CN116911985A

CN116911985A - Product recommendation method, device, equipment and storage medium

Info

Publication number: CN116911985A
Application number: CN202310913358.2A
Authority: CN
Inventors: 邵玉杰
Original assignee: Bank of China Ltd
Current assignee: Bank of China Ltd
Priority date: 2023-07-24
Filing date: 2023-07-24
Publication date: 2023-10-20

Abstract

The application provides a product recommendation method, device, equipment and storage medium, which can be used in the financial field or other fields. The method comprises the following steps: aiming at a first user group of each type of product, determining the similarity of any two users according to the historical behavior data of each user; obtaining similarity ranking of any user and other users; determining similar users of any user according to the similarity ranking of any user; determining the product which is not purchased by the user in the product types corresponding to any user as a product to be recommended; determining the interest degree value of any user to any product to be recommended according to the user set of any product to be recommended purchased, the similarity between other users in the first user group and any user, and the similarity between any user and any user; and determining target products from all the products to be recommended according to the interest degree value of any user on each product to be recommended, and recommending the target products to any user. The scheme improves the accuracy and the customer range of product recommendation.

Description

Product recommendation method, device, equipment and storage medium

Technical Field

The present application relates to the financial field or other fields, and in particular, to a product recommendation method, apparatus, device, and storage medium.

Background

The loan product is one of the most important financial services provided by the bank, can effectively meet the fund requirements of borrowers and the accumulated fund requirements of the bank, and can effectively improve the fund condition of the borrowers. However, different loan products have different characteristics and are suitable for different people. Therefore, how to recommend proper loan products to borrowers is a highly desirable issue.

At present, when recommending loan products to borrowers, mainly when the borrowers go to the counter on line to transact business, sales staff determine the loan products suitable for the borrowers according to experience, introduce the loan products to the borrowers in a verbal marketing mode, and promote the borrowers to purchase.

However, the prior art recommends loan products with lower accuracy and a smaller customer range.

Disclosure of Invention

The application provides a product recommending method, device, equipment and storage medium, which are used for solving the problems of low accuracy and small client range of recommending loan products in the prior art.

In a first aspect, the present application provides a product recommendation method, including:

determining a first user group meeting purchase conditions for each type of product;

for each first user group, determining the similarity of any two users according to the historical behavior data of each user;

Aiming at any user in each first user group, arranging other users in the first user group according to the sequence of the similarity with the any user from high to low, and obtaining the similarity ranking of the any user;

for any user in each first user group, determining the users with the preset number of names in the similarity ranking as similar users of any user according to the similarity ranking of the any user;

for any user in each first user group, determining products which are not purchased by any user in the product types corresponding to the any user as products to be recommended;

determining the interest degree value of any user to any product to be recommended according to the user set of any product to be recommended purchased, the similarity between other users in the first user group and any user and the similar users of any user aiming at any user in each first user group and any product to be recommended corresponding to any user;

determining target products from all the products to be recommended according to the interest degree value of any user to each product to be recommended aiming at any user in each first user group;

Recommending the target product to any user in each first user group aiming at the any user.

In a second aspect, the present application provides a product recommendation device, comprising:

the first processing module is used for determining a first user group meeting purchasing conditions for each type of product;

the second processing module is used for determining the similarity of any two users according to the historical behavior data of each user aiming at each first user group;

the third processing module is used for aiming at any user in each first user group, arranging other users in the first user group according to the sequence of the similarity with the any user from high to low, and obtaining the similarity ranking of the any user;

a fourth processing module, configured to determine, for any user in each first user group, a user in a preset number of names in the similarity rank as a similar user of the any user according to the similarity rank of the any user;

a fifth processing module, configured to determine, for any user in each first user group, a product that is not purchased by the any user in a product category corresponding to the any user as a product to be recommended;

A sixth processing module, configured to determine, for any user in each first user group and any product to be recommended corresponding to the any user, a value of interest degree of the any user on the any product to be recommended according to a user set that has purchased the any product to be recommended, a similarity between other users in the first user group and the any user, and a similar user of the any user;

a seventh processing module, configured to determine, for any user in each first user group, a target product from all the products to be recommended according to the interest level value of the any user for each product to be recommended;

and the recommending module is used for recommending the target product to any user in each first user group aiming at any user.

In a third aspect, the present application provides an electronic device comprising: a processor, and a memory communicatively coupled to the processor;

the memory stores computer-executable instructions;

the processor executes computer-executable instructions stored in the memory to implement the method provided by the first aspect and each possible design.

