CN114329231B - Object feature processing method, device, electronic device and storage medium - Google Patents
Object feature processing method, device, electronic device and storage mediumInfo
- Publication number
- CN114329231B CN114329231B CN202111679571.9A CN202111679571A CN114329231B CN 114329231 B CN114329231 B CN 114329231B CN 202111679571 A CN202111679571 A CN 202111679571A CN 114329231 B CN114329231 B CN 114329231B
- Authority
- CN
- China
- Prior art keywords
- similar
- data
- target
- sample
- nodes
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
本公开提供了一种对象特征处理方法、装置、电子设备及存储介质，涉及数据处理领域，尤其涉及人工智能技术领域。具体实现方案为：获取目标对象的对象数据，其中，对象数据包括目标对象的行为数据和目标对象的属性数据，目标对象的活跃度低于预定对象范围内的平均活跃度；基于对象数据，预测目标对象的对象特征向量。
The present disclosure provides a method, apparatus, electronic device, and storage medium for processing object features, relating to the field of data processing, and more particularly, to the field of artificial intelligence technology. A specific implementation scheme comprises: obtaining object data of a target object, wherein the object data includes behavioral data and attribute data of the target object, and the target object's activity level is lower than the average activity level within a predetermined object range; and predicting an object feature vector of the target object based on the object data.
Description
Technical Field
The disclosure relates to the technical field of data processing, in particular to the technical field of artificial intelligence, and specifically relates to an object feature processing method, an object feature processing device, electronic equipment and a storage medium.
Background
In the related art, when recommending resources for a user object, a method is generally adopted in which an operation history (e.g., search, click) of the user on the resources is obtained, then a similar user is found through the vector representation of the user and the vector representations of other users, and then the preference of the current user is predicted according to the use condition of the resources of the similar user, so as to recommend the related resources. However, when the method is adopted to recommend the resources, the current user is required to have rich operation histories, otherwise, the vector representation corresponding to the user object is incomplete, so that the similar users found according to the vector representation are inaccurate, the accuracy of the resources recommended to the user is low, and the recommendation effect is poor.
Disclosure of Invention
The present disclosure provides a method, apparatus, device and storage medium for object feature processing.
According to one aspect of the disclosure, an object feature processing method is provided, which comprises the steps of obtaining object data of a target object, wherein the object data comprises behavior data of the target object and attribute data of the target object, the activity of the target object is lower than the average activity in a preset object range, and predicting an object feature vector of the target object based on the object data.
Optionally, the method further comprises the steps of determining a target similar object similar to the target object based on the object feature vector and the similar feature vector of the similar object, determining resource content to be pushed based on behavior data of the target similar object, and pushing the resource content to the target object.
Optionally, determining the target similar object similar to the target object based on the object feature vector and the similar feature vector of the similar object includes obtaining a euclidean distance between the object feature vector and the similar feature vector of the similar object, and selecting as the target similar object a similar object in which the euclidean distance is less than a predetermined distance threshold.
Optionally, predicting the object feature vector of the target object based on the object data comprises predicting the object feature vector of the target object based on the object data by adopting a meta-learning network model, wherein the meta-learning network model is obtained by training based on a plurality of groups of sample object data, the plurality of groups of sample object data comprise the object data of the sample object and the feature vector of the sample object, and the activity of the sample object is higher than the average activity.
Optionally, the method further comprises the steps of constructing an object resource graph, wherein the object resource graph is generated based on the walking route of a plurality of objects, the object resource graph comprises a plurality of object nodes and resource nodes which have one or more layers of association relation with the plurality of object nodes, head nodes and long tail nodes aiming at the objects are screened out from the object resource graph, the number of the neighbor nodes of the head nodes is larger than or equal to a preset threshold value, the number of the neighbor nodes of the long tail nodes is smaller than the preset threshold value, and the objects corresponding to part or all of the head nodes are used as sample objects.
Optionally, the object data of the sample object includes behavior data of the sample object and attribute data of the sample object.
