CN111179031B

CN111179031B - Training method, device and system for commodity recommendation model

Info

Publication number: CN111179031B
Application number: CN201911337899.5A
Authority: CN
Inventors: 刘正夫
Original assignee: 4Paradigm Beijing Technology Co Ltd
Current assignee: 4Paradigm Beijing Technology Co Ltd
Priority date: 2019-12-23
Filing date: 2019-12-23
Publication date: 2023-09-26
Anticipated expiration: 2039-12-23
Also published as: CN111179031A

Abstract

The invention discloses a training method, a training device and a training system for a commodity recommendation model, wherein the training method comprises the following steps: acquiring interaction data of a target user for executing various interaction operations on a target commodity and an initial sample set for training a commodity recommendation model; training an initial commodity recommendation model by adopting a preset machine learning algorithm based on an initial sample set; combining the interactive data of various interactive operations according to the initial commodity recommendation model and the initial sample set to obtain combined interactive data; based on a preset graphic neural network, obtaining the user characteristics of each target user and the commodity characteristics of each target commodity according to the combined interaction data; constructing a new sample set according to the user characteristics, commodity characteristics and the initial sample set; based on the new sample set, training a final commodity recommendation model by adopting a machine learning algorithm.

Description

Training method, device and system for commodity recommendation model

Technical Field

The present invention relates to the technical field of commodity recommendation, and more particularly, to a training method of a commodity recommendation model, a training device of a commodity recommendation model, a system including at least one computing device and at least one storage device, and a readable storage medium.

Background

In the internet age, a large amount of interactive data between users and goods is generated every day. The interactive data contains rich information, and can be used for constructing a commodity recommendation model for accurate marketing, so that a user can be better served. In the current industry, it is common practice to recommend models to first extract features and then construct training samples to train the model.

In the prior art, when features for training a recommendation model are extracted, only specified interactive operations executed by a user on commodities are generally considered, and other interactive operations executed by the user on the commodities are rarely considered, so that the accuracy of recommending the commodities to the user through the recommendation model is low.

Disclosure of Invention

The invention aims to provide a new technical scheme for training a commodity recommendation model.

According to a first aspect of the present invention, there is provided a training method of a commodity recommendation model, including:

acquiring interaction data of a target user for executing various interaction operations on a target commodity and an initial sample set for training a commodity recommendation model, wherein each original sample in the initial sample set comprises a plurality of selected features and a label, and the label represents whether the target user executes specified interaction operations on the target commodity when the corresponding original sample is generated;

Training an initial commodity recommendation model by adopting a preset machine learning algorithm based on the initial sample set;

combining the interactive data of various interactive operations according to the initial commodity recommendation model and the initial sample set to obtain combined interactive data;

based on a preset graphic neural network, obtaining the user characteristics of each target user and the commodity characteristics of each target commodity according to the combined interaction data;

constructing a new sample set according to the user features, the commodity features and the initial sample set;

and training a final commodity recommendation model by adopting the machine learning algorithm based on the new sample set.

Optionally, the merging the interaction data of the multiple interaction operations according to the initial commodity recommendation model and the initial sample set, where obtaining the merged interaction data includes:

based on the initial commodity recommendation model, obtaining a prediction matching degree between each target user and each target commodity according to the selected characteristics;

constructing a prediction matching degree matrix according to the prediction matching degree between each target user and each target commodity;

respectively constructing an interaction matrix corresponding to each interaction operation according to the interaction data of each interaction operation;

And training a preset machine learning model according to the prediction matching degree matrix and each interaction matrix to obtain the combined interaction data.

Optionally, training a preset machine learning model according to the prediction matching degree matrix and each interaction matrix, and obtaining the combined interaction data includes:

taking undetermined parameters of the machine learning model as variables, and constructing an expression combining the interaction matrixes according to the interaction matrixes;

constructing a first loss function according to the predictive matching degree matrix and the expression of the parallel interaction matrix;

solving the first loss function, and determining the value of the undetermined parameter of the machine learning model to obtain the combined interaction matrix;

and obtaining the merged interaction data according to the merged interaction matrix.

Optionally, the obtaining the merged interaction data according to the merged interaction matrix includes:

determining a total number of elements and a non-null rate in the interaction matrix of the specified interaction operation; wherein the non-null rate is the ratio of the number of non-zero elements to the total number of elements;

determining a score threshold based on the total number of elements and the non-null rate;

and adjusting element values with values smaller than or equal to the score threshold value in the combined interaction matrix to be a first set value, and adjusting element values larger than the score threshold value to be a second set value, so as to obtain the combined interaction data.

Optionally, the expression of the merged interaction matrix is expressed as:

wherein X is _i An interaction matrix for the ith interaction operation; w (w) _i And b is the machine learning modelAnd (5) undetermined parameters, wherein P is the merging interaction matrix.

Optionally, the first loss function is expressed as:

wherein K is _i,j For the element value of the ith row and the jth column in the prediction matching degree matrix, P _i,j The element values of the ith row and the jth column in the combined interaction matrix are obtained; m is the number of rows of the matrix and n is the number of columns of the matrix.

Optionally, the obtaining, based on the preset graphic neural network, the user characteristics of each target user and the commodity characteristics of each target commodity according to the combined interaction data includes:

respectively constructing a graphic neural network corresponding to each target user and each target commodity according to the combined interaction data; wherein the number of layers of each graph neural network is the same;

training the graphic neural network of each target user and each target commodity according to the combined interaction data, and obtaining the value of each target user on the hidden layer of the corresponding graphic neural network as the user characteristic of the corresponding target user; and obtaining the value of each target commodity on the hidden layer of the corresponding graph neural network as the commodity characteristic of the corresponding target commodity.