In a fourth aspect, the application provides a computer-readable storage medium having stored therein computer-executable instructions which, when executed by a processor, are adapted to carry out the method provided by the first aspect and each possible design.

The application provides a product recommendation method, a device, equipment and a storage medium, wherein the method comprises the following steps: determining a first user group meeting purchase conditions for each type of product; for each first user group, determining the similarity of any two users according to the historical behavior data of each user; aiming at any user in each first user group, arranging other users in the first user group according to the sequence of the similarity with any user from high to low to obtain the similarity ranking of any user; for any user in each first user group, determining the users with the preset number of names in the similarity ranking as similar users of any user according to the similarity ranking of any user; for any user in each first user group, determining products which are not purchased by any user in the product types corresponding to any user as products to be recommended; determining the interest degree value of any user to any product to be recommended according to the user set of any product to be recommended purchased, the similarity between other users in the first user group and any user and the similar users of any user aiming at any user in each first user group and any product to be recommended corresponding to any user; determining target products from all the products to be recommended according to the interest degree value of any user to each product to be recommended aiming at any user in each first user group; recommending target products to any user in each first user group. According to the technical scheme, the product recommendation is carried out on the crowd meeting the same purchasing condition, so that risks are avoided to a certain extent. Secondly, the scheme is established on the basis that each user has transacted the product, has transacted records of the product and has high evaluation on the product, and the reliability of recommendation is improved. Finally, the technical scheme uses a collaborative filtering algorithm to find out similar users according to the similarity of the users, and recommends users with high similarity to each other, so that the reliability of recommendation is improved, and the determined target product is more persuasive.

Drawings

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

Fig. 1 is a schematic flow chart of a first embodiment of a product recommendation method according to an embodiment of the present application;

fig. 2 is a schematic flow chart of a second embodiment of a product recommendation method according to the embodiment of the present application;

fig. 3 is a schematic structural diagram of a product recommendation device according to an embodiment of the present application;

fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Specific embodiments of the present application have been shown by way of the above drawings and will be described in more detail below. The drawings and the written description are not intended to limit the scope of the inventive concepts in any way, but rather to illustrate the inventive concepts to those skilled in the art by reference to the specific embodiments.

Detailed Description

Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples do not represent all implementations consistent with the application. Rather, they are merely examples of apparatus and methods consistent with aspects of the application as detailed in the accompanying claims.

It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with related laws and regulations and standards, and provide corresponding operation entries for the user to select authorization or rejection.

It should be noted that the product recommendation method, device, equipment and storage medium provided by the application can be used in the financial field and can also be used in any field except the financial field, and the application fields of the product recommendation method, device, equipment and storage medium are not limited by the application.

First, the terms related to the present application will be described:

collaborative filtering algorithm (English: userCF): in short, the preference of a community of interest, which has a common experience, is used to recommend information to users in this community that they may be interested in. For example, when user a needs personalized recommendation, other users having similar interests to user a can be found first, and then those other users like, but products that user a has not heard are recommended to a, which is called a collaborative filtering algorithm based on user.

Next, a specific application scenario of the present application is described:

the method takes the customer demand as a starting point, and is an important ring in enterprise work for accurately recommending products for users, so that service conversion can be effectively improved, transaction is promoted, and the awareness of enterprises is realized and improved. Taking banking as an example, loan products are one of the most important financial services provided by banks, can help banks accumulate funds, can improve the funds condition of borrowers, and can solve the funds requirements of borrowers. At present, the way to recommend loan products for borrowers is mainly by the sales staff orally marketing the loan products when the borrowers transact business.

However, the above manner has the following technical problems:

1. such screening marketing is not targeted, resulting in a low success rate of recommendations.

2. Sales personnel can only contact clients who transact business, and the popularization of the sales personnel is narrow and not enough.

3. The loan products recommended by sales personnel are often determined according to own experience or task indexes formulated by companies, and the accuracy is not high.

In summary, the prior art has the problems of low accuracy of recommending loan products and small customer range.

Based on the above, the application provides a product recommendation method, which can classify users according to each type of products, determine user groups meeting the purchase conditions of each type of products, and then calculate the similarity between any two users in each user group according to the historical behavior data of the users. Therefore, according to the historical behavior data of the users, the products purchased and loved by any user can be recommended to other users who have high similarity and do not purchase the products. By way of example, assuming that the two users AB have extremely high similarity through calculation, the user A can recommend products which are loved by the user A but not purchased by the user B to the user B, so that accuracy of product recommendation is effectively improved, and the range of clients is enlarged.

