CN112633978B - Method, device and equipment for building graphic neural network model and method, device and equipment for recommending commodities - Google Patents
Method, device and equipment for building graphic neural network model and method, device and equipment for recommending commodities Download PDFInfo
- Publication number
- CN112633978B CN112633978B CN202011528670.2A CN202011528670A CN112633978B CN 112633978 B CN112633978 B CN 112633978B CN 202011528670 A CN202011528670 A CN 202011528670A CN 112633978 B CN112633978 B CN 112633978B
- Authority
- CN
- China
- Prior art keywords
- commodities
- merchants
- relations
- neural network
- network model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000003062 neural network model Methods 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 57
- 239000000203 mixture Substances 0.000 claims abstract description 17
- 238000010276 construction Methods 0.000 claims abstract description 13
- 239000013598 vector Substances 0.000 claims description 55
- 230000006399 behavior Effects 0.000 claims description 45
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000005295 random walk Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 11
- 238000004891 communication Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 9
- 238000004590 computer program Methods 0.000 description 8
- 239000011159 matrix material Substances 0.000 description 6
- 238000001914 filtration Methods 0.000 description 5
- 230000009466 transformation Effects 0.000 description 5
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000012552 review Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Finance (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Accounting & Taxation (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the technical field of information recommendation and discloses a graph neural network model construction method for commodity recommendation. The method comprises the following steps: acquiring behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants; acquiring a plurality of different compositions according to behavior data, relations among commodities, relations among merchants, relations among users and relations among commodities and merchants; and obtaining a graph neural network model according to the multipath different patterns. The graph neural network model obtained by the scheme considers the attribute of multiple dimensions, so that the output result of the graph neural network model accords with the behavior of a user when being used for commodity recommendation, and the experience of the user when obtaining commodity recommendation results is improved. The application also discloses a graphic neural network model construction device and equipment for commodity recommendation, and a method, a device and equipment for commodity recommendation.
Description
Technical Field
The present invention relates to the technical field of information recommendation, and for example, to a method for building a neural network model for commodity recommendation, a method, a device and equipment for commodity recommendation.
Background
Prior art commodity recommendation typically employs some classical recommendation algorithms, such as: matrix decomposition, collaborative filtering, recommendation according to heat ranking and other algorithms. The recommendation algorithm in the prior art can only recommend the commodity to the user according to a single attribute, and cannot obtain output results for recommending the commodity according to attributes of multiple dimensions.
Disclosure of Invention
The following presents a simplified summary in order to provide a basic understanding of some aspects of the disclosed embodiments. This summary is not an extensive overview, and is intended to neither identify key/critical elements nor delineate the scope of such embodiments, but is intended as a prelude to the more detailed description that follows.
The embodiment of the disclosure provides a graph neural network model construction method for commodity recommendation, a method, a device and equipment for commodity recommendation, so that an output result for commodity recommendation can be obtained according to various dimension attributes.
In some embodiments, the graph neural network model construction method for commodity recommendation includes:
acquiring behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants;
acquiring a multipath different composition according to the behavior data, the relation between commodities, the relation between merchants, the relation between users and the relation between commodities and merchants;
and obtaining a graph neural network model according to the multipath different patterns.
In some embodiments, the method for merchandise recommendation includes: and recommending commodities to the user according to the graph neural network model constructed by the method.
In some embodiments, the apparatus comprises: the system comprises a first processor and a first memory storing first program instructions, wherein the first processor is configured to execute the graph neural network model construction method for commodity recommendation when executing the first program instructions.
In some embodiments, the apparatus comprises: the system comprises a first processor and a first memory storing first program instructions, the first processor being configured to perform the method for merchandise recommendation described above when the first program instructions are executed.
In some embodiments, the apparatus comprises: the graphic neural network model construction device for commodity recommendation.
In some embodiments, the apparatus comprises: the device for recommending commodities.
The method, the device and the equipment for constructing the graphic neural network model for commodity recommendation provided by the embodiment of the disclosure can realize the following technical effects: the multi-path different composition is obtained through the behavior data of the users, the relationships among commodities, the relationships among merchants, the relationships among the users and the relationships among the commodities and the merchants, and the graph neural network model is obtained according to the multi-path different composition, so that the obtained graph neural network model is more in line with the behavior of the users when the output result is used for commodity recommendation due to the fact that the attributes of multiple dimensions are considered, and the experience of the users when the users obtain commodity recommendation results is improved.
