CN114418699A - Product recommendation method, device, equipment, medium and program product - Google Patents

Abstract

The disclosure provides a product recommendation method which can be applied to the technical field of big data or the financial field. The method comprises the following steps: obtaining a product recommendation strategy pool, wherein the product recommendation strategy pool comprises m first product recommendation strategies, and m is an integer greater than or equal to 2; polling and allocating users to m first product recommendation strategies based on a product recommendation strategy pool and a first traffic segmentation method; generating a product recommendation list based on a first product recommendation strategy distributed by a user and user access historical data, wherein the user access historical data comprises a user historical transaction product list and a user historical browsing product list; acquiring user access data, wherein the user access data is associated with a product recommendation list; and evaluating the m first product recommendation strategies based on a user access data set, wherein the user access data set comprises a set of all user access data in the test period. The present disclosure also provides a product recommendation apparatus, device, storage medium and program product.

Description

Product recommendation method, device, equipment, medium and program product

Technical Field

The present disclosure relates to the field of big data technology or finance, and in particular, to a product recommendation method, apparatus, device, medium, and program product.

Background

At present, when a user-based personalized recommendation is performed on products sold online, a product recommendation strategy is made according to transaction behaviors and browsing behaviors of a client and consideration of certain time factors, such as historical transaction time and historical browsing time of the user, a product recommendation list is generated based on the made product recommendation strategy, and then the product recommendation is performed according to the product recommendation list.

In the course of implementing the disclosed concept, the inventors found that there are at least the following problems in the prior art:

because the method for making the product recommendation strategy is single and lacks of a monitoring and evaluating mechanism, the effectiveness of the made product recommendation strategy still needs to be improved.

Disclosure of Invention

In view of the foregoing, embodiments of the present disclosure provide a product recommendation method, apparatus, device, medium, and program product that optimize a product recommendation policy and improve product recommendation effectiveness.

According to a first aspect of the present disclosure, there is provided a product recommendation method including: obtaining a product recommendation strategy pool, wherein the product recommendation strategy pool comprises m first product recommendation strategies, and m is an integer greater than or equal to 2; polling and distributing users to the m first product recommendation strategies based on the product recommendation strategy pool and a first traffic segmentation method; generating a product recommendation list based on a first product recommendation strategy distributed by a user and user access historical data, wherein the user access historical data comprises a user historical transaction product list and a user historical browsing product list; acquiring user access data, wherein the user access data is associated with a product recommendation list; and evaluating the m first product recommendation strategies based on a user access data set, wherein the user access data set comprises a set of all user access data in a test period.

According to an embodiment of the present disclosure, the first product recommendation policy is generated based on a product base policy and a first dynamic addition factor.

According to an embodiment of the present disclosure, the product base policy includes: and calculating a basic product recommendation score for the to-be-recommended products contained in the to-be-recommended product list, wherein the basic product recommendation score is calculated based on the user transaction time factor, the user browsing time factor, the transaction product similarity and the browsing product similarity.

According to an embodiment of the present disclosure, the user transaction time factor and the user browsing time factor are calculated based on a time factor algorithm, and the time factor algorithm includes: calculating a transaction time interval between a historical product transaction date and a current visit date, wherein the historical product transaction date is obtained based on the user historical transaction product list; calculating the browsing time interval between the browsing date of the historical products and the current visiting date, wherein the browsing date of the historical products is obtained based on the historical browsing product list of the user; acquiring the user transaction time factor based on the transaction time interval and the product deadline attribute; and acquiring the user browsing time factor based on the browsing time interval and the product term attribute.

According to an embodiment of the disclosure, the method for generating the transaction product similarity and the browsing product similarity comprises the following steps: calculating the similarity between the product to be traded and the historical trading product of the user based on the product similarity index and the preset similarity index configuration, and acquiring the similarity of the trading product; and calculating the similarity between the product to be traded and the historical browsing product of the user based on the product similarity index and the preset similarity index configuration, and acquiring the similarity of the browsing product.

According to the embodiment of the disclosure, the product similarity index comprises two or more of product period, sales channel, risk level, purchase amount, product type, transaction currency and limit state.

According to an embodiment of the present disclosure, the first dynamic addition factor comprises at least one trending factor, including a star trending factor, an age trending factor, or a risk assessment trending factor.

According to an embodiment of the present disclosure, the star popularity factor is determined based on historical access data of users on the same star level; the age trending factor is determined based on historical access data of users in the same age group; the risk evaluation popularity factor is determined based on historical access data of users with the same risk evaluation grade, wherein the user star grade, the user age group and the user risk evaluation grade are determined based on preset evaluation rules.

According to an embodiment of the present disclosure, the first product recommendation policy based on the product base policy and the first dynamic addition factor generation includes: obtaining a first strategy weight vector, wherein the first strategy weight vector comprises a product basic strategy weight and a first dynamic addition factor weight; and generating the first product recommendation policy based on the product base policy, the first dynamic addition factor, and the first policy weight vector.

According to an embodiment of the present disclosure, the evaluating the m first product recommendation policies based on the user access data set further includes: calculating a strategy evaluation index value based on the user access data set, wherein the strategy evaluation index value comprises a product basic strategy evaluation index value and a first product recommendation strategy evaluation index value; calculating a first policy validity index value based on the product base policy evaluation index value and the first product recommendation policy evaluation index value; and when a first strategy effectiveness index value corresponding to the kth first product recommendation strategy is larger than or equal to a preset threshold value, marking the kth first product recommendation strategy as an effective strategy, wherein k is larger than or equal to 1 and is smaller than or equal to m, and k is an integer.

According to an embodiment of the present disclosure, when the first policy validity index values corresponding to all the first product recommendation policies are smaller than a preset threshold, the method further includes: executing the product recommendation strategy updating method for j times until a second strategy effectiveness index value corresponding to a second product recommendation strategy is greater than or equal to a preset threshold value, marking the second product recommendation strategy as an effective strategy, wherein j is an integer greater than or equal to 1, and the product recommendation strategy updating method comprises the following steps: obtaining at least one of a second strategy weight vector or a second dynamic addition factor, wherein the second strategy weight vector is obtained by adjusting the relative proportion of the product basic strategy weight and the first dynamic addition factor weight in the first strategy weight vector, and the second dynamic addition factor is obtained by adjusting the type of the hot degree factor contained in the first dynamic addition factor; generating a second product recommendation strategy based on the second strategy weight vector and/or a second dynamic addition factor and a product basic strategy; and evaluating the second product recommendation strategy based on the same evaluation method as the first product recommendation strategy.

According to an embodiment of the present disclosure, when n valid policies are included, the method further includes: and sorting the n effective strategies from large to small according to the first strategy effectiveness index value, and taking the sorted first effective strategy as an optimal strategy, wherein n is more than or equal to 2 and less than or equal to m and is an integer.

According to an embodiment of the present disclosure, when there are q effective strategies that are ordered and juxtaposed first, where q satisfies 2 ≦ q ≦ n and q is an integer, the method further includes: executing a flow segmentation updating method for i times until a first and only effective strategy exists, and marking the effective strategy as an optimal strategy, wherein i is an integer greater than or equal to 1, wherein the flow segmentation updating method comprises the following steps: adjusting the flow distribution proportion in the first flow segmentation method to obtain a second flow segmentation method; allocating user polling to the q effective strategies based on the second traffic segmentation method; acquiring second strategy effectiveness index values of the q effective strategies, wherein the second strategy effectiveness index values are calculated based on the same method as the first strategy effectiveness index values; and sequencing the q effective strategies from large to small according to a second strategy effectiveness index value.

