Product Recommendation Method, Device, Computer Equipment and Storage Medium

Abstract

The present application relates to a product pushing method, apparatus, computer device, storage medium, and computer program product. The method includes: obtaining sample portrait data of a plurality of sample objects, each sample portrait data covering multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data being the same; determining the repetition degree corresponding to each object attribute type; obtaining target portrait data of a target object, and determining the allocation ratio of each product recommendation algorithm according to the repetition degree corresponding to each object attribute type and the target object attribute types corresponding to the valid data included in the target portrait data; and determining target virtual products recommended to the target object among all virtual products through each product recommendation algorithm according to the allocation ratio of each product recommendation algorithm. Using this method can improve the accuracy of product recommendation.

Description

Technical Field

[0001] The present application relates to the field of big data technology, and in particular, to a product recommendation method, device, computer device, storage medium, and computer program product. Background Art

[0002] With the development of computer software technology and the popularization of Internet technology, there is an increasing amount of information data on the Internet. Therefore, how to extract data of interest to users from the vast amount of information data and recommend it to users is a problem that many enterprises are very concerned about. And the accurate recommendation of products will directly affect the transaction volume of enterprise products.

[0003] In the prior art, product recommendation methods include algorithms based on content recommendation and collaborative filtering algorithms, etc. However, traditional recommendation methods have the disadvantage of inaccurate recommended products. Therefore, there is an urgent need for a product recommendation method at present. Summary of the Invention

[0004] Based on this, it is necessary to provide a product recommendation method, device, computer device, computer-readable storage medium, and computer program product for the above technical problems.

[0005] In a first aspect, the present application provides a product recommendation method. The method includes:

[0006] Obtain the sample portrait data of multiple sample objects, each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0007] Determine the repetition degree corresponding to each object attribute type, and the repetition degree is used to characterize the occurrence frequency of each subtype valid data of the corresponding object attribute type in different sample portrait data;

[0008] Obtain the target portrait data of the target object, and determine the allocation ratio of each product recommendation algorithm according to the repetition degree corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data;

[0009] According to the allocation ratio of each product recommendation algorithm, determine the target virtual product recommended to the target object among all virtual products through each product recommendation algorithm.

[0010] In one of the embodiments, determining the repetition degree corresponding to each object attribute type includes:

[0011] For each object attribute type, determine the occurrence times of each subtype of the object attribute type in all sample portrait data according to whether the corresponding subtype valid data of the object attribute type is included in each sample portrait data;

[0012] Determine the repeatability of each object attribute type according to the occurrence times of the valid data of each subtype in each object attribute type.

[0013] In one embodiment, determine the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data, including:

[0014] Determine the specified object attribute type used as the input data type for each product recommendation algorithm according to the adaptation degree between each product recommendation algorithm and each object attribute type;

[0015] Determine the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to the target object attribute type and the repeatability of the specified object attribute type corresponding to each product recommendation algorithm.

[0016] In one embodiment, determine the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to the target object attribute type and the repeatability of the specified object attribute type corresponding to each product recommendation algorithm, including:

[0017] Cluster multiple sample portrait data to obtain multiple clustering types, and determine the target clustering type to which the target portrait data belongs;

[0018] Determine the first allocation ratio corresponding to each product recommendation algorithm according to the repeatability of the specified object attribute type corresponding to each product recommendation algorithm and the repeatability of each target object attribute type;

[0019] Determine the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type;

[0020] Integrate the first allocation ratio and the second allocation ratio corresponding to each product recommendation algorithm to obtain the allocation ratio of each product recommendation algorithm.

[0021] In one embodiment, determine the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type, including:

[0022] Obtain multiple second allocation ratio reference groups, and each second allocation ratio reference group includes the reference second allocation ratio corresponding to each product recommendation algorithm;

[0023] For each second allocation ratio reference group and each target sample object belonging to the target clustering type, according to the second allocation ratio reference group, simulate and recommend virtual products to the target sample object through each product recommendation algorithm, so as to obtain the true scores of each target sample object for each virtual product and the predicted scores of each virtual product recommended for each target sample object by each product recommendation algorithm;

[0024] Calculate the evaluation scores of the second allocation ratio reference groups based on the true scores of each target sample object for each virtual product and the predicted scores of each virtual product recommended for each target sample object by each product recommendation algorithm. The evaluation scores are used to measure the recommendation accuracy when using the second allocation ratio reference for product recommendation;

[0025] Based on the evaluation scores of each second allocation ratio reference group, select the target second allocation ratio reference group from all second allocation ratio reference groups, and use the target second allocation ratio reference group as the corresponding second allocation ratio of each product recommendation algorithm under the target clustering type.

[0026] In one embodiment, according to the allocation ratio of each product recommendation algorithm, determine the target virtual products recommended to the target object among all virtual products through each product recommendation algorithm, including:

[0027] Obtain the product recommendation rankings of each product recommendation algorithm for all virtual products;

[0028] Traverse each product recommendation algorithm from high to low according to the allocation ratio of each product recommendation algorithm;

[0029] For the current product recommendation algorithm currently traversed and the current remaining virtual products that have not been recommended among all virtual products, determine the current recommended quantity according to the allocation ratio of the current product recommendation algorithm and the total quantity of all current remaining virtual products;

[0030] Select and recommend the current recommended quantity of virtual products from all current remaining virtual products according to the product recommendation ranking corresponding to the current product recommendation algorithm, and repeat the above process until each product recommendation algorithm has been traversed.

[0031] In a second aspect, the present application also provides a product recommendation device. The device includes:

[0032] A first acquisition module, configured to acquire the sample portrait data of multiple sample objects. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0033] A determination module, configured to determine the repeatability corresponding to each object attribute type. The repeatability is used to characterize the occurrence frequency of the valid data of each subtype of the corresponding object attribute type in different sample portrait data;

[0034] A second acquisition module, configured to acquire the target portrait data of the target object, and determine the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data;

[0035] A recommendation module, configured to determine, according to the allocation ratios of various product recommendation algorithms and through the various product recommendation algorithms, target virtual products to be recommended to a target object from all virtual products.

[0036] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0037] Obtain the sample portrait data of multiple sample objects respectively. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0038] Determine the repetition degree corresponding to each object attribute type. The repetition degree is used to represent the occurrence frequency of valid data of each subtype of the corresponding object attribute type in different sample portrait data;

[0039] Obtain the target portrait data of the target object, and determine the allocation ratios of various product recommendation algorithms according to the repetition degree corresponding to each object attribute type and the target object attribute types corresponding to the valid data included in the target portrait data;

[0040] According to the allocation ratios of various product recommendation algorithms and through the various product recommendation algorithms, determine target virtual products to be recommended to the target object from all virtual products.