The following describes the technical scheme of the present application and how the technical scheme of the present application solves the above technical problems in detail with specific embodiments. The following embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. Embodiments of the present application will be described below with reference to the accompanying drawings.

Fig. 1 is a schematic flow chart of a first embodiment of a product recommendation method according to an embodiment of the present application. As shown in fig. 1, the product recommendation method may include the steps of:

s101, determining a first user group meeting purchase conditions for each type of product.

The execution body of the embodiment of the application can be electronic equipment or a product recommendation device arranged in the electronic equipment. The product recommendation device may be implemented by software or by a combination of software and hardware. For ease of understanding, hereinafter, an execution body will be described as an example of an electronic device.

In this step, since each product has its own purchase condition, only the user who satisfies the purchase condition can purchase the product. Therefore, the products can be classified according to the purchase conditions, each type of product has the same purchase conditions, and then the users are classified according to the purchase conditions of each type of product, so that the first user group corresponding to each type of product meets the purchase conditions of the type of product. Therefore, the crowd can be roughly classified, the similarity of the users in each first user group is higher, the similarity of the users in different first user groups is lower, and the similar users are determined only from the same first user group, so that the calculation amount is greatly reduced.

Illustratively, the loan products of a bank are exemplified as different kinds, for example, a small amount loan, a medium amount loan, and a large amount loan, according to the loan amount. The purchase conditions corresponding to different types of loan products are loan amount, that is, the loan amount corresponding to the type of loan products is satisfied by comprehensively evaluating the credit, running water, property and the like of the user, that is, the purchase conditions of the type of loan are satisfied.

S102, determining the similarity of any two users according to the historical behavior data of each user aiming at each first user group.

In this step, after determining the first user group of each product, the similarity between each user and other users needs to be calculated in each first user group to determine the users similar to the user. The similarity between the two users is determined mainly according to historical behavior data of the users. And determining that each user has a product which is fed back forward through historical behavior data of the users, so as to determine the similarity between the users.

The historical behavior data of the user can be product data which is purchased by the user and fed back forward. By way of example, the historical behavioral data may be reviews of the product by the user in the purchase order, or evaluation articles written by the user on the product in a network platform, or the like. Alternatively, a product with positive feedback may be understood as a product that is liked by the user.

Illustratively, for each first user group, the favorite products of the respective users may be represented by table 1.

TABLE 1

User' s	Product(s)
		A	a、b、d、e
B	a、c、d
		C	a、b、e
D	c、d、e

As shown in table 1, where A, B, C, D in uppercase represents users in the same user group, a, b, c, d, e in lowercase is the same kind of product. Referring to table 1, user a likes product a, product b, product d, product e; user C likes product a, product b, product e.

Alternatively, the formula may be based on:and calculating the similarity between any two users in the first user group. Wherein u and v are any two users, w, in the first user group _uv For the similarity of user u and user v, N _(u) For the product set with user u having had forward feedback, N _(v) There is a product set that has been forward fed back for user v.

Illustratively, in connection with table 1, the favorite product set of user a is { a, B, d }, and the favorite product set of user B is { a, c }, then the similarity between user a and user B is calculated by the above formula:

s103, aiming at any user in each first user group, arranging other users in the first user group according to the sequence of the similarity with any user from high to low, and obtaining the similarity rank of any user.

In this step, after the similarity between any two users in the first user group is calculated, the similar user of each user may be determined. When determining the similarity of any user, the other users in the first user group need to be ranked according to the similarity of the other users in the first user group, and the similarity ranking is obtained, wherein the users in the similarity ranking are ranked according to the sequence from high to low of the similarity of the users with any user.

Illustratively, the similarity between users in the first user group may be represented by table 2:

TABLE 2

As shown in table 2, A, B, C, D in the first row and first column represent users in the same user group. From the similarities between the users shown in table 2, a corresponding similarity rank for each user is determined. Wherein, the similarity ranking of user a is: user B- (user C, user D), the similarity rank of user B is: user A-user D-user C, user C's similarity rank is: user a- (user B, user D), the similarity rank of user D is: user B-user a-user C. Wherein the ranking of the users in brackets is the same, that is, the second names are juxtaposed.

S104, aiming at any user in each first user group, according to the similarity rank of any user, determining the users with the preset number of names in the similarity rank as similar users of any user.

In this step, since the similarity ranking of any user is obtained by sorting the similarity between other users and the user in the order from high to low, the users with the preset number of names ranked before are users with high similarity to any user. Thus, after the similarity rank of any user is obtained, the users that are at the top preset number of names may be determined to be similar users of any user.

The preset number may be positively correlated with the number of users in the first user group, that is, the larger the number of users in the first user group, the larger the preset number.