According to another aspect of the disclosure, an object feature processing device is provided, which comprises an acquisition module, a prediction module and a prediction module, wherein the acquisition module is used for acquiring object data of a target object, the object data comprises behavior data of the target object and attribute data of the target object, the activity of the target object is lower than the average activity in a preset object range, and the prediction module is used for predicting an object feature vector of the target object based on the object data.
Optionally, the device further comprises a first determining module, a second determining module and a pushing module, wherein the first determining module is used for determining a target similar object similar to the target object based on the object feature vector and the similar feature vector of the similar object, the second determining module is used for determining resource content to be pushed based on behavior data of the target similar object, and the pushing module is used for pushing the resource content to the target object.
Optionally, the first determining module comprises an obtaining unit for obtaining the euclidean distance between the object feature vector and the similar feature vector of the similar object, and a selecting unit for selecting the similar object with the euclidean distance smaller than the preset distance threshold value as the target similar object.
Optionally, the prediction module comprises a prediction unit, and the prediction unit is used for predicting the object feature vector of the target object by adopting a meta-learning network model based on object data, wherein the meta-learning network model is obtained by training based on a plurality of groups of sample object data, the plurality of groups of sample object data comprise object data of a sample object and the feature vector of the sample object, and the activity of the sample object is higher than the average activity.
Optionally, the prediction module further comprises a construction unit, a screening unit and a processing unit, wherein the construction unit is used for constructing an object resource graph, the object resource graph is generated based on the walking route of a plurality of objects, the object resource graph comprises a plurality of object nodes and resource nodes which have one or more layers of association relation with the plurality of object nodes, the screening unit is used for screening head nodes and long tail nodes aiming at the objects from the object resource graph, the number of neighbor nodes of the head nodes is larger than or equal to a preset threshold value, the number of neighbor nodes of the long tail nodes is smaller than the preset threshold value, and the processing unit is used for taking the objects corresponding to part or all of the head nodes as sample objects.
Optionally, the object data of the sample object includes behavior data of the sample object and attribute data of the sample object.
According to yet another aspect of the present disclosure, there is provided an electronic device comprising at least one processor, and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform any one of the methods described above.
According to yet another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform any of the methods described above.
According to a further aspect of the present disclosure there is provided a computer program product comprising a computer program which, when executed by a processor, implements any of the methods described above.
It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification.
Drawings
The drawings are for a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:
FIG. 1 is a flow chart of an object feature processing method provided in accordance with an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of an object feature processing method provided in accordance with an alternative embodiment of the present disclosure;
FIG. 3 is a first-order neighbor schematic diagram of a head node u;
FIG. 4 is a schematic diagram of a regression model;
FIG. 5 is a block diagram of an object feature processing apparatus provided in accordance with an embodiment of the present disclosure;
fig. 6 is a block diagram of an electronic device for implementing an object feature processing method of an embodiment of the present disclosure.
Detailed Description
Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
Description of the terms
Meta-Learning (Meta-Learning), the ability to learn how to learn, rather than Learning a specific task. Through meta-learning, the algorithm has learning ability (i.e., a priori knowledge) to quickly learn with only a small amount of data in the face of a new task.
U2u, user to User, emphasizes that the User and User on the social platform establish a connection and association.
Based on collaborative filtering algorithm of users, user-based Collaborative Filtering, UCF for short, an algorithm for recommending based on a group of users with the same interests generally needs three steps, including collecting information which can represent the interests of the users, nearest neighbor searching and generating recommendation results.
Topk, the top k objects are arranged in a set of ranks.
Dropout, when training a large neural network, randomly makes some neurons not participate in the current training, so as to avoid the overfitting caused by excessive noise of model learning, namely, after the model is trained to a certain degree, the test error obtained on the training set is far greater than the error obtained on the test set.
Full Connectivity (FC) network architecture, a most basic neural network or deep neural network layer, with each node of the full connectivity layer being connected to all nodes of the previous layer.