The method, the device and the equipment for commodity recommendation provided by the embodiment of the disclosure can realize the following technical effects: the method comprises the steps of obtaining behavior data of a user, relations among commodities, relations among merchants, relations among the users and relations among the commodities and the merchants, inputting the relations among the commodities and the merchants into a graph neural network model to obtain relevance scores of any two nodes in the commodities, the merchants and the users, recommending the commodities to the users according to the relevance scores, and considering attributes of various dimensions, so that the recommended commodities are more in line with the demands of the users, and experience of the users when obtaining commodity recommendation results is improved.
The foregoing general description and the following description are exemplary and explanatory only and are not restrictive of the application.
Drawings
One or more embodiments are illustrated by way of example and not limitation in the figures of the accompanying drawings, in which like references indicate similar elements, and in which like reference numerals refer to similar elements, and in which:
FIG. 1 is a schematic diagram of a method for building a neural network model for commodity recommendation according to an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of a method for merchandise recommendation provided by an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of a neural network model building apparatus for commodity recommendation according to an embodiment of the present disclosure;
fig. 4 is a schematic diagram of an apparatus for merchandise recommendation provided in an embodiment of the present disclosure.
Detailed Description
So that the manner in which the features and techniques of the disclosed embodiments can be understood in more detail, a more particular description of the embodiments of the disclosure, briefly summarized below, may be had by reference to the appended drawings, which are not intended to be limiting of the embodiments of the disclosure. In the following description of the technology, for purposes of explanation, numerous details are set forth in order to provide a thorough understanding of the disclosed embodiments. However, one or more embodiments may still be practiced without these details. In other instances, well-known structures and devices may be shown simplified in order to simplify the drawing.
The terms first, second and the like in the description and in the claims of the embodiments of the disclosure and in the above-described figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate in order to describe embodiments of the present disclosure. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion.
The term "plurality" means two or more, unless otherwise indicated.
In the embodiment of the present disclosure, the character "/" indicates that the front and rear objects are an or relationship. For example, A/B represents: a or B.
The term "and/or" is an associative relationship that describes an object, meaning that there may be three relationships. For example, a and/or B, represent: a or B, or, A and B.
Referring to fig. 1, an embodiment of the present disclosure provides a method for constructing a neural network model for commodity recommendation, including:
step S101, acquiring behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants; the behavior data comprises one or more of historical purchasing behavior data of the user on the commodity, historical collecting behavior data of the user on the commodity and historical comment behavior data of the user on the commodity; the relationship between the goods includes one or more of being purchased by the same user, being collected by the same user, being reviewed by the same user, being of the same type of goods, belonging to the same brand, or belonging to a companion good; the relationships between merchants include: the method comprises the steps that the commodities of the same type exist, the commodities purchased by the same user exist, the commodities collected by the same user exist and the commodities reviewed by the same user exist; the relationship between the item and the merchant includes: the merchant includes the same type of merchandise or the merchant includes the merchandise; the relationship between users includes: friend relationships, and the like.
Step S102, acquiring a multipath different composition according to the behavior data of the users, the relation between commodities, the relation between merchants, the relation between the users and the relation between the commodities and the merchants.
And step S103, obtaining a graph neural network model according to the multipath different composition.
By adopting the graph neural network model construction method for commodity recommendation provided by the embodiment of the invention, the multipath different patterns are obtained through the behavior data of the users, the relationship between commodities, the relationship between merchants, the relationship between the users and the relationship between the commodities and the merchants, and the graph neural network model is obtained according to the multipath different patterns, so that the obtained graph neural network model accords with the behavior of the users when the output result is used for commodity recommendation due to the consideration of the attribute of various dimensions, and the experience of the users when obtaining commodity recommendation results is improved.
Optionally, acquiring the multipath heterograph according to the behavior data of the user, the relationship between commodities, the relationship between merchants, the relationship between the users, and the relationship between the commodities and the merchants, including: taking behavior data of users, relations among commodities, relations among merchants, relations among users and relations among commodities and merchants as edges, and taking the users, the commodities or the merchants as nodes to construct a multipath different composition.
Optionally, the types of edges corresponding to the behavior data of the user include: purchase type, collection type, comment type, etc.; optionally, the types of edges corresponding to the relationship between the commodities include: purchased by the same user, collected by the same user, reviewed by the same user, belonging to the same commodity type, belonging to the same brand or belonging to a matched commodity, etc.; optionally, the types of edges corresponding to the relationship between merchants include: the method comprises the steps that commodities purchased by the same user exist, commodities collected by the same user exist, commodities reviewed by the same user exist, the commodities belong to the same type and the like; optionally, belonging to the same type comprises: belonging to household appliances, clothing, etc.; optionally, the types of edges corresponding to the relationship between the article and the merchant include: the merchant includes the same type of merchandise or the merchant includes the merchandise, etc.; optionally, the types of edges corresponding to the relationship between the users include: a buddy type, etc.