[0041] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, the following steps are implemented:

[0042] Obtain the sample portrait data of multiple sample objects respectively. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0043] Determine the repetition degree corresponding to each object attribute type. The repetition degree is used to represent the occurrence frequency of valid data of each subtype of the corresponding object attribute type in different sample portrait data;

[0044] Obtain the target portrait data of the target object, and determine the allocation ratios of various product recommendation algorithms according to the repetition degree corresponding to each object attribute type and the target object attribute types corresponding to the valid data included in the target portrait data;

[0045] According to the allocation ratios of various product recommendation algorithms and through the various product recommendation algorithms, determine target virtual products to be recommended to the target object from all virtual products.

[0046] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0047] Obtain the sample portrait data of each of multiple sample objects. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0048] Determine the repeatability corresponding to each object attribute type. The repeatability is used to characterize the occurrence frequency of each subtype of valid data of the corresponding object attribute type in different sample portrait data;

[0049] Obtain the target portrait data of the target object, and determine the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data;

[0050] According to the allocation ratio of each product recommendation algorithm, through each product recommendation algorithm, determine the target virtual product to be recommended to the target object among all virtual products.

[0051] For the above product recommendation method, device, computer device, storage medium and computer program product, obtain the sample portrait data of each of multiple sample objects. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same; determine the repeatability corresponding to each object attribute type. The repeatability is used to characterize the occurrence frequency of each subtype of valid data of the corresponding object attribute type in different sample portrait data; obtain the target portrait data of the target object, and determine the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data; according to the allocation ratio of each product recommendation algorithm, through each product recommendation algorithm, determine the target virtual product to be recommended to the target object among all virtual products. By this method, the accuracy of product recommendation can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic flowchart of the product recommendation method in an embodiment;

[0053] Figure 2 It is a schematic flowchart of the product recommendation method in another embodiment;

[0054] Figure 3 It is a structural block diagram of the product recommendation device in an embodiment;

[0055] Figure 4 It is an internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0057] It can be understood that the terms "first", "second", etc. used in the present application can be used in this article to describe various professional terms. However, unless otherwise specified, these professional terms are not restricted by these terms. These terms are only used to distinguish one professional term from another. For example, without departing from the scope of the present application, the third preset threshold and the fourth preset threshold may be the same or different.

[0058] In one embodiment, as Figure 1 shown, a product recommendation method is provided. In this embodiment, an example is given where this method is applied to a terminal. Among them, the terminal can be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. It can be understood that this method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0059] 101. Obtain the sample portrait data of each of multiple sample objects. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0060] 102. Determine the repeatability corresponding to each object attribute type. The repeatability is used to characterize the occurrence frequency of valid data of each subtype of the corresponding object attribute type in different sample portrait data;

[0061] 103. Obtain the target portrait data of the target object, and determine the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data;

[0062] 104. According to the allocation ratio of each product recommendation algorithm, determine the target virtual product recommended to the target object among all virtual products through each product recommendation algorithm.

[0063] Among them, the sample object refers to the object that receives the recommended product, such as a customer handling business in a bank. The types of sample portrait data can be multiple.

[0064] The object attribute type refers to the data of different dimensions of the sample object and the target object. For example, in the product recommendation scenario of a bank, the object attribute type can be the customer's basic information, the customer's personal assets and liabilities, the customer's interaction behavior with financial products, etc.

[0065] Each sample portrait data covers multiple object attribute types, and the same coverage range of the object attribute types corresponding to different sample portrait data means that the object attribute types in each sample portrait data are the same. For example, there are sample portrait data A, B, and C. Among them, the sample portrait data A has data of object attribute type α and data of object attribute type β. Then, both the sample portrait data B and C have data of object attribute type α and data of object attribute type β.

[0066] Each object attribute type includes data of each subtype. The valid data refers to whether each subtype of data is included in the object attribute type data. For example, the object attribute type α data includes α1 subtype data, α2 subtype data, and α3 subtype data; the object attribute type β data includes β1 subtype data, β2 subtype data, and β3 subtype data; if the object attribute type α data in the sample portrait data B is α1 subtype data and α3 subtype data, and the object attribute type β data in the sample portrait data B is β2 subtype data and β3 subtype data; then there is no valid data of the α2 subtype in the object attribute type α and the β1 subtype data in the object attribute type β in the sample portrait data B. The sample portrait data B has valid data of the α1 subtype and α3 subtype of the object attribute type α, and valid data of the β2 subtype and β3 subtype of the object attribute type β.

[0067] The target object refers to the object that receives the recommended product, such as the target customer who handles business in a bank. The target portrait data refers to the data with the same object attribute types as those in the sample portrait data. For example, if the sample portrait data has data of object attribute type α and data of object attribute type β, then the data of object attribute type α and data of object attribute type β of the target object are obtained as the target portrait data.

[0068] The target object attribute type refers to the object attribute type of the valid data in the target portrait data.

[0069] The virtual product refers to the digital products and services in the corresponding product recommendation scenario, which can be obtained by exchanging virtual resources or real resources.

[0070] The target virtual product refers to the virtual product selected from all virtual products based on the analysis of the target portrait data of the target object.

[0071] For the types of product recommendation algorithms, the embodiments of the present invention do not specifically limit them, including but not limited to: ranking list (popular) algorithm, content-based recommendation algorithm, and collaborative filtering algorithm.

[0072] The allocation ratio refers to the ratio of virtual products that each product recommendation algorithm needs to recommend; for example, if 10 types of target virtual products need to be recommended to the target object, and the allocation ratio of the collaborative filtering algorithm is 1 / 5, then 2 types of target virtual products are recommended to the target object through the collaborative filtering algorithm.

[0073] Specifically, after determining the allocation ratio of each product recommendation algorithm, select the target recommended products from all virtual recommended products according to the size of the allocation ratio and recommend them to the target object. For example, there are three product recommendation algorithms A, B, and C, the number of types of target virtual products is 10, and the number of types of all virtual products is 100; the recommendation ratio of product recommendation algorithm A is 50%, the recommendation ratio of product recommendation algorithm B is 30%, and the recommendation ratio of product recommendation algorithm C is 20%. Then, first, according to the recommendation ratio of 50% of product recommendation algorithm A, 5 types of virtual products are selected from 100 types as the corresponding target virtual products of product recommendation algorithm A and recommended to the target object. Then, 3 types of virtual products are selected from the 95 types of virtual products not recommended by product recommendation algorithm A as the corresponding target virtual products of product recommendation algorithm B and recommended to the target object. Finally, 2 types of virtual products are selected from the 92 types of virtual products not recommended by both product recommendation algorithm A and product recommendation algorithm B as the corresponding target virtual products of product recommendation algorithm C and recommended to the target object.