In an embodiment of the present disclosure, there is provided an object feature processing method, and fig. 1 is a flowchart of the object feature processing method provided according to an embodiment of the present disclosure, as shown in fig. 1, where the flowchart includes the following steps:
step S102, obtaining object data of a target object, wherein the object data comprises behavior data of the target object and attribute data of the target object, and the activity of the target object is lower than the average activity in a preset object range;
step S104, based on the object data, predicting the object feature vector of the target object.
Through the steps, the behavior data and the attribute data of the target object are obtained, so that the characteristics of the low-liveness target object can be obtained more comprehensively. In addition, when the target object is predicted, the adopted object data not only comprises the behavior data of the object but also comprises the attribute data of the object, so that vector representation of the low-activity object (the object with activity lower than average activity in a preset object range) can be more comprehensive, the low-activity object can be further recommended more accurately, and the problems of inaccurate recommendation and poor recommendation effect of the low-activity object are solved.
As an alternative embodiment, after the prediction result of the object feature vector of the target object is obtained, various operations may be performed, for example, the following operations may be adopted, namely, determining a target similar object similar to the target object based on the object feature vector and the similar feature vector of the similar object, determining resource content to be pushed based on the behavior data of the target similar object, and pushing the resource content to the target object. By the method, the similarity of the feature vectors is utilized to obtain the high-activity users similar to the low-activity users, and then the high-activity users are recommended to the low-activity users according to the behavior data of the high-activity users, so that the accurate recommendation of the low-activity users can be realized, and the use interests of the users are improved.
As an alternative embodiment, when determining a target similar object similar to the target object based on the object feature vector and the similar feature vector of the similar object, various manners may be adopted, for example, a manner may be adopted in which a euclidean distance between the object feature vector and the similar feature vector of the similar object is obtained, and a similar object whose euclidean distance is smaller than a predetermined distance threshold value among the similar objects is selected as the target similar object. The Euclidean distance between the feature vectors is obtained, and the user with the Euclidean distance smaller than the preset threshold value is selected as a similar user of the target user, so that the similarity between the users can be quantitatively and accurately determined, and therefore, the high-activity user similar to the low-activity user can be accurately found, the low-activity user is recommended according to the behavior data of the similar high-activity user, the accuracy of recommendation is improved, and the recommending effect is optimized. It should be noted that, the use of euclidean distance to characterize the similarity between users is only an alternative embodiment, and other ways of representing the similarity between users, such as cosine distance values, may also be used in other embodiments of the present application.
As an alternative embodiment, when predicting the object feature vector of the target object based on the object data, a plurality of prediction modes may be adopted, for example, may be implemented based on the prediction mode of the artificial intelligent network model. For example, the object feature vector of the target object may be predicted by using a meta-learning network model based on object data, where the meta-learning network model is trained based on a plurality of sets of sample object data, and the plurality of sets of sample object data includes object data of a sample object and feature vectors of the sample object, and the liveness of the sample object is higher than the average liveness. Because the meta learning network model is trained by object data of sample objects with higher liveness than average liveness and feature vectors of the sample objects, the model can learn various richer features more efficiently, and the feature vectors of users with low liveness are predicted by using the model, so that the obtained result is more accurate.
As an alternative embodiment, when determining the sample object, for example, a method may be adopted for constructing an object resource graph, wherein the object resource graph is generated based on the walking route of a plurality of objects, the object resource graph comprises a plurality of object nodes and resource nodes with one or more layers of association relation with the plurality of object nodes, head nodes and long tail nodes aiming at the objects are screened out from the object resource graph, the number of neighbor nodes of the head nodes is larger than or equal to a preset threshold value, the number of neighbor nodes of the long tail nodes is smaller than a preset threshold value, and the objects corresponding to part or all of the head nodes are taken as the sample objects. By determining the sample object in the mode, the sample object can be ensured to be all the user with higher activity, more behavior data can be provided for the subsequent model learning process, so that the high efficiency and high accuracy of model learning are ensured, and the accuracy of the prediction result of the feature vector for the user with low activity is further improved. In addition, the richness of the features corresponding to each object can be comprehensively and systematically described in a mode of constructing the object resource diagram. And screening head nodes and long tail nodes aiming at the object based on the object resource graph, wherein the number of neighbor nodes of the head nodes is larger than or equal to a preset threshold value, and the number of neighbor nodes of the long tail nodes is smaller than the preset threshold value. Based on the above determination modes of the head node and the long tail node aiming at the object, the characteristics of the user can be determined quantitatively and standard to a certain extent, so that the sample for training the meta-learning network model is relatively accurate and efficient.