Optionally, the multi-path iso-graph is composed of a plurality of nodes and a plurality of edges.
Optionally, attribute information of the user, attribute information of the commodity and attribute information of the merchant are acquired. Optionally, the attribute information of the user includes: age, sex, location area, etc. of the user; the attribute information of the commodity includes: price, brand, place of origin, etc. Optionally, the attribute information of the merchant includes: location areas, audience groups, merchant types, etc.
Optionally, preprocessing the behavior data of the user, the relationship between commodities, the relationship between merchants, the relationship between the users and the relationship between the commodities and the merchants is further included after the behavior data of the user, the relationship between the commodities, the relationship between the merchants, the relationship between the users and the relationship between the commodities and the merchants are obtained.
Optionally, preprocessing behavior data of the user, a relationship between commodities, a relationship between merchants, a relationship between the users, a relationship between the commodities and the merchants, including: the behavior data of the user, the relationship between commodities, the relationship between merchants, the relationship between the user, the relationship between the commodity and the merchant are processed into data in the format of < node one type > < node two type > < edge type >, < node one type > < attribute 1> < attribute 2> < attribute 3>, < attribute P >, and the like, wherein P is a positive integer.
Optionally, a multi-way iso-graph is constructed from the preprocessed data. Optionally, behavior data of the user within a preset time period is acquired. In some embodiments, the preset time period is one week.
Optionally, the attribute of age, sex, region, price, brand, place of origin, etc. is quantified to obtain the feature vector of each node.
In some embodiments, in the case that the node is a user, the attribute such as age, sex, region, user ID (Identity Document, identification number) and the like of the user is quantized to obtain the feature vector of the node.
In some embodiments, in the case that the node is a commodity, the attribute of the price, brand, place of origin, etc. of the commodity is quantized to obtain the feature vector of the node.
In some embodiments, in the case that the node is a merchant, the attribute of the region where the merchant is located, the type of merchant, the audience group, and the like is quantized to obtain the feature vector of the node. Therefore, the feature vector dimension obtained by quantification of each attribute is different, and various interactions such as purchase, collection, comment and the like exist in the data, so that the description by using the multi-path different composition better meets the requirements of users.
Optionally, obtaining the graph neural network model according to the multiple heterogeneous graphs includes: acquiring characteristic embedded vectors of all nodes and edge embedded vectors of all nodes in a multipath different composition; acquiring an aggregate feature vector of each node according to the feature embedding vector and the edge embedding vector; solving an inner product of the aggregate feature vectors of any two nodes to obtain a correlation score of the two nodes; obtaining a loss value of the correlation score; under the condition that the loss value does not meet the preset condition, the feature embedded vector and the edge embedded vector are adjusted according to the loss value until the loss value meets the preset condition; and under the condition that the loss value meets the preset condition, obtaining a graph neural network model, wherein the graph neural network model is used for solving the inner product of the aggregate eigenvectors of any two nodes in the model, and obtaining the correlation scores of the two nodes.
Optionally, feature embedding and edge embedding are respectively performed on each node to obtain feature embedding vectors and edge embedding vectors of each node.
Alternatively, by calculating X i ＝D z x i Obtaining characteristic embedded vectors of all nodes; wherein X is i Embedding vectors for the features of node i, D z Is x i The feature transformation matrix of the feature embedding layer of the corresponding node type z; z is node type, x i Is the initial feature vector of the node i, i is more than or equal to 0.
Alternatively, by calculating v i ＝M k U i a i,k Obtaining an edge embedding vector of each node; wherein v is i Embedding vectors for edges of nodes, M k Feature transformation matrix of feature embedding layer for edge type k, U i Embedding edges for all edges of node i; a, a i,k The edge-embedded weights on edge type k are embedded for the edges of node i.
Optionally, the feature embedded vector and the edge embedded vector of each node are weighted and summed to obtain an aggregate feature vector of each node.
Alternatively, by calculating e i ＝αX i +βv i Obtaining an aggregate feature vector of each node; wherein e i To aggregate feature vectors, α is the weight of the feature embedded vector, β is the weight of the edge embedded vector, X i Embedding vectors for the features of node i, v i Vectors are embedded for the edges of each node, alternatively α+β=1.
Optionally, the inner product of the aggregate feature vectors of any two nodes is calculated to obtain a correlation score between the two nodes.
