generating method and related device for convolution network according to relationship graph |
-
2020
- 2020-11-25 CN CN202011342170.XA patent/CN112464042B/en active Active
-
2021
- 2021-03-31 WO PCT/CN2021/084223 patent/WO2021213156A1/en active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110363282A (en) * | 2019-06-06 | 2019-10-22 | 中国科学院信息工程研究所 | A kind of network node label Active Learning Method and system based on figure convolutional network |
| CN111199474A (en) * | 2019-12-16 | 2020-05-26 | 北京淇瑀信息科技有限公司 | Risk prediction method and device based on network diagram data of two parties and electronic equipment |
| CN111553215A (en) * | 2020-04-20 | 2020-08-18 | 深圳云天励飞技术有限公司 | Personnel association method and device, and graph convolution network training method and device |
| CN111932308A (en) * | 2020-08-13 | 2020-11-13 | 中国工商银行股份有限公司 | Data recommendation method, device and equipment |
| CN111931076A (en) * | 2020-09-22 | 2020-11-13 | 平安国际智慧城市科技股份有限公司 | Method and device for carrying out relationship recommendation based on authorized directed graph and computer equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021213156A1 (en) | 2021-10-28 |
| CN112464042A (en) | 2021-03-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112464042B (en) | Task label generating method and related device for convolution network according to relationship graph | |
| CN114048331A (en) | Knowledge graph recommendation method and system based on improved KGAT model | |
| CN113761153B (en) | Picture-based question-answering processing method and device, readable medium and electronic equipment | |
| US11373117B1 (en) | Artificial intelligence service for scalable classification using features of unlabeled data and class descriptors | |
| US11423307B2 (en) | Taxonomy construction via graph-based cross-domain knowledge transfer | |
| US10147103B2 (en) | System and method for a scalable recommender system using massively parallel processors | |
| CN111259647A (en) | Question and answer text matching method, device, medium and electronic equipment based on artificial intelligence | |
| CN111859872A (en) | Text labeling method and device | |
| JP2017527013A (en) | Adaptive characterization as a service | |
| CN112966701A (en) | Method and device for classifying objects | |
| CN111563158A (en) | Text sorting method, sorting device, server and computer-readable storage medium | |
| CN113761190A (en) | Text recognition method and device, computer readable medium and electronic equipment | |
| US20200210851A1 (en) | Augmenting neural networks | |
| CN112269875B (en) | Text classification method, device, electronic equipment and storage medium | |
| CN108509179B (en) | Method for detecting human face and device for generating model | |
| CN113448821B (en) | Method and device for identifying engineering defects | |
| Bielecki et al. | Estimation of execution time for computing tasks | |
| CN111667018A (en) | Object clustering method and device, computer readable medium and electronic equipment | |
| CN110851600A (en) | Text data processing method and device based on deep learning | |
| CN113010759B (en) | Cluster set processing method and device, computer readable medium and electronic equipment | |
| CN116168267A (en) | Training method, device, equipment and medium for multi-task neural network model | |
| CN114898426A (en) | Synonym label aggregation method, device, equipment and storage medium | |
| CN111274818B (en) | Word vector generation method and device | |
| CN111291196B (en) | Knowledge graph perfecting method and device, and data processing method and device | |
| CN112101015A (en) | Method and device for identifying multi-label object |
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 |