As an alternative embodiment, the object data of the sample object may include a variety of data, for example, behavior data of the sample object, such as click, collection, forwarding, comment, etc. of the resource by the user, and attribute data of the sample object, such as gender, age, education level, consumption level, etc. of the user. Through the data, interest preferences and attribute characteristics of the users can be reflected in multiple directions, so that the prediction model is trained efficiently and comprehensively by utilizing the data, the prediction result of the model is more accurate, high-activity users similar to low-activity users can be found conveniently, recommendation is completed, the accuracy of recommendation for the low-activity users is greatly improved, and the recommendation effect is also improved.
Based on the above embodiments and optional embodiments, an optional implementation is provided, and is described below.
Aiming at users with low liveness, whether the vector representation of the user is obtained or the user is characterized according to the population attribute characteristics of the user, the low liveness user is inaccurately characterized, and the method is not suitable for application scenes of a large number of users with low liveness, and further the problems of inaccurate recommendation and poor recommendation effect are caused.
In the related art user collaborative filtering recall method (for example, a novel is recommended to a user), a user click sequence is adopted or a vector representation of the user is constructed based on a wandering graph model (namely, the object resource graph) to calculate cosine distance as similarity of the user, similar users are obtained, and resource recall is performed according to a reading list of the similar users. However, the method has better effect on the representation of the User vector with rich behaviors, but has poorer effect on the representation of the low-activity User vector with sparse behaviors (offline evaluation u2u (User to User, u2u for short), and when UCF (User-based Collaborative Filtering, UCF for short) is directly adopted on line for recall, no obvious benefit is obtained in experiments.
In the implementation process of the method, based on collaborative filtering recall of users, a user history reading list is constructed through clicking preference of the users on articles or resources, user vector representation is obtained according to the user history reading list, and similarity among users is calculated through the user vectors. And finding topk the nearest neighbor users according to the similarity, predicting the favorite articles or resources of the current user according to the similarity weight of the adjacent users and the preference of the users for the articles or resources, and calculating to obtain an ordered article list for recommendation. This approach is the most basic user recall approach, which is only applicable to recommended scenes where users with a large number of rich behaviors exist, and is not applicable to the more vertical analogies of users with sparse behaviors.
For another example, for a sparse-behaved user, a user-item graph model (i.e., the above-mentioned object resource graph) is typically introduced, a vector representation of the user is obtained based on composition wander, and then the recommendation of the item or resource is performed using the above-mentioned collaborative filtering recall method based on the user. In the practice process, the method finds that for the sparse-behavior user, the vector representation is not accurate enough, so that the recall resource of the user is deviated, and the method is not suitable for the novel recommendation scene.
However, for sparse-behaved users, cluster recalls may also be made by user demographics, etc. But only adopt the population attribute characteristic of the user to carry out clustering recall, the user behavior is characterized by coarser, and a large number of sparse users can influence the clustering effect.