Alternatively, by calculating r i,j ＝e i T ·e j Obtaining a correlation score between the node i and the node j; wherein r is i,j Scoring the correlation between node i and node j, e i T Transpose of aggregate feature vector for node i, e j Is the aggregate feature vector for node j.
Optionally, the penalty value of the relevance score is obtained by a BPR (Bayesian Personalized Ranking ) penalty function.
Alternatively, the process may be carried out in a single-stage,by calculation ofObtaining a loss value of the correlation score, wherein L bpr Loss value for correlation score, N s Sigma is a Sixmoid (S-shaped growth curve) function for the first normalization parameter; r is (r) i,j Scoring the correlation of a pair of positive samples, r f,g A correlation score for a pair of negative samples. Optionally, the positive samples are correlation scores calculated from two nodes in the same path; the negative samples are correlation scores calculated from two nodes that are not in the same path.
Alternatively, the local minimum of the loss value of the correlation score is obtained by a random gradient descent algorithm.
Optionally, in case the local minimum of the loss values does not reach within the set threshold range, embedding the feature transformation matrix D in the vector of features z Adjusting the feature transformation matrix M embedded in the edge-embedded vector k And adjusting until the local minimum value of the loss value reaches the set threshold range. And obtaining the graph neural network model under the condition that the local minimum value of the loss value reaches the set threshold range. Therefore, the model is perfectly trained through the BPR loss function and the random gradient descent algorithm, and the model can be optimized, so that the correlation score calculated through the graph neural network model is more accurate.
Optionally, obtaining the feature embedded vector of each node and the edge embedded vector of each node in the multipath heterograph includes: extracting gamma paths from the multipath heterograms by using a random walk algorithm, wherein gamma is more than or equal to 2; each path includes nodes and edges in the multi-path iso-graph; taking any node in the path as a target node, and extracting neighbor nodes of w target nodes in the path containing the target nodes; acquiring edges between the neighbor nodes and the target nodes, wherein w is more than 0; performing feature embedding on the target node to obtain a feature embedding vector of the target node until all nodes in the path are subjected to feature embedding; and performing edge embedding on the target node to obtain an edge embedding vector of the target node until all nodes in the path are subjected to edge embedding.
Optionally, extracting gamma paths in the multipath heterograms by a random walk algorithm, wherein gamma is more than or equal to 2; each path has a length t.
Optionally, the procedure of transferring the feature vector of the neighboring node j to the node i on the k-type edge is as follows:
by calculation ofObtaining the embedding of the node i on each type of edge; wherein U is j→i,k Information transfer for edge embedding of node j into node i on edge type k, C i,j For the second normalization parameter, W k For converting matrices for dimensions on different edge types, U j,k For node j edge embedding on the edge of the k-type.
Optionally U i ＝(U i,1 ,U i,2 ,…U i,k )，U i The U is obtained by embedding and splicing the edges of all edge types on the node i i,k The edges on node i are embedded for different types of edges.
Alternatively, by calculating a i,k ＝softmax(W k tanh(W k U k )) T Obtaining an edge embedded weight value of the node i on the edge embedded edge type k; wherein W is k For converting matrices for dimensions on different edge types, U k For vector representations on the same node for different edge types, a i,k The edge-embedded weights on edge type k are embedded for the edges of node i.
Alternatively, by calculating v i ＝M k U i a i,k Obtaining an edge embedding vector of each node; wherein v is i Embedding vectors for edges of nodes, M k Feature transformation matrix of feature embedding layer for edge type k, U i Embedding edges for all edges of node i; a, a i,k The edge-embedded weights on edge type k are embedded for the edges of node i.
The embodiment of the disclosure provides a method for commodity recommendation, which comprises the following steps: and recommending commodities to the user according to the graph neural network model constructed by the method.
As shown in conjunction with fig. 2, an embodiment of the present disclosure provides another method for merchandise recommendation, comprising:
step S201, acquiring behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants in a preset time period;
step S202, inputting behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants into a graph neural network model for calculation to obtain relevance scores;
and step S203, recommending the commodity to the user according to the relevance score.
By adopting the method for commodity recommendation provided by the embodiment of the disclosure, the following technical effects can be achieved: the method comprises the steps of obtaining behavior data of a user, relations among commodities, relations among merchants, relations among the users and relations among the commodities and the merchants, inputting the relations among the commodities and the merchants into a graph neural network model to obtain relevance scores of any two nodes in the commodities, the merchants and the users, recommending the commodities to the users according to the relevance scores, and considering attributes of various dimensions, so that the recommended commodities are more in line with the demands of the users, and experience of the users when obtaining commodity recommendation results is improved.
