CN118861681A - Product recommendation model training method, product recommendation method and device - Google Patents

Abstract

The invention discloses a training method of a product recommendation model, a product recommendation method, a device, electronic equipment and a storage medium. The training method of the product recommendation model comprises the following steps: acquiring user characteristics, product characteristics and cross behavior characteristics; determining a user feature vector and a product feature vector based on the user feature, the product feature, and the cross-behavior feature; and determining a loss value of a loss function of the product recommendation model based on the user feature vector and the product feature vector, and adjusting model parameters based on the loss value of the loss function of the product recommendation model until a training stop condition is met, so as to obtain a trained product recommendation model. According to the technical scheme, the user feature vector and the product feature vector with high dimensions are generated through the user feature, the product feature and the cross behavior feature, and further product recommendation model training is carried out based on the user feature vector and the product feature vector, so that the accuracy of the product recommendation model is effectively improved.

Description

Product recommendation model training method, product recommendation method and device

Technical Field

The present invention relates to the field of artificial intelligence technology, and in particular to a product recommendation model training method, a product recommendation method, a device, an electronic device and a storage medium.

Background Art

With the development of artificial intelligence technology, machine learning is being used more and more widely in the field of product recommendations.

In the process of implementing the present invention, it is found that there are at least the following technical problems in the prior art: the training scheme of the existing product recommendation model has the problem of low model prediction accuracy.

Summary of the invention

The present invention provides a product recommendation model training method, a product recommendation method, a device, an electronic device and a storage medium to improve the accuracy of the product recommendation model.

According to one aspect of the present invention, a method for training a product recommendation model is provided, comprising:

Obtain user characteristics, product characteristics, and cross-behavior characteristics;

Determine a user feature vector and a product feature vector based on the user feature, the product feature and the cross-behavior feature;

The loss value of the loss function of the product recommendation model is determined based on the user feature vector and the product feature vector, and the model parameters are adjusted based on the loss value of the loss function of the product recommendation model until a training stop condition is met, thereby obtaining a trained product recommendation model.

According to another aspect of the present invention, there is provided a product recommendation method, comprising:

Obtain user characteristics of target users, product characteristics of target users, and cross-behavior characteristics of target users;

Input the user characteristics of the target user, the product characteristics of the target user, and the cross-behavior characteristics of the target user into the product recommendation model to obtain the user characteristic vector of the target user and the product characteristic vector of the target user;

A product recommendation result is determined based on the user feature vector of the target user and the product feature vector of the target user, wherein the product recommendation model is the product recommendation model described in any embodiment of the present invention.

According to another aspect of the present invention, there is provided a training device for a product recommendation model, comprising:

Feature data acquisition module, used to acquire user features, product features and cross-behavior features;

A feature vector determination module, used to determine a user feature vector and a product feature vector based on the user feature, the product feature and the cross-behavior feature;

The product recommendation model training module is used to determine the loss value of the loss function of the product recommendation model based on the user feature vector and the product feature vector, and adjust the model parameters based on the loss value of the loss function of the product recommendation model until the training stop condition is met, thereby obtaining a trained product recommendation model.

According to another aspect of the present invention, there is provided a product recommendation device, comprising:

A target user feature acquisition module is used to acquire user features of the target user, product features of the target user, and cross-behavior features of the target user;

A feature vector prediction module is used to input the user features of the target user, the product features of the target user, and the cross-behavior features of the target user into the product recommendation model to obtain the user feature vector of the target user and the product feature vector of the target user;

The product recommendation result determination module is used to determine the product recommendation result based on the user feature vector of the target user and the product feature vector of the target user, wherein the product recommendation model is the product recommendation model described in any embodiment of the present invention.

According to another aspect of the present invention, there is provided an electronic device, the electronic device comprising:

at least one processor;

and a memory communicatively coupled to the at least one processor;

In which, the memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the product recommendation model described in any embodiment of the present invention or the product recommendation method described in any embodiment of the present invention.

According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the training method of the product recommendation model described in any embodiment of the present invention or the product recommendation method described in any embodiment of the present invention when executed.