Based on the above, in an alternative embodiment of the present disclosure, an object feature processing scheme is provided. FIG. 2 is a flow chart of an object feature processing method provided in accordance with an alternative embodiment of the present disclosure, as shown in FIG. 2, the flow including the following processes:
(1) Constructing a user-resource isomorphic diagram (the same as the above-mentioned object resource diagram, taking a user as an object for example) by using the novel reading history, screening users and resources with high node degree as head nodes, and offline training a meta-learning network model, wherein the method can comprise the following steps:
constructing a basic model, namely taking a novel user reading history as a training sample, composing a wandering, and acquiring vector representations of users and resources by a word2vec algorithm to serve as the basic model;
Screening head nodes, namely constructing a user-resource isomorphic diagram based on the original sample, screening head nodes (the number of neighbors is more than 5 and comprises users and resources), and obtaining head node vectors produced by the model;
obtaining user population attribute characteristics, namely adopting knowledge distillation ideas to solve the vector representation problem of too sparse users, and introducing the user population attribute characteristics (gender, age, education degree and consumption level);
The construction of the meta-learning task, which is to randomly sample the first-order neighbors of the head node (the number of neighbor samples of each node is not more than 5), as shown in fig. 3, fig. 3 is a schematic diagram of the first-order neighbors of the head node u, and the construction of the meta-learning task includes constructing a test set by the node itself, constructing a training set by the neighbors of the node, and model learning how to characterize the knowledge of the node itself by the node neighbors (for example, it can be represented by a 3-layer Fully Connected (FC) network [512,128,32 ]. In the optional embodiment, the feature vector of the long-tail user is mainly optimized, so that only head nodes of the user are used in a training set and a testing set, and all neighbor nodes used in the training process are head nodes; summing the Euclidean distance between each node type output vector in the test set and the training set and the basic model vector, constructing a training sample as shown in fig. 4, wherein the training sample comprises 1-order neighbor node aggregate vectors (average pooling) with input features as nodes, splicing user population attribute features, the model fitting target is the vector produced by the model, the loss function can be a model, and fig. 4 is a schematic structural diagram of a meta-learning network model in an alternative embodiment of the disclosure;
The network parameter updating comprises the steps of iterating process element learning, constructing a sub-network based on element network parameters in each training, calculating loss for training nodes and returning gradient updated sub-network parameters, calculating test loss by using the updated sub-network, repeating the process k times (k=5), and finally optimizing the test loss by gradient and updating the element network parameters.
(2) The method for offline prediction of long-tail user node vector representation by using the meta-learning network model obtained by training in the step (1) can adopt the following steps:
screening long-tail user nodes, namely screening long-tail node users from the user-resource isomorphic diagram (for example, the total number of clicking strokes of the novel resource in 30 days is not more than 5);
the meta-test (meta-test) task is constructed by obtaining vector representation of long-tail user nodes, constructing the meta-test (meta-test) task in a similar mode to a training set, wherein the testing set is a long-tail user node to be predicted, the training set is a neighbor of the node, and the vector representation of the user nodes is finally obtained in a similar calculation process to training.
(3) An online build meta-collaboration (meta-ucf) recall path may be implemented as follows:
Combining the updated long tail node vector with a user vector produced by the original model, and selecting a user with click history of >5 as a core user to construct a u2u similarity matrix;
And the online newly added element is cooperated (metaUCF) to recall that each user recalls the most similar first fifty users, and the novel resources clicked by the similar users are recalled by adopting a voting algorithm.
The above-mentioned embodiments provide an effective low-activity user vector representation and user recommendation method for a vertical class scenario where a large number of low-activity users exist, and the method combines meta-learning and recommendation service scenarios, thereby proving the feasibility of transfer learning in the recommendation scenario.
In an embodiment of the present disclosure, there is further provided an object feature processing apparatus, and fig. 5 is a block diagram of the structure of the object feature processing apparatus provided according to an embodiment of the present disclosure, as shown in fig. 5, and the apparatus includes an acquisition module 51 and a prediction module 52, and the apparatus is described below.
The system comprises an acquisition module 51 for acquiring object data of a target object, wherein the object data comprises behavior data of the target object and attribute data of the target object, and the activity of the target object is lower than the average activity in a preset object range, and a prediction module 52 connected to the acquisition module 51 for predicting an object feature vector of the target object based on the object data.
As an alternative embodiment, the device further comprises a first determining module, a second determining module and a pushing module, and the device is described below.
The device comprises a prediction module, a first determination module, a second determination module and a pushing module, wherein the prediction module is used for predicting a target object, the first determination module is connected with the prediction module and is used for determining a target similar object similar to the target object based on the object feature vector and the similar feature vector of the similar object, the second determination module is connected with the first determination module and is used for determining resource content to be pushed based on behavior data of the target similar object, and the pushing module is connected with the second determination module and is used for pushing the resource content to the target object.