According to an embodiment of the present disclosure, after obtaining the optimal policy, the method further includes: distributing a full amount of users to the optimal strategy; calculating an optimal policy effectiveness index value based on the same method as the first policy effectiveness index value; and evaluating the optimal strategy based on the optimal strategy effectiveness index value and a preset evaluation period.

According to the embodiment of the disclosure, the user access data comprises page statistical data and access behavior data, wherein the page statistical data at least comprises contact point identification, product recommendation strategies, strategy calling amount and product recommendation amount; the access behavior data is obtained based on at least one of a user transaction behavior and a user browsing behavior.

According to an embodiment of the present disclosure, the policy evaluation index includes a recommended click rate and a purchase conversion rate.

A second aspect of the present disclosure provides a product recommendation device, including: the system comprises a first obtaining module, a second obtaining module and a third obtaining module, wherein the first obtaining module is configured to obtain a product recommendation strategy pool, the product recommendation strategy pool comprises m first product recommendation strategies, and m is an integer greater than or equal to 2; the first processing module is configured to distribute user polling to the m first product recommendation strategies based on the product recommendation strategy pool and a first traffic segmentation method; the second processing module is configured to generate a product recommendation list based on the first product recommendation strategy distributed by the user and user access historical data, wherein the user access historical data comprises a user historical transaction product list and a user historical browsing product list; the second acquisition module is configured to acquire user access data, and the user access data is associated with the product recommendation list; and the third processing module is configured to evaluate the m first product recommendation strategies based on a user access data set, wherein the user access data set comprises a set of all user access data in a test period.

A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the above-described product recommendation method.

The fourth aspect of the present disclosure also provides a computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the above-mentioned product recommendation method.

A fifth aspect of the present disclosure also provides a computer program product comprising a computer program which, when executed by a processor, implements the above-described product recommendation method.

According to the method provided by the embodiment of the disclosure, the flow of the user is segmented by configuring a plurality of product recommendation strategies, and the user is allocated to different product recommendation strategies in a polling manner according to the proportion, so that the evaluation of each product recommendation strategy can be realized according to the actual operation effect, and compared with a single product recommendation strategy lacking an evaluation mechanism, the method is beneficial to improving the effectiveness of product recommendation and is beneficial to improving the transaction rate in the actual sale of the product.

Drawings

The foregoing and other objects, features and advantages of the disclosure will be apparent from the following description of embodiments of the disclosure, which proceeds with reference to the accompanying drawings, in which:

fig. 1 schematically shows an application scenario diagram of a product recommendation method according to an embodiment of the present disclosure.

FIG. 2 schematically shows a flow diagram of a product recommendation method according to an embodiment of the disclosure.

FIG. 3 schematically shows a flow chart of a time factor algorithm according to an embodiment of the present disclosure.

Fig. 4 schematically shows a flowchart of a method of generating a deal product similarity and a browse product similarity according to an embodiment of the present disclosure.

FIG. 5 schematically illustrates a flow diagram of a method of generating a first product recommendation policy based on a product base policy and a first dynamic addition factor, according to an embodiment of the disclosure.

FIG. 6 schematically illustrates a flow chart of a method of evaluating m first product recommendation policies based on a user access data set, according to an embodiment of the disclosure.

FIG. 7 schematically shows a flow diagram of a product recommendation policy update method according to an embodiment of the present disclosure.

Fig. 8 schematically shows a flow chart of a method of screening an optimal strategy according to an embodiment of the present disclosure.

Fig. 9 schematically shows a flowchart of a method of screening an optimal policy according to another embodiment of the present disclosure.

Fig. 10 schematically shows a flow chart of a traffic split updating method according to an embodiment of the present disclosure.

FIG. 11 schematically illustrates a flow chart of a method of evaluating an optimal policy according to an embodiment of the present disclosure.

Fig. 12 schematically shows a block diagram of a product recommendation device according to an embodiment of the present disclosure.

FIG. 13 schematically illustrates a block diagram of an electronic device suitable for implementing a method of product recommendation in accordance with an embodiment of the present disclosure.

Detailed Description

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is illustrative only and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present disclosure.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.

All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It is noted that the terms used herein should be interpreted as having a meaning that is consistent with the context of this specification and should not be interpreted in an idealized or overly formal sense.

Where a convention analogous to "at least one of A, B and C, etc." is used, in general such a construction is intended in the sense one having skill in the art would understand the convention (e.g., "a system having at least one of A, B and C" would include but not be limited to systems that have a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.).

At present, when a user-based personalized recommendation is performed on products sold on line, for example, financial products sold by a financial institution client, a product recommendation strategy is made according to transaction behaviors and browsing behaviors of a client and consideration of certain time factors, such as historical transaction time and historical browsing time of the user, a product recommendation list is generated based on the made product recommendation strategy, and then, product recommendation is performed according to the product recommendation list. The method for making the product recommendation strategy is single and lacks of a monitoring and evaluating mechanism, and the effectiveness and the accuracy of the made product recommendation strategy are still to be improved. On the other hand, how to dynamically monitor the effectiveness of the product recommendation strategy in real time is also an urgent problem to be solved in order to find the optimal strategy among various feasible product recommendation strategies and continuously evaluate the effectiveness of the optimal strategy.

In view of the foregoing problems of the prior art, an embodiment of the present disclosure provides a product recommendation method, including: obtaining a product recommendation strategy pool, wherein the product recommendation strategy pool comprises m first product recommendation strategies, and m is an integer greater than or equal to 2; polling and distributing users to the m first product recommendation strategies based on the product recommendation strategy pool and a first traffic segmentation method; generating a product recommendation list based on a first product recommendation strategy distributed by a user and user access historical data, wherein the user access historical data comprises a user historical transaction product list and a user historical browsing product list; acquiring user access data, wherein the user access data is associated with a product recommendation list; and evaluating the m first product recommendation strategies based on a user access data set, wherein the user access data set comprises a set of all user access data in a test period. By configuring a plurality of product recommendation strategies, segmenting the flow of the user, and allocating the user to different product recommendation strategies in a polling mode according to the proportion, the evaluation of each product recommendation strategy can be achieved according to the actual operation effect, the effectiveness of the product recommendation strategies is improved, and the transaction rate in the actual sale of the product is improved. Further, after the first product recommendation strategy is evaluated, effective strategies can be screened out, and the effective strategies are sorted to obtain the optimal strategy. Further, when an effective policy is not acquired, a product recommendation policy update method may be executed to acquire the effective policy; when more than one effective policy is ranked first, a traffic split update method may be performed to obtain the optimal policy. Furthermore, after the optimal strategy is obtained, the optimal strategy can be continuously operated and monitored based on a total number of users, so that the effectiveness of the optimal strategy can be continuously evaluated.

It should be noted that the product recommendation method, apparatus, device, medium, and program product provided in the embodiments of the present disclosure may be used in the big data technology in the product recommendation related aspect, and may also be used in various fields other than the big data technology, such as the financial field. The application fields of the product recommendation method, device, equipment, medium and program product provided by the embodiments of the present disclosure are not limited.

It should be noted that, in the technical solution of the present disclosure, the processes of collecting, storing, using, processing, transmitting, providing, disclosing and applying the personal information of the related users all conform to the regulations of the related laws and regulations, and necessary security measures are taken without violating the customs of the public order. In the technical scheme of the disclosure, before the personal information of the user is acquired or collected, the authorization or the consent of the user is acquired.

The above-described operations for carrying out at least one of the objects of the present disclosure will be described with reference to the accompanying drawings and description thereof.