[0074] The method provided by the embodiments of the present invention determines the allocation ratio of each product recommendation algorithm by using the target portrait data of the target object, the repetition degree corresponding to each object attribute type, and the target object attribute type corresponding to the valid data included in the target portrait data. Therefore, according to the allocation ratio of each product recommendation algorithm, the target virtual products to be recommended to the target object can be determined from all virtual products through each product recommendation algorithm. By constructing sample portrait data, a conceptual model of the target object is constructed, and the allocation ratio of each product recommendation algorithm is given for the target object. Through the effective combination of multiple product recommendation algorithms, problems such as low recommendation accuracy of a single product recommendation algorithm are avoided, thereby improving the recommendation accuracy of virtual products and the recommendation quality of products. In addition, the allocation ratio of each product recommendation algorithm can also be trained to obtain a more refined allocation ratio strategy formula with the change of the respective sample portrait data of the sample object or the increase of data dimensions.

[0075] Combined with the content of the above embodiments, in one embodiment, determining the repetition degree corresponding to each object attribute type includes:

[0076] For each object attribute type, determine the number of types of each subtype of the object attribute type in all sample portrait data according to whether the corresponding subtype valid data of the object attribute type is included in each sample portrait data;

[0077] Determine the repeatability of each object attribute type according to the occurrence times of the valid data of each subtype in each object attribute type.

[0078] Among them, the number of object attribute types is determined according to the sample portrait data. For example, if there are 3 object attribute types in all sample portrait data, then the number of object attribute types is 3. Sort all object attribute types, and all object attribute types are divided into the first object attribute type, the second object attribute type, and the third object attribute type.

[0079] For each object attribute type, determining the occurrence times of each subtype of the object attribute type in all sample portrait data according to whether the corresponding subtype valid data of the object attribute type is included in each sample portrait data means that: for any object attribute type S, the subtypes corresponding to the object attribute type S are S1, S2, S3, S4, S5. If there is valid data of the subtype of the object attribute type S in all sample portrait data, and the object attribute type S has n subtypes of valid data, then the occurrence times of each subtype of the object attribute type S in all sample portrait data is n.

[0080] For example, there are 5 sample portrait data. In the first sample portrait data, there are valid data of subtype S1 and subtype S2 of the object attribute type S; in the second sample portrait data, there are valid data of subtype S2 and subtype S3 of the object attribute type S; in the third sample portrait data, there are valid data of subtype S1 and subtype S2 of the object attribute type S; in the fourth sample portrait data, there are valid data of subtype S3 and subtype S4 of the object attribute type S; in the fifth sample portrait data, there are valid data of subtype S1, subtype S2, subtype S4, and subtype S5 of the object attribute type S. Then the occurrence times of the valid data of subtype S1 of the object attribute type S is 1 + 0 + 1 + 0 + 1 = 3, the occurrence times of the valid data of subtype S2 of the object attribute type S is 1 + 1 + 1 + 0 + 1 = 4, the occurrence times of the valid data of subtype S3 of the object attribute type S is 0 + 1 + 1 + 0 + 1 = 3, the occurrence times of the valid data of subtype S4 of the object attribute type S is 0 + 0 + 0 + 1 + 1 = 2, and the occurrence times of the valid data of subtype S5 of the object attribute type S is 0 + 0 + 0 + 0 + 1 = 1.

[0081] Specifically, according to the occurrence times corresponding to each object attribute type, determine the repetition degree of each object attribute type, including: determine the repetition degree of each subtype in each object attribute type according to the occurrence times of the valid data of each subtype in each object attribute type; determine the repetition degree of each object attribute type according to the repetition degree of each subtype in each object attribute type.

[0082] Among them, determining the repetition degree of each subtype in each object attribute type according to the occurrence times of the valid data of each subtype in each object attribute type includes:

[0083]

[0084] In formula (1), a i refers to the i-th subtype of the object attribute type a, ω(a i ) refers to the repetition degree of the valid data of the i-th subtype of the object attribute type a, N(a i ) refers to the occurrence times of the valid data of the i-th subtype of the object attribute type a in all sample portrait data, N(a j ) refers to the occurrence times of the j-th subtype a of the object attribute type a j in all sample portrait data, and k refers to the number of valid data of k subtypes of the object attribute type a in all sample portrait data, and k is a positive integer.

[0085] Determining the repetition degree of each object attribute type according to the repetition degree of each subtype in each object attribute type includes:

[0086]

[0087] In formula (2), ω(a j ) refers to the repetition degree of the valid data of the j-th subtype of the object attribute type a, ω(a) refers to the repetition degree of the object attribute type a; k refers to the number of valid data of k subtypes of the object attribute type a in all sample portrait data, and k is a positive integer.

[0088] The method provided by the embodiment of the present invention determines the repetition degree of each subtype of each object attribute type in all sample portrait data, determines the repetition degree of each subtype in each object attribute type, and thus determines the repetition degree of each object attribute type. According to the repetition degree of the object attribute type, the importance of each object attribute type in all sample portrait data can be determined, and further the accuracy of determining the allocation ratio of each product recommendation algorithm according to the corresponding repetition degree of each object attribute type can be improved.

[0089] Combined with the content of the above embodiments, in one embodiment, according to the repetition degree corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data, determining the allocation ratio of each product recommendation algorithm includes:

[0090] Determine the specified object attribute type of each product recommendation algorithm as the input data type according to the degree of adaptation between each product recommendation algorithm and each object attribute type;

[0091] Determine the allocation ratio of each product recommendation algorithm according to the repetition degree corresponding to the target object attribute type and the repetition degree of the specified object attribute type corresponding to each product recommendation algorithm.

[0092] Among them, determining the specified object attribute type of each product recommendation algorithm as the input data type according to the degree of adaptation between each product recommendation algorithm and each object attribute type means: taking the object attribute type with the highest degree of adaptation to product recommendation algorithm A as the corresponding specified object attribute type of this product recommendation algorithm. For example, there are 3 product recommendation algorithms, namely product recommendation algorithm A, product recommendation algorithm B, and product recommendation algorithm C; among them, the object attribute type a has the highest degree of adaptation to product recommendation algorithm A; the object attribute type b has the highest degree of adaptation to product recommendation algorithm B, and the object attribute type c has the highest degree of adaptation to product recommendation algorithm C. Then the specified object attribute type corresponding to product recommendation algorithm A is object attribute type a, the specified object attribute type corresponding to product recommendation algorithm B is object attribute type b; the specified object attribute type corresponding to product recommendation algorithm C is object attribute type c.