As an optional embodiment, the first determining module comprises an obtaining unit, a selecting unit and a determining unit, wherein the obtaining unit is used for obtaining Euclidean distances between the object feature vectors and similar feature vectors of similar objects, and the selecting unit is used for selecting similar objects with Euclidean distances smaller than a preset distance threshold value in the similar objects as target similar objects.
As an optional embodiment, the prediction module may include a prediction unit, configured to predict an object feature vector of a target object using a meta learning network model based on object data, where the meta learning network model is obtained by training based on a plurality of sets of sample object data, and the plurality of sets of sample object data includes object data of a sample object and feature vectors of the sample object, and activity of the sample object is higher than average activity.
As an alternative embodiment, the prediction module may further include a construction unit, a screening unit, and a processing unit, and the prediction module is described below.
The system comprises a construction unit, a screening unit and a processing unit, wherein the construction unit is used for constructing an object resource graph, the object resource graph is generated based on the walking route of a plurality of objects, the object resource graph comprises a plurality of object nodes and resource nodes which have one or more layers of association relation with the plurality of object nodes, the screening unit is connected with the construction unit and is used for screening head nodes and long tail nodes aiming at the objects from the object resource graph, the number of the neighbor nodes of the head nodes is larger than or equal to a preset threshold value, the number of the neighbor nodes of the long tail nodes is smaller than the preset threshold value, and the processing unit is connected with the screening unit and is used for taking the objects corresponding to part or all of the head nodes as sample objects.
As an alternative embodiment, the object data of the sample object comprises behavior data of the sample object and attribute data of the sample object.
In the technical scheme of the disclosure, the acquisition, storage, application and the like of the related user personal information all conform to the regulations of related laws and regulations, and the public sequence is not violated.
According to embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.
Fig. 6 illustrates a schematic block diagram of an example electronic device 600 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 6, the electronic device (or device 600) includes a computing unit 601 that can perform various suitable actions and processes according to a computer program stored in a Read Only Memory (ROM) 602 or a computer program loaded from a storage unit 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 may also be stored. The computing unit 601, ROM 602, and RAM 603 are connected to each other by a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.
Various components in the device 600 are connected to the I/O interface 605, including an input unit 606, e.g., keyboard, mouse, etc., an output unit 607, e.g., various types of displays, speakers, etc., a storage unit 608, e.g., magnetic disk, optical disk, etc., and a communication unit 609, e.g., network card, modem, wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.
The computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 601 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the respective methods and processes described above, for example, the object feature processing method. For example, in some embodiments, the object feature processing method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 600 via the ROM 602 and/or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the object feature processing method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the object feature processing method in any other suitable way (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include being implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be a special or general purpose programmable processor, operable to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on 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 (EPROM or 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.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user, for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback), and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a Local Area Network (LAN), a Wide Area Network (WAN), and the Internet.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel or sequentially or in a different order, provided that the desired results of the technical solutions of the present disclosure are achieved, and are not limited herein.
The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.
Claims (13)
1. An object feature processing method, comprising:
acquiring object data of a target object, wherein the object data comprises behavior data of the target object and attribute data of the target object, and the activity of the target object is lower than the average activity in a preset object range;
Predicting an object feature vector of the target object based on the object data;
The method comprises the steps of predicting the object feature vector of the target object based on the object data, and predicting the object feature vector of the target object by adopting a meta-learning network model based on the object data, wherein the meta-learning network model is trained based on a plurality of groups of sample object data, the plurality of groups of sample object data comprise object data of sample objects and feature vectors of the sample objects, and the activity of the sample objects is higher than the average activity.
2. The method of claim 1, wherein the method further comprises:
Determining a target similar object similar to the target object based on the object feature vector and the similar feature vector of the similar object;
Determining the resource content to be pushed based on the behavior data of the target similar object;
and pushing the resource content to the target object.
3. The method of claim 2, wherein the determining a target similar object that is similar to the target object based on the object feature vector and a similar feature vector of a similar object comprises:
acquiring Euclidean distance between the object feature vector and the similar feature vector of the similar object;
and selecting the similar object with the Euclidean distance smaller than a preset distance threshold value from the similar objects as the target similar object.