Fig. 1 schematically shows an application scenario diagram of a product recommendation method according to an embodiment of the present disclosure.

As shown in fig. 1, the application scenario 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 serves as a medium for providing communication links between the terminal devices 101, 102, 103 and the server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.

The user may use the terminal devices 101, 102, 103 to interact with the server 105 via the network 104 to receive or send messages or the like. The terminal devices 101, 102, 103 may have installed thereon various communication client applications, such as shopping-like applications, web browser applications, search-like applications, instant messaging tools, mailbox clients, social platform software, etc. (by way of example only).

The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.

The server 105 may be a server providing various services, such as a background management server (for example only) providing support for websites browsed by users using the terminal devices 101, 102, 103. The background management server may analyze and perform other processing on the received data such as the user request, and feed back a processing result (e.g., a webpage, information, or data obtained or generated according to the user request) to the terminal device.

It should be noted that the product recommendation method provided by the embodiment of the present disclosure may be generally executed by the server 105. Accordingly, the product recommendation device provided by the embodiments of the present disclosure may be generally disposed in the server 105. The product recommendation method provided by the embodiment of the present disclosure may also be executed by a server or a server cluster different from the server 105 and capable of communicating with the terminal devices 101, 102, 103 and/or the server 105. Accordingly, the product recommendation device provided by the embodiment of the present disclosure may also be disposed in a server or a server cluster different from the server 105 and capable of communicating with the terminal devices 101, 102, 103 and/or the server 105.

It should be understood that the number of terminal devices, networks, and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

Hereinafter, a product recommendation method of the disclosed embodiment will be described in detail through fig. 2 to 11 based on the scenario described in fig. 1.

FIG. 2 schematically shows a flow diagram of a product recommendation method according to an embodiment of the disclosure.

As shown in fig. 2, the product recommendation method of this embodiment includes operations S210 to S250.

In operation S210, a pool of product recommendation policies is obtained.

According to embodiments of the present disclosure, the product recommendation policy pool may be preconfigured. It is to be understood that the pool of product recommendation policies may contain m first product recommendation policies, m being an integer greater than or equal to 2. Wherein the first product recommendation policy may be an unevaluated product recommendation policy. The first product recommendation policy may be evaluated during subsequent evaluations to facilitate tuning and optimizing the policy.

In operation S220, a user is polled and allocated to the m first product recommendation policies based on the product recommendation policy pool and the first traffic splitting method.

According to the embodiment of the disclosure, the user can be polled and distributed to the m first product recommendation strategies according to the preset flow distribution proportion by using a flow splitting method. The first traffic segmentation method may be a traffic segmentation method of a preset traffic distribution proportion, and a typical first traffic segmentation method may include average traffic segmentation, that is, users are averagely routed to each category according to the number of categories of the first product recommendation policy, so that the effectiveness of each first product recommendation policy is balanced and evaluated without considering traffic factors.

In some specific embodiments, assuming that the product recommendation policy pool has 4 first product recommendation policies, the flow distribution proportion of the 4 product recommendation policies is 100%. According to the user access condition, the flow distribution can be carried out in a polling mode, for example, the user of the first access is routed to the product recommendation strategy Y0, the user of the second access is routed to the product recommendation strategy Y1, and the user of the third access is routed to the product recommendation strategy Y₃The fourth visited user is routed to the product recommendation policy Y4, the fifth visited user is routed to the product recommendation policy Y0, and so on, and polling is performed in sequence, so that the proportion of all visited users routed to the product recommendation policies is ensured to be the same, namely, the product recommendation policies are aimed atSlightly Y0, Y1, Y₂，Y₃The respective flow distribution was 25%. Therefore, the influence of the user flow can be eliminated, and the effectiveness of each first product recommendation strategy can be evaluated under the same user flow proportion.

In operation S230, a product recommendation list is generated based on the first product recommendation policy assigned by the user and the user access history data.

According to the embodiment of the disclosure, after a user logs in a product recommendation platform, user access history data can be obtained based on a user identifier, and the user access history data at least can comprise a user historical transaction product list and a user historical browsing product list. To evaluate the first product recommendation policy, after configuring the first product recommendation policy, a product recommendation list for the current user may be generated based on the first product recommendation policy actually assigned by the user and the user access history data. It will be appreciated that one or more recommended products, which may be products on sale, may be included in the product recommendation list, wherein the products on sale may be determined based on the current user visit time point. The historical transaction or browsing information of the user can be obtained by the user accessing the historical data, so that the actual recommended product can be obtained by combining the product recommendation strategy.

In operation S240, user access data associated with a product recommendation list is acquired.

In operation S250, the m first product recommendation policies are evaluated based on the user access data set.

According to the embodiment of the disclosure, in the process of accessing the product recommendation platform by the user, the user access data can be acquired in real time, and further, the set of all the user access data in the test period, namely the user access data set, can be acquired.

It will be appreciated that the user access data may be obtained based on a buried point technique, which may be based on user contact settings. The contact point means a point which can be contacted by a user in the aspects of vision, hearing, smell, feeling and the like and a related channel of a product recommendation platform, such as a financial institution, more specifically, a website of the financial institution, a worker, an intelligent terminal and an online channel. Embodiments of the present disclosure may enable product recommendation through touch point marketing. The meaning of contact marketing is that existing or potential contacts are created and packaged, so that a user can pay attention to the contact marketing system, and the value of products and services provided by the product recommendation platform is sensed, and therefore the user can identify the product recommendation platform. The contacts may generally include an inline contact and an inline contact. The contacts involved in the embodiments of the present disclosure may include online contacts, for example, a mobile banking page, a menu, and a login IP. It is understood that the user may access the product recommendation list through a touch point. In embodiments of the present disclosure, the touch points may be set to pages that are frequently visited by the user. In some specific embodiments, in a business scenario of a mobile banking, the contact may be set to a financial product purchase success page, a monthly bill financial product recommendation page, a monthly bill my assets page, a guess you like financial product page, a financial product expiration reminding page, and the like. The user can browse the recommended financing products on the contact page, click the financing products to view detailed information, or purchase the financing products. It will be appreciated that after a user accesses a contact, the system may record user information (including but not limited to a customer number) and record the contact identification and contact name that the user is asking.

According to an embodiment of the present disclosure, the buried point may be set based on the user contact point. The term "buried point" means that information is collected by a specific process in an application to track the usage of the application, and then used to further optimize the product or provide data support for operation. In an embodiment of the present disclosure, the user access data collected from the buried points includes page statistics and access behavior data. The page statistical data at least comprises contact point identification, product recommendation strategies, strategy calling amount and product recommendation number. The access behavior data is obtained based on at least one of the user transaction behavior and the user browsing behavior, and the access behavior data may include page dwell time, page jump rate, recommended product click number, recommended product purchase number, and the like. In addition, information such as information material types and interface names can be collected through the buried points.

Therefore, the overall calling condition of each product recommendation strategy in the test period can be counted based on the user access data set, and the strategy calling is converted into the actual user click rate and purchase rate conditions, so that the evaluation of each first product recommendation strategy is realized.

According to the embodiment of the disclosure, the flow of the user is segmented by configuring the plurality of product recommendation strategies, and the user is allocated to different product recommendation strategies in a polling manner according to the proportion, so that the evaluation of each product recommendation strategy can be realized according to the actual operation effect, the effectiveness of the product recommendation strategies is favorably improved, and the transaction rate in the actual sale of the product is improved.

According to an embodiment of the present disclosure, the first product recommendation policy is generated based on a product base policy and a first dynamic addition factor.