The technical solution of the embodiment of the present invention obtains user features, product features and cross-behavior features, and then determines the user feature vector and product feature vector based on the user features, product features and cross-behavior features, and then determines the loss value of the loss function of the product recommendation model based on the user feature vector and product feature vector, and adjusts the model parameters based on the loss value of the loss function of the product recommendation model until the training stop condition is met, thereby obtaining a trained product recommendation model. The above technical solution generates high-dimensional user feature vectors and product feature vectors through user features, product features and cross-behavior features, and then trains the product recommendation model based on the user feature vector and product feature vector, which effectively improves the accuracy of the product recommendation model.

It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

FIG1 is a flow chart of a method for training a product recommendation model according to a first embodiment of the present invention;

FIG2 is a flow chart of a method for training a product recommendation model according to a second embodiment of the present invention;

FIG3 is a flow chart of a method for training a product recommendation model according to Embodiment 3 of the present invention;

FIG4 is a flow chart of a method for training a product recommendation model according to an embodiment of the present invention;

FIG5 is a flow chart of a product recommendation method provided according to a fourth embodiment of the present invention;

FIG6 is a schematic diagram of the structure of a training device for a product recommendation model according to Embodiment 5 of the present invention;

FIG7 is a schematic diagram of the structure of a product recommendation device provided according to Embodiment 6 of the present invention;

FIG8 is a schematic diagram of the structure of an electronic device for implementing the product recommendation model training method or product recommendation method according to an embodiment of the present invention.

DETAILED DESCRIPTION

In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices. The acquisition, storage, use, processing, etc. of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

Embodiment 1

FIG1 is a flow chart of a method for training a product recommendation model provided in the first embodiment of the present invention. This embodiment is applicable to product recommendations in multiple scenarios, such as newspaper main interface recommendations or newspaper drop-down interface recommendations. The method can be executed by a training device for a product recommendation model. The training device for a product recommendation model can be implemented in the form of hardware and/or software. The training device for a product recommendation model can be configured in a terminal and/or a server. As shown in FIG1 , the method includes:

S110, obtaining user characteristics, product characteristics, and cross-behavior characteristics.

In this embodiment, user features refer to basic features of users. For example, in the newspaper recommendation scenario, user features may include but are not limited to information such as age, gender, and occupation of newspaper subscribers. Specifically, multiple basic tags of users may be obtained based on multi-dimensional portrait information of users, and user features may be generated based on the association between the basic tags.

Product features refer to the basic features of users browsing and/or purchasing products. For example, in the newspaper recommendation scenario, product features may include but are not limited to information such as newspaper type, newspaper price, and newspaper store. Specifically, the behavior data of users browsing and/or purchasing products can be mined and analyzed to obtain product features.

Cross-behavior features refer to the cross-behavior features of users and products. For example, in the newspaper and periodical scenario, cross-behavior features may include the amount of newspapers and periodicals browsed by users, the newspaper and periodical categories most frequently browsed by users, and the time intervals for browsing each newspaper and periodical category. Specifically, the behavior features and preference features of user product browsing behavior and product order behavior data can be mined to obtain cross-behavior features.

S120: Determine a user feature vector and a product feature vector based on the user feature, the product feature, and the cross-behavior feature.

Among them, user feature vector refers to the vectorized representation of user features, and product feature vector refers to the vectorized representation of product features.

Specifically, a user feature vector may be generated based on user features, or based on user features and cross-behavior features; a product feature vector may be generated based on product features, or based on product features and cross-behavior features, without specific limitation herein.

S130. Determine a loss value of a loss function of the product recommendation model based on the user feature vector and the product feature vector, and adjust model parameters based on the loss value of the loss function of the product recommendation model until a training stop condition is met, thereby obtaining a trained product recommendation model.

Exemplarily, the loss function can be Sample Softmax Loss or other loss functions, which are not specifically limited here. It should be noted that Sample Softmax Loss can optimize the traditional Softmax function while ensuring the cross entropy loss function, thereby saving the training time of the product recommendation model. Specifically, the user feature vector and the product feature vector can be substituted into the Sample Softmax Loss to obtain the loss value, and then the Adam optimizer can be used in the product recommendation model training process to gradually optimize the various network parameters based on the loss value, and gradually optimize the model indicators such as ACC, AUC and F1 by adjusting the hyperparameters until the training stop conditions are met to obtain the trained product recommendation model.