4. The method of claim 1, wherein the method further comprises:
constructing an object resource graph, wherein the object resource graph is generated based on the walking route of a plurality of objects, and the object resource graph comprises a plurality of object nodes and resource nodes with one or more layers of association relations with the plurality of object nodes;
Screening head nodes and long tail nodes aiming at an object from the object resource graph, wherein the number of neighbor nodes of the head nodes is larger than or equal to a preset threshold value, and the number of neighbor nodes of the long tail nodes is smaller than the preset threshold value;
And taking the objects corresponding to part or all of the head nodes as the sample objects.
5. The method according to any one of claims 1 to 4, wherein the object data of the sample object includes behavior data of the sample object and attribute data of the sample object.
6. An object feature processing apparatus comprising:
The system comprises an acquisition module, a storage module and a storage module, wherein the acquisition module is used for acquiring object data of a target object, the object data comprise behavior data of the target object and attribute data of the target object, and the activity of the target object is lower than the average activity in a preset object range;
a prediction module for predicting an object feature vector of the target object based on the object data;
The prediction module comprises a prediction unit and a prediction unit, wherein the prediction unit is used for predicting an object feature vector of the target object by adopting a meta-learning network model based on the object data, the meta-learning network model is trained based on a plurality of groups of sample object data, the plurality of groups of sample object data comprise object data of a sample object and the feature vector of the sample object, and the activity of the sample object is higher than the average activity.
7. The apparatus of claim 6, wherein the apparatus further comprises:
a first determining module, configured to determine a target similar object similar to the target object based on the object feature vector and a similar feature vector of the similar object;
The second determining module is used for determining the resource content to be pushed based on the behavior data of the target similar object;
and the pushing module is used for pushing the resource content to the target object.
8. The apparatus of claim 7, wherein the first determination module comprises:
an obtaining unit, configured to obtain a euclidean distance between the object feature vector and a similar feature vector of the similar object;
and the selecting unit is used for selecting the similar object with the Euclidean distance smaller than a preset distance threshold value in the similar objects as the target similar object.
9. The apparatus of claim 6, wherein the prediction module further comprises:
The construction unit is used for constructing an object resource graph, wherein the object resource graph is generated based on the walking route of a plurality of objects, and comprises a plurality of object nodes and resource nodes with one or more layers of association relations with the plurality of object nodes;
A screening unit, configured to screen a head node and a long tail node for an object from the object resource graph, where the number of neighbor nodes of the head node is greater than or equal to a predetermined threshold, and the number of neighbor nodes of the long tail node is less than the predetermined threshold;
And the processing unit is used for taking the objects corresponding to part or all of the head nodes as the sample objects.
10. The apparatus according to any one of claims 6 to 9, wherein the object data of the sample object includes behavior data of the sample object and attribute data of the sample object.
11. An electronic device, comprising:
at least one processor, and
A memory communicatively coupled to the at least one processor, wherein,
The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 5.