In embodiments of the present disclosure, the product base policy may primarily consider the customer's own transaction behavior, i.e., internal influencing factors. The recommendation method is formulated according to the transaction behavior and the browsing behavior of the client, and by combining certain time factors (such as the historical transaction time and the historical browsing time of the user) and the consideration of the similarity between the product to be recommended and the historical transaction product of the user or the historical browsing product of the user.

Specifically, the product base policy may include: and calculating a basic product recommendation score for the product to be recommended contained in the product list to be recommended, wherein the basic product recommendation score is calculated based on the user transaction time factor, the user browsing time factor, the transaction product similarity and the browsing product similarity. And calculating the user transaction time factor and the user browsing time factor based on a time factor algorithm. Wherein the time factor may be determined according to the transaction day on which the customer purchased the product or the time interval between the browsing day of browsing the product and the current page visited by the customer. The specific time factor value can be set according to the length of the interval time. For example, the shorter the interval time length, the larger the time factor. The similarity of the transaction products or the browsing products is the similarity of the products to be recommended and the historical transaction products or the historical browsing products of the user. Therefore, after the user accesses the product recommendation system, the user historical transaction product list and the user historical browsing list can be inquired based on the user question data, and whether transaction information and browsing information exist or not can be judged. A time factor may then be calculated for the user's historical trading of products and/or the user's historical viewing of products. Furthermore, the similarity between the user historical transaction products or the user historical browsing products and the products in the product list to be recommended can be calculated, and the product recommendation list is generated based on the comprehensive calculation of the user transaction time factor, the user browsing time factor, the transaction product similarity and the browsing product similarity.

FIG. 3 schematically shows a flow chart of a time factor algorithm according to an embodiment of the present disclosure.

As shown in fig. 3, the time factor algorithm of this embodiment includes operations S310 to S340.

In operation S310, a transaction time interval between the historical product transaction day and the current visit day is calculated.

According to an embodiment of the present disclosure, the historical product trading date is obtained based on the user historical trading product list.

In operation S320, a browsing time interval between the browsing date of the historical product and the current visiting day is calculated.

According to an embodiment of the present disclosure, the historical product browsing date is obtained based on the user historical browsing product list.

In operation S330, the user transaction time factor is acquired based on the transaction time interval and the product term attribute.

In operation S340, the user browsing time factor is acquired based on the browsing time interval and the product term attribute.

In some specific embodiments, the product to be recommended may be a financial product, such as a financial dish. The product similarity index may include two or more of product term, sale channel, risk level, purchase amount, product type, transaction currency, and limit state.

According to a specific embodiment of the present disclosure, the user transaction time factor and the user browsing time factor may be configured based on the term attribute of the financial product.

In a typical example, the term attribute of the financial product may be classified as money-alive management, 0-6 months, 6-12 months, 12 months or more.

For the user transaction time factor, based on the term attribute of the financial product, it may be configured based on the following valuations:

for a financial product whose deadline attribute is money management (which can be redeemed at any time), a user trading time factor of 100 may be assigned for trading time intervals of 0-1 month between the historical product trading day and the current visit day. Similarly, a trade time interval of 90 assignments for 1-3 months, a trade time interval of 3-6 months of 80 assignments, a trade time interval of 6-12 months of 70 assignments, a trade time interval of 12-24 months of 60 assignments, and a trade time interval of more than 24 months of 40 assignments.

For a financial product with a term attribute of 0-6 months, the trading time interval between the historical product trading day and the current access day is 0-6 months, the user trading time factor can be assigned 100, the trading time interval is assigned 80 for 6-12 months, the trading time interval is assigned 60 for 12-24 months, and the trading time interval is assigned 40 for more than 24 months.

For a financial product with a term attribute of 6-12 months, the trading time interval between the historical product trading day and the current visit day is 0-12 months, the time factor of the trading product can be assigned to 100, the time is assigned to 80 from 12 months to 24 months, and the time is assigned to 40 above 24 months.

For a financial product with a term attribute of more than 12 months, the trading time interval between the historical product trading day and the current access day is 0-12 months, the user trading time factor can be assigned 100, the trading time interval is assigned 80 between 12 months and 24 months, the trading time interval is assigned 60 between 24 months and 36 months, and the trading time interval is assigned 40 between 36 months.

For the user browsing time factor, it can be understood that since browsing the product only indicates interest in the product to a certain extent, and the product is not necessarily purchased, the value can be obtained by multiplying the result by 50% according to the same calculation mode as the trading product time factor. Specifically, based on the term attribute of the financial product, the configuration may be based on the following assignments:

for a financial product whose attributes are money management (which can be redeemed at any time), a user's view time factor of 50 may be assigned for viewing time intervals of 0-1 month between the historical product view day and the current visit day. The browsing time interval is a value of 1-3 months 45. A value of 40 for a browsing time interval of 3-6 months, a value of 35 for a browsing time interval of 6-12 months, a value of 30 for a browsing time interval of 12-24 months, and a value of 20 for a browsing time interval of more than 24 months.

For a financial product with an attribute of 0-6 months, the browsing time interval between the browsing date of the historical product and the current access date is 0-6 months, the user browsing time factor can be assigned 50, the browsing time interval is assigned 40 for 6-12 months, the browsing time interval is assigned 30 for 12-24 months, and the browsing time interval is assigned 20 for more than 24 months.

For a financial product with an attribute of 6-12 months, the browsing time interval between the browsing date of the historical product and the current access date is 0-12 months, the user browsing time factor can be assigned 50, the browsing time interval is assigned 40 from 12 months to 24 months, and the browsing time interval is assigned 20 above 24 months.

For a financial product with the attribute of more than 12 months, the browsing time interval between the browsing date of the historical product and the current visiting date is 0-12 months, and the user browsing time factor can be assigned 50. The value 40 for a browsing interval of 12 months to 24 months, the value 30 for a browsing interval of 24 months to 36 months, and the value 20 for a browsing interval of more than 36 months.

Fig. 4 schematically shows a flowchart of a method of generating a deal product similarity and a browse product similarity according to an embodiment of the present disclosure.

As shown in fig. 4, the method for generating the transaction product similarity and the browsing product similarity according to this embodiment includes operations S410 to S420.

In operation S410, the similarity between the product to be recommended and the historical transaction product of the user is calculated based on the product similarity index and the preset similarity index configuration, and the transaction product similarity is obtained.

In operation S420, the similarity between the product to be recommended and the historical browsing product of the user is calculated based on the product similarity index and a preset similarity index configuration, and the similarity between the browsing product and the historical browsing product of the user is obtained.

According to embodiments of the present disclosure, product similarity may have a number of influencing factors. Accordingly, the relevant influence factor may be set as a product similarity index. In some specific embodiments, the preset configuration may be performed by a method of directly assigning similarity indexes, and when multiple similarity indexes are included, joint evaluation may also be performed based on a preset calculation method, where a typical calculation method may include calculating and summing after configuring a weight for each product similarity index, and may also be to multiply multiple similarity index assignments.

In some specific embodiments, financial institution online financial product recommendations are used as examples of application scenarios. The product similarity index may include two or more of product term, sales channel, risk level, purchase amount, product type, transaction currency, and limit status.