Optionally, the recommending the commodity to the user according to the graphic neural network model comprises the following steps: acquiring behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants; inputting behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants into a graph neural network model for calculation to obtain relevance scores; and recommending the commodity to the user according to the relevance score. The behavior data comprises one or more of historical purchasing behavior data of the user on the commodity, historical collecting behavior data of the user on the commodity and historical comment behavior data of the user on the commodity; the relationship between the goods includes one or more of being purchased by the same user, being collected by the same user, being reviewed by the same user, being of the same type of goods, belonging to the same brand, or belonging to a companion good; the relationships between merchants include: the method comprises the steps that the commodities of the same type exist, the commodities purchased by the same user exist, the commodities collected by the same user exist and the commodities reviewed by the same user exist; the relationship between the item and the merchant includes: the merchant includes the same type of merchandise or the merchant includes the merchandise; the relationship between users includes: friend relationships, and the like.
Optionally, acquiring behavior data of the user, a relationship between commodities, a relationship between merchants, a relationship between the user and a relationship between the commodities and the merchants in a preset time period, and inputting the behavior data of the user, the relationship between the commodities, the relationship between the merchants, the relationship between the users and the relationship between the commodities and the merchants into the graph neural network model to obtain a correlation score of the behavior data of the user. Optionally, the graph neural network model calculates an inner product of the aggregate feature vectors of any two nodes in the model to obtain a correlation score of any two nodes.
Optionally, selecting a commodity corresponding to the correlation score meeting a preset condition such as the highest score from the obtained correlation scores, and recommending the commodity to the user.
Optionally, recommending the commodity to the user according to the relevance score comprises: under the condition that nodes i and j in the graph neural network model are users and commodities, recommending n commodities with highest relevance scores to the users according to the idea of collaborative filtering, wherein n is a positive integer; under the condition that a node i and a node j in the graph neural network model are users, and the correlation degree of the node i and the node j is highest in score, recommending the commodity purchased, collected or reviewed by the user j to the user i according to the idea of collaborative filtering; under the condition that a node i and a node j in the graph neural network model are commodities, and the correlation degree of the node i and the node j is highest in score, recommending the commodity j to a user who purchases, collects or reviews the commodity i according to the idea of collaborative filtering; under the condition that the node i is a user and the node j is a merchant in the graph neural network model, and the correlation degree score of the node i and the node j is highest, recommending the commodity of the merchant j to the user i according to the idea of collaborative filtering.
Referring to fig. 3, an embodiment of the present disclosure provides a neural network model building apparatus for commodity recommendation, including a first processor (processor) 100 and a first memory (memory) 101 storing first program instructions. Optionally, the apparatus may further comprise a first communication interface (Communication Interface) 102 and a first bus 103. The first processor 100, the first communication interface 102, and the first memory 101 may communicate with each other through the first bus 103. The first communication interface 102 may be used for information transfer. The first processor 100 may call the first program instructions in the first memory 101 to perform the neural network model building method for commodity recommendation of the above-described embodiment.
Further, the first program instructions in the first memory 101 described above may be implemented in the form of software functional units and may be stored in a computer readable storage medium when sold or used as a separate product.
The first memory 101 is used as a computer readable storage medium for storing a software program, a computer executable program, such as program instructions/modules corresponding to the methods in the embodiments of the present disclosure. The first processor 100 executes the functional application and the data processing by executing the program instructions/modules stored in the first memory 101, that is, implements the neural network model building method for commodity recommendation in the above-described embodiment.
The first memory 101 may include a first storage program area and a first storage data area, wherein the first storage program area may store a first operating system, a first application program required for at least one function; the first storage data area may store data created according to the use of the terminal device, etc. Further, the first memory 101 may include a high-speed random access memory, and may also include a nonvolatile memory.
By adopting the graphic neural network model construction device for commodity recommendation provided by the embodiment of the disclosure, the multipath different patterns are obtained through the behavior data of the users, the relationships among commodities, the relationships among merchants, the relationships among the users and the relationships among the commodities and the merchants, and the graphic neural network model is obtained according to the multipath different patterns, so that the obtained graphic neural network model accords with the behavior of the users when the output result is used for commodity recommendation due to the consideration of the attribute of various dimensions, and the experience of the users when obtaining commodity recommendation results is improved.
As shown in connection with fig. 4, an embodiment of the present disclosure provides an apparatus for merchandise recommendation, including a second processor (processor) 200 and a second memory (memory) 201 storing second program instructions. Optionally, the apparatus may further comprise a second communication interface (Communication Interface) 202 and a second bus 203. The second processor 200, the second communication interface 202, and the second memory 201 may communicate with each other through the second bus 203. The second communication interface 202 may be used for information transfer. The second processor 200 may call the second program instructions in the second memory 201 to perform the neural network model building method for commodity recommendation of the above-described embodiment.