12. A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the method of any one of claims 1 to 5.
13. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111679571.9A CN114329231B (en) | 2021-12-31 | 2021-12-31 | Object feature processing method, device, electronic device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111679571.9A CN114329231B (en) | 2021-12-31 | 2021-12-31 | Object feature processing method, device, electronic device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114329231A CN114329231A (en) | 2022-04-12 |
| CN114329231B true CN114329231B (en) | 2025-09-12 |
Family
ID=81023756
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111679571.9A Active CN114329231B (en) | 2021-12-31 | 2021-12-31 | Object feature processing method, device, electronic device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114329231B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117132367B (en) * | 2023-10-20 | 2024-02-06 | 腾讯科技（深圳）有限公司 | Service processing method, device, computer equipment, storage medium and program product |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111931045A (en) * | 2020-07-30 | 2020-11-13 | 北京邮电大学 | Heterogeneous information network cold start recommendation method and device based on meta-learning |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106980999A (en) * | 2016-01-19 | 2017-07-25 | 阿里巴巴集团控股有限公司 | The method and apparatus that a kind of user recommends |
| US10803386B2 (en) * | 2018-02-09 | 2020-10-13 | Twitter, Inc. | Matching cross domain user affinity with co-embeddings |
| CN109635201B (en) * | 2018-12-18 | 2020-07-31 | 苏州大学 | Heterogeneous social network cross-platform associated user account mining method |
| CN109902753B (en) * | 2019-03-06 | 2023-01-13 | 深圳市珍爱捷云信息技术有限公司 | User recommendation model training method and device, computer equipment and storage medium |
| CN110162695B (en) * | 2019-04-09 | 2022-04-26 | 中国科学院深圳先进技术研究院 | Information pushing method and equipment |
| US20210110306A1 (en) * | 2019-10-14 | 2021-04-15 | Visa International Service Association | Meta-transfer learning via contextual invariants for cross-domain recommendation |
| CN111177575B (en) * | 2020-04-07 | 2020-07-24 | 腾讯科技（深圳）有限公司 | A content recommendation method, apparatus, electronic device and storage medium |
| CN111898031B (en) * | 2020-08-14 | 2024-04-05 | 腾讯科技（深圳）有限公司 | Method and device for obtaining user portrait |
| CN113342963B (en) * | 2021-04-29 | 2022-03-04 | 山东大学 | Service recommendation method and system based on transfer learning |
| CN113449205B (en) * | 2021-08-30 | 2021-11-09 | 四川省人工智能研究院(宜宾) | Recommendation method and system based on metadata enhancement |
-
2021
- 2021-12-31 CN CN202111679571.9A patent/CN114329231B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111931045A (en) * | 2020-07-30 | 2020-11-13 | 北京邮电大学 | Heterogeneous information network cold start recommendation method and device based on meta-learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114329231A (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112765477B (en) | Information processing method and device, information recommendation method and device, electronic equipment and storage medium | |
| CN111651671B (en) | User object recommendation method, device, computer equipment and storage medium | |
| CN112989146B (en) | Method, apparatus, device, medium and program product for recommending resources to target user | |
| CN111814056A (en) | Supplier recommendation method based on information processing and related equipment | |
| CN112836125A (en) | A recommendation method and system based on knowledge graph and graph convolutional network | |
| CN115470401B (en) | Resource recommendation method, device, electronic device and storage medium | |
| CN112000763A (en) | Method, device, equipment and medium for determining competition relationship of interest points | |
| CN117251626B (en) | Course recommendation method, system and electronic device | |
| CN113254795B (en) | Training method and device for recommendation model | |
| CN118071400A (en) | Application method and system based on graph computing technology in information consumption field | |
| JP2023025147A (en) | Method and apparatus for training label recommendation model and method and apparatus for acquiring label | |
| CN112905885A (en) | Method, apparatus, device, medium, and program product for recommending resources to a user | |
| CN115456707A (en) | Method, device and electronic device for providing product recommendation information | |
| CN113407808B (en) | Method, device and computer equipment for determining the applicability of graph neural network models | |
| CN115510327B (en) | Training method, resource recommendation method and device for click-through rate prediction model | |
| CN114547416B (en) | A media resource sorting method and electronic device | |
| CN114329231B (en) | Object feature processing method, device, electronic device and storage medium | |
| CN115062779A (en) | Event prediction method and device based on dynamic knowledge graph | |
| CN119211648A (en) | Resource recommendation method, deep learning model training method, device and equipment | |
| CN112818228A (en) | Method, device, equipment and medium for recommending object to user | |
| CN117668351B (en) | Recommendation method, model training method, device, electronic device and storage medium | |
| WO2025043793A1 (en) | Recommendation result optimization method and apparatus for recommendation model, and device and medium | |
| CN114880599B (en) | Information ordering method and device, electronic equipment and storage medium | |
| CN116562982A (en) | Offline article recommendation method, device, equipment and storage medium | |
| CN112288195B (en) | Method, device and electronic device for predicting user behavior of center based on friend behavior |
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 |