Optionally, the training the graphic neural network of each target user and each target commodity according to the combined interaction data includes:

for each target commodity and each target user, respectively taking the parameters to be determined of the corresponding graph neural network as variables to construct an expression of the hidden layer; wherein, each graphic neural network is the same in the undetermined parameters of the same layer;

constructing an expression of a distance between each target commodity and each target user according to the expression of the hidden layer of each target commodity and the expression of the hidden layer of each target user;

constructing a second loss function according to the expression of the distance between each target commodity and each target user and the combined interaction data;

and solving the second loss function, and determining the value of the undetermined parameter of each graph neural network.

Optionally, taking each target commodity and each target user as target nodes in turn;

the expression of the hidden layer of the target node is expressed as:

h ⁰ ＝x

wherein, thereinIs the value of the hidden layer of the target node of the kth layer, x is the initial value of the target node, N (v) represents the node connected with the target node in the graph neural network, sigma is the activation function, W _k And B _k Are all undetermined coefficients of the k layer of the graph neural network.

Optionally, the expression of the second loss function is expressed as:

L2＝∑ _ij (y _ij -P _ij ) ²

wherein P is _ij Element values corresponding to the ith target user and the jth target commodity in the synthetic interaction matrix; y is _ij And the distance between the ith target user and the jth target commodity is the distance.

Optionally, the machine learning algorithm is a GBDT algorithm.

Optionally, the constructing a new sample set according to the user feature, the commodity feature and the initial sample set includes:

and adding the user characteristics of the corresponding target user and the commodity characteristics of the corresponding target commodity in each initial sample to obtain a new sample.

Optionally, the method further comprises:

acquiring a feature value of a selected feature of at least one candidate commodity corresponding to a user to be recommended and new interaction data of the user to be recommended for executing the plurality of interaction operations on each candidate commodity;

obtaining user characteristics of the user to be recommended and commodity characteristics of each candidate commodity according to the new interaction data;

based on the final commodity recommendation model, obtaining recommendation scores of the candidate commodities corresponding to the candidate commodities according to the feature values of the selected features of the candidate commodities corresponding to the user to be recommended, the user features of the user to be recommended and the commodity features of the candidate commodities respectively;

Selecting the candidate commodities with the recommendation score meeting preset recommendation conditions as recommended commodities to be recommended to the user to be recommended.

Optionally, the step of selecting the candidate commodity with the recommendation score meeting the preset recommendation condition as the recommended commodity to be recommended to the user to be recommended includes:

sorting the candidate commodities in a descending order according to the recommendation score, and acquiring the sorting order of each candidate commodity;

selecting candidate commodities with the sorting order conforming to a preset sorting range, and recommending the candidate commodities to the user to be recommended as recommended commodities.

Optionally, the method further comprises:

the candidate good, and the order of ordering of each candidate good, is shown.

According to a second aspect of the present invention, there is provided a training apparatus of a commodity recommendation model, comprising:

the system comprises a data acquisition module, a commodity recommendation module and a commodity recommendation module, wherein the data acquisition module is used for acquiring interaction data of a target user for executing various interaction operations on a target commodity and an initial sample set for training a commodity recommendation model, each initial sample in the initial sample set comprises a plurality of selected characteristics and a label, and the label represents whether the target user executes specified interaction operations on the target commodity when the corresponding initial sample is generated;

The initial training module is used for training an initial commodity recommendation model by adopting a preset machine learning algorithm based on the initial sample set;

the data merging module is used for merging the interactive data of various interactive operations according to the initial commodity recommendation model and the initial sample set to obtain merged interactive data;

the feature generation module is used for obtaining the user features of each target user and the commodity features of each target commodity according to the combined interaction data based on a preset graphic neural network;

the sample construction module is used for constructing a new sample set according to the user characteristics, the commodity characteristics and the initial sample set; the method comprises the steps of,

and the final training module is used for training a final commodity recommendation model by adopting the machine learning algorithm based on the new sample set.

Optionally, the data merging module is further configured to:

Optionally, the expression of the merged interaction matrix is expressed as:

wherein X is _i An interaction matrix for the ith interaction operation; w (w) _i And b is a pending parameter of the machine learning model, and P is the merge interaction matrix.

Optionally, the first loss function is expressed as:

Optionally, the feature generation module is further configured to:

the expression of the hidden layer of the target node is expressed as:

h ⁰ ＝x

wherein, thereinIs the value of the hidden layer of the target node of the k-th layer, x is the initial value of the target node, and N (v) represents the value of the hidden layer of the target nodeNodes connected with the target node in the graph neural network, sigma is an activation function, W _k And B _k Are all undetermined coefficients of the k layer of the graph neural network.

Optionally, the expression of the second loss function is expressed as:

L2＝∑ _ij (y _ij -P _ij ) ²

Optionally, the machine learning algorithm is a GBDT algorithm.

Optionally, the sample construction module is further configured to:

Optionally, the method further comprises:

the module is used for acquiring a characteristic value of a selected characteristic of at least one candidate commodity corresponding to a preset user to be recommended and new interaction data of the user to be recommended for executing the various interaction operations on each candidate commodity;

a module for obtaining the user characteristics of the user to be recommended and the commodity characteristics of each candidate commodity according to the new interaction data;

based on the final commodity recommendation model, obtaining recommendation scores of the candidate commodities corresponding to the candidate commodities according to the feature values of the selected features of the candidate commodities corresponding to the user to be recommended, the user features of the user to be recommended and the commodity features of the candidate commodities;

And the module is used for selecting the candidate commodities of which the recommendation scores meet preset recommendation conditions and recommending the candidate commodities to the user to be recommended as recommended commodities.

Optionally, the selecting the candidate commodity with the recommendation score meeting the preset recommendation condition as the recommended commodity to be recommended to the user to be recommended includes:

Optionally, the method further comprises:

the system also includes means for displaying the candidate good, and a ranking order for each candidate good.