It should be appreciated that assuming that there are multiple users under the same ranking in the similarity ranking, then multiple users are determined to be similar users for either user at the same time.

For example, referring to table 2, assuming that the preset number is 2, for any user in each first user group, according to the similarity ranking of any user, the users in the first two in the similarity ranking are determined as similar users of any user. That is, the similar user set of user A is { B, C, D }, the similar user set of user B is { A, D }, the similar user set of user C is { A, B, D }, and the similar user set of user D is { A, B }.

S105, determining products which are not purchased by any user in the product types corresponding to any user as products to be recommended for any user in each first user group.

In this step, since the users have been categorized according to the purchase conditions of different kinds of products in the foregoing, that is, each user corresponds to one or more kinds of products. Therefore, the product which is not purchased by the user in the product category corresponding to the user can be determined to be the product to be recommended according to the product which is already purchased by the user, so that the target product which can be recommended to the user can be obtained from the product to be recommended later.

S106, determining the interest degree value of any user to any product to be recommended according to the user set of any product to be recommended purchased, the similarity between other users in the first user group and any user and the similarity between the other users in the first user group and any user aiming at any user in each first user group and any product to be recommended corresponding to any user.

In this step, for any user and any product to be recommended corresponding to the user, the interest level value of any user on any product to be recommended may be determined according to the user set that purchased any product to be recommended, the similarity between other users in the first user group and any user, and the similarity between any user, so that whether the any product to be recommended can be recommended to any user is determined according to the interest level value.

In one possible implementation manner, for any user in each first user group and any product to be recommended corresponding to any user, according to the formula:p(u,i)＝∑ _v∈S(u,k) ∩N(i)w _uv and determining the interest degree value of any user on any product to be recommended. Wherein i is a product to be recommended, u is any user, p (u, i) is the interest degree value of the user u on the product i, S (u, k) is a similar user of the user u, N (i) is a user set after the product i is purchased, and v is a user in the first user group.

In another possible implementation manner, for any user in each first user group and any product to be recommended corresponding to any user, according to the formula: p (u, i) = Σ _v∈S(u,k) ∩N(i)w _uv r _vi And determining the interest degree value of any user on any product to be recommended. Wherein r is _vi Interest to user v in product i. Since the implicit feedback data of a single behavior is used in this implementation, r _vi ＝1。

By way of example, referring to tables 1 and 2, it can be seen that user a has no purchasing behavior for product c (i.e., the product to be recommended), the user most similar to user a is BCD, and the user having positive feedback behavior for product c is BD. Thus, the users who purchased the product c (user B, user D) are selected from the similar users of the user a, and then the similarity between the user a and them is calculated and summed, while r is set _vi ＝1：

Likewise, user B has no over-purchase for product B and product e, user C has no over-purchase for product C and product D, user D has no over-purchase for product a and product B, and so on. It can be derived that:

s107, determining target products from all the products to be recommended according to the interest degree value of any user for each product to be recommended aiming at any user in each first user group.

In this step, the higher the value of the degree of interest, the more likely the representative user is interested in the product. Therefore, according to the interest degree value of each product to be recommended by any user, the product with the high interest degree value can be determined as the target product.

In one possible implementation, S107 may be implemented by steps (1) to (3):

and (1) determining the maximum interested degree value from the interested degree value of each user to each product to be recommended aiming at any user in each first user group.

In order to improve the accuracy of the determined target product, a threshold level of interest may be set according to the level of interest value of any user for each product to be recommended. For example, the level of interest threshold may be determined from a maximum level of interest value.

And (2) multiplying the maximum interesting degree value by a product of preset proportions to determine the interesting degree threshold value.

In the step (2), the preset proportion can be 3/4, 2/3, 1/2 and the like, and can be determined according to actual requirements, and the larger the preset proportion is, the smaller the number of target products is determined subsequently, and the higher the accuracy is; otherwise, the more the number of target products is determined later, the lower the accuracy is.

And (3) determining the product to be recommended, of which the interest degree value is larger than the interest degree threshold value, as a target product for any user in each first user group.

For example, referring to tables 1 and 2, assuming that the preset ratio is 3/4, for user a, product c is the target product; for the user B, the product e is the target product; for user C, product D is the target product, and for user D, product a is the target product.

S108, recommending target products to any user in each first user group.

In this step, after determining the target product of each user, the target product is recommended to the user to promote the transaction of the product.

In one possible implementation manner, for any user in each first user group, recommendation information of the target product is sent to terminal equipment of any user. Specifically, the recommendation information can be pushed to the terminal equipment of the user in the form of push information of the application program, so that the recommended client range is enlarged, and the success rate of recommendation is indirectly increased.