In addition, the second program instructions in the second memory 201 may be implemented in the form of software functional units and may be stored in a computer readable storage medium when sold or used as a separate product.
The second memory 201 is used as a computer readable storage medium for storing a software program, a computer executable program, such as program instructions/modules corresponding to the methods in the embodiments of the present disclosure. The second processor 200 executes the functional application and the data processing by executing the program instructions/modules stored in the second memory 201, that is, implements the neural network model building method for commodity recommendation in the above-described embodiment.
The second memory 201 may include a second storage program area and a second storage data area, wherein the second storage program area may store a second operating system, a second application program required for at least one function; the second storage data area may store data created according to the use of the terminal device, and the like. In addition, the second memory 201 may include a high-speed random access memory, and may also include a nonvolatile memory.
By adopting the device for commodity recommendation provided by the embodiment of the disclosure, the correlation scores of any two nodes in the commodity, the merchant and the user can be obtained by acquiring the behavior data of the user, the relationship between the commodities, the relationship between the merchants, the relationship between the users and the relationship between the commodities and the merchant and inputting the relationship into the graph neural network model, and the commodity recommendation is carried out on the user according to the correlation scores.
The embodiment of the disclosure provides equipment, which comprises the graph neural network model construction device for commodity recommendation. The device acquires the multipath different composition according to the behavior data of the user, the relationship between commodities, the relationship between merchants, the relationship between the users and the relationship between the commodities and the merchants, and acquires the graph neural network model according to the multipath different composition, so that the acquired graph neural network model is more in line with the behavior of the user when being used for commodity recommendation due to the fact that the attribute of multiple dimensions is considered, and the experience of the user when acquiring commodity recommendation results is improved.
Optionally, the device comprises a computer, a server, or the like.
The embodiment of the disclosure provides equipment, which comprises the device for recommending commodities. The method comprises the steps of obtaining the behavior data of a user, the relation between commodities, the relation between merchants, the relation between the users and the relation between the commodities and the merchants, inputting the relation between the commodities and the merchants into a graph neural network model to obtain the relevance scores of any two nodes in the commodities, the merchants and the users, recommending the commodities to the users according to the relevance scores, and considering the attribute of multiple dimensions to enable the recommended commodities to meet the requirements of the users more, so that the experience of the users when obtaining commodity recommendation results is improved
Optionally, the device comprises a mobile phone, a tablet, a computer, etc.
The embodiment of the disclosure provides a computer-readable storage medium storing computer-executable instructions configured to perform the above-described graph neural network model construction method for commodity recommendation.
The disclosed embodiments provide a computer program product comprising a computer program stored on a computer readable storage medium, the computer program comprising first program instructions which, when executed by a computer, cause the computer to perform the above-described graph neural network model building method for commodity recommendation.
Another computer-readable storage medium is provided by embodiments of the present disclosure, storing computer-executable instructions configured to perform the above-described method for merchandise recommendation.
Another computer program product is provided according to an embodiment of the present disclosure, the computer program product comprising a computer program stored on a computer readable storage medium, the computer program comprising second program instructions which, when executed by a computer, cause the computer to perform the method for merchandise recommendation described above.
The computer readable storage medium may be a transitory computer readable storage medium or a non-transitory computer readable storage medium.
Embodiments of the present disclosure may be embodied in a software product stored on a storage medium, including one or more instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of a method according to embodiments of the present disclosure. And the aforementioned storage medium may be a non-transitory storage medium including: a plurality of media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or a transitory storage medium.
The above description and the drawings illustrate embodiments of the disclosure sufficiently to enable those skilled in the art to practice them. Other embodiments may involve structural, logical, electrical, process, and other changes. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the sequence of operations may vary. Portions and features of some embodiments may be included in, or substituted for, those of others. Moreover, the terminology used in the present application is for the purpose of describing embodiments only and is not intended to limit the claims. As used in the description of the embodiments and the claims, the singular forms "a," "an," and "the" (the) are intended to include the plural forms as well, unless the context clearly indicates otherwise. Similarly, the term "and/or" as used in this application is meant to encompass any and all possible combinations of one or more of the associated listed. Furthermore, when used in this application, the terms "comprises," "comprising," and/or "includes," and variations thereof, mean that the stated features, integers, steps, operations, elements, and/or components are present, but that the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof is not precluded. Without further limitation, an element defined by the phrase "comprising one …" does not exclude the presence of other like elements in a process, method or apparatus comprising such elements. In this context, each embodiment may be described with emphasis on the differences from the other embodiments, and the same similar parts between the various embodiments may be referred to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method sections disclosed in the embodiments, the description of the method sections may be referred to for relevance.