According to a third aspect of the present invention there is provided a system comprising at least one computing device and at least one storage device, wherein the at least one storage device is adapted to store instructions for controlling the at least one computing device to perform a method according to the first aspect of the present invention.

According to a fourth aspect of the present invention there is provided a computer readable storage medium having stored thereon a computer program which when executed by a processor implements a method according to the first aspect of the present invention.

In the embodiment of the invention, the user characteristics of each target user and the commodity characteristics of each target commodity are obtained based on the graph neural network through the interactive data of the target user for executing various interactive operations on the target commodity, and the final commodity recommendation model is trained and obtained according to the user characteristics and the commodity characteristics, so that the final commodity recommendation model can learn useful information more easily, and the recommendation effect of the final commodity recommendation model can be improved.

Other features of the present invention and its advantages will become apparent from the following detailed description of exemplary embodiments of the invention, which proceeds with reference to the accompanying drawings.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description, serve to explain the principles of the invention.

Fig. 1 is a block diagram of one example of a hardware configuration of an electronic device that may be used to implement an embodiment of the invention.

FIG. 2 is a flow chart of a training method of a commodity recommendation model according to an embodiment of the present invention;

FIG. 3 is a flow chart of the steps for obtaining merged data according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of the results of the neural network of FIG. 4, according to an embodiment of the present invention;

FIG. 5 is a block schematic diagram of a training apparatus for a commodity recommendation model according to an embodiment of the present invention;

fig. 6 is a block schematic diagram of a system according to an embodiment of the invention.

Detailed Description

Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that: the relative arrangement of the components and steps, numerical expressions and numerical values set forth in these embodiments do not limit the scope of the present invention unless it is specifically stated otherwise.

The following description of at least one exemplary embodiment is merely exemplary in nature and is in no way intended to limit the invention, its application, or uses.

Techniques, methods, and apparatus known to one of ordinary skill in the relevant art may not be discussed in detail, but are intended to be part of the specification where appropriate.

In all examples shown and discussed herein, any specific values should be construed as merely illustrative, and not a limitation. Thus, other examples of exemplary embodiments may have different values.

It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further discussion thereof is necessary in subsequent figures.

Various embodiments and examples according to embodiments of the present invention are described below with reference to the accompanying drawings.

< hardware configuration >

Fig. 1 is a block diagram showing a hardware configuration of an electronic device 1000 in which an embodiment of the present invention can be implemented.

The electronic device 1000 may be a laptop, desktop, cell phone, tablet, etc. As shown in fig. 1, the electronic device 1000 may include a processor 1100, a memory 1200, an interface device 1300, a communication device 1400, a display device 1500, an input device 1600, a speaker 1700, a microphone 1800, and the like. The processor 1100 may be a central processing unit CPU, a microprocessor MCU, or the like. The memory 1200 includes, for example, ROM (read only memory), RAM (random access memory), nonvolatile memory such as a hard disk, and the like. The interface device 1300 includes, for example, a USB interface, a headphone interface, and the like. The communication device 1400 can be capable of wired or wireless communication, and specifically can include Wifi communication, bluetooth communication, 2G/3G/4G/5G communication, and the like. The display device 1500 is, for example, a liquid crystal display, a touch display, or the like. The input device 1600 may include, for example, a touch screen, keyboard, somatosensory input, and the like. A user may input/output voice information through the speaker 1700 and microphone 1800.

The electronic device shown in fig. 1 is merely illustrative and is in no way meant to limit the invention, its application or uses. The memory 1200 of the electronic device 1000 is configured to store instructions for controlling the processor 1100 to operate to perform the training method of any one of the commodity recommendation models provided in the embodiments of the present invention. It will be appreciated by those skilled in the art that although a plurality of devices are shown for the electronic apparatus 1000 in fig. 1, the present invention may relate to only some of the devices thereof, for example, the electronic apparatus 1000 relates to only the processor 1100 and the storage device 1200. The skilled person can design instructions according to the disclosed solution. How the instructions control the processor to operate is well known in the art and will not be described in detail here.

< method example >

In this embodiment, a training method of a commodity recommendation model is provided. The training method of the commodity recommendation model can be implemented by the electronic equipment. The electronic device may be an electronic device 1000 as shown in fig. 1.

As shown in fig. 2, the training method of the commodity recommendation model of the present embodiment may include the following steps S2100 to S2600:

In step S2100, interaction data of the target user for performing various interaction operations on the target commodity and an initial sample set for training a commodity recommendation model are obtained.

Each original sample in the initial sample set comprises a plurality of selected characteristics and labels, and the labels represent whether a target user performs specified interaction operation on a target commodity when the corresponding original sample is generated.

The interaction data may indicate whether the target user performs a corresponding interaction operation with respect to the target commodity, which may include, for example, purchasing, clicking, or searching.

In one embodiment of the invention, the interaction data corresponding to each interaction may be a data table as shown in tables 1-3 below. In the data table, when a target user executes corresponding interactive operation on a target commodity, the value of a corresponding position is 1; when the target user does not execute the corresponding interactive operation on the target commodity, the value of the corresponding position is 0.

For example, table 1 may be an interaction data table that is interactively operated as purchased. Table 2 may be an interaction data table in which the interaction is a click; table 3 may be an interaction data table that is interactively operated as a search. Wherein U represents the user number of the target user, and P represents the commodity number of the target commodity.

TABLE 1

TABLE 2

TABLE 3 Table 3

In one embodiment of the present invention, the specified interaction may be one of the aforementioned interactions, and may be specifically specified in advance according to an application scenario or specific requirements. For example, the specified interactive operation may be a purchase.