The S102-S108 may be implemented by an improved collaborative filtering algorithm, that is, the specific implementation process of the improved collaborative filtering algorithm is the specific content in S102-S108, which is not described herein.

According to the product recommendation method provided by the embodiment of the application, aiming at each type of product, a first user group meeting the purchase condition is determined; for each first user group, determining the similarity of any two users according to the historical behavior data of each user; aiming at any user in each first user group, arranging other users in the first user group according to the sequence of the similarity with any user from high to low to obtain the similarity ranking of any user; for any user in each first user group, determining the users with the preset number of names in the similarity ranking as similar users of any user according to the similarity ranking of any user; for any user in each first user group, determining products which are not purchased by any user in the product types corresponding to any user as products to be recommended; determining the interest degree value of any user to any product to be recommended according to the user set of any product to be recommended purchased, the similarity between other users in the first user group and any user and the similar users of any user aiming at any user in each first user group and any product to be recommended corresponding to any user; determining target products from all the products to be recommended according to the interest degree value of any user to each product to be recommended aiming at any user in each first user group; recommending target products to any user in each first user group. According to the technical scheme, the product recommendation is carried out on the crowd meeting the same purchasing condition, so that risks are avoided to a certain extent. Secondly, the scheme is established on the basis that each user has transacted the product, has transacted records of the product and has high evaluation on the product, and the reliability of recommendation is improved. Then, the technical scheme uses a collaborative filtering algorithm to find out similar users according to the similarity of the users, and mutually recommends users with high similarity, so that the reliability of recommendation is improved, and the determined target product is more persuasive and interpretable. Finally, the technical scheme does not need to be manually processed, and manpower and material resources are effectively saved.

In practice, many users do not have positive feedback behavior with respect to each other for the same product, i.e., many times |N _(u) ∩N _(v) The similarity between these users is 0, |=0. In order to save time and resources for computing the similarity of the users and reduce the computation complexity, it may be determined that the first user group satisfies |N _(u) ∩N _(v) The user pair of |=0, and the similarity between other users except the user pair is calculated, and the specific implementation process can refer to the following embodiments.

Based on the embodiment shown in fig. 1, fig. 2 is a schematic flow chart of a second embodiment of a product recommendation method according to the embodiment of the present application. As shown in fig. 2, the step S102 may include the following steps:

s201, determining a second user group with forward feedback to any product according to historical behavior data of the user who purchases any product aiming at any product in each product type.

In this step, in order to simplify the calculation process of S102 and reduce the user range of the calculation similarity, it is necessary to determine that each user has a product that is fed back in forward direction, so that the user pairs that do not generate positive feedback actions on the same product in the first user group are determined later.

In one possible implementation manner, S201 may be implemented through steps (4) to (5):

and (4) establishing an inverted list from the product to the user according to the historical behavior data of the user who purchases any product aiming at each type of product.

Illustratively, the inverted list is used to maintain a list of users who have not performed positive feedback on the product, and based on table 1, the inverted list can be represented by table 3:

TABLE 3 Table 3

Product(s)	User' s
		a	A、B、C
b	A、C
		c	B、D
d	A、B、D
		e	A、C、D

Referring to table 3, the left column is the same type of product and the right column is the second group of users who like each product.

And (5) determining a second user group with forward feedback to any product according to the inverted list aiming at any product of each product type.

S202, determining the number of second user groups simultaneously containing any two users according to the second user groups of each product corresponding to the first user groups for any two users in each first user group.

In this step, after determining the second user group of each product, it may be determined whether any two users in the first user group like at least one product at the same time, that is, determine the number of second user groups including any two users at the same time, and if the number is 0, it represents that any two users do not like products at the same time.

In one possible implementation, for any two users of each first user group, according to the formula: c (C) _[u][v] ＝|N _(u) ∩N _(v) And determining the number of the second user groups simultaneously containing any two users.

Wherein u and v are any two users in the first user group, C _[u][v] N being the number of second user groups comprising user u and user v _(u) For a second group of users comprising user u, N _(v) Is the second user group comprising user v.

For example, assuming that user u and user v belong to the second user group corresponding to K products in Table 3 at the same time, then there is C _[u][v] ＝K。

For example, based on table 3, the number of the second user groups including any two users at the same time obtained by performing the calculation according to the above formula may be represented by table 4.