The technical solution of the embodiment of the present invention obtains user features, product features and cross-behavior features, and then determines the user feature vector and product feature vector based on the user features, product features and cross-behavior features, and then determines the loss value of the loss function of the product recommendation model based on the user feature vector and product feature vector, and adjusts the model parameters based on the loss value of the loss function of the product recommendation model until the training stop condition is met, thereby obtaining a trained product recommendation model. The above technical solution generates high-dimensional user feature vectors and product feature vectors through user features, product features and cross-behavior features, and then trains the product recommendation model based on the user feature vector and product feature vector, which effectively improves the accuracy of the product recommendation model.

Embodiment 2

Figure 2 is a flow chart of a method for training a product recommendation model provided in Embodiment 2 of the present invention. The method of this embodiment can be combined with the various optional schemes in the method for training a product recommendation model provided in the above embodiments. The training method for the product recommendation model provided in this embodiment is further optimized. Optionally, the determining of a user feature vector and a product feature vector based on the user features, the product features and the cross-behavior features includes: determining a user feature vector based on the user features and the cross-behavior features; determining a product feature vector based on the product features.

As shown in FIG. 2 , the method includes:

S210, obtaining user characteristics, product characteristics, and cross-behavior characteristics.

S220: Determine a user feature vector based on the user feature and the cross-behavior feature.

S230: Determine a product feature vector based on the product features.

S240. Determine a loss value of a loss function of the product recommendation model based on the user feature vector and the product feature vector, and adjust model parameters based on the loss value of the loss function of the product recommendation model until a training stop condition is met, thereby obtaining a trained product recommendation model.

In this embodiment, the user feature vector and the product feature vector can be generated based on two independent neural networks, that is, the user features and cross-behavior features are input into one neural network to obtain the user feature vector, and the product features are input into another neural network to obtain the product feature vector.

Specifically, determining a user feature vector based on user features and cross-behavior features includes: inputting the user features and cross-behavior features into a first feature vector layer to obtain a first user feature; inputting the first user feature into a first input layer to obtain a second user feature; inputting the second user feature into a first hidden layer to obtain a third user feature; and inputting the third user feature into a first output layer to obtain a user feature vector.

Among them, the first feature vector layer can be a feature embedding layer, and the first hidden layer can include two hidden layers or other number of hidden layers, which are not specifically limited here. Each neuron in the hidden layer is connected to the neurons in the previous layer, and the activation function of the hidden layer can adopt the SELU function or other activation functions, which are not specifically limited here. It should be noted that the SELU function has good convergence, can effectively suppress overfitting, has good training performance, can converge quickly and effectively suppress gradient explosion.

Exemplarily, the user features and cross-behavior features may be inputted into an independent neural network together. The independent neural network includes a feature vector layer, an input layer, a hidden layer, and an output layer connected in sequence. The output layer finally outputs a user feature vector.

Specifically, determining a product feature vector based on product features includes: inputting the product feature into the second feature vector layer to obtain a first product feature; inputting the first product feature into the second input layer to obtain a second product feature; inputting the second product feature into the second hidden layer to obtain a third product feature; and inputting the third product feature into the second output layer to obtain a product feature vector.

The second feature vector layer may be a feature embedding layer, and the second hidden layer may include two hidden layers or other numbers of hidden layers, which are not specifically limited herein.

Exemplarily, the product features may be input into another independent neural network, which includes a feature vector layer, an input layer, a hidden layer, and an output layer connected in sequence, and the output layer finally outputs a product feature vector.

The technical solution of the embodiment of the present invention realizes vectorized representation of users and products respectively by determining user feature vectors based on user features and cross-behavior features and determining product feature vectors based on product features.