In one embodiment of the invention, the initial sample set may be a data table as shown in Table 4 below. The selected features of each of the original samples may include features f 1-f 4, and each of the original samples corresponds to a combination of a target user and a target commodity, and specifically may be a combination of a user number corresponding to a target user and a commodity number of a target commodity.

When the target user executes the appointed interaction behavior aiming at the target commodity, the value of the corresponding label position in the table 4 is 1; when the target user does not execute the specified interaction with respect to the target commodity, the value of the corresponding tag position in table 4 is 0.

TABLE 4 Table 4

User number	Shang Pinhao	f1	f2	f3	f4	Label (Label)
							U1	P1	2	1	3	11	1
U1	P2	4	2	5	12	0
							U2	P3	20	10	15	10	1
U3	P2	25	13	12	23	1
							U4	P3	34	32	13	22	0
U5	P2	52	17	15	27	1
							U6	P1	29	83	32	23	0
U2	P4	96	27	36	27	1
							U1	P4	25	32	35	22	0

Step S2200, training an initial commodity recommendation model by adopting a preset machine learning algorithm based on the initial sample set.

The machine learning algorithm of the present embodiment may be a recommendation algorithm, and may be any one of GBDT algorithm, DNN algorithm, LR algorithm, collaborative filtering algorithm, graphX algorithm, for example.

Step S2300, combining the interactive data of the multiple interactive operations according to the initial commodity recommendation model and the initial sample set to obtain combined interactive data.

In one embodiment of the present invention, combining the interactive data of the plurality of interactions according to the initial commodity recommendation model and the initial sample set to obtain the combined interactive data may include steps S2310 to S2340 shown in fig. 3:

step S2310, obtaining the predicted matching degree between each target user and each target commodity according to the selected characteristics based on the initial commodity recommendation model.

Specifically, the selected feature of each original sample may be input into the initial commodity recommendation model to obtain a corresponding predicted matching degree, that is, a predicted matching degree between the target user and the target commodity corresponding to the original sample.

Step S2320, a prediction matching degree matrix is constructed according to the prediction matching degree between each target user and each target commodity.

The predicted match between each target user and each target commodity may be as shown in table 5, with the value of each location being the predicted match between the corresponding target user and the target commodity. Wherein U represents the user number of the target user, and P represents the commodity number of the target commodity.

TABLE 5

The prediction matching degree matrix obtained according to table 5 above can be expressed as:

step S2330, constructing interaction matrixes corresponding to the interactions according to the interaction data of each interaction operation.

Specifically, the construction mode of the interaction matrix is the same as that of the prediction matching degree matrix. In the interaction matrix of each interaction operation, the target users and the target commodities corresponding to the elements in the same position are the same.

According to the data table of the purchasing operation in table 1, the interaction matrix for obtaining the purchasing operation can be constructed as follows:

according to the data table of the clicking operation in table 2, the construction of the interaction matrix for obtaining the clicking operation may be:

according to the data table of the search operation in table 3, the interaction matrix constructed to obtain the search operation may be:

step S2340, training a preset machine learning model according to the prediction matching degree matrix and each interaction matrix to obtain combined interaction data.

In one embodiment of the present invention, training a preset machine learning model according to the prediction matching degree matrix and each interaction matrix, to obtain the merged interaction data may include steps S2341 to S2344 as follows:

in step S2341, the undetermined parameters of the machine learning model are used as variables, and the expression combining the interaction matrix is constructed according to the interaction matrix.

In one embodiment of the invention, the expression of the merged interaction matrix may be expressed as:

wherein X is _i An interaction matrix for the ith interaction operation; w (w) _i And b is a pending parameter of the machine learning model, and P is a combined interaction matrix.

Step S2342, constructing a first loss function according to the expression of the prediction matching degree matrix and the parallel interaction matrix.

In one embodiment of the invention, the first loss function may be expressed as:

wherein K is _i,j To predict the element value of the ith row and the jth column in the matching degree matrix, P _i,j Element values of the ith row and the jth column in the interaction matrix are combined; m is the number of rows of the matrix and n is the number of columns of the matrix.

Step S2343, solving the first loss function, and determining the value of the undetermined parameter of the machine learning model to obtain the combined interaction matrix.

In one embodiment of the present invention, a random gradient descent optimization algorithm may be used to optimize the first loss function so that the value of the first loss function L1 is minimized, resulting in a trained w _i And b, further obtaining a combined interaction matrix P.

In one example of the present invention, the 1 st interaction matrix is the purchase interaction matrix, the 2 nd interaction matrix is the click interaction matrix, the 3 rd interaction matrix is the search interaction matrix, and the first loss function is solved to obtain w ₁ ＝0.4,w ₂ ＝0.1,w ₃ =0.1, b=0.1. Then, by p=sigmoid (b+Σ _i＝1 w _i *X _i ) The merged interaction matrix P may be obtained as:

step S2344, obtaining the merged interaction data according to the merged interaction matrix.

In one embodiment of the present invention, obtaining the merged interaction data from the merged interaction matrix may include steps S2344-1 to S2344-3 as follows:

step S2344-1, determining the total number of elements and the non-null rate in the interaction matrix for the specified interaction.

Wherein the non-null rate is the ratio of the number of non-zero elements to the total number of elements.

In particular, where the specified interactive operation is a purchase, it may be to determine the total number of elements and the non-null rate in the interactive matrix of the purchase operation.

Interaction matrix in a purchasing operationThe total number of elements is 24 and the number of non-zero elements is 9, then the non-null rate may be 9/24=0.375.

Step S2344-2, determining a score threshold based on the total number of elements and the non-null rate.

In one embodiment of the invention, the element values in the composite interaction matrix may be ordered in descending order in advance, and the ordering value for each element determined.