Table 4:

	A	B	C	D
					A	0	2	3	1
B	2	0	1	2
					C	3	1	0	1
D	1	2	1	0

as shown with reference to table 4, A, B, C, D in the first column and first row are users in the first user group. As can be seen from table 4, there are 2 second user groups containing both user a and user B, and 1 second user group containing both user C and user B.

S203, calculating the similarity of two users with the number larger than 0 in the first user group for each first user group.

In this step, if the number of second user groups including any two users is 0 for any two users of each first user group, it is indicated that the similarity between the two users is 0, and subsequent calculation is not required. That is, we only need to calculate the similarity between two users who have commonly liked products in the first user group, wherein the number of the second user groups containing the two users is greater than 0.

In one possible implementation, for each first user group, according to the formula:and calculating the similarity of two users with the number larger than 0 in the first user group, wherein u and v are the two users with the number larger than 0 in the first user group.

In the embodiment, the inverted list of the favorite products of the users is established, and then the similarity between the users with the commonly favorite products in the first user group is calculated according to the cosine similarity formula, so that similar users are found, the calculated amount and the calculated time are effectively reduced, and the calculation resources are saved.

Based on the product recommendation method shown in the above-described embodiment, the method is illustrated next by way of one example.

Specifically, the product recommendation method may include the steps of:

user data collection: the transaction records of the users and the feedback of each user on the product are also recorded, and the data format is from web heterogeneous data, which is the data source of the technical scheme.

Data extraction and integration: and the handling records of all users are also processed by each user to sort the feedback of the products, the products with good user feedback are selected, the products with poor experience are eliminated, meanwhile, the collected data are required to be refined and integrated, integrated into the form of an EXCEL table and compressed, and the subsequent operation of the data is convenient.

Classification: and classifying the users aiming at each type of product to obtain the user group of each type of product.

Personalized recommendation is carried out based on a collaborative filtering algorithm of the user: and calculating the similarity between users by using a cosine similarity algorithm, screening out user pairs which do not generate positive feedback actions on the same product by using an inverted list, calculating the interest degree of each user on the products which do not have positive feedback actions by using a collaborative filtering algorithm of the users, and finally setting an interest degree threshold according to the integral interest degree to conduct personalized recommendation, recommending the products exceeding the threshold, and not recommending the products exceeding the threshold. Meanwhile, the recommended result can be stored in a database for storage.

The classification and personalized recommendation process based on the collaborative filtering algorithm of the user can be realized through a model, namely, the product recommendation method can be integrated in one model in advance, and the purpose of recommending the product to the user can be realized only by decompressing the EXCEL form and introducing the EXCEL form into the model in batches.

The method has perfect steps, collects information from the user behavior records, extracts the information and classifies the user groups, calculates the information by a collaborative filtering algorithm, and finally sets a threshold value according to the interest degree of the user product so as to recommend the information, thereby realizing full-flow coverage. Can help vast clients with difficult selection to handle loan products suitable for application. By using the scheme to recommend loan products, the recommendation accuracy can be greatly improved. Because the algorithm used by the application is the personalized recommendation according to the behavior generated by each user, the method has strong interpretability, and the obtained recommendation result is very reliable. And (3) regularly maintaining a back algorithm by a programmer, regularly training the model by using fresh data, and displaying a recommended result by a service person. Programmers and operators do not need to do a great deal of work, so that the work efficiency is improved in a straight line, and meanwhile, the method is very simple and practical, and the loss of human resources and financial resources of each unit can be reduced to a great extent, so that potential intangible benefits are brought.

The following are examples of the apparatus of the present application that may be used to perform the method embodiments of the present application. For details not disclosed in the embodiments of the apparatus of the present application, please refer to the embodiments of the method of the present application.

Fig. 3 is a schematic structural diagram of a product recommendation device according to an embodiment of the present application. As shown in fig. 3, the product recommendation device 300 includes:

a first processing module 301, configured to determine, for each product type, a first user group that satisfies a purchase condition;

a second processing module 302, configured to determine, for each first user group, a similarity of any two users according to historical behavior data of each user;

the third processing module 303 is configured to, for any user in each first user group, rank other users in the first user group in order from high to low in similarity with any user, and obtain a similarity ranking of any user;

a fourth processing module 304, configured to determine, for any user in each first user group, a user with a preset number of names in the similarity ranking as a similar user of any user according to the similarity ranking of any user;

a fifth processing module 305, configured to determine, for any user in each first user group, a product that is not purchased by any user in the product category corresponding to any user as a product to be recommended;

A sixth processing module 306, configured to determine, for any user in each first user group and any product to be recommended corresponding to any user, a value of interest degree of any user for any product to be recommended according to a user set of any product to be recommended purchased, similarity between other users in the first user group and any user, and similar users of any user;

a seventh processing module 307, configured to determine, for any user in each first user group, a target product from all the products to be recommended according to the interest level value of any user for each product to be recommended;

a recommending module 308, configured to recommend a target product to any user in each first user group.