In one particular example, the product term may include money transfer, 0-6 months, 6-12 months, 12 months or more. The sale channel can contain self-operation sale or sale, and correspondingly, the self-operation/sale label can be set to comprise self-operation and sale. The purchase amount can include 0-1 ten thousand, 1-5 ten thousand, 5-50 ten thousand, 50-500 ten thousand and over 500 ten thousand. The risk level may include low risk, moderate risk, high risk. The product types may include fixed income classes, equity classes, commodities (e.g., precious metals, etc.), and financial derivatives classes, mixed classes. The currency may include RMB, U.S. dollars, British pounds, Euro. The status of credit may include remaining credit and temporary non-credit. Based on the product similarity index, the preset similarity index configuration may include: 1) if the product deadline of the product to be recommended is consistent with the product deadline of the user historical transaction product (or the user historical browse product), the deadline similarity is 1; and if the product time limit is not consistent, the similarity of the time limit is 0. 2) If the product to be recommended is consistent with the sales channel of the user historical transaction product (or the user historical browse product), the similarity of the sales channel is 1; if the sales channels are not consistent, the similarity of the sales channels is 0.5. 3) If the purchase amount of the product to be recommended is different from the purchase amount of the user historical transaction product (or the user historical browsing product), and the purchase amounts of the product to be recommended and the user historical browsing product are in adjacent areas, configuring the similarity of the purchase amounts to be 0.75; if the purchase amount of the product to be recommended is completely the same as the purchase amount of the user historical transaction product (or the user historical browsing product), configuring the similarity of the purchase amount to be 1; in other cases, the purchase amount similarity is set to 0. 4) If the risk levels of the product to be recommended and the user historical transaction product (or the user historical browsing product) are in adjacent areas, configuring the similarity of the risk levels to be 0.80; if the risk level of the product to be recommended is completely the same as the risk level of the user historical transaction product (or the user historical browsing product), the similarity of the risk levels is 1; otherwise, the configuration risk level similarity is 0. 5) If the product to be recommended is consistent with the product type of the user historical transaction product (or the user historical browsing product), the similarity of the product types is 1, otherwise, the similarity is 0. 6) If the product to be recommended is consistent with the transaction currency of the user historical transaction product (or the user historical browsing product), the similarity of the transaction currency is 1, and if not, the similarity is 0. 7) If the credit status of the product to be recommended is still credit, the credit status similarity is 1; if the status of the product quota to be recommended is temporary quota, the similarity of the quota status is 0.

Therefore, the similarity between the product to be recommended and the historical trading product of the user can be calculated based on the product similarity index and the preset similarity index configuration, and the similarity of the trading product is obtained; and calculating the similarity between the product to be recommended and the historical browsing product of the user based on the product similarity index and the preset similarity index configuration, and acquiring the similarity of the browsing product. In an exemplary method for calculating the similarity of products, whether the similarity of products is traded or browsed, the preset assignment configured by the index of the similarity of products may be multiplied to calculate the similarity of products, that is, the similarity of product terms, the similarity of sales channel, the similarity of purchase amount, the similarity of risk level, the similarity of product types, the similarity of trading currencies and the similarity of state. In another exemplary method for calculating the product similarity, the sum of the product similarity indexes may be divided by a total value of all the product similarity indexes 1. It is understood that the assignment and calculation methods described above are merely illustrative examples, and that the assignment and calculation methods may be adjusted based on actual operating conditions. It can be understood that if the product to be recommended is a product that has been traded once, for example, the product identifiers are the same, then the similarity of the traded product is directly determined to be 1 without performing the similarity calculation of the traded product, and if the product to be recommended is a product that has been browsed once, for example, the product identifiers are the same, then the similarity of the browsed product is directly determined to be 1 without performing the similarity calculation of the browsed product. When the product to be recommended is a product that has neither been traded nor viewed, the similarity of the traded product and the similarity of the viewed product can be obtained according to the index configuration and calculation method listed in the embodiment of the disclosure.

According to the embodiment of the disclosure, after the user transaction time factor, the user browsing time factor, the transaction product similarity and the browsing product similarity are obtained, the product basic recommendation score can be comprehensively calculated for each product to be recommended according to the product identification. For example, the user transaction time factor, the user browsing time factor, the transaction product similarity and the browsing product similarity are respectively given certain weights and then multiplied or added to calculate the product basic recommendation score.

In one example, the product base recommendation policy Y₀The product basis recommendation score sigma Y can be calculated by the formula (1)₀：

∑Y₀＝a×c_a+b×c_bFormula (1)

Where a is a user transaction time factor, b is a user browsing time factor, c_aIs the similarity of the transaction products, c_bIs the browsing product similarity.

It should be noted that, when a plurality of user historical transaction products and/or a plurality of user historical browsing products are included, the time factor of each user transaction product × the transaction product similarity, and the time factor of the user browsing product × the browsing product similarity may be calculated respectively, and the product basic recommendation score is obtained by averaging and calculating after the calculation.

After the basic recommendation scores of the products to be recommended are obtained, ranking can be performed on the products to be recommended based on the scores. When a user accesses a touch point, a specified number of products before the total rank can be displayed as recommended products based on the configuration of the corresponding touch point page, and the recommending mode can be used as a product basic recommending strategy Y₀。

According to the embodiment of the disclosure, the first dynamic addition factor introduces external reference factors, the external reference factors are certain attribute dimensions which have an association relation with purchasing behavior, and the transaction and browsing behaviors of the user can be predicted through product transaction preferences of other users which are the same as the user in the related attribute dimensions. Taking the online financial product sale of the financial institution as an example, when a user purchases a financial product, the risk preference and the asset allocation requirement of the financial product are directly related to the income or the asset level of the user, and the financial product purchase preference of other users knowing the same asset level, the same age bracket or the same risk evaluation level is considered, the first dynamic addition factor can comprise a popularity factor, and the popularity factor can be the same asset level, the same age bracket or the popularity of the user purchasing the product with the same risk evaluation level.

In an embodiment of the disclosure, the first dynamic addition factor comprises at least one trending factor, including a star trending factor, an age trending factor, or a risk assessment trending factor. According to embodiments of the present disclosure, the trending factor may be determined based on user historical access data. Specifically, the star popularity factor may be determined based on historical access data of users at the same star level; the age trending factor may be determined based on historical access data of users of the same age group; the risk assessment popularity factor can be determined based on historical access data of users with the same risk assessment grade, wherein the user star rating, the user age group and the user risk assessment grade are determined based on preset assessment rules.

In one example, taking a bank as an example, the user star rating is primarily related to the user's size of financial assets in the bank's last month and day. For example, at the beginning of each natural month, the bank may rate the star rating based on the user's previous month, day, and average financial asset size at the bank. The new user performs the first star rating at the beginning of the next month of the account opening. The financial assets which are brought into the star rating range comprise the deposit of foreign currency of a user at a bank and the sum of the market values of assets such as financing, fund, insurance, national debt, third party deposit and account transaction. For example, the personal customer star level may include a personal bank level, a seven star level, a six star level, a five star level, a four star level, a three star level, a two star level, and a one star level from high to low, and each star level corresponds to the previous month, day, and month of the financial asset size, for example, the following correspondence exists: private bank level: the monthly and daily average financial assets are more than 800 ten thousand yuan (inclusive); seven-star level: the monthly-daily average financial assets are 600 ten thousand yuan (including) to 800 ten thousand yuan (not including); six-star grade: the monthly-daily average financial assets are 100 ten thousand yuan (including) to 600 ten thousand yuan (not including); five-star level: the monthly financial assets are 20 ten thousand yuan (including) to 100 ten thousand yuan (not including); four-star level: 5-20 ten thousand yuan (including) of monthly financial assets (not including); three-star grade: 1 ten thousand yuan (including) to 5 ten thousand yuan (not including) of financial assets are averaged every month and day; two-star grade: the monthly and daily average financial assets are less than 1 ten thousand yuan (no balance is zero); a star stage: the monthly and daily balance of the financial assets is zero. The method can acquire the sales ranking information of the financial products purchased by the same-star user in the last month, and obtain the star popularity factor of the same-star user according to the sales ranking. In some typical examples, the specific values of the star-level heat factors may be labeled by assignment. For example, the top 20 financing products can be selected according to the sales ranking of the financing products purchased by the same user on star level, the star-level popularity factor is sequentially valued at 100, 95, 90, 85, 80, 75, 70, 65, 60, 55, 50, 45, 40, 35, 30, 25, 20, 15, 10, 5, and the star-level popularity factor of other financing products is valued at 0.