The sorting threshold is determined according to the total number of elements N and the non-null rate p, and may be, for example, N x p, where λ is a coefficient of the null filling rate, and λ may be, for example, 2. From the ranking threshold, a score threshold may be determined. The score threshold may be an element value with a ranking value of N x p.

For example, the element value of ranking value N x λ x p is 0.52, and then the score threshold may also be 0.52.

Step S2344-3, the element values of which the values are smaller than or equal to the score threshold value in the combined interaction matrix are adjusted to be the first set value, and the element values larger than the score threshold value are adjusted to be the second set value, so that the combined interaction data are obtained.

In one embodiment of the present invention, the first setting value and the second setting value may be set in advance according to an application scenario or specific requirements, for example, the first setting value may be 1, and the second setting value may be 0. Then, the element value greater than the score threshold in the merged interaction matrix P may be adjusted to 1 and the element value less than or equal to the score threshold may be adjusted to 0.

In the case of a score threshold of 0.52, the matrix will be mergedAfter the element values in the table are adjusted, the obtained merged interaction data can be shown in the following table 6:

TABLE 6

/>

Wherein U represents the user number of the target user, and P represents the commodity number of the target commodity.

Step S2400, based on a preset graphic neural network, obtaining the user characteristics of each target user and the commodity characteristics of each target commodity according to the combined interaction data.

In one embodiment of the present invention, obtaining the user characteristics of each target user and the commodity characteristics of each target commodity according to the merged interaction data based on the preset graph neural network may include steps S2410 to S2420 as follows:

Step S2410, respectively constructing a graphic neural network corresponding to each target user and each target commodity according to the merged interaction data.

Wherein the number of layers of each graph neural network is the same.

Specifically, each target user and each target commodity are taken as root nodes respectively, and a graph neural network corresponding to each target user and a graph neural network corresponding to each target commodity are established. In the case of N1 target users and N2 target commodities, n1+n2 graph neural networks may be constructed, with the root node of each graph neural network being different. The k-th layer in the graphic neural network may be a target commodity (or target user) corresponding to the first set value with the target user (or target commodity) of the k-1-th layer in the merged interaction data.

For example, in the case where the number of layers of the graph neural network is five, for the graph neural network having the target user U1 as the root node, the first layer of the graph neural network may be the target user U1, the second layer may be all target items corresponding to the first set value with the target user U1 in the merged interaction data, the third layer may be all target items corresponding to the first set value with the target items of the second layer in the merged interaction data, and the fourth layer may be all target items corresponding to the first set value with the target items of the third layer in the merged interaction data, and the schematic structural diagram of the graph neural network may be as shown in fig. 4.

Step S2420, training the graphic neural network of each target user and each target commodity according to the combined interaction data, and obtaining the value of each target user on the hidden layer of the corresponding graphic neural network as the user characteristic of the corresponding target user; and obtaining the value of each target commodity on the hidden layer of the corresponding graph neural network as the commodity characteristic of the corresponding target commodity.

In one embodiment of the present invention, training the graphic neural network of each target user and each target commodity according to the merged interaction data includes steps S2421 to S2424 as follows:

step S2421, for each target commodity and each target user, respectively using the parameters to be determined of the corresponding graph neural network as variables, to construct the expression of the hidden layer.

Wherein each of the graph neural networks is identical in pending parameter at the same layer.

In one embodiment of the present invention, each target commodity and each target user are taken as target nodes in turn, and the expression of the hidden layer of the target nodes can be expressed as follows:

h ⁰ ＝x

wherein, thereinA hidden layer value that is a target node of a kth layer, x is an initial value of the target node, N (v) represents a node connected to the target node in the graph neural network, σ is an activation function (which may be, for example, but not limited to, a ReLU), W _k And B _k Are all undetermined coefficients of the k layer of the graph neural network.

Specifically, the expression of the hidden layer of the i-th target user may be expressed as:

wherein, thereinIs the value of the hidden layer of the kth layer in the graph neural network of the ith target user, x _i Is the initial value of the ith target user, N (i) represents the node connected with the ith target user in the graph neural network, sigma is the activation function, W _k And B _k Are all undetermined coefficients of the k layer of the graph neural network.

The expression of the hidden layer of the j-th target commodity can be expressed as:

wherein, thereinIs the value of the hidden layer of the kth layer in the graphic neural network of the jth target commodity, x _i Is the initial value of the jth target commodity, N (i) represents the node connected with the jth target commodity in the graphic neural network, sigma is the activation function, W _k And B _k Are all undetermined coefficients of the k layer of the graph neural network.

Step S2422, an expression of the distance between each target commodity and each target user is constructed from the expression of the hidden layer of each target commodity and the expression of the hidden layer of each target user.

In one embodiment of the invention, the distance between each target commodity and each target user may be a cosine distance.

Then, the distance y between the ith target user and the jth target commodity _ij The expression of (2) can be expressed as:

wherein, the liquid crystal display device comprises a liquid crystal display device,value of hidden layer of kth layer in graph neural network for ith target user,/>Is the value of the hidden layer of the kth layer in the graphic neural network of the jth target commodity.

Step S2423, constructing a second loss function according to the expression of the distance between each target commodity and each target user, and the merged interaction data.

In one embodiment of the present invention, the expression of the second loss function L2 may be expressed as:

L2＝∑ _ij (y _ij -P _ij ) ²

wherein P is _ij The data corresponding to the ith target user and the jth target commodity in the interaction data are combined; y is _ij Is the firstDistance between i target users and j target commodity.

Step S2424, solve the second loss function, and determine the value of the pending parameter for each neural network.

And obtaining the value of the parameter to be determined of each graph neural network by solving the second loss function. According to each graphic neural network, the value of the hidden layer of the k layer in the graphic neural network of each target user and the value of the hidden layer of the k layer in the graphic neural network of each target commodity can be obtained.