In one possible implementation, the second processing module 302 is specifically configured to:

and determining a second user group with forward feedback on any product according to the historical behavior data of the user who purchased any product aiming at any product in each product type.

And determining the number of the second user groups simultaneously containing any two users according to the second user groups of each product corresponding to the first user groups aiming at any two users in each first user group.

And calculating the similarity of two users with the number larger than 0 in the first user group aiming at each first user group.

for any two users of each first user group, according to the formula: c (C) _[u][v] ＝|N _(u) ∩N _(v) The number of the second user groups simultaneously containing any two users is determined; wherein u and v are any two users in the first user group, C _[u][v] N being the number of second user groups comprising user u and user v _(u) For a second group of users comprising user u, N _(v) Is the second user group comprising user v.

for each first user group, according to the formula:calculating the similarity of two users with the number greater than 0 in the first user group, wherein u and v are the number of the first user groupTwo users at 0, w _uv The similarity between user u and user v.

In one possible implementation, the seventh processing module 307 is specifically configured to:

and determining the maximum interest degree value from the interest degree values of the users to the recommended products according to any user in each first user group.

And multiplying the maximum interest degree value by a product of preset proportions to determine the interest degree threshold value.

And determining the product to be recommended, of which the interest degree value is larger than the interest degree threshold value, as a target product for any user in each first user group.

In one possible implementation, the recommendation module 308 is specifically configured to:

and for any user in each first user group, sending the recommendation information of the target product to the terminal equipment of any user.

In one possible implementation, the sixth processing module 306 is specifically configured to:

for any user in each first user group and any product to be recommended corresponding to any user, according to the formula: p (u, i) = Σ _v∈S(u,k) ∩N(i)w _uv And determining the interest degree value of any user on any product to be recommended. Wherein i is a product to be recommended, u is any user, p (u, i) is the interest degree value of the user u on the product i, S (u, k) is a similar user of the user u, N (i) is a user set after the product i is purchased, and v is a user in the first user group.

and establishing an inverted list from the product to the user according to the historical behavior data of the user who purchases any product aiming at each type of product.

And determining a second user group with forward feedback for any product according to the inverted list aiming at any product of each product type.

The product recommendation device provided by the embodiment of the application can be used for executing the product recommendation method in any of the above embodiments, and the implementation principle and the technical effect are similar, and are not repeated here.

It should be noted that, it should be understood that the division of the modules of the above apparatus is merely a division of a logic function, and may be fully or partially integrated into a physical entity or may be physically separated. And these modules may all be implemented in the form of software calls through the processing elements. Or may be implemented entirely in hardware. The method can also be realized in a form of calling software by a processing element, and the method can be realized in a form of hardware by a part of modules. In addition, all or part of the modules may be integrated together or may be implemented independently. The processing element here may be an integrated circuit with signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in a software form.

Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application. As shown in fig. 4, the electronic device 400 may include: the product recommendation method provided in any of the foregoing embodiments is implemented by the processor 401, the memory 402, and computer program instructions stored in the memory 402 and executable on the processor 401, when the processor 401 executes the computer program instructions.

Alternatively, the above-mentioned devices of the electronic apparatus 400 may be connected by a system bus.

The memory 402 may be a separate memory unit or may be a memory unit integrated into the processor. The number of processors is one or more.

Optionally, the electronic device 400 may also include interfaces to interact with other devices.

It is to be appreciated that the processor 401 may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method disclosed in connection with the present application may be embodied directly in a hardware processor for execution, or in a combination of hardware and software modules in a processor for execution.

The system bus may be a peripheral component interconnect standard (peripheral component interconnect, PCI) bus or an extended industry standard architecture (extended industry standard architecture, EISA) bus, among others. The system bus may be classified into an address bus, a data bus, a control bus, and the like. For ease of illustration, the figures are shown with only one bold line, but not with only one bus or one type of bus. The memory may include random access memory (random access memory, RAM) and may also include non-volatile memory (NVM), such as at least one disk memory.

All or part of the steps for implementing the method embodiments described above may be performed by hardware associated with program instructions. The foregoing program may be stored in a readable memory. The program, when executed, performs steps including the method embodiments described above. And the aforementioned memory (storage medium) includes: read-only memory (ROM), RAM, flash memory, hard disk, solid state disk, magnetic tape, floppy disk, optical disk (optical disc), and any combination thereof.