[0093] Specifically, calculating the repetition degree corresponding to the target object attribute type includes: taking the subtype of the valid data in the target object attribute type data as the target subtype, and determining the repetition degree of each target subtype according to the occurrence times of the valid data of each target subtype; determining the repetition degree of the target object attribute type according to the repetition degrees of the various target subtypes of the target object attribute type.

[0094] Among them, determining the repetition degree of each target subtype according to the occurrence times of the valid data of each target subtype includes:

[0095]

[0096] In formula (3), m i refers to the i-th target subtype of the target object attribute type m, ω(m i ) refers to the repetition degree of the valid data of the i-th target subtype of the target object attribute type m, N(m i ) refers to the occurrence times of the valid data of the i-th subtype of the target object attribute type m in the target portrait data, N(m jRefers to the j-th subtype m of the target object attribute type m j The number of occurrences in all sample portrait data, and k refers to the number of valid data of k subtypes of the target object attribute type m in all sample portrait data, where k is a positive integer.

[0097] Among them, determining the repeatability of the target object attribute type according to the repeatability of each target subtype of the target object attribute type includes:

[0098]

[0099] In formula (4), ω(m i ) refers to the repeatability of the i-th target subtype valid data of the target object attribute type m, ω(m) refers to the repeatability of the target object attribute type m, k refers to the number of valid data of k subtypes of the target object attribute type m in all sample portrait data, where k is a positive integer, and n refers to the number of valid data of n subtypes of the target object attribute type m in all target portrait data.

[0100] The method provided by the embodiment of the present invention can determine the allocation ratio of each product recommendation algorithm through the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data, so that product recommendations for the target object can be made according to the allocation ratio of each product recommendation algorithm.

[0101] Combined with the content of the above embodiments, in one embodiment, determining the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to the target object attribute type and the repeatability of the specified object attribute type corresponding to each product recommendation algorithm includes:

[0102] Cluster multiple sample portrait data to obtain multiple clustering types, and determine the target clustering type to which the target portrait data belongs;

[0103] Determine the first allocation ratio corresponding to each product recommendation algorithm according to the repeatability of the specified object attribute type corresponding to each product recommendation algorithm and the repeatability of each target object attribute type;

[0104] Determine the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type;

[0105] Integrate the first allocation ratio and the second allocation ratio corresponding to each product recommendation algorithm to obtain the allocation ratio of each product recommendation algorithm.

[0106] Among them, for the method of clustering multiple sample portrait data to obtain multiple clustering types and determining the target clustering type to which the target portrait data belongs, the embodiments of the present invention do not make specific limitations on it, including but not limited to using the K-Means clustering algorithm. Processing the sample portrait data by using the K-Means clustering algorithm includes:

[0107] Step 1. The initialization operation is to select k mass points as the initial clustering centers μ = μ1, μ2,..., μ k .

[0108] Step 2. For each sample χ in the sample portrait dataset i , calculate its Euclidean distance to the k clustering centers, as shown in formula (5), and assign it to the class corresponding to the centroid with the closest distance.

[0109] Step 3. Recalculate the clustering centers according to formula (6).

[0110] Step 4. Repeat the operations in the above Steps 2 and 3 continuously to continuously construct a relationship graph between the sum of squared errors (formula 7) and the value of K.

[0111]

[0112]

[0113] When ,

[0114] In formulas (5), (6), and (7), D(χ, μ) is the Euclidean distance between the non-mass point χ and the mass point μ, μ c is the mean vector of the mass point c c , and SSE(C) is the sum of the squares of the Euclidean distances from the non-mass point χ to the mass point μ, that is, the sum of squared errors.

[0115] Step 5. The selection of the value of k in the clustering algorithm has a greater impact on K-means. Common methods for selecting the value of k include the elbow method. When k is less than the true number of clusters, an increase in the value of k will cause a large drop in the sum of squared errors (SSE), and the aggregation degree of each cluster increases. When k is equal to or greater than the true number of clusters, the aggregation degree return of each cluster will become smaller, and the drop in the sum of squared errors (SSE) tends to level off. By continuously repeating the K-Means clustering algorithm, draw the elbow method relationship graph to obtain the most suitable value of k and the final change graph of the clusters.

[0116] Step 6: The clustering results obtained by the K-Means clustering algorithm in Step 5 can be analyzed for group characteristics through a visualization analysis tool (BI) or a basic Excel chart. First, it is necessary to map the clusters back to the dataset and view the data frame. By comparing the averages of all variables on each cluster, the attributes of the clusters can be extracted. The analysis and extraction of the cluster attributes are combined with the RFM (Recency Frequency Monetary) model, and intuitively analyzed based on aspects such as the basic information of the target object, the virtual resources owned by the target object, and the interaction behavior of the target object with respect to the recommended products. Finally, the data related to the basic information of the target object, the virtual resources owned by the target object, and the interaction behavior of the target object with respect to the recommended products are selected, and the data in multiple aspects are analyzed to determine the product recommendation algorithm with the highest adaptation degree for each aspect of the data. For example, the content recommendation algorithm has the highest adaptation degree for the data where the target object has sufficient interaction behavior with the recommended product, and the collaborative filtering algorithm has the highest adaptation degree for the data of the basic information of the target object. Among them, the data where the recommended product has sufficient interaction behavior is a type of data of an object attribute type (such as object attribute type a), and the data of the basic information of the target object is a type of data of another object attribute type (such as object attribute type b).

[0117] The specified object attribute type refers to the object attribute type corresponding to each product recommendation algorithm and having the same type as the target object attribute type. The repeatability of the specified object attribute type refers to the repeatability of the object attribute type calculated based on all sample portrait data, that is, the repeatability of the object attribute type calculated in the above formula (2).

[0118] Specifically, according to the repeatability of the specified object attribute type corresponding to each product recommendation algorithm and the repeatability of each target object attribute type, determine the first allocation ratio corresponding to each product recommendation algorithm, including:

[0119]

[0120] In formula (8), ω(m) refers to the repeatability of the specified object attribute type m, and ω(m') refers to the repeatability of the target object attribute type m' of the same type as the specified object attribute type m; k refers to the number of valid sub-types of the data of the specified object attribute type m in all sample portrait data, and k is a positive integer; ω(m i ) refers to the repeatability of the i-th valid sub-type of the target object attribute type m, ω(m j) refers to the repeatability of the valid data of the j-th target subtype of the specified object attribute type m, n refers to the number of subtypes of valid data of the target object attribute type m in all target portrait data; η refers to the first allocation ratio corresponding to the product recommendation algorithm with the highest adaptation degree to the specified object attribute type m.

[0121] Multiply the first allocation ratio and the second allocation ratio corresponding to each product recommendation algorithm to obtain a multiplication result, and perform normalization processing on each multiplication result to obtain the recommendation ratio of each product recommendation algorithm.