Further, taking the value of each target user at the hidden layer of the corresponding graph neural network as the user characteristic of the corresponding target user; and taking the value of each target commodity in the hidden layer of the corresponding graph neural network as the commodity characteristic of the corresponding target commodity.

Specifically, the value of the hidden layer of the kth layer in the graph neural network of the ith target user can beAs the user characteristics of the ith target user, the value of the hidden layer of the kth layer in the graphic neural network of the jth target commodity is +.>As the commodity feature of the j-th target commodity.

In one embodiment of the present invention, based on a preset graph neural network, the user characteristics of each target user obtained according to the combined interaction data shown in table 6 may be as shown in table 7, and the commodity characteristics of each target commodity obtained according to the combined interaction data shown in table 6 may be as shown in table 8:

TABLE 7

User number	User features
		U1	(0.1,0.2,0,0.4)
U2	(1,0.4,0.2,0.6)
		U3	(0.3,0.6,0,0.2)
U4	(0.4,0.8,0.3,0.3)
		U5	(0.1,0.1,0.2,0.1)
U6	(0.9,0.7,0.2,0.6)

TABLE 8

Shang Pinhao	Commodity characteristics
		P1	(0.3,0.2,0.1,0.5)
P2	(0.3,0.1,0.1,0.8)
		P3	(0.7,0.8,0.2,0.5)
P4	(0.3,0.6,0.4,0.4)

Step S2500, constructing a new sample set according to the user characteristics, commodity characteristics and the initial sample set.

In one embodiment of the invention, constructing a new sample set based on the user characteristics, the commodity characteristics, and the initial sample set comprises:

In one embodiment of the present invention, the user characteristics in table 7 and the commodity characteristics in table 8 may be added to the sample data table in table 4, and a new sample data table as shown in table 9 may be obtained.

TABLE 9

Step S2600, training a final commodity recommendation model by using a machine learning algorithm based on the new sample set.

In one embodiment of the present invention, the initial commodity recommendation model may be retrained based on the new sample set to obtain the final commodity recommendation model.

In one embodiment of the invention, the commodity recommendation can be performed to the user according to the final commodity recommendation model. Specifically, the method may further include steps S3100 to S3400 as follows:

step S3100, obtaining a feature value of the selected feature of the at least one candidate commodity corresponding to the preset to-be-recommended user, and new interaction data of the to-be-recommended user for executing multiple interaction operations on each candidate commodity.

Step S3200, obtaining user characteristics of the user to be recommended and commodity characteristics of each candidate commodity according to the new interaction data.

The step of obtaining the user characteristics of the user to be recommended and the commodity characteristics of the candidate commodity according to the new interaction data may include:

combining the new interactive data of the various interactive operations according to the final commodity recommendation model to obtain new combined interactive data; based on the graphic neural network, the user characteristics of the user to be recommended and the commodity characteristics of the candidate commodity are obtained according to the new combined interaction data. Reference may be made specifically to the aforementioned step 2300 and step S2400, and no further description is given here.

Step S3300, based on the final commodity recommendation model, obtaining recommendation scores of the candidate commodities corresponding to the to-be-recommended users according to the feature values of the selected features of the to-be-recommended users corresponding to the candidate commodities, the user features of the to-be-recommended users and the commodity features of the candidate commodities.

Specifically, the feature value of the selected feature of each candidate commodity corresponding to each user to be recommended, the user feature of the user to be recommended and the commodity feature of the candidate commodity are input into a final commodity recommendation model, so that the recommendation score of the user to be recommended corresponding to each candidate commodity can be obtained.

And S3400, selecting candidate commodities with recommendation scores meeting preset recommendation conditions as recommended commodities to be recommended to the user to be recommended.

In one embodiment of the present invention, the step of selecting candidate commodities whose recommendation score meets a preset recommendation condition as recommended commodities to be recommended to the user to be recommended includes steps S3410 and S3420 as follows:

step S3410, sorting the candidate commodities in a descending order according to the recommendation score, and acquiring the sorting order of each candidate commodity.

Step S3420, selecting candidate commodities with the sorting order conforming to a preset sorting range as recommended commodities to be recommended to the user to be recommended.

In one embodiment of the present invention, the ranking range may be set in advance according to the application scenario or specific requirements, for example, the ranking range may be 1-3, and then candidate commodities with ranking order of 1, 2, and 3 may be selected and recommended to the user to be recommended as recommended commodities.

In one embodiment of the present invention, the method may further comprise:

and displaying the candidate commodities and the sorting order of each candidate commodity for selection by the user to be recommended.

< device example >

In this embodiment, a training apparatus 5000 of a commodity recommendation model is provided, as shown in fig. 5, and includes a data acquisition module 5100, an initial training module 5200, a data merging module 5300, a feature generation module 5400, a sample construction module 5500, and a final training module 5600.

The data acquisition module 5100 is configured to acquire interaction data of a target user performing various interactions with respect to a target commodity, and an initial sample set for training a commodity recommendation model. Each original sample in the initial sample set comprises a plurality of selected characteristics and labels, and the labels represent whether a target user performs specified interaction operation on a target commodity when the corresponding original sample is generated.

The initial training module 5200 is configured to train an initial commodity recommendation model by using a preset machine learning algorithm based on an initial sample set.

The data merging module 5300 is configured to merge interaction data of multiple kinds of interaction operations according to an initial commodity recommendation model and an initial sample set, so as to obtain merged interaction data.

The feature generation module 5400 is configured to obtain a user feature of each target user and a commodity feature of each target commodity according to the combined interaction data based on a preset neural network.

The sample construction module 5500 is configured to construct a new sample set according to the user characteristics, the commodity characteristics, and the initial sample set.

The final training module 5600 is configured to train a final commodity recommendation model using a machine learning algorithm based on the new sample set.