Those of skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. The skilled artisan may use different methods for each particular application to achieve the described functionality, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. It will be clearly understood by those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.
In the embodiments disclosed herein, the disclosed methods, articles of manufacture (including but not limited to devices, apparatuses, etc.) may be practiced in other ways. For example, the apparatus embodiments described above are merely illustrative, and for example, the division of the units may be merely a logical function division, and there may be additional divisions when actually implemented, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. In addition, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interface, device or unit indirect coupling or communication connection, which may be in electrical, mechanical or other form. The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to implement the present embodiment. In addition, each functional unit in the embodiments of the present disclosure may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.
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 embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). 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. In the description corresponding to the flowcharts and block diagrams in the figures, operations or steps corresponding to different blocks may also occur in different orders than that disclosed in the description, and sometimes no specific order exists between different operations or steps. For example, two consecutive operations or steps may actually be performed substantially in parallel, they may sometimes be performed in reverse order, which may be dependent on the functions involved. 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.
Claims (7)
1. A method for constructing a graphic neural network model for commodity recommendation is characterized in that,
acquiring behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants;
acquiring a multipath different composition according to the behavior data, the relation between commodities, the relation between merchants, the relation between users and the relation between commodities and merchants;
obtaining a graph neural network model according to the multipath different composition;
acquiring a multipath different composition according to the behavior data, the relation between commodities, the relation between merchants, the relation between users and the relation between commodities and merchants, wherein the multipath different composition comprises the following steps: taking the behavior data, the relation among commodities, the relation among merchants, the relation among users and the relation among commodities and merchants as edges, and taking the users, the commodities or the merchants as nodes to construct a multipath different composition;
obtaining the graph neural network model according to the multipath heterograph, comprising: acquiring characteristic embedded vectors of all nodes and edge embedded vectors of all the nodes in the multipath different composition; acquiring an aggregate feature vector of each node according to the feature embedding vector and the edge embedding vector; solving an inner product of the aggregate feature vectors of any two nodes to obtain a correlation score of the two nodes; obtaining a loss value of the correlation score; under the condition that the loss value does not meet a preset condition, adjusting the characteristic embedded vector and the edge embedded vector according to the loss value until the loss value meets the preset condition; obtaining the graph neural network model under the condition that the loss value meets the preset condition;
the obtaining of the feature embedded vector of each node and the edge embedded vector of each node in the multipath heterogram comprises the following steps: extracting gamma paths from the multipath heterograms by using a random walk algorithm, wherein gamma is more than or equal to 2; each path includes nodes and edges in the multi-path iso-graph; taking any node in the path as a target node, and extracting w neighbor nodes of the target node from the path containing the target node; acquiring edges between the neighbor nodes and the target node, wherein w is more than 0; performing feature embedding on the target node to obtain a feature embedding vector of the target node until all nodes in the path are subjected to feature embedding; and performing edge embedding on the target node to obtain an edge embedding vector of the target node until all nodes in the path are subjected to edge embedding.
2. A method for commodity recommendation, wherein commodity recommendation is performed for a user according to a graph neural network model constructed according to the method of claim 1.
3. The method of claim 2, wherein making a commodity recommendation to a user based on the graph neural network model comprises:
acquiring behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants;
inputting behavior data of users, relations between commodities, relations between merchants, relations between users and relations between commodities and merchants into the graph neural network model for calculation to obtain relevance scores;
and recommending the commodity to the user according to the relevance score.
4. A graphic neural network model construction device for commodity recommendation, comprising a first processor and a first memory storing first program instructions, wherein the first processor is configured to execute the graphic neural network model construction method for commodity recommendation of claim 1 when executing the first program instructions.
5. An apparatus for merchandise recommendation comprising a second processor and a second memory storing second program instructions, wherein the second processor is configured to perform the method for merchandise recommendation of claim 2 or 3 when executing the second program instructions.
6. An electronic device comprising the graphic neural network model construction device for commodity recommendation according to claim 4.