Similarly, a value may be assigned to the age trending factor purchased by a customer of the same age group: according to the sales ranking condition of the financial products purchased by users in the same age group, the age popularity factors of the top 20 financial products are sequentially set to be 100, 95, 90, 85, 80, 75, 70, 65, 60, 55, 50, 45, 40, 35, 30, 25, 20, 15, 10 and 5, and the age popularity factors of other financial products are set to be 0. Wherein, the age groups are divided as follows: young people: 18-35 years old; the young and the middle-aged: 36-50 years old; in middle age: 51-65 years old; old people: over 65 years old.

Similarly, the top 20 financing products can be selected according to the sales ranking condition of the financing products purchased by the users with the same risk assessment level, the risk assessment popularity factor is sequentially valued as 100, 95, 90, 85, 80, 75, 70, 65, 60, 55, 50, 45, 40, 35, 30, 25, 20, 15, 10, 5, and the risk assessment popularity factor of other financing products is valued as 0.

Therefore, multiple first dynamic addition factors can be constructed based on selection and combination of the popularity factors, multiple first product recommendation strategies are obtained, and each first product recommendation strategy is evaluated in actual operation.

It can be understood that the above-mentioned popularity factor can be adjusted according to the actual operation process, and the adjustment includes adding a popularity factor associated with the purchase preference of the user product of other attributes to optimize the construction of the first dynamic addition factor. It is noted that in embodiments of the present disclosure, a trending factor may be associated with user information. User consent or authorization may be obtained when it comes to obtaining user information, for example, prior to obtaining the user's information. For example, a request may be made to the user to obtain user information before obtaining the age trending factor. The step of obtaining the age trending factor is performed in the event that the user information is available with user consent or authorization.

FIG. 5 schematically illustrates a flow diagram of a method of generating a first product recommendation policy based on a product base policy and a first dynamic addition factor, according to an embodiment of the disclosure.

As shown in fig. 5, the method for generating the first product recommendation policy based on the product base policy and the first dynamic addition factor of the embodiment includes operations S510 to S520.

In operation S510, a first policy weight vector is obtained, where the first policy weight vector includes a product basis policy weight and a first dynamic addition factor weight.

In operation S520, the first product recommendation policy is generated based on the product base policy, the first dynamic addition factor and the first policy weight vector.

According to an embodiment of the present disclosure, when generating a first product recommendation policy, a first policy weight vector may be set, wherein the first policy weight vector includes a product base policy weight value and a first dynamic addition policy weight value. Therefore, the product basic strategy score can be obtained through the product basic recommendation score and the product basic strategy weight value; and obtaining a dynamic addition factor score through the first dynamic addition factor and the first dynamic addition strategy weight value. Therefore, after the product identification to be recommended is obtained, the total recommendation score of each product to be recommended can be calculated by combining the product basic strategy score and the dynamic addition factor score, and according to the ranking, the products with the specified number ranked at the top are displayed as the recommended products and serve as the first product recommendation strategy. In one specific example, the top 2 or top 10 total ranked products may be displayed as recommended products, depending on the user's settings for accessing the touchpoint page.

In one example, the first product recommendation policy Y₁The first product recommendation score Σ Y can be calculated by equation (2)₁：

∑Y₁＝(a×c_a+b×c_b)*λ₁+d*λ₂Formula (2)

Where a is a user transaction time factor, b is a user browsing time factor, c_aIs the similarity of the transaction products, c_bIs a browsing of product similarity, λ₁Is the product basis strategy weight value, d is the star-level hot-degree factor assignment, λ₂Is the first dynamic additive factor weight. In one example, the importance of the product base policy and the first dynamic addition factor may be considered to the same extent, and λ may be set₁＝λ₂＝50％。

Obtaining a first product recommendation score Sigma Y of each product to be recommended₁The products to be recommended may then be ranked based on the score. When in useWhen a user accesses a touch point, the appointed number of products before the total rank can be displayed as recommended products based on the configuration of the corresponding touch point page, and the recommended mode can be used as a first product recommendation strategy Y₁。

Similarly, the first product recommendation policy Y2 may calculate the first product recommendation score Σ Y2 by equation (2):

∑Y₂＝(a×c_a+b×c_b)*λ₁+e*λ₃formula (2)

Where a is a user transaction time factor, b is a user browsing time factor, c_aIs the similarity of the transaction products, c_bIs a browsing of product similarity, λ₁Is the product basis policy weight value, e is the star-level hot-door factor assignment, λ₃Is the first dynamic additive factor weight. In one example, the importance of the product base policy and the first dynamic addition factor may be considered to the same extent, and λ may be set₁＝λ₃＝50％。

Obtaining a first product recommendation score Sigma Y of each product to be recommended₂The products to be recommended may then be ranked based on the score. When a user accesses a touch point, a specified number of products before the total rank can be displayed as recommended products based on the configuration of the corresponding touch point page, and the recommended mode can be used as a first product recommendation strategy Y₂。

Similarly, the first product recommendation policy Y₃The first product recommendation score Σ Y can be calculated by equation (3)₃：

∑Y₃＝(a×c_a+b×c_b)*λ₁+(d+e)*λ₄Formula (3)

Where a is a user transaction time factor, b is a user browsing time factor, c_aIs the similarity of the transaction products, c_bIs to browse the similarity of products₄Is the product base policy weight value, e is the star-level hot-factor assignment, and₂is the first dynamic additive factor weight. In one example, the importance of the product base policy and the first dynamic addition factor may be considered to the same extent, and λ may be set₁＝λ₄＝50％。

Obtaining a first product recommendation score Sigma Y of each product to be recommended₃The products to be recommended may then be ranked based on the score. When a user accesses a touch point, a specified number of products before the total rank can be displayed as recommended products based on the configuration of the corresponding touch point page, and the recommended mode can be used as a first product recommendation strategy Y₃。

It is understood that when the first dynamic addition factor introduces more hot factors, the first product recommendation strategy Y can be followed₁Y2, Y3 similar method generates a first product recommendation strategy that introduces a new heat factor

FIG. 6 schematically illustrates a flow chart of a method of evaluating m first product recommendation policies based on a user access data set, according to an embodiment of the disclosure.

As shown in fig. 6, the product recommendation method of this embodiment includes operations S610 to S640, or operations S610 to S630, S650.

In operation S610, a policy evaluation index value is calculated based on a user access data set.

In operation S620, a first policy effectiveness index value is calculated based on the product base policy evaluation index value and the first product recommendation policy evaluation index value.

In operation S630, it is determined whether a first policy validity index value corresponding to the kth first product recommendation policy is greater than or equal to a preset threshold, where k is greater than or equal to 1 and less than or equal to m and is an integer.

When the first policy validity index value corresponding to the kth first product recommendation policy is greater than or equal to the preset threshold, operation S640 is performed.

In operation S640, the kth first product recommendation policy is marked as a valid policy.