In one embodiment of the invention, the data merge module 5300 may also be configured to:

based on an initial commodity recommendation model, obtaining a prediction matching degree between each target user and each target commodity according to the selected characteristics;

and training a preset machine learning model according to the prediction matching degree matrix and each interaction matrix to obtain combined interaction data.

In one embodiment of the present invention, training a preset machine learning model according to the prediction matching degree matrix and each interaction matrix, and obtaining the merged interaction data includes:

constructing a first loss function according to the expression of the prediction matching degree matrix and the parallel interaction matrix;

solving a first loss function, and determining the value of undetermined parameters of a machine learning model to obtain a combined interaction matrix;

and obtaining the combined interaction data according to the combined interaction matrix.

In one embodiment of the present invention, obtaining consolidated interaction data from a consolidated interaction matrix includes:

Determining the total number of elements and the non-null rate in the interaction matrix of the specified interaction operation; wherein the non-null rate is the ratio of the number of non-zero elements to the total number of elements;

determining a score threshold according to the total number of elements and the non-null rate;

and adjusting the element values of which the values are smaller than or equal to the score threshold value in the combined interaction matrix to a first set value, and adjusting the element values which are larger than the score threshold value to a second set value to obtain the combined interaction data.

In one embodiment of the invention, the expression of the merged interaction matrix is expressed as:

In one embodiment of the invention, the first loss function is expressed as:

In one embodiment of the invention, the feature generation module 5400 may also be used to:

In one embodiment of the invention, training the graphic neural network for each target user and each target commodity based on the consolidated interaction data includes:

for each target commodity and each target user, respectively taking the parameters to be determined of the corresponding graph neural network as variables to construct an expression of the hidden layer; wherein, each graphic neural network is the same in the undetermined parameter of the same layer;

and solving a second loss function, and determining the value of the undetermined parameter of each graph neural network.

In one embodiment of the invention, each target commodity and each target user are taken as target nodes in turn;

the expression of the hidden layer of the target node is expressed as:

h ⁰ ＝x

wherein, thereinA hidden layer value, which is a target node of a kth layer, x is an initial value of the target node, N (v) represents a node connected to the target node in the graph neural network, σ is an activation function, W _k And B _k Are all undetermined coefficients of the k layer of the graph neural network.

In one embodiment of the invention, the expression of the second loss function is expressed as:

L2＝∑ _ij (y _ij -P _ij ) ²

wherein P is _ij Synthesizing element values corresponding to the ith target user and the jth target commodity in the interaction matrix; y is _ij Is the distance between the ith target user and the jth target commodity.

In one embodiment of the invention, the machine learning algorithm is a GBDT algorithm.

In one embodiment of the invention, the sample construction module 5500 may also be used to:

In one embodiment of the present invention, the training device 5000 of the commodity recommendation model may further include:

the method comprises the steps of acquiring a characteristic value of a selected characteristic of at least one candidate commodity corresponding to a user to be recommended and new interaction data of the user to be recommended for executing various interaction operations on each candidate commodity;

A module for obtaining user characteristics of the user to be recommended and commodity characteristics of each candidate commodity according to the new interaction data;

based on the final commodity recommendation model, obtaining recommendation scores of the users to be recommended corresponding to each candidate commodity according to the feature values of the selected features of the users to be recommended corresponding to each candidate commodity, the user features of the users to be recommended and the commodity features of the candidate commodities;

and the module is used for selecting candidate commodities with recommendation scores meeting preset recommendation conditions and recommending the candidate commodities to the user to be recommended as recommended commodities.

In one embodiment of the present invention, selecting candidate commodities whose recommendation score meets a preset recommendation condition as recommended commodities to a user to be recommended includes:

selecting candidate commodities with the sorting order conforming to a preset sorting range, and recommending the candidate commodities to a user to be recommended as recommended commodities.

the system includes means for displaying candidate items, and a ranking order for each candidate item.

Those skilled in the art will appreciate that the training device 5000 of the merchandise recommendation model may be implemented in various ways. For example, the training device 5000 of the commodity recommendation model may be implemented by an instruction configuration processor. For example, instructions may be stored in the ROM and when the device is started, the instructions are read from the ROM into the programmable device to implement the training apparatus 5000 of the merchandise recommendation model. For example, the training device 5000 of the merchandise recommendation model may be solidified into a dedicated device (e.g., ASIC). The training device 5000 of the commodity recommendation model may be divided into units independent of each other, or they may be implemented by being combined together. The training apparatus 5000 of the commodity recommendation model may be implemented by one of the above-described various implementations, or may be implemented by a combination of two or more of the above-described various implementations.

In this embodiment, the training device 5000 of the commodity recommendation model may have various implementation forms, for example, the training device 5000 of the commodity recommendation model may be any functional module running in a software product or an application program that provides a model training service, or an external embedded part, a plug-in part, a patch part, etc. of the software product or the application program, or may be the software product or the application program itself.

< System >

In this embodiment, as shown in fig. 6, a system 6000 of at least one computing device 6100 and at least one storage device 6200 is also provided. The at least one storage 6200 is configured to store executable instructions; the instructions are for controlling at least one computing device 6100 to perform a training method of a commodity recommendation model according to any embodiment of the present invention.

In this embodiment, the system 6000 may be a mobile phone, a tablet computer, a palm computer, a desktop computer, a notebook computer, a workstation, a game console, or a distributed system formed by a plurality of devices.

< computer-readable storage Medium >

In this embodiment, there is also provided a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the training method of the commodity recommendation model according to any of the embodiments of the present invention.

The present invention may be an apparatus, method and/or computer program product. The computer program product may include a computer readable storage medium having computer readable program instructions embodied thereon for causing a processor to implement aspects of the present invention.