7. An electronic device comprising the apparatus for merchandise recommendation of claim 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011528670.2A CN112633978B (en) | 2020-12-22 | 2020-12-22 | Method, device and equipment for building graphic neural network model and method, device and equipment for recommending commodities |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011528670.2A CN112633978B (en) | 2020-12-22 | 2020-12-22 | Method, device and equipment for building graphic neural network model and method, device and equipment for recommending commodities |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112633978A CN112633978A (en) | 2021-04-09 |
| CN112633978B true CN112633978B (en) | 2024-03-08 |
Family
ID=75321637
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011528670.2A Active CN112633978B (en) | 2020-12-22 | 2020-12-22 | Method, device and equipment for building graphic neural network model and method, device and equipment for recommending commodities |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112633978B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113239232B (en) * | 2021-05-17 | 2023-12-26 | 北京达佳互联信息技术有限公司 | System, method and device for recommending graph neural network, electronic equipment and storage medium |
| CN113409121B (en) * | 2021-06-29 | 2022-02-15 | 南京财经大学 | Cross-border e-commerce recommendation method based on heterogeneous graph expression learning |
| CN116611897B (en) * | 2023-07-19 | 2023-10-13 | 北京快益通科技有限公司 | Message reminding method and system based on artificial intelligence |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110209926A (en) * | 2019-04-22 | 2019-09-06 | 北京三快在线科技有限公司 | Merchant recommendation method, device, electronic equipment and readable storage medium storing program for executing |
| WO2020073534A1 (en) * | 2018-10-12 | 2020-04-16 | 平安科技（深圳）有限公司 | Pushing method and apparatus based on re-clustering, and computer device and storage medium |
| CN111695965A (en) * | 2020-04-26 | 2020-09-22 | 清华大学 | Product screening method, system and equipment based on graph neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11062198B2 (en) * | 2016-10-31 | 2021-07-13 | Microsoft Technology Licensing, Llc | Feature vector based recommender system |
-
2020
- 2020-12-22 CN CN202011528670.2A patent/CN112633978B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020073534A1 (en) * | 2018-10-12 | 2020-04-16 | 平安科技（深圳）有限公司 | Pushing method and apparatus based on re-clustering, and computer device and storage medium |
| CN110209926A (en) * | 2019-04-22 | 2019-09-06 | 北京三快在线科技有限公司 | Merchant recommendation method, device, electronic equipment and readable storage medium storing program for executing |
| CN111695965A (en) * | 2020-04-26 | 2020-09-22 | 清华大学 | Product screening method, system and equipment based on graph neural network |
Non-Patent Citations (1)
| Title |
|---|
| Yifan Wang等.DisenHAN: Disentangled Heterogeneous Graph Attention Network for Recommendation.《CIKM'20 :proceedings of the 29th ACM international conference on information&knowledge managemant》.2020,第1605-1614页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112633978A (en) | 2021-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112633978B (en) | Method, device and equipment for building graphic neural network model and method, device and equipment for recommending commodities | |
| CN108805598B (en) | Similarity information determination method, server and computer-readable storage medium | |
| CN108665333A (en) | Method of Commodity Recommendation, device, electronic equipment and storage medium | |
| CN110909176A (en) | Data recommendation method and device, computer equipment and storage medium | |
| CN111611472A (en) | Binding recommendation method and system based on graph convolution neural network | |
| CN111506801A (en) | Sequencing method and device for sub-applications in application App | |
| Cumby et al. | Predicting customer shopping lists from point-of-sale purchase data | |
| CN106874503B (en) | Method and device for acquiring recommended data | |
| CN112801425B (en) | Method and device for determining information click rate, computer equipment and storage medium | |
| CN110197390A (en) | A kind of recommended method and system based on the correlation rule degree of association and economic value | |
| Biau et al. | Statistical analysis of k-nearest neighbor collaborative recommendation | |
| CN111461827B (en) | Push method and device for product evaluation information | |
| CN114638704A (en) | Illegal fund transfer identification method and device, electronic equipment and storage medium | |
| CN112989211B (en) | Method and system for determining information similarity | |
| CN110008348A (en) | The method and apparatus for carrying out network insertion in conjunction with node and side | |
| CN110134861B (en) | Recommendation method based on item categories and user preferences | |
| CN110020918B (en) | Recommendation information generation method and system | |
| CN112669127B (en) | Method, device and equipment for commodity recommendation | |
| CN110348947A (en) | Object recommendation method and device | |
| Zhao | Synergy between customer segmentation and personalization | |
| CN111984698B (en) | Information prediction method, device and storage medium | |
| CN111651456B (en) | Potential user determination method, service pushing method and device | |
| CN114519600A (en) | Graph neural network CTR estimation algorithm fusing adjacent node variances | |
| CN113344642A (en) | Method, device, electronic equipment and readable storage medium for predicting return on investment of advertisement | |
| CN113065892A (en) | Information pushing method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |