Product information recommendation method and device and medium

Abstract

The embodiment of the invention provides a product information recommendation method and device and a medium. The method comprises the steps of obtaining the similarity of historically purchased products among historical user information through historical purchase information, and selecting target historical user information similar to a target user through the similarity of historically purchased products among the historical user information. Wherein the historical purchase information comprises multiple pieces of historical user information and historical purchase product information of a historical user corresponding to each piece of historical user information in the multiple pieces of historical user information. Therefore, according to the embodiment of the invention, the purchase preference of the target user is measured by utilizing the purchase preference of the target historical user information through the historical purchase product similarity between the historical users, and the historical purchase product information of the target historical user information is recommended to the target user. The mode of considering the purchase preference of the target user is helpful for improving the recommendation accuracy, so that the recommended product conforms to the expectation of the target user, and the user experience is stronger.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a product information recommendation method, device and medium. Background Art

[0002] With the explosive growth of global information, it is becoming increasingly important to obtain effective information from massive amounts of information. In particular, financial institutions are often faced with the need to recommend products to users to help them make choices. At present, when financial institutions recommend products to users, they often recommend hot-selling products to users. However, different users may have different needs for products, which may result in low accuracy of recommended products, that is, the recommended products do not meet user expectations and the user experience is poor. Summary of the invention

[0003] The embodiments of the present application provide a product recommendation method, device and medium for improving the recommendation accuracy of recommended products, making the recommended products meet user expectations, and improving user experience.

[0004] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0005] In a first aspect, the present application provides a product information recommendation method, the method comprising:

[0006] Acquire historical purchase information, wherein the historical purchase information includes a plurality of historical user information and historical purchase product information of a historical user corresponding to each of the plurality of historical user information;

[0007] Based on the historical purchase information, user similarity information is obtained; the user similarity information indicates the similarity of historical purchase products of multiple historical users corresponding to the multiple historical user information;

[0008] Based on the user similarity information, target historical user information similar to the target user is determined; and the historical purchase product information of the target historical user information is recommended to the target user.

[0009] Optionally, if the user similarity information is a user relationship graph, obtaining the user similarity information based on the historical purchase information includes:

[0010] Based on the historical purchase information, construct a user-purchased product bipartite graph; wherein the user nodes in the user-purchased product bipartite graph are nodes corresponding to the historical user information, and the product nodes are nodes corresponding to the historical purchased product information;

[0011] According to the user-purchased product bipartite graph, the user relationship graph is obtained; the nodes in the user relationship graph are nodes corresponding to historical user information, two connected user nodes have the same historical purchased product information, and the edge connecting the two user nodes includes a weight value, and the weight value is related to the similarity of the historical purchased product information of two historical users corresponding to the two user nodes connected by the edge;

[0012] The determining target historical user information similar to the target user based on the user similarity information includes:

[0013] The target historical user information is determined according to the user relationship graph.

[0014] Optionally, based on the user relationship graph, historical user information similar to the target user is screened out, including:

[0015] According to the user relationship graph, a similarity matrix of historical user information is obtained; the elements in the similarity matrix are used to describe the similarity of historical purchased products of two historical user information;

[0016] Acquire indirect influence information of historical user information; the indirect influence information indicates that there is no same historical purchase product information between two historical user information, and the two historical user information have the same historical purchase product information as the third historical user information;

[0017] adjusting the similarity matrix according to the indirect impact information of the historical user information;

[0018] The target historical user information is determined according to the adjusted similarity matrix.

[0019] Optionally, the indirect influence information of the historical user information is an indirect influence factor δ (-1 < δ < 1), and the indirect influence factor is used to indicate whether the indirect influence between the historical user information has a positive effect on the similarity matrix;

[0020] The adjusting the similarity matrix according to the indirect influence information of the historical user information includes:

[0021] According to the indirect influence factor δ (-1 < δ < 1), the similarity matrix is ​​adjusted; the adjusted similarity matrix is: S ′=A-δ A 2 ;

[0022] S′ is the adjusted similarity matrix, and A is the similarity matrix.

[0023] Optionally, acquiring a similarity matrix of historical user information according to the user relationship graph includes:

[0024] For the historical user information corresponding to any two historical user nodes in the user relationship graph, the historical user information corresponding to any two historical user nodes is first historical user information u and second historical user information v;

[0025] According to the historical purchase product information N(u) of the first historical user information u and the historical purchase product information N(v) of the second historical user information v, the number of products purchased together by the first historical user information u and the second historical user information v is obtained, N(u)∩N(v), and the total number of products purchased by the first historical user node information u and the second historical user node information v is obtained, N(u)||N(v);

[0026] according to Obtaining a similarity Suv between the first historical user information u and the second historical user information v;

[0027] The similarity matrix is ​​obtained according to the similarity between the historical user information corresponding to every two historical nodes in the user relationship graph.

[0028] Optionally, determining the target historical user information according to the adjusted similarity matrix includes:

[0029] Based on the adjusted similarity matrix, K historical user information with the highest similarity to the target user is screened out; K ≥ 1, and K is an integer;

[0030] The historical user information most similar to the target user is determined from the K historical user information, and the most similar historical user information is used as the target historical user information.

[0031] In a second aspect, an embodiment of the present application provides a product information recommendation device, the device comprising:

[0032] An acquisition unit, configured to acquire historical purchase information, wherein the historical purchase information includes a plurality of historical user information and historical purchase product information of a historical user corresponding to each of the plurality of historical user information;

[0033] A processing unit, configured to obtain user similarity information based on the historical purchase information; the user similarity information indicates similarities of historically purchased products of multiple historical users corresponding to the multiple historical user information;

[0034] A determination unit is used to determine target historical user information similar to the target user based on the user similarity information; and recommend the historical purchase product information of the target historical user information to the target user.

[0035] Optionally, the processing unit is specifically configured to:

[0036] Based on the historical purchase information, construct a user-purchased product bipartite graph; wherein the user nodes in the user-purchased product bipartite graph are user nodes corresponding to the historical users, and the product nodes are nodes corresponding to the products purchased by the historical users;

[0037] According to the user-purchased product bipartite graph, a user relationship graph is obtained; the nodes in the user relationship graph are nodes corresponding to the historical users, two connected user nodes have the same purchased products, and the edge connecting the two user nodes includes a weight value, and the weight value is related to the similarity of the products purchased by two users corresponding to the two user nodes adjacent to the edge;

[0038] The determining unit is specifically configured to filter out users similar to the target user based on the user relationship graph.

[0039] Optionally, the determining unit is further configured to:

[0040] According to the user relationship graph, a similarity matrix of the historical users is obtained; the elements in the similarity matrix are used to describe the similarity between the products purchased by two historical users;

[0041] Based on the indirect influence between historical users, the similarity matrix is ​​adjusted; there is no direct relationship between the two historical users with the indirect influence, and the two historical users are both two users who have a direct relationship with a third historical user;

[0042] Based on the adjusted similarity matrix, historical users similar to the target user are screened out.

[0043] In a third aspect, an embodiment of the present application provides a computer storage medium for storing a computer program, which, when executed, is used to implement the method as described in any one of the first aspects.

[0044] In a fourth aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when executed on at least one computing device, enables the at least one computing device to implement a method as described in any one of the first aspects.

[0045] In summary, the embodiments of the present application provide a product information recommendation method, device and medium. When executing the method, the similarity of historically purchased products between historical user information can be obtained through historical purchase information, and the target historical user information similar to the target user can be selected through the similarity of historically purchased products between historical information. Among them, the historical purchase information includes multiple historical user information, and historical purchase product information of historical users corresponding to each historical user information in the multiple historical user information. Therefore, the embodiments of the present application measure the purchase preference of the target user by the purchase preference of the target historical user information through the similarity of historically purchased products between historical users, and recommend the historical purchase product information of the target historical user information to the target user. Compared with the method of directly recommending hot-selling products to the target user, this method of considering the purchase preference of the target user has higher recommendation accuracy, and the recommended products meet the expectations of the target user, so the user experience is stronger. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0047] Figure 1 A system architecture diagram of a product recommendation method provided in an embodiment of the present application;

[0048] Figure 2 A flowchart of a method for recommending product information provided in an embodiment of the present application;

[0049] Figure 3 A schematic diagram of an implementation of a recommended method provided in an embodiment of the present application;

[0050] Figure 4 A schematic diagram of a bipartite graph provided in an embodiment of the present application;

[0051] Figure 5 A schematic diagram of a user relationship provided in an embodiment of the present application;

[0052] Figure 6 A schematic diagram of a user relationship diagram provided in an embodiment of the present application;

[0053] Figure 7 A schematic diagram of the structure of a product information recommendation device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] As mentioned above, different users may have different demands for products. If a single method is used to recommend hot-selling products to users, the accuracy of the recommended products may be low, that is, the recommended products may not meet user expectations and the user experience may be poor.

[0056] In view of this, an embodiment of the present application provides a product information recommendation method. Through historical purchase information, the similarity of historically purchased products between historical user information can be obtained, and through the similarity of historically purchased products between historical information, target historical user information similar to the target user is selected. Among them, the historical purchase information includes multiple historical user information, and historical purchase product information of historical users corresponding to each historical user information in the multiple historical user information. Therefore, the embodiment of the present application measures the purchase preference of the target user by the purchase preference of the target historical user information through the similarity of historically purchased products between historical users, and recommends the historical purchase product information of the target historical user information to the target user. This method of considering the purchase preference of the target user is more accurate in recommendation than the method of directly recommending hot-selling products to the target user, and the recommended products meet the expectations of the target user, so the user experience is stronger. The product recommendation method provided by the embodiment of the present application is described in detail below in conjunction with the accompanying drawings.

[0057] First, the application scenarios of the product recommendation method provided in the embodiments of the present application are introduced.

[0058] Attached Figure 1 A system architecture diagram of a product recommendation method provided in an embodiment of the present application. The system architecture 100 includes a terminal device 101, a network 102, and a server 103. The network 102 is used to provide a medium for a communication link between the user terminal 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0059] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication user terminal applications can be installed on the terminal device 101. In the embodiment of the present application, the terminal device 101 can be a user terminal such as a mobile phone, a tablet computer, a notebook, etc. that integrates financial display services.

[0060] The server 103 may be a server that provides various services, such as a background management server (only as an example) that provides support for the terminal device 101. The background management server may obtain historical purchase information, and based on the historical purchase information, determine target historical user information similar to the target user; and send the historical purchase product information of the target historical user information to the terminal device 101 to recommend the product information to the target user.

[0061] It should be noted that the product information recommendation method provided in the embodiment of the present invention is generally executed by the server 103, and accordingly, the computing device is generally arranged in the server 103. It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. According to actual needs, there may be any number of terminal devices, networks and servers.

[0062] Attached Figure 2 This is a flowchart of a method for recommending product information provided in an embodiment of the present application. The method can be applied to Figure 1 The system architecture 100 shown is specifically applied at a server 103. The method includes:

[0063] S210. Server 103 obtains historical purchase information.

[0064] Historical purchase information refers to information about products purchased by historical users in the previous period. In the embodiment of the present application, the historical purchase information includes historical user information and historical product purchase information of historical users corresponding to the historical user information. The historical user information is used to uniquely identify the historical user.

[0065] The historical product purchase information includes historical product purchases. In addition, the historical product purchase information may also include information such as the number of historical product purchases, the time of historical product purchases, etc., which is not specifically limited in the embodiments of the present application.

[0066] For example, Table 1 is a table of historically purchased products.

[0067] Table 1

[0068]

[0069] The historical user information may be a user ID code (Identitier, ID). Table 1 uses A1 to A5 to represent different historical user IDs, which are used to uniquely identify historical users A1 to A5. The historical purchase product information of historical user A1 includes product 1 and product 2, the historical purchase product information of historical user A2 includes product 1 and product 3, the historical purchase product information of historical user A3 includes product 2 and product 4, the historical purchase product information of historical user A4 includes product 3 and product 4, and the historical purchase product information of historical user A5 includes product 1 and product 4.

[0070] In an embodiment of the present application, the server 103 may obtain the historical purchase information from the terminal device 101, or the server may also obtain the historical purchase information from other devices, which is not specifically limited in the embodiment of the present application.

[0071] S220. The server 103 obtains user similarity information based on the historical purchase information.

[0072] User similarity information refers to the similarity of products purchased by historical users. User similarity information can be user similarity or user dissimilarity. User similarity means that the products purchased by two historical users are the same. User dissimilarity means that the products purchased by two historical users are different.

[0073] For example, historical user A1 corresponding to historical user information A1 purchased product 1 and product 2, historical user A2 corresponding to historical user information A2 purchased product 1 and product 3, and historical user A5 corresponding to historical user information A5 purchased product 1 and product 4. Historical user A3 corresponding to historical user information A3 purchased product 2 and product 4, and historical user A3 corresponding to historical user information A4 purchased product 3 and product 4. Thus, user similarity information can be obtained, for example: historical user A1, historical user A2 and historical user A4 are similar, and all purchased product A1, historical user A1 is similar to historical user A3, and both purchased product 2, and historical user A1 is not similar to historical user A4.

[0074] Furthermore, the user similarity information also includes user similarity, wherein the user similarity refers to the degree of similarity between products purchased by two historical users. In one example, the more similar the products purchased by two historical users are, the higher the user similarity is, otherwise the user similarity is lower.

[0075] In an embodiment of the present application, the server 103 can determine user similarity by calculating the number of identical products purchased between historical user information, or by calculating the ratio of the number of identical products purchased between historical user information to the total number, or by other means, which are not specifically limited in the embodiment of the present application.

[0076] S230. The server 103 determines target historical user information based on the user similarity information.

[0077] The target historical user information refers to historical user information similar to the target user.

[0078] For example, as shown in Table 1, historical user A1, historical user A2 and historical user A4 are similar and all purchased product A1. Therefore, if the target user is historical user A1, the target historical user information is the historical user information corresponding to historical user A2 and historical user A4.

[0079] In one implementation, the target historical user information most similar to the target user can be calculated based on the K nearest neighbor algorithm. Specifically, based on user similarity, the similarity between the target user and the historical users in the historical purchase information is obtained, and the historical user information of the K historical users with the highest similarity is obtained. K is a positive integer. The historical user information of the K historical users is used as the target historical user information.

[0080] S240: Recommend products corresponding to the historical product purchase information of the target historical user information to the target user.

[0081] The server 103 recommends products corresponding to the historically purchased products in the target historical user information to the target user.

[0082] For example, as shown in Table 1, the target user is A1, where the target historical user information is the historical user information corresponding to historical users A2 and A3. Since historical user A1 purchased product 1 and product 2. Historical user A2 purchased product 1 and product 3. Historical user A5 purchased product 1 and product 4. The products that historical user A1 did not purchase are product 3 and product 4. Therefore, the server 103 can recommend product 3 and product 4 to the target user A1.

[0083] The embodiment of the present application provides a product information recommendation method, which can obtain the similarity of historically purchased products between historical user information through historical purchase information, and select target historical user information similar to the target user through the similarity of historically purchased products between historical information. Among them, the historical purchase information includes multiple historical user information, and historical product purchase information of historical users corresponding to each historical user information in the multiple historical user information. Therefore, the embodiment of the present application measures the purchase preference of the target user by using the purchase preference of the target historical user information through the similarity of historically purchased products between historical users, and recommends the historical product purchase information of the target historical user information to the target user. Compared with the method of directly recommending hot-selling products to the target user, this method of considering the purchase preference of the target user has a higher recommendation accuracy, and the recommended products meet the expectations of the target user, so the user experience is stronger.

[0084] Furthermore, user similarity information can be represented by a user relationship graph. The following is an implementation method for obtaining a user relationship graph. Figure 3 , which is a schematic diagram of an implementation of a recommended method provided in an embodiment of the present application, and the implementation includes:

[0085] S310, construct a user-purchased product bipartite graph.

[0086] The user-purchased product bipartite graph uses each historical user information as a user node and each purchased product information as a product node. The historical user purchase record corresponding to the historical user information corresponds to an edge, which connects the user node and the product node, indicating that the user corresponding to the user node purchased the product corresponding to the product node. Among them, x1, x2, x3, ... represent the product nodes purchased by the user in history, y1, y2, y3, ... represent the user nodes, and the constructed bipartite graph is as follows: Figure 4 shown.

[0087] 320. Calculate user similarity.

[0088] User similarity refers to the degree of similarity between historical users who have purchased products together.

[0089] The embodiments of the present application may calculate user similarity in a variety of ways, such as through a cosine similarity algorithm, a correlation similarity algorithm, and an adjusted cosine similarity algorithm.

[0090] For example: For any two historical users, historical user u and historical user v, W uv It is used to represent the similarity between historical user u and historical user v. N(u) represents the investment products purchased by historical user u, N(v) represents the investment products purchased by historical user v, and N(u)∩N(v) represents the common investment products purchased by historical user u and historical user v. Then the historical user similarity between historical user u and historical user v can be calculated using the historical user similarity S uv Represents, then, the historical user similarity can be calculated by formula (1), and obtained as:

[0091]

[0092] The above formula (1) is the cosine similarity.

[0093] S330, single-mode projection.

[0094] In order to more accurately analyze the relationship between users, we can perform a single-mode projection of the bipartite graph onto the user nodes. After projection, we get the user relationship graph. The edges between each user node represent the products they purchased together. In other words, the user relationship graph helps to discover the similarities between users. The weight of the edge in the user relationship graph represents the user similarity. The higher the user similarity, the greater the weight.

[0095] For example, Figure 4 As shown, the user-product bipartite graph is projected onto the user node in a single mode to obtain the user relationship graph as shown in Figure 5 As shown in the figure, y1, y2 and y4 have the same purchased products, and y4, y5 and y6 have the same purchased products. The weights between user nodes are positively correlated with user similarity. The higher the user similarity, the higher the weight.

[0096] By performing a single-mode projection on the user node, the user-purchased product bipartite graph is transformed into a user relationship graph, and then the weight of the user relationship graph is used to represent the similarity between users. Finally, the matrix A is used to represent the similarity matrix between users, and the following similarity matrix formula can be obtained:

[0097]

[0098] S340. Calculate the indirect influence of users.

[0099] Indirect influence means that there is no direct relationship between two historical users, and both historical users have a direct relationship with a third historical user, so the relationship between the two users is called indirect influence. Among them, the direct relationship between two historical users means that the two historical users have purchased products in common, and the historical users have no direct relationship means that the two historical users have no products purchased in common.

[0100] For example, Figure 6 A user relationship diagram is provided in the embodiment of the present application. Figure 6 As shown, Figure 6 In (a), historical user a and historical user b have a direct connection, historical user a and historical user c have a direct connection, b and c have no direct connection, but historical user b and historical user c have an indirect connection through a. Figure 6 In (b), historical users d, e, and f all have direct relationships and indirect relationships. Figure 6 In (c), historical user m and historical user n have no direct connection, but through intermediate historical user k and intermediate historical user g, historical user m and historical user n have an indirect relationship.

[0101] Among them, the indirect influence can be represented by the indirect influence factor δ (-1<δ<1). When -1<δ<0, it means that the indirect influence has a negative effect on the similarity. When 0<δ<1, it means that the indirect influence has a positive effect on the similarity. In specific applications, it is determined based on the actual situation whether the indirect influence factor corresponding to the indirect influence between historical users is positive or negative.

[0102] S350: Adjust the similarity matrix.

[0103] The server 103 may include indirect influences between users and process the user similarity matrix, thereby capturing deeper relationships between users, rather than just users who directly purchase the same product, thereby improving the accuracy of recommendations.

[0104] Specifically, the indirect influence factor δ (-1 < δ < 1), S' is used to represent the similarity of the indirect influence, then:

[0105] S′=A-δA 2

[0106] That is, S′ is a matrix that considers indirect influence and user similarity.

[0107] S360: Obtain target historical user information, and recommend historically purchased products of the target historical user information to the target user.

[0108] According to S′, the user who is most similar to the target user can be calculated based on the K nearest neighbor algorithm, and the products corresponding to the similar users are recommended to the target user.

[0109] The embodiment of the present application constructs a bipartite graph of users and purchased products, and mines potential implicit relationships among users based on the connectivity of the graph. By using a matrix to represent the influence between users, an indirect influence matrix of different paths is constructed to analyze the impact of indirect influence on recommendations. Indirect influence factors are introduced to analyze the impact of indirect influence on the collaborative filtering similarity algorithm, optimize the similarity between users, and improve the accuracy of recommendations.

[0110] In addition, the present application also provides a product information recommendation device. Figure 7 A schematic diagram of a product information recommendation device provided in an embodiment of the present application. The device 700 includes:

[0111] The acquisition unit 701 is used to acquire historical purchase information, wherein the historical purchase information includes a plurality of historical user information and historical purchase product information of a historical user corresponding to each of the plurality of historical user information;

[0112] Processing unit 702 is used to obtain user similarity information based on the historical purchase information; the user similarity information indicates the similarity of historical purchase products of multiple historical users corresponding to the multiple historical user information;

[0113] The determination unit 703 is used to determine target historical user information similar to the target user based on the user similarity information; and recommend the historical purchase product information of the target historical user information to the target user.

[0114] Optionally, the processing unit 702 is specifically configured to:

[0115] Based on the historical purchase information, construct a user-purchased product bipartite graph; wherein the user nodes in the user-purchased product bipartite graph are user nodes corresponding to the historical users, and the product nodes are nodes corresponding to the products purchased by the historical users;

[0116] According to the user-purchased product bipartite graph, a user relationship graph is obtained; the nodes in the user relationship graph are nodes corresponding to the historical users, two connected user nodes have the same purchased products, and the edge connecting the two user nodes includes a weight value, and the weight value is related to the similarity of the products purchased by two users corresponding to the two user nodes adjacent to the edge;

[0117] The determination unit 703 is also used to: obtain a similarity matrix of the historical users according to the user relationship graph; the elements in the similarity matrix are used to describe the similarity of two historical users purchasing products; based on the indirect influence between historical users, adjust the similarity matrix; there is no direct relationship between the two historical users with the indirect influence, and the two historical users are two users who have a direct relationship with a third historical user; based on the adjusted similarity matrix, filter out historical users who are similar to the target user.

[0118] Optionally, the indirect influence information of the historical user information is an indirect influence factor δ (-1 < δ < 1), and the indirect influence factor is used to indicate whether the indirect influence between the historical user information has a positive effect on the similarity matrix;

[0119] The determining unit 703 is further configured to:

[0120] According to the indirect influence factor δ (-1 < δ < 1), the similarity matrix is ​​adjusted; the adjusted similarity matrix is: S ′=A-δ A 2 ;

[0121] S′ is the adjusted similarity matrix, and A is the similarity matrix.

[0122] Optionally, the determining unit 703 is further configured to:

[0123] For the historical user information corresponding to any two historical user nodes in the user relationship graph, the historical user information corresponding to any two historical user nodes is first historical user information u and second historical user information v;

[0124] According to the historical purchase product information N(u) of the first historical user information u and the historical purchase product information N(v) of the second historical user information v, the number of products purchased together by the first historical user information u and the second historical user information v is obtained, N(u)∩N(v), and the total number of products purchased by the first historical user node information u and the second historical user node information v is obtained, N(u)||N(v);

[0125] according to Obtain the similarity Suv between the first historical user information u and the second historical user information v; and obtain the similarity matrix according to the similarity of the historical user information corresponding to every two historical nodes in the user relationship graph.

[0126] Optionally, the determination unit 703 is further used to: based on the adjusted similarity matrix, screen out K historical user information with the highest similarity to the target user; K ≥ 1, and K is an integer; determine the historical user information most similar to the target user from the K historical user information, and use the most similar historical user information as the target historical user information.

[0127] The embodiment of the present application provides a product information recommendation device, which can obtain the similarity of historically purchased products between historical user information through historical purchase information, and select target historical user information similar to the target user through the similarity of historically purchased products between historical information. The historically purchased product information of the target historical user information meets the purchase preference of the target user, so the historically purchased product information of the target historical user information is recommended to the target user, the recommendation accuracy is higher, the recommended products meet the user's expectations, and the user experience is improved.

[0128] It is understandable that in the specific implementation of the present application, related data such as user information and historical user information are involved. When the above embodiments of the present application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0129] For the convenience of description, the above system is described as being divided into various modules according to their functions. Of course, when implementing the present invention, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0130] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0131] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0132] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0133] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0134] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A product information recommendation method, characterized in that: The method comprises: Acquire historical purchase information, wherein the historical purchase information includes a plurality of historical user information and historical purchase product information of a historical user corresponding to each of the plurality of historical user information; Based on the historical purchase information, user similarity information is obtained; the user similarity information indicates the similarity of historical purchase products of multiple historical users corresponding to the multiple historical user information; Based on the user similarity information, target historical user information similar to the target user is determined; and the historical purchase product information of the target historical user information is recommended to the target user.

2. The method according to claim 1, characterized in that If the user similarity information is a user relationship graph, the acquiring of user similarity information based on the historical purchase information includes: Based on the historical purchase information, construct a user-purchased product bipartite graph; wherein the user nodes in the user-purchased product bipartite graph are nodes corresponding to the historical user information, and the product nodes are nodes corresponding to the historical purchased product information; According to the user-purchased product bipartite graph, the user relationship graph is obtained; the nodes in the user relationship graph are nodes corresponding to historical user information, two connected user nodes have the same historical purchased product information, and the edge connecting the two user nodes includes a weight value, and the weight value is related to the similarity of the historical purchased product information of two historical users corresponding to the two user nodes connected by the edge; The determining target historical user information similar to the target user based on the user similarity information includes: The target historical user information is determined according to the user relationship graph.

3. The method according to claim 2, characterized in that According to the user relationship graph, historical user information similar to the target user is filtered out, including: According to the user relationship graph, a similarity matrix of historical user information is obtained; the elements in the similarity matrix are used to describe the similarity of historical purchased products of two historical user information; Acquire indirect influence information of historical user information; the indirect influence information indicates that there is no same historical purchase product information between two historical user information, and the two historical user information have the same historical purchase product information as the third historical user information; adjusting the similarity matrix according to the indirect impact information of the historical user information; The target historical user information is determined according to the adjusted similarity matrix.

4. The method according to claim 3, characterized in that The indirect influence information of the historical user information is an indirect influence factor δ (-1 < δ < 1), and the indirect influence factor is used to indicate whether the indirect influence between the historical user information has a positive effect on the similarity matrix; The adjusting the similarity matrix according to the indirect influence information of the historical user information includes: According to the indirect influence factor δ (-1 < δ < 1), adjusting the similarity matrix; The adjusted similarity matrix is: S ′=A-δ A 2 ; S′ is the adjusted similarity matrix, and A is the similarity matrix.

5. The method according to claim 3, characterized in that: The obtaining of a similarity matrix of historical user information according to the user relationship graph includes: For the historical user information corresponding to any two historical user nodes in the user relationship graph, the historical user information corresponding to any two historical user nodes is first historical user information u and second historical user information v; According to the historical purchase product information N(u) of the first historical user information u and the historical purchase product information N(v) of the second historical user information v, the number of products purchased together by the first historical user information u and the second historical user information v is obtained, N(u)∩N(v), and the total number of products purchased by the first historical user node information u and the second historical user node information v is obtained, N(u)||N(v); according to Obtaining a similarity Suv between the first historical user information u and the second historical user information v; The similarity matrix is ​​obtained according to the similarity between the historical user information corresponding to every two historical nodes in the user relationship graph.

6. The method according to any one of claims 3 to 5, characterized in that: The step of determining the target historical user information according to the adjusted similarity matrix includes: Based on the adjusted similarity matrix, K historical user information with the highest similarity to the target user is screened out; K ≥ 1, and K is an integer; The historical user information most similar to the target user is determined from the K historical user information, and the most similar historical user information is used as the target historical user information.

7. A product information recommendation device, characterized in that: The device comprises: An acquisition unit, configured to acquire historical purchase information, wherein the historical purchase information includes a plurality of historical user information and historical purchase product information of a historical user corresponding to each of the plurality of historical user information; A processing unit, configured to obtain user similarity information based on the historical purchase information; the user similarity information indicates similarities of historically purchased products of multiple historical users corresponding to the multiple historical user information; A determination unit is used to determine target historical user information similar to the target user based on the user similarity information; and recommend the historical purchase product information of the target historical user information to the target user.

8. The device according to claim 7, characterized in that The processing unit is specifically used for: Based on the historical purchase information, construct a user-purchased product bipartite graph; wherein the user nodes in the user-purchased product bipartite graph are user nodes corresponding to the historical users, and the product nodes are nodes corresponding to the products purchased by the historical users; According to the user-purchased product bipartite graph, a user relationship graph is obtained; the nodes in the user relationship graph are nodes corresponding to the historical users, two connected user nodes have the same purchased products, and the edge connecting the two user nodes includes a weight value, and the weight value is related to the similarity of the products purchased by two users corresponding to the two user nodes adjacent to the edge; The determining unit is specifically configured to filter out users similar to the target user based on the user relationship graph.

9. The device according to claim 8, characterized in that The determining unit is further configured to: According to the user relationship graph, a similarity matrix of the historical users is obtained; the elements in the similarity matrix are used to describe the similarity between the products purchased by two historical users; Based on the indirect influence between historical users, the similarity matrix is ​​adjusted; there is no direct relationship between the two historical users with the indirect influence, and the two historical users are both two users who have a direct relationship with a third historical user; Based on the adjusted similarity matrix, historical users similar to the target user are screened out.

10. A computer storage medium, characterized in that: Used to store a computer program, which, when executed, is used to implement the method according to any one of claims 1 to 6.

Images