The computer readable storage medium may be a tangible device that can hold and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: portable computer disks, hard disks, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static Random Access Memory (SRAM), portable compact disk read-only memory (CD-ROM), digital Versatile Disks (DVD), memory sticks, floppy disks, mechanical coding devices, punch cards or in-groove structures such as punch cards or grooves having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media, as used herein, are not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., optical pulses through fiber optic cables), or electrical signals transmitted through wires.

The computer readable program instructions described herein may be downloaded from a computer readable storage medium to a respective computing/processing device or to an external computer or external storage device over a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmissions, wireless transmissions, routers, firewalls, switches, gateway computers and/or edge servers. The network interface card or network interface in each computing/processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium in the respective computing/processing device.

Computer program instructions for carrying out operations of the present invention may be assembly instructions, instruction Set Architecture (ISA) instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, c++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions may be executed entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, aspects of the present invention are implemented by personalizing electronic circuitry, such as programmable logic circuitry, field Programmable Gate Arrays (FPGAs), or Programmable Logic Arrays (PLAs), with state information for computer readable program instructions, which can execute the computer readable program instructions.

Various aspects of the present invention are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer-readable program instructions.

These computer readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer readable medium having the instructions stored therein includes an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.

The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions. It is well known to those skilled in the art that implementation by hardware, implementation by software, and implementation by a combination of software and hardware are all equivalent.

The foregoing description of embodiments of the invention has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described. The terminology used herein was chosen in order to best explain the principles of the embodiments, the practical application, or the technical improvements in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein. The scope of the invention is defined by the appended claims.

Claims

1. A training method of a commodity recommendation model, comprising:

training a final commodity recommendation model by adopting the machine learning algorithm based on the new sample set;

combining the interactive data of the multiple interactive operations according to the initial commodity recommendation model and the initial sample set, wherein the obtaining combined interactive data comprises the following steps:

2. The method of claim 1, wherein training a preset machine learning model according to the prediction matching degree matrix and each interaction matrix to obtain the merged interaction data comprises:

constructing a first loss function according to the expression of the prediction matching degree matrix and the combined interaction matrix;

3. The method of claim 2, the obtaining the merged interaction data from the merged interaction matrix comprising:

4. The method of claim 2, the expression of the merged interaction matrix being expressed as:

5. The method of claim 2, the first loss function being expressed as:

wherein K is _i, For the element value of the ith row and the jth column in the prediction matching degree matrix, P _i, The element values of the ith row and the jth column in the combined interaction matrix are obtained; m is the number of rows of the matrix and n is the number of columns of the matrix.

6. The method of claim 1, wherein the obtaining, based on the preset graphic neural network, the user characteristic of each target user and the commodity characteristic of each target commodity according to the merged interaction data includes:

7. The method of claim 6, the training the graphic neural network for each target user and each target commodity according to the consolidated interaction data comprising:

8. The method of claim 7, taking each target commodity and each target user in turn as target nodes;

the expression of the hidden layer of the target node is expressed as:

h ⁰ ＝x

wherein, thereinIs the value of the hidden layer of the target node of the k-th layer, x isAn initial value of the target node, N (v) represents a node connected with the target node in the graph neural network, sigma is an activation function, W _k And B _k Are all undetermined coefficients of the k layer of the graph neural network.

9. The method of claim 7, the expression of the second loss function being expressed as:

L2＝∑ _ij (y _ij -P _ij ) ²

wherein P is _ij The element values corresponding to the ith target user and the jth target commodity in the combined interaction matrix are obtained; y is _ij And the distance between the ith target user and the jth target commodity is the distance.

10. The method of claim 1, the machine learning algorithm being a GBDT algorithm.

11. The method of claim 1, the constructing a new sample set from the user characteristic, the commodity characteristic, and the initial sample set comprising:

12. The method of claim 1, further comprising:

13. The method of claim 12, wherein the step of selecting the candidate commodity whose recommendation score meets a preset recommendation condition as the recommended commodity is recommended to the user to be recommended comprises:

14. The method of claim 13, further comprising:

the candidate good, and the order of ordering of each candidate good, is shown.

15. A training device for a commodity recommendation model, comprising:

the final training module is used for training a final commodity recommendation model by adopting the machine learning algorithm based on the new sample set;

the data merging module is further configured to:

16. The apparatus of claim 15, wherein training a preset machine learning model according to the prediction matching degree matrix and each interaction matrix to obtain the merged interaction data comprises:

17. The apparatus of claim 16, the deriving the consolidated interaction data from the consolidated interaction matrix comprising:

18. The apparatus of claim 16, the expression of the merged interaction matrix expressed as:

19. The apparatus of claim 16, the first loss function expressed as:

20. The apparatus of claim 15, the feature generation module further to:

21. The apparatus of claim 20, the training the graph neural network for each target user and each target commodity according to the consolidated interaction data comprising:

22. The apparatus of claim 21, taking each target commodity and each target user in turn as target nodes;

the expression of the hidden layer of the target node is expressed as:

h ⁰ ＝x

23. The apparatus of claim 21, the expression of the second loss function expressed as:

L2＝Σ _ij (y _ij -P _ij ) ²

24. The apparatus of claim 15, the machine learning algorithm being a GBDT algorithm.

25. The apparatus of claim 15, the sample construction module further to:

26. The apparatus of claim 15, further comprising:

27. The apparatus of claim 26, wherein the selecting the candidate merchandise whose recommendation score meets the preset recommendation condition as the recommended merchandise to be recommended to the user to be recommended comprises:

28. The apparatus of claim 27, further comprising:

29. A system comprising at least one computing device and at least one storage device, wherein the at least one storage device is to store instructions to control the at least one computing device to perform the method of any one of claims 1 to 14.

30. A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method of any of claims 1 to 14.