[0122] The method provided by the embodiment of the present invention can determine the allocation ratio of each product recommendation algorithm through the repeatability corresponding to the target object attribute type and the repeatability of the specified object attribute type corresponding to each product recommendation algorithm.

[0123] Combined with the content of the above embodiments, in one embodiment, determining the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type includes:

[0124] Obtain multiple second allocation ratio reference groups, and each second allocation ratio reference group includes the reference second allocation ratio corresponding to each product recommendation algorithm;

[0125] For each second allocation ratio reference group and each target sample object belonging to the target clustering type, according to the second allocation ratio reference group, simulate and recommend virtual products to the target sample object through each product recommendation algorithm to obtain the true scores of each target sample object for each virtual product and the predicted scores of recommending each virtual product to each target sample object through each product recommendation algorithm;

[0126] According to the true scores of each target sample object for each virtual product and the predicted scores of recommending each virtual product to each target sample object through each product recommendation algorithm, calculate the evaluation score of the second allocation ratio reference group, and the evaluation score is used to measure the recommendation accuracy when using the second allocation ratio reference for product recommendation;

[0127] According to the evaluation scores of each second allocation ratio reference group, select the target second allocation ratio reference group from all second allocation ratio reference groups, and use the target second allocation ratio reference group as the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type.

[0128] Among them, the corresponding second allocation ratios of the product recommendation algorithms under the target clustering type are determined when clustering all sample portrait data through the clustering algorithm. For example, after clustering all sample portrait data through the clustering algorithm, 3 types of clustering type data are obtained. Each type of clustering type data corresponds to multiple second allocation ratio reference groups, and each second allocation ratio reference group includes 3 types of second allocation ratios. For example, there are three product recommendation algorithms, namely the leaderboard (popular) recommendation algorithm, the collaborative filtering algorithm, and the content-based recommendation method. Among them, the type of the second allocation ratio corresponding to the collaborative filtering algorithm is α, the type of the second allocation ratio corresponding to the content-based recommendation method is β, and the type of the second allocation ratio corresponding to the leaderboard (popular) recommendation algorithm is 1 - α - β, where α + β < 1, 0 ≤ α, 0 ≤ β. The n second allocation ratio reference groups can be: the 1st second allocation ratio reference group (α1, β1, 1 - α1 - β1), the 2nd second allocation ratio reference group (α2, β2, 1 - α2 - β2),..., the nth second allocation ratio reference group (αn, βn, 1 - αn - βn), where n is an integer greater than 2.

[0129] Among them, the collaborative filtering algorithm finds the most similar object set for the target object by calculating the similarity between target objects, and recommends the corresponding products of the object set to the target object. The similarity comparison mainly uses the modified cosine similarity, considering the rating scales of different objects. The content-based recommendation method refers to calculating the similar products of the products with high ratings by the target object, and then recommending the similar products to the target object.

[0130] Specifically, according to the true ratings of each target sample object for each virtual product and the predicted ratings of each virtual product recommended for each target sample object by each product recommendation algorithm, calculate the evaluation score of each second allocation ratio reference group, including:

[0131]

[0132] In formula (9), r s,ui is the true rating of the target sample object u for the i-th type of virtual product when using the second allocation ratio reference group s, is the predicted rating of the i-th type of virtual product recommended for the target sample object u by each product recommendation algorithm when using the second allocation ratio reference group s, and E p represents the test value, and RMSE(s) is the evaluation score of the second allocation ratio reference group s.

[0133] Selecting a target second allocation ratio reference group from all second allocation ratio reference groups means: according to the evaluation scores of each second allocation ratio reference group, selecting the second allocation ratio reference group with the highest evaluation score from the evaluation scores of all second allocation ratio reference groups as the target second allocation ratio reference group, and using this target second allocation ratio reference group as the corresponding second allocation ratio of each product recommendation algorithm under the target clustering type.

[0134] The method provided by the embodiments of the present invention can determine the target virtual product recommended to the target object among all virtual products according to the second allocation ratio and each product recommendation algorithm by determining the corresponding second allocation ratio of each product recommendation algorithm under the target clustering type.

[0135] Combined with the content of the above embodiments, in one embodiment, according to the allocation ratio of each product recommendation algorithm, determining the target virtual product recommended to the target object among all virtual products through each product recommendation algorithm includes:

[0136] Obtaining the product recommendation rankings of each product recommendation algorithm for all virtual products;

[0137] Traversing each product recommendation algorithm from high to low according to the allocation ratio of each product recommendation algorithm;

[0138] For the currently traversed current product recommendation algorithm and the current remaining virtual products that have not been recommended among all virtual products, determining the current recommended quantity according to the allocation ratio of the current product recommendation algorithm and the total quantity of all current remaining virtual products;

[0139] Selecting and recommending the current recommended quantity of virtual products from all current remaining virtual products according to the product recommendation ranking corresponding to the current product recommendation algorithm, and repeating the above process until all product recommendation algorithms have been traversed.

[0140] Specifically, obtaining the product recommendation rankings of each product recommendation algorithm for all virtual products means sorting all virtual products according to the sorting method of each product recommendation algorithm for all virtual products to obtain the virtual product sorting queue of each product recommendation algorithm. When the allocation ratio of each product recommendation algorithm is determined, the corresponding virtual products are selected from the virtual product sorting queue of each product recommendation algorithm according to the allocation ratio of each product recommendation algorithm as the target virtual products, and finally the target virtual products are recommended to the target object.

[0141] For example, the product recommendation algorithms are the recommendation algorithm for Product A, the recommendation algorithm for Product B, and the recommendation algorithm for Product C. The number of virtual products is 200, and the number of target virtual products is 20. Among them, the allocation ratio of the recommendation algorithm for Product A is 50%, the allocation ratio of the recommendation algorithm for Product B is 30%, and the allocation ratio of the recommendation algorithm for Product C is 20%. Then, the number of target virtual products recommended by the recommendation algorithm for Product A is 10, the number of target virtual products recommended by the recommendation algorithm for Product B is 6, and the number of target virtual products recommended by the recommendation algorithm for Product C is 4. First, select 20 virtual products in descending order of sorting from the virtual product sorting queue of the recommendation algorithm for Product A as the corresponding first target virtual products of the recommendation algorithm for Product A. Secondly, select 6 virtual products in descending order of sorting from the virtual product sorting queue of the recommendation algorithm for Product B and not including the first target virtual products as the corresponding second target virtual products of the recommendation algorithm for Product B. Finally, select 4 virtual products in descending order of sorting from the virtual product sorting queue of the recommendation algorithm for Product C and not including the first target virtual products and the second target virtual products as the corresponding third target virtual products of the recommendation algorithm for Product C. Recommend the first target virtual products, the second target virtual products, and the third target virtual products to the target object as the target virtual products.

[0142] The method provided by the embodiment of the present invention can improve the accuracy of the recommended products by determining the target virtual products recommended to the target object among all virtual products through each product recommendation algorithm according to the allocation ratio of each product recommendation algorithm, and can avoid the problems existing in a single product recommendation algorithm and improve the overall quality of product recommendation by recommending products to the target object according to multiple recommendation algorithms.

[0143] Combined with the content of the above embodiment, in one embodiment, as Figure 2 shown, a product recommendation method includes:

[0144] 201. Obtain the sample portrait data of each of multiple sample objects. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0145] 202. For each object attribute type, determine the number of occurrences of each subtype of the object attribute type in all sample portrait data according to whether the corresponding subtype valid data is included in each sample portrait data;

[0146] 203. Determine the repeatability of each object attribute type according to the number of occurrences of the valid data of each subtype in each object attribute type; the repeatability is used to characterize the occurrence frequency of the corresponding object attribute type having the valid data of each subtype in different sample portrait data;

[0147] 204. Obtain the target portrait data of the target object, and determine the specified object attribute type of each product recommendation algorithm as the input data type according to the adaptation degree between each product recommendation algorithm and each object attribute type.

[0148] 205. Cluster multiple sample portrait data to obtain multiple clustering types, and determine the target clustering type to which the target portrait data belongs.

[0149] 206. Determine the first allocation ratio of each product recommendation algorithm according to the repetition degree of the specified object attribute type corresponding to each product recommendation algorithm and the repetition degree of each target object attribute type.

[0150] 207. Determine the second allocation ratio of each product recommendation algorithm under the target clustering type.

[0151] 208. Integrate the first allocation ratio and the second allocation ratio of each product recommendation algorithm to obtain the allocation ratio of each product recommendation algorithm.

[0152] 209. According to the allocation ratio of each product recommendation algorithm, through each product recommendation algorithm, determine the target virtual product recommended to the target object among all virtual products.

[0153] The method provided by the embodiment of the present invention analyzes and processes the sample portrait data of multiple sample objects respectively, constructs a corresponding concept model of the sample object group, and gives the allocation ratio strategy of each product recommendation algorithm for the sample object group. Through the effective combination of multiple product recommendation algorithms, the existing problems of a single algorithm are avoided, so that the overall quality of the recommendation can be improved. The allocation ratio of the product recommendation algorithm for different target objects can also be trained to a more refined allocation ratio strategy formula due to the change of input parameters or the increase of mining dimensions. In addition, the method provided by the embodiment of the present invention gives a recommendation method of a hybrid product recommendation algorithm, and can also increase the number of hybrid product recommendation algorithms, thereby improving the accuracy of product recommendation.

[0154] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are sequentially shown according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0155] Based on the same inventive concept, an embodiment of the present application further provides a product recommendation device for implementing the product recommendation method involved above. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the product recommendation device provided below can refer to the limitations on the product recommendation method in the above text, and will not be repeated here.

[0156] In one embodiment, as Figure 3 shown, a product recommendation device is provided, including: a first acquisition module 301, a determination module 302, a second acquisition module 303, and a recommendation module 304, where:

[0157] The first acquisition module 301 is configured to acquire the sample portrait data of multiple sample objects. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0158] The determination module 302 is configured to determine the repeatability corresponding to each object attribute type. The repeatability is used to characterize the occurrence frequency of the valid data of each subtype of the corresponding object attribute type in different sample portrait data;

[0159] The second acquisition module 303 is configured to acquire the target portrait data of the target object, and determine the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data;

[0160] The recommendation module 304 is configured to determine the target virtual product recommended to the target object among all virtual products through each product recommendation algorithm according to the allocation ratio of each product recommendation algorithm.

[0161] In one embodiment, the determination module 302 includes:

[0162] The first determination sub-module is used to, for each object attribute type, determine the occurrence times of each subtype of the object attribute type in all sample portrait data according to whether the corresponding subtype valid data of the object attribute type is included in each sample portrait data;

[0163] The second determination sub-module is used to determine the repeatability of each object attribute type according to the occurrence times of the valid data of each subtype in each object attribute type.

[0164] In one embodiment, the second acquisition module 303 includes:

[0165] The third determination sub-module is used to determine the specified object attribute type of each product recommendation algorithm as the input data type according to the adaptation degree between each product recommendation algorithm and each object attribute type;

[0166] The fourth determination sub-module is used to determine the allocation ratio of each product recommendation algorithm according to the repeatability of the target object attribute type and the repeatability of the specified object attribute type corresponding to each product recommendation algorithm.

[0167] In one embodiment, the fourth determination sub-module includes:

[0168] The first determination unit is used to cluster multiple sample portrait data to obtain multiple clustering types and determine the target clustering type to which the target portrait data belongs;

[0169] The second determination unit is used to determine the first allocation ratio corresponding to each product recommendation algorithm according to the repeatability of the specified object attribute type corresponding to each product recommendation algorithm and the repeatability of each target object attribute type;

[0170] The third determination unit is used to determine the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type;

[0171] The integration unit is used to integrate the first allocation ratio and the second allocation ratio corresponding to each product recommendation algorithm to obtain the allocation ratio of each product recommendation algorithm.

[0172] In one embodiment, the third determination unit includes:

[0173] The acquisition subunit is used to acquire multiple second allocation ratio reference groups, and each second allocation ratio reference group includes the reference second allocation ratio corresponding to each product recommendation algorithm;

[0174] An acquisition subunit, configured to, for each second allocation ratio reference group and each target sample object belonging to the target clustering type, according to the second allocation ratio reference group, simulate and recommend virtual products to the target sample objects through various product recommendation algorithms, so as to obtain the true scores of each target sample object for each virtual product and the predicted scores of recommending each virtual product to each target sample object by various product recommendation algorithms;

[0175] A calculation subunit, configured to calculate the evaluation score of the second allocation ratio reference group according to the true scores of each target sample object for each virtual product and the predicted scores of recommending each virtual product to each target sample object by various product recommendation algorithms, where the evaluation score is used to measure the recommendation accuracy when using the second allocation ratio reference for product recommendation;

[0176] A selection subunit, configured to select a target second allocation ratio reference group from all second allocation ratio reference groups according to the evaluation score of each second allocation ratio reference group, and use the target second allocation ratio reference group as the corresponding second allocation ratio of each product recommendation algorithm under the target clustering type.

[0177] In one embodiment, the recommendation module 304 includes:

[0178] A sorting sub-module, configured to obtain the product recommendation rankings of various product recommendation algorithms for all virtual products;

[0179] A traversal sub-module, configured to traverse various product recommendation algorithms from high to low according to the allocation ratios of various product recommendation algorithms;

[0180] A fifth determination sub-module, configured to, for the currently traversed current product recommendation algorithm and the current remaining virtual products that have not been recommended among all virtual products, determine the current recommended quantity according to the allocation ratio of the current product recommendation algorithm and the total quantity of all current remaining virtual products;

[0181] A recommendation sub-module, configured to select and recommend the current recommended quantity of virtual products from all current remaining virtual products according to the corresponding product recommendation ranking of the current product recommendation algorithm, and repeat the above process until all product recommendation algorithms have been traversed.

[0182] Each module in the above product recommendation device can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor in the computer device in hardware form or be independent of it, or be stored in the memory in the computer device in software form, so that the processor can call and execute the operations corresponding to the above respective modules.

[0183] In one embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 4As shown in the figure. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. The computer program, when executed by the processor, implements a product recommendation method. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0184] Those skilled in the art can understand that Figure 4 the structure shown in the figure is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0185] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the following steps are implemented:

[0186] Obtain the sample portrait data of each of multiple sample objects. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0187] Determine the repeatability corresponding to each object attribute type. The repeatability is used to characterize the occurrence frequency of valid data of each subtype of the corresponding object attribute type in different sample portrait data;

[0188] Obtain the target portrait data of the target object. According to the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data, determine the allocation ratio of each product recommendation algorithm;

[0189] Determine target virtual products recommended to the target object among all virtual products through each product recommendation algorithm according to the allocation ratio of each product recommendation algorithm.

[0190] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0191] For each object attribute type, determine the occurrence times of each subtype of the object attribute type in all sample portrait data according to whether the corresponding subtype valid data of the object attribute type is included in each sample portrait data;

[0192] Determine the repeatability of each object attribute type according to the occurrence times of the valid data of each subtype in each object attribute type.

[0193] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0194] Determine the specified object attribute type of each product recommendation algorithm as the input data type according to the adaptation degree between each product recommendation algorithm and each object attribute type;

[0195] Determine the allocation ratio of each product recommendation algorithm according to the repeatability of the corresponding target object attribute type and the repeatability of the specified object attribute type corresponding to each product recommendation algorithm.

[0196] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0197] Cluster multiple sample portrait data to obtain multiple clustering types, and determine the target clustering type to which the target portrait data belongs;

[0198] Determine the first allocation ratio corresponding to each product recommendation algorithm according to the repeatability of the specified object attribute type corresponding to each product recommendation algorithm and the repeatability of each target object attribute type;

[0199] Determine the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type;

[0200] Integrate the first allocation ratio and the second allocation ratio corresponding to each product recommendation algorithm to obtain the allocation ratio of each product recommendation algorithm.

[0201] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0202] Obtain multiple second allocation ratio reference groups, and each second allocation ratio reference group includes the reference second allocation ratio corresponding to each product recommendation algorithm;

[0203] For each second allocation ratio reference group and each target sample object belonging to the target clustering type, according to the second allocation ratio reference group, through each product recommendation algorithm, virtual products are simulated and recommended to the target sample objects to obtain the true scores of each target sample object for each virtual product, and the predicted scores of each virtual product recommended to each target sample object by each product recommendation algorithm;

[0204] According to the true scores of each target sample object for each virtual product and the predicted scores of each virtual product recommended to each target sample object by each product recommendation algorithm, calculate the evaluation score of the second allocation ratio reference group, and the evaluation score is used to measure the recommendation accuracy when using the second allocation ratio reference for product recommendation;

[0205] According to the evaluation score of each second allocation ratio reference group, select the target second allocation ratio reference group from all second allocation ratio reference groups, and use the target second allocation ratio reference group as the corresponding second allocation ratio of each product recommendation algorithm under the target clustering type.

[0206] In one embodiment, when the processor executes the computer program, the following steps are also implemented:

[0207] Obtain the product recommendation rankings of each product recommendation algorithm for all virtual products;

[0208] Traverse each product recommendation algorithm from high to low according to the allocation ratio of each product recommendation algorithm;

[0209] For the current product recommendation algorithm currently traversed and the current remaining virtual products that have not been recommended among all virtual products, determine the current recommended quantity according to the allocation ratio of the current product recommendation algorithm and the total quantity of all current remaining virtual products;

[0210] Select and recommend the current recommended quantity of virtual products from all current remaining virtual products according to the product recommendation ranking corresponding to the current product recommendation algorithm, and repeat the above process until each product recommendation algorithm has been traversed.

[0211] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0212] Obtain the sample portrait data of multiple sample objects, each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same;

[0213] Determine the repeatability corresponding to each object attribute type, and the repeatability is used to characterize the occurrence frequency of each subtype valid data of the corresponding object attribute type in different sample portrait data;

[0214] Obtain the target portrait data of the target object, and determine the allocation ratio of each product recommendation algorithm according to the repetition degree corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data;

[0215] According to the allocation ratio of each product recommendation algorithm, through each product recommendation algorithm, determine the target virtual products recommended to the target object among all virtual products.

[0216] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0217] For each object attribute type, determine the number of occurrences of each subtype of the object attribute type in all sample portrait data according to whether the sample portrait data contains the valid data of the subtype corresponding to the object attribute type;

[0218] According to the number of occurrences of the valid data of each subtype in each object attribute type, determine the repetition degree of each object attribute type.

[0219] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0220] According to the adaptation degree between each product recommendation algorithm and each object attribute type, determine the specified object attribute type of each product recommendation algorithm as the input data type;

[0221] According to the repetition degree corresponding to the target object attribute type and the repetition degree of the specified object attribute type corresponding to each product recommendation algorithm, determine the allocation ratio of each product recommendation algorithm.

[0222] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0223] Cluster multiple sample portrait data to obtain multiple clustering types, and determine the target clustering type to which the target portrait data belongs;

[0224] According to the repetition degree of the specified object attribute type corresponding to each product recommendation algorithm and the repetition degree of each target object attribute type, determine the first allocation ratio corresponding to each product recommendation algorithm;

[0225] Determine the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type;

[0226] Integrate the first allocation ratio and the second allocation ratio corresponding to each product recommendation algorithm to obtain the allocation ratio of each product recommendation algorithm.

[0227] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0228] Obtain multiple second allocation ratio reference groups, where each second allocation ratio reference group includes the corresponding reference second allocation ratio of each product recommendation algorithm;

[0229] For each second allocation ratio reference group and each target sample object belonging to the target clustering type, according to the second allocation ratio reference group, through each product recommendation algorithm, simulate recommending virtual products to the target sample object to obtain the true scores of each target sample object for each virtual product, and the predicted scores of recommending each virtual product to each target sample object through each product recommendation algorithm;

[0230] According to the true scores of each target sample object for each virtual product and the predicted scores of recommending each virtual product to each target sample object through each product recommendation algorithm, calculate the evaluation score of the second allocation ratio reference group. The evaluation score is used to measure the recommendation accuracy when using the second allocation ratio reference for product recommendation;

[0231] According to the evaluation scores of each second allocation ratio reference group, select the target second allocation ratio reference group from all second allocation ratio reference groups, and use the target second allocation ratio reference group as the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type.

[0232] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0233] Obtain the product recommendation rankings of each product recommendation algorithm for all virtual products;

[0234] Traverse each product recommendation algorithm from high to low according to the allocation ratio of each product recommendation algorithm;

[0235] For the current product recommendation algorithm currently traversed and the current remaining virtual products that have not been recommended among all virtual products, determine the current recommended quantity according to the allocation ratio of the current product recommendation algorithm and the total quantity of all current remaining virtual products;

[0236] Select and recommend the current recommended quantity of virtual products from all current remaining virtual products according to the product recommendation ranking corresponding to the current product recommendation algorithm, and repeat the above process until each product recommendation algorithm has been traversed.

[0237] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0238] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with the relevant laws, regulations, and standards of relevant countries and regions.

[0239] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0240] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0241] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A product recommendation method, characterized in that, The method includes: Obtaining sample portrait data of multiple sample objects, each sample portrait data covering multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data being the same; Determining the repetition degree corresponding to each object attribute type, where the repetition degree is used to characterize the occurrence frequency of valid data of each subtype of the corresponding object attribute type in different sample portrait data; Obtaining target portrait data of a target object, and determining the allocation ratio of each product recommendation algorithm according to the repetition degree corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data; According to the allocation ratio of each product recommendation algorithm, determining, through each product recommendation algorithm, target virtual products recommended to the target object among all virtual products; The determining the allocation ratio of each product recommendation algorithm according to the repetition degree corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data includes: Determining, according to the adaptation degree between each product recommendation algorithm and each object attribute type, the specified object attribute type that each product recommendation algorithm uses as the input data type; Clustering multiple sample portrait data to obtain multiple clustering types, and determining the target clustering type to which the target portrait data belongs; determining the first allocation ratio corresponding to each product recommendation algorithm according to the repetition degree of the specified object attribute type corresponding to each product recommendation algorithm and the repetition degree of each target object attribute type; determining the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type; and integrating the first allocation ratio and the second allocation ratio corresponding to each product recommendation algorithm to obtain the allocation ratio of each product recommendation algorithm.

2. The method according to claim 1, wherein The determining the repetition degree corresponding to each object attribute type includes: For each object attribute type, determining the occurrence times of each subtype of the object attribute type in all sample portrait data according to whether the corresponding subtype valid data of the object attribute type is included in each sample portrait data; Determining the repetition degree of each object attribute type according to the occurrence times of the valid data of each subtype in each object attribute type.

3. The method according to claim 1, wherein The determining the second allocation ratio corresponding to each product recommendation algorithm under the target clustering type includes: Obtaining multiple second allocation ratio reference groups, each second allocation ratio reference group including the reference second allocation ratio corresponding to each product recommendation algorithm; For each second allocation ratio reference group and each target sample object belonging to the target clustering type, simulating the recommendation of virtual products to the target sample object according to the second allocation ratio reference group through each product recommendation algorithm, so as to obtain the true scores of each target sample object for each virtual product and the predicted scores of each virtual product recommended to each target sample object by each product recommendation algorithm; Calculate the evaluation score of the second allocation ratio reference group based on the true scores of each target sample object for each virtual product and the predicted scores of each virtual product recommended for each target sample object by each product recommendation algorithm. The evaluation score is used to measure the recommendation accuracy when using the second allocation ratio reference for product recommendation; Select the target second allocation ratio reference group from all second allocation ratio reference groups according to the evaluation score of each second allocation ratio reference group, and use the target second allocation ratio reference group as the corresponding second allocation ratio of each product recommendation algorithm under the target clustering type.

4. The method according to claim 1, characterized in that, Determine the target virtual products recommended to the target object among all virtual products through each product recommendation algorithm according to the allocation ratio of each product recommendation algorithm, including: Obtain the product recommendation rankings of each product recommendation algorithm for all virtual products; Traverse each product recommendation algorithm from high to low according to the allocation ratio of each product recommendation algorithm; For the current product recommendation algorithm currently traversed and the current remaining virtual products that have not been recommended among all virtual products, determine the current recommended quantity according to the allocation ratio of the current product recommendation algorithm and the total quantity of all current remaining virtual products; Select the current recommended quantity of virtual products from all current remaining virtual products according to the product recommendation ranking corresponding to the current product recommendation algorithm and recommend them. Repeat the above process until each product recommendation algorithm has been traversed.

5. A product recommendation device, characterized in that, The device includes: A first acquisition module for acquiring the sample portrait data of multiple sample objects. Each sample portrait data covers multiple object attribute types, and the coverage ranges of the object attribute types corresponding to different sample portrait data are the same; A determination module for determining the repeatability corresponding to each object attribute type. The repeatability is used to characterize the occurrence frequency of each subtype of valid data of the corresponding object attribute type in different sample portrait data; A second acquisition module for acquiring the target portrait data of the target object, and determining the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to each object attribute type and the target object attribute type corresponding to the valid data included in the target portrait data; A recommendation module for determining the target virtual products recommended to the target object among all virtual products through each product recommendation algorithm according to the allocation ratio of each product recommendation algorithm; The second acquisition module includes: A third determination sub-module for determining the specified object attribute type of each product recommendation algorithm as the input data type according to the adaptation degree between each product recommendation algorithm and each object attribute type; A fourth determination sub-module for determining the allocation ratio of each product recommendation algorithm according to the repeatability corresponding to the target object attribute type and the repeatability of the specified object attribute type corresponding to each product recommendation algorithm; The fourth determination sub-module includes: A first determination unit for clustering multiple sample portrait data to obtain multiple clustering types and determining the target clustering type to which the target portrait data belongs; A second determination unit, configured to determine a first allocation ratio corresponding to each product recommendation algorithm according to the repetition degree of the specified object attribute type corresponding to each product recommendation algorithm and the repetition degree of each target object attribute type; A third determination unit, configured to determine a second allocation ratio corresponding to each product recommendation algorithm under the target clustering type; An integration unit, configured to integrate the first allocation ratio and the second allocation ratio corresponding to each product recommendation algorithm to obtain the allocation ratio of each product recommendation algorithm.

6. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 4 are implemented.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images