According to an embodiment of the present disclosure, the policy evaluation index value includes a product base policy evaluation index value and a first product recommendation policy evaluation index value. It will be appreciated that the first product recommendation strategy should have a higher effectiveness than the product base strategy. Therefore, the product basic strategy evaluation index value can be calculated based on the user access data set, and then the first strategy effectiveness index value is calculated by comparing the product basic strategy evaluation index value with the first product recommendation strategy evaluation index value so as to represent the improvement degree of the product recommendation strategy effectiveness after the first dynamic factor is introduced. For example, whether the corresponding first product recommendation strategy is an effective strategy may be determined by whether the ratio of the first product recommendation strategy evaluation index value to the product basis strategy evaluation index value is greater than a preset threshold. In one example, the threshold may be set to 150%, that is, when the ratio of the first product recommendation policy evaluation index value to the product base policy evaluation index value corresponding to a certain first product recommendation policy is greater than or equal to 150%, the first product recommendation policy is marked as an effective policy.

According to embodiments of the present disclosure, the policy evaluation index may include a recommended click through rate and a purchase conversion rate. Wherein the recommended click rate may be calculated based on the number of clicks/recommendations for recommended products in the user access data. The purchase conversion rate may be calculated based on the number of post-click purchases/number of recommendations for the recommended product in the user access data. It can be understood that when performing policy evaluation, an auxiliary reference index, such as a deal amount, can also be introduced to evaluate the product from multiple dimensions, so as to realize screening of effective policies and even optimal policies.

According to the embodiment of the present disclosure, when the first policy validity index values corresponding to all the first product recommendation policies are less than the preset threshold, operation S650 is performed.

In operation S650, the method for updating the product recommendation policy is executed j times, until a second policy validity index value corresponding to the second product recommendation policy is greater than or equal to a preset threshold value, the second product recommendation policy is marked as a valid policy, where j is an integer greater than or equal to 1.

According to the embodiment of the disclosure, when the first policy effectiveness index values corresponding to all the first product recommendation policies are smaller than the preset threshold value, the second policy effectiveness index value can be obtained by updating the product recommendation policies so as to evaluate the product recommendation policies again. The updating process may be performed in a loop until a second policy validity index value greater than or equal to a preset threshold value is obtained.

FIG. 7 schematically shows a flow diagram of a product recommendation policy update method according to an embodiment of the present disclosure.

As shown in fig. 7, the product recommendation policy updating method of this embodiment includes operations S710 to S730.

In operation S710, at least one of a second policy weight vector or a second dynamic addition factor is obtained.

In operation S720, a second product recommendation policy is generated based on the second policy weight vector and/or the obtained second dynamic addition factor, and the product base policy.

In operation S730, the second product recommendation policy is evaluated based on the same evaluation method as the first product recommendation policy.

According to embodiments of the present disclosure, the first product recommendation policy may be adjusted by adjusting the specific content of the policy weight vector or the dynamic addition factor. And obtaining a second strategy weight vector by adjusting the relative proportion of the product basic strategy weight and the first dynamic addition factor weight in the first strategy weight vector. On the other hand, the second dynamic addition factor may be obtained by adjusting the kind of the thermal factor included in the first dynamic addition factor. In one example, in the first policy weight vector, the product basis policy weight and the first dynamic addition factor weight are both 50%, and in the first policy weight vector, the actual condition may be evaluated based on the operation condition, the product basis policy weight may be adjusted to 30%, and the first dynamic addition factor weight may be both 70%. In another example, adjusting the category of the popularity factor included in the first dynamic addition factor may include adding a new popularity factor category, such as a new income popularity factor, which is a popularity factor for users to purchase financial products in the same income interval of the same year, and the value of the popularity factor may be similar to other popularity factors. Thus, a second product recommendation policy may be generated based on the second policy weight vector and/or the second dynamic addition factor, and the product base policy. Further, the generation method of the second product recommendation policy may be the same as the generation method of the first product recommendation policy, and is not described herein again. It is understood that after the generation method of the second product recommendation policy is obtained, the second product recommendation policy may be evaluated based on the same evaluation method as the first product recommendation policy.

According to embodiments of the present disclosure, an effective policy may include more than one. To maximize the functionality of product recommendation strategies, embodiments of the present disclosure include the step of screening for optimal strategies. It is understood that when only one effective strategy is obtained through evaluation, the effective strategy is taken as the optimal strategy. Furthermore, when the effective strategies comprise n types, wherein n is more than or equal to 2 and less than or equal to m and n is an integer, the method also comprises the step of screening the optimal strategy from the n types of effective strategies.

Fig. 8 schematically shows a flow chart of a method of screening an optimal strategy according to an embodiment of the present disclosure.

As shown in fig. 8, the method of screening an optimal policy of this embodiment includes operation S810.

In operation S810, the n effective policies are sorted from large to small according to the first policy effectiveness index value, and the first effective policy is sorted as the optimal policy.

According to another embodiment of the disclosure, when there are q effective strategies that are ranked and juxtaposed first, where q satisfies that q is greater than or equal to 2 and less than or equal to n and q is an integer, policy effectiveness index values of different effective strategies may be updated by updating a traffic segmentation method to implement screening of an optimal strategy.

Fig. 9 schematically shows a flowchart of a method of screening an optimal policy according to another embodiment of the present disclosure.

As shown in fig. 9, the method of screening an optimal policy of the another embodiment includes operation S910.

In operation S910, the traffic split updating method is executed i times until there exists a first and only effective policy in the sequence, and the effective policy is marked as the optimal policy, where i is an integer greater than or equal to 1.

Fig. 10 schematically shows a flow chart of a traffic split updating method according to an embodiment of the present disclosure.

As shown in fig. 10, the traffic slice updating method of this embodiment includes operations S1010 to S1040.

In operation S1010, a flow distribution ratio in the first flow splitting method is adjusted to obtain a second flow splitting method.

In operation S1020, user polls are allocated to the q effective policies based on the second traffic slicing method.

In operation S1030, second policy validity index values of the q valid policies are acquired, the second policy validity index values being calculated based on the same method as the first policy validity index values.

In operation S1040, the q effective policies are sorted from large to small according to the second policy effectiveness index value.

According to the embodiment of the disclosure, the flow distribution proportion in the first flow splitting method can be adjusted to obtain the second flow splitting method. For example, the first product recommendation policy includes q product recommendation policies, and the first product recommendation policies are respectively marked as Y assuming that q is 4₀，Y₁，Y₂，Y₃In the first flow segmentation method, the flow proportion allocated to each first product recommendation strategy is 25% respectively. When the second flow segmentation method is obtained, the flow distribution proportion of each first product recommendation strategy can be adjusted to Y₀10% of Y₁30% of Y₂30% of Y₃30%, thereby increasing the evaluation of the dynamic addition factor. After the second traffic segmentation method is obtained, the user may be polled and allocated to q effective strategies based on the second traffic segmentation method, and the second strategy effectiveness index is calculated based on a method the same as the first strategy effectiveness index, which is not described herein again. After the q effective strategies are sorted from large to small according to the second strategy effectiveness index value, the optimal strategy can be obtained.

According to the embodiment of the disclosure, after the optimal strategy is obtained, the optimal strategy can be continuously evaluated based on the full amount of user access data, and the optimal strategy recommendation effect is verified again by comparing the operation results of the product basic recommendation strategy and the optimal strategy in a fixed time interval (for example, one month).

FIG. 11 schematically illustrates a flow chart of a method of evaluating an optimal policy according to an embodiment of the present disclosure.

As shown in fig. 11, the method for evaluating the optimal policy of this embodiment includes operations S1110 to S1130.

In operation S1110, a full number of users are assigned to the optimal policy.

In operation S1120, an optimal policy effectiveness index value is calculated based on the same method as the first policy effectiveness index value.

In operation S1130, the optimal policy is evaluated based on the optimal policy validity index value and a preset evaluation period.

According to the embodiment of the disclosure, the flow of the user is segmented by configuring a plurality of product recommendation strategies, and the user is allocated to different product recommendation strategies in a polling manner according to the proportion so as to evaluate each product recommendation strategy, which is beneficial to improving the effectiveness of the product recommendation strategies, thereby improving the transaction rate in the actual sale of the product. In the process, the optimal strategy can be obtained by methods of updating the traffic segmentation, updating the product recommendation strategy association factors and the like, so that continuous operation monitoring of the optimal strategy by using a full number of users is facilitated, and the effectiveness of the optimal strategy is continuously evaluated. The method can improve the penetration rate and success rate of online product marketing, and can recommend suitable online products to users in time when the users have product purchase requirements or regularly, improve the transaction operation convenience of the users, improve the viscosity of the users, and help different users to perform personalized product configuration planning.

Based on the product recommendation method, the disclosure also provides a product recommendation device. The apparatus will be described in detail below with reference to fig. 12.

Fig. 12 schematically shows a block diagram of a product recommendation device according to an embodiment of the present disclosure.

As shown in fig. 12, the product recommendation apparatus 1200 of this embodiment includes a first acquisition module 1210, a first processing module 1220, a second processing module 1230, a second acquisition module 1240, and a third processing module 1250.

The first obtaining module 1210 is configured to obtain a product recommendation policy pool, where the product recommendation policy pool includes m first product recommendation policies, and m is an integer greater than or equal to 2.

The first processing module 1220 is configured to assign user polls to the m first product recommendation policies based on the pool of product recommendation policies and a first traffic splitting method.

The second processing module 1230 is configured to generate a product recommendation list based on the first product recommendation policy assigned by the user and user access history data, wherein the user access history data includes a user history transaction product list and a user history browsing product list.

The second obtaining module 1240 is configured to obtain user access data associated with the product recommendation list.

The second processing module 1250 is configured to evaluate the m first product recommendation policies based on a user access data set, wherein the user access data set includes a set of all user access data in a test period.

According to the embodiment of the present disclosure, any plurality of the first obtaining module 1210, the first processing module 1220, the second processing module 1230, the second obtaining module 1240 and the third processing module 1250 may be combined into one module to be implemented, or any one of them may be split into a plurality of modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of the other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the first obtaining module 1210, the first processing module 1220, the second processing module 1230, the second obtaining module 1240 and the third processing module 1250 may be implemented at least in part as a hardware circuit, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or may be implemented by hardware or firmware in any other reasonable manner of integrating or packaging a circuit, or implemented by any one of three implementations of software, hardware and firmware, or by a suitable combination of any several of them. Alternatively, at least one of the first acquiring module 1210, the first processing module 1220, the second processing module 1230, the second acquiring module 1240 and the third processing module 1250 may be at least partially implemented as a computer program module, which when executed may perform the corresponding functions.

FIG. 13 schematically illustrates a block diagram of an electronic device suitable for implementing a method of product recommendation in accordance with an embodiment of the present disclosure.

As shown in fig. 13, an electronic apparatus 900 according to an embodiment of the present disclosure includes a processor 901 which can perform various appropriate actions and processes in accordance with a program stored in a Read Only Memory (ROM)902 or a program loaded from a storage portion 908 into a Random Access Memory (RAM) 903. Processor 901 may comprise, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or associated chipset, and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), among others. The processor 901 may also include on-board memory for caching purposes. The processor 901 may comprise a single processing unit or a plurality of processing units for performing the different actions of the method flows according to embodiments of the present disclosure.

In the RAM 903, various programs and data necessary for the operation of the electronic apparatus 900 are stored. The processor 901, the ROM 902, and the RAM 903 are connected to each other through a bus 904. The processor 901 performs various operations of the method flows according to the embodiments of the present disclosure by executing programs in the ROM 902 and/or the RAM 903. Note that the programs may also be stored in one or more memories other than the ROM 902 and the RAM 903. The processor 901 may also perform various operations of the method flows according to embodiments of the present disclosure by executing programs stored in the one or more memories.

Electronic device 900 may also include input/output (I/O) interface 905, input/output (I/O) interface 905 also connected to bus 904, according to an embodiment of the present disclosure. The electronic device 900 may also include one or more of the following components connected to the I/O interface 905: an input portion 906 including a keyboard, a mouse, and the like; an output section 907 including components such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 908 including a hard disk and the like; and a communication section 909 including a network interface card such as a LAN card, a modem, or the like. The communication section 909 performs communication processing via a network such as the internet. The drive 910 is also connected to the I/O interface 905 as necessary. A removable medium 911 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 910 as necessary, so that a computer program read out therefrom is mounted into the storage section 908 as necessary.

The present disclosure also provides a computer-readable storage medium, which may be contained in the apparatus/device/system described in the above embodiments; or may exist separately and not be assembled into the device/apparatus/system. The computer-readable storage medium carries one or more programs which, when executed, implement a product recommendation method according to an embodiment of the present disclosure.

According to embodiments of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example but is not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to embodiments of the present disclosure, a computer-readable storage medium may include the ROM 902 and/or the RAM 903 described above and/or one or more memories other than the ROM 902 and the RAM 903.

Embodiments of the present disclosure also include a computer program product comprising a computer program containing program code for performing the method illustrated in the flow chart. When the computer program product runs in a computer system, the program code is used for causing the computer system to realize the product recommendation method provided by the embodiment of the disclosure.

The computer program performs the above-described functions defined in the system/apparatus of the embodiments of the present disclosure when executed by the processor 901. The systems, apparatuses, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the present disclosure.

In one embodiment, the computer program may be hosted on a tangible storage medium such as an optical storage device, a magnetic storage device, or the like. In another embodiment, the computer program may also be transmitted, distributed in the form of a signal on a network medium, and downloaded and installed through the communication section 909 and/or installed from the removable medium 911. The computer program containing program code may be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 909, and/or installed from the removable medium 911. The computer program, when executed by the processor 901, performs the above-described functions defined in the system of the embodiment of the present disclosure. The systems, devices, apparatuses, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the present disclosure.

In accordance with embodiments of the present disclosure, program code for executing computer programs provided by embodiments of the present disclosure may be written in any combination of one or more programming languages, and in particular, these computer programs may be implemented using high level procedural and/or object oriented programming languages, and/or assembly/machine languages. The programming language includes, but is not limited to, programming languages such as Java, C + +, python, the "C" language, or the like. The program code may execute entirely on the user computing device, partly on the user device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).

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

Those skilled in the art will appreciate that various combinations and/or combinations of features recited in the various embodiments and/or claims of the present disclosure can be made, even if such combinations or combinations are not expressly recited in the present disclosure. In particular, various combinations and/or combinations of the features recited in the various embodiments and/or claims of the present disclosure may be made without departing from the spirit or teaching of the present disclosure. All such combinations and/or associations are within the scope of the present disclosure.

The embodiments of the present disclosure have been described above. However, these examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments are described separately above, this does not mean that the measures in the embodiments cannot be used in advantageous combination. The scope of the disclosure is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be devised by those skilled in the art without departing from the scope of the present disclosure, and such alternatives and modifications are intended to be within the scope of the present disclosure.

Claims (20)

1. A method for recommending products, comprising:

obtaining a product recommendation strategy pool, wherein the product recommendation strategy pool comprises m first product recommendation strategies, and m is an integer greater than or equal to 2;

polling and distributing users to the m first product recommendation strategies based on the product recommendation strategy pool