The electronic device provided by the embodiment of the application can be used for executing the product recommendation method provided by any of the method embodiments, and the implementation principle and the technical effect are similar, and are not repeated here.

Embodiments of the present application provide a computer-readable storage medium having stored therein computer-executable instructions that, when executed on a computer, cause the computer to perform the above-described product recommendation method.

The computer readable storage medium described above may be implemented by any type of volatile or non-volatile memory device or combination thereof, such as static random access memory, electrically erasable programmable read-only memory, magnetic memory, flash memory, magnetic disk or optical disk. A readable storage medium can be any available medium that can be accessed by a general purpose or special purpose computer.

In the alternative, a readable storage medium is coupled to the processor such that the processor can read information from, and write information to, the readable storage medium. In the alternative, the readable storage medium may be integral to the processor. The processor and the readable storage medium may reside in an application specific integrated circuit (Application Specific Integrated Circuits, ASIC). The processor and the readable storage medium may reside as discrete components in a device.

Embodiments of the present application also provide a computer program product, where the computer program product includes a computer program, where the computer program is stored in a computer readable storage medium, where at least one processor may read the computer program from the computer readable storage medium, and where the at least one processor may implement the product recommendation method when executing the computer program. Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. This application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the application pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.

It is to be understood that the application is not limited to the precise arrangements and instrumentalities shown in the drawings, which have been described above, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the application is limited only by the appended claims.

Claims

1. A method of product recommendation, comprising:

2. The method of claim 1, wherein determining, for each first group of users, a similarity of any two users based on historical behavioral data for each user, comprises:

determining a second user group with forward feedback on any product according to historical behavior data of a user who purchases the any product aiming at any product in each product type;

for any two users in each first user group, determining the number of second user groups simultaneously containing any two users according to the second user groups of each product corresponding to the first user groups;

and calculating the similarity of the two users with the number larger than 0 in the first user group aiming at each first user group.

3. The method according to claim 2, wherein the determining, for any two users in each first user group, the number of second user groups simultaneously including the any two users according to the second user group of each product corresponding to the first user group includes:

4. A method according to claim 3, wherein said calculating, for each first group of users, a similarity of said two users of said first group of users having a number greater than 0 comprises:

for each first user group, according to the formula:calculating the similarity of the two users with the number greater than 0 in the first user group, wherein u and v are the two users with the number greater than 0 in the first user group, and w _uv The similarity between user u and user v.

5. The method according to any one of claims 1 to 4, wherein the determining, for any user in each first user group, a target product from all products to be recommended according to the interest level value of the any user for each product to be recommended includes:

determining a maximum interest degree value from interest degree values of any user in each product to be recommended aiming at any user in each first user group;

Multiplying the maximum interest degree value by a product of a preset proportion to determine an interest degree threshold;

and determining the product to be recommended, of which the interest degree value is larger than the interest degree threshold value, as the target product for any user in each first user group.

6. The method according to any one of claims 1 to 4, wherein said recommending the target product to any user of each first user group for said any user comprises:

and for any user in each first user group, sending the recommendation information of the target product to terminal equipment of any user.

7. The method according to claim 4, wherein the determining, for any user in each first user group and any product to be recommended corresponding to the any user, the interest level value of the any user for the any product to be recommended according to the user set that purchased the any product to be recommended, the similarity between other users in the first user group and the any user, and the similarity between the other users in the first user group and the any user, includes:

for any user in each first user group and any product to be recommended corresponding to any user, according to the formula: p (u, i) = Σ _v∈S(u,k) ∩N(i)w _uv Determining the interest degree value of any user on any product to be recommended; wherein i is a product to be recommended, u is the user, p (u, i) is the interest degree value of the user u on the product i, S (u, k) is the similar user of the user u, N (i) is the user set after the product i is purchased, and v is the user in the first user group.

8. The method according to any one of claims 2 to 4, wherein the determining, for any one of the products of each category, a second group of users for whom there is positive feedback based on historical behavior data of users who purchased the any one of the products, comprises:

for each type of product, establishing a reverse list from the product to the user according to the historical behavior data of the user who purchased any product;

9. A product recommendation device, comprising:

10. An electronic device, comprising: a processor, and a memory communicatively coupled to the processor;

the memory stores computer-executable instructions;

the processor executes computer-executable instructions stored in the memory to implement the method of any one of claims 1 to 8.

11. A computer readable storage medium having stored therein computer executable instructions which when executed by a processor are adapted to carry out the method of any one of claims 1 to 8.