Embodiment 3

Figure 3 is a flow chart of a method for training a product recommendation model provided in the third embodiment of the present invention. The method of this embodiment can be combined with the various optional schemes in the method for training a product recommendation model provided in the above embodiments. The training method for the product recommendation model provided in this embodiment has been further optimized. Optionally, before determining the user feature vector and the product feature vector based on the user feature, the product feature and the cross-behavior feature, it also includes: obtaining a recommendation scene feature; inputting the recommendation scene feature into the third input layer to obtain a first scene feature; inputting the first scene feature into the third hidden layer to obtain a second scene feature; inputting the second scene feature into the third output layer to obtain a scene feature vector; accordingly, determining the loss value of the loss function of the product recommendation model based on the user feature vector and the product feature vector includes: determining the loss value of the loss function of the product recommendation model based on the user feature vector, the product feature vector and the scene feature vector.

As shown in FIG3 , the method includes:

S310, obtaining user characteristics, product characteristics, and cross-behavior characteristics.

S320, obtaining recommended scene features; inputting the recommended scene features into the third input layer to obtain first scene features; inputting the first scene features into the third hidden layer to obtain second scene features; inputting the second scene features into the third output layer to obtain a scene feature vector.

S330: Determine a user feature vector and a product feature vector based on the user feature, the product feature, and the cross-behavior feature.

S340. Determine the loss value of the loss function of the product recommendation model based on the user feature vector, the product feature vector and the scene feature vector, and adjust the model parameters based on the loss value of the loss function of the product recommendation model until the training stop condition is met, thereby obtaining a trained product recommendation model.

In this embodiment, the recommendation scenario feature refers to feature information related to the business scenario. For example, the recommendation scenario feature can be a main interface feature in a newspaper recommendation scenario or a drop-down interface feature in a newspaper recommendation scenario. The scenario feature vector is a vectorized representation of the recommendation scenario feature.

It should be noted that by introducing recommendation scenario features, the product recommendation model can meet the needs of multiple business scenarios, thereby providing personalized product recommendations for different business scenarios, avoiding resource waste and duplication of work, and improving the reusability and versatility of the product recommendation model.

In some optional embodiments, before determining the user feature vector and the product feature vector based on the user features, product features and cross-behavior features, it also includes: obtaining a product behavior sequence; inputting the product behavior sequence into a fourth input layer to obtain a first product behavior feature; inputting the first product behavior feature into a fourth hidden layer to obtain a second product behavior feature; and inputting the second product behavior feature into a fourth output layer to obtain a product feature.

The product behavior sequence refers to a continuous behavior sequence of multiple products obtained according to the user behavior trajectory. For example, the product behavior sequence may include but is not limited to the first product information browsed at the first moment, the second product information browsed at the second moment, and the third product information browsed at the third moment.

It should be noted that by vectorizing the product behavior sequence and using the product features after the vectorization of the product behavior sequence as the input of the product recommendation model and transmitting it to the independent neural network on the product side, the training and fitting of the product recommendation model can be strengthened and the vectorization results of the product recommendation model can be optimized.

Exemplarily, FIG4 is a flow chart of a training method for a product recommendation model provided according to an embodiment of the present invention. Specifically, user features, product features, cross-behavior features, and recommendation scene features are obtained. Then, the user features and cross-behavior features can be input into the feature embedding layer, input layer, two hidden layers, and output layer connected in sequence to obtain a user feature vector; then, the product features can be input into the feature embedding layer, input layer, two hidden layers, and output layer connected in sequence to obtain a product feature vector, wherein the number of neurons in the input layer, two hidden layers, and output layer are 256, 128, 64, and 32, respectively. In addition, the recommendation scene features can be input into the input layer, hidden layer, and output layer connected in sequence to obtain a scene feature vector. Then, based on the user feature vector, the product feature vector, the scene feature vector, and the labels corresponding to each vector, the loss value of the Sample Softmax Loss of the product recommendation model is determined, and the model parameters are adjusted based on the loss value of the Sample Softmax Loss until the training stop condition is met to obtain a trained product recommendation model. In this embodiment, the behavior data of users browsing and/or purchasing products can be mined and analyzed to obtain product features, and the product behavior sequence can also be input into the input layer, hidden layer and output layer connected in sequence to obtain product features.

The technical solution of the embodiment of the present invention introduces recommendation scenario features to enable the product recommendation model to meet the needs of multiple business scenarios, thereby providing personalized product recommendations for different business scenarios, avoiding resource waste and duplication of work, and improving the reusability and versatility of the product recommendation model.

Embodiment 4

FIG5 is a flowchart of a product recommendation method provided by Embodiment 4 of the present invention. This embodiment is applicable to product recommendation in multiple scenarios, such as main interface recommendation or pull-down interface recommendation. The method can be executed by a product recommendation device, which can be implemented in the form of hardware and/or software, and can be configured in a terminal and/or a server. As shown in FIG5 , the method includes:

S410: Obtain user characteristics of the target user, product characteristics of the target user, and cross-behavior characteristics of the target user.

S420: Input the user characteristics of the target user, the product characteristics of the target user, and the cross-behavior characteristics of the target user into a product recommendation model to obtain a user characteristic vector of the target user and a product characteristic vector of the target user.

S430. Determine a product recommendation result based on the user feature vector of the target user and the product feature vector of the target user, wherein the product recommendation model is the product recommendation model described in any embodiment.

In this embodiment, the target user may be any user to whom product recommendation is to be made. The number of product feature vectors of the target user may be multiple, that is, product feature vectors corresponding to multiple products that the target user is interested in may be obtained.

Exemplarily, the user features, product features, and cross-behavior features of the target user can be input into the product recommendation model to obtain the user feature vector of the target user and multiple product feature vectors of the target user. Further, the similarity between the user feature vector of the target user and each product feature vector of the target user is calculated, and the products are sorted based on the similarity to obtain a set of product vectors that the target user is interested in. Furthermore, the products in the product vector set can be re-sorted according to the product sales volume or product views in the business scenario, and truncated according to the number of recommended products required by the business scenario to obtain the final product recommendation result. In this embodiment, the above-mentioned similarity calculation or search can be performed based on the constructed index. Specifically, an index can be constructed based on Faiss to improve retrieval efficiency, wherein the number of nearest neighbor TOPK can be set to 100, the search accuracy can be set to 0.8, and the retrieval method can be IVFx Flat.

The technical solution of the embodiment of the present invention obtains the user feature vector of the target user and the product feature vector of the target user by inputting the user features of the target user, the product features of the target user and the cross-behavior features of the target user into the product recommendation model, and then determines the product recommendation result based on the user feature vector of the target user and the product feature vector of the target user, thereby realizing personalized product recommendation for the target user. In addition, the product recommendation model can provide personalized product recommendation results for different business scenarios, while ensuring high-quality recommendation services, avoiding the situation of developing one model for one scenario, saving human resources, hardware resources and software resources, and ensuring commonalities and differences under multi-scenario recommendations.

Embodiment 5

FIG6 is a schematic diagram of the structure of a training device for a product recommendation model provided in Embodiment 5 of the present invention. As shown in FIG6 , the device includes:

A feature data acquisition module 510 is used to acquire user features, product features, and cross-behavior features;

A feature vector determination module 520, configured to determine a user feature vector and a product feature vector based on the user feature, the product feature and the cross-behavior feature;

The product recommendation model training module 530 is used to determine the loss value of the loss function of the product recommendation model based on the user feature vector and the product feature vector, and adjust the model parameters based on the loss value of the loss function of the product recommendation model until the training stop condition is met, thereby obtaining a trained product recommendation model.

The technical solution of the embodiment of the present invention obtains user features, product features and cross-behavior features, and then determines the user feature vector and product feature vector based on the user features, product features and cross-behavior features, and then determines the loss value of the loss function of the product recommendation model based on the user feature vector and product feature vector, and adjusts the model parameters based on the loss value of the loss function of the product recommendation model until the training stop condition is met, thereby obtaining a trained product recommendation model. The above technical solution generates high-dimensional user feature vectors and product feature vectors through user features, product features and cross-behavior features, and then trains the product recommendation model based on the user feature vector and product feature vector, which effectively improves the accuracy of the product recommendation model.

In some optional implementations, the feature vector determination module 520 includes:

A user feature vector determining unit, configured to determine a user feature vector based on the user feature and the cross-behavior feature;

The product feature vector determining unit is used to determine the product feature vector based on the product features.

In some optional implementations, the user feature vector determination unit may also be specifically configured to:

Inputting the user feature and the cross-behavior feature into a first feature vector layer to obtain a first user feature;

Inputting the first user feature into a first input layer to obtain a second user feature;

Inputting the second user feature into the first hidden layer to obtain a third user feature;

The third user feature is input into the first output layer to obtain a user feature vector.

In some optional implementations, the product feature vector determination unit may also be specifically configured to:

Inputting the product feature into the second feature vector layer to obtain the first product feature;

Inputting the first product feature into a second input layer to obtain a second product feature;

Inputting the second product feature into the second hidden layer to obtain a third product feature;

The third product feature is input into the second output layer to obtain a product feature vector.

In some optional implementations, the training device for the recommendation model further includes:

A scene feature vector determination module is used to obtain a recommended scene feature; input the recommended scene feature into a third input layer to obtain a first scene feature; input the first scene feature into a third hidden layer to obtain a second scene feature; input the second scene feature into a third output layer to obtain a scene feature vector;

Accordingly, the product recommendation model training module 530 includes:

A loss value determining unit is used to determine the loss value of the loss function of the product recommendation model based on the user feature vector, the product feature vector and the scene feature vector.

In some optional implementations, the training device for the recommendation model further includes:

The product feature acquisition unit is used to acquire a product behavior sequence; input the product behavior sequence into the fourth input layer to obtain a first product behavior feature; input the first product behavior feature into the fourth hidden layer to obtain a second product behavior feature; input the second product behavior feature into the fourth output layer to obtain a product feature.

The product recommendation model training device provided in the embodiment of the present invention can execute the product recommendation model training method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

Embodiment 6

FIG7 is a schematic diagram of the structure of a product recommendation device provided in Embodiment 6 of the present invention. As shown in FIG7 , the device includes:

A target user feature acquisition module 610 is used to acquire user features of the target user, product features of the target user, and cross-behavior features of the target user;

A feature vector prediction module 620 is used to input the user features of the target user, the product features of the target user, and the cross-behavior features of the target user into the product recommendation model to obtain the user feature vector of the target user and the product feature vector of the target user;

The product recommendation result determination module 630 is used to determine the product recommendation result based on the user feature vector of the target user and the product feature vector of the target user, wherein the product recommendation model is the product recommendation model described in any embodiment of the present invention.

The technical solution of the embodiment of the present invention obtains the user feature vector of the target user and the product feature vector of the target user by inputting the user features of the target user, the product features of the target user and the cross-behavior features of the target user into the product recommendation model, and then determines the product recommendation result based on the user feature vector of the target user and the product feature vector of the target user, thereby realizing personalized product recommendation for the target user. In addition, the product recommendation model can provide personalized product recommendation results for different business scenarios, while ensuring high-quality recommendation services, avoiding the situation of developing one model for one scenario, saving human resources, hardware resources and software resources, and ensuring commonalities and differences under multi-scenario recommendations.

Embodiment 7

FIG8 shows a block diagram of an electronic device 10 that can be used to implement an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and/or required herein.

As shown in FIG8 , the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other via a bus 14. The I/O interface 15 is also connected to the bus 14.

A number of components in the electronic device 10 are connected to the I/O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks.

The processor 11 may be a variety of general and/or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as a training method for a product recommendation model, which includes:

Obtain user characteristics, product characteristics, and cross-behavior characteristics;

Determine a user feature vector and a product feature vector based on the user feature, the product feature and the cross-behavior feature;

The loss value of the loss function of the product recommendation model is determined based on the user feature vector and the product feature vector, and the model parameters are adjusted based on the loss value of the loss function of the product recommendation model until a training stop condition is met, thereby obtaining a trained product recommendation model.

In some embodiments, the training method of the product recommendation model may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and/or installed on the electronic device 10 via the ROM 12 and/or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the training method of the product recommendation model described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the training method of the product recommendation model in any other appropriate manner (e.g., by means of firmware).

Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and/or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions/operations specified in the flow chart and/or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device 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 trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the 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, voice input, or tactile input).

The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may 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), a blockchain network, and the Internet.

A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A method for training a product recommendation model, comprising: Obtain user characteristics, product characteristics, and cross-behavior characteristics; Determine a user feature vector and a product feature vector based on the user feature, the product feature and the cross-behavior feature; The loss value of the loss function of the product recommendation model is determined based on the user feature vector and the product feature vector, and the model parameters are adjusted based on the loss value of the loss function of the product recommendation model until a training stop condition is met, thereby obtaining a trained product recommendation model. 2. The method according to claim 1, characterized in that the determining of the user feature vector and the product feature vector based on the user feature, the product feature and the cross-behavior feature comprises: Determine a user feature vector based on the user feature and the cross-behavior feature; A product feature vector is determined based on the product features. 3. The method according to claim 2, characterized in that the determining of the user feature vector based on the user feature and the cross-behavior feature comprises: Inputting the user feature and the cross-behavior feature into a first feature vector layer to obtain a first user feature; Inputting the first user feature into a first input layer to obtain a second user feature; Inputting the second user feature into the first hidden layer to obtain a third user feature; The third user feature is input into the first output layer to obtain a user feature vector. 4. The method according to claim 2, characterized in that the determining of the product feature vector based on the product feature comprises: Inputting the product feature into the second feature vector layer to obtain the first product feature; Inputting the first product feature into a second input layer to obtain a second product feature; Inputting the second product feature into the second hidden layer to obtain a third product feature; The third product feature is input into the second output layer to obtain a product feature vector. 5. The method according to claim 1, characterized in that before determining the user feature vector and the product feature vector based on the user feature, the product feature and the cross-behavior feature, it also includes: Obtain recommended scene features; Inputting the recommended scene feature into the third input layer to obtain a first scene feature; Inputting the first scene feature into a third hidden layer to obtain a second scene feature; Inputting the second scene feature into the third output layer to obtain a scene feature vector; Accordingly, the determining the loss value of the loss function of the product recommendation model based on the user feature vector and the product feature vector includes: A loss value of a loss function of the product recommendation model is determined based on the user feature vector, the product feature vector, and the scene feature vector. 6. The method according to claim 1, characterized in that before determining the user feature vector and the product feature vector based on the user feature, the product feature and the cross-behavior feature, it also includes: Get product behavior sequence; Inputting the product behavior sequence into the fourth input layer to obtain a first product behavior feature; Inputting the first product behavior feature into the fourth hidden layer to obtain a second product behavior feature; The second product behavior feature is input into the fourth output layer to obtain product features. 7. A product recommendation method, comprising: Obtain user characteristics of target users, product characteristics of target users, and cross-behavior characteristics of target users; Input the user characteristics of the target user, the product characteristics of the target user, and the cross-behavior characteristics of the target user into the product recommendation model to obtain the user characteristic vector of the target user and the product characteristic vector of the target user; The product recommendation result is determined based on the user feature vector of the target user and the product feature vector of the target user, wherein the product recommendation model is the product recommendation model described in any one of claims 1 to 6. 8. A training device for a product recommendation model, comprising: Feature data acquisition module, used to acquire user features, product features and cross-behavior features; A feature vector determination module, used to determine a user feature vector and a product feature vector based on the user feature, the product feature and the cross-behavior feature; The product recommendation model training module is used to determine the loss value of the loss function of the product recommendation model based on the user feature vector and the product feature vector, and adjust the model parameters based on the loss value of the loss function of the product recommendation model until the training stop condition is met, thereby obtaining a trained product recommendation model. 9. An electronic device, characterized in that the electronic device comprises: at least one processor; and a memory communicatively coupled to the at least one processor; In which, the memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the product recommendation model described in any one of claims 1 to 6, or execute the product recommendation method described in claim 7. 10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the training method of the product recommendation model described in any one of claims 1 to 6, or to execute the product recommendation method described in claim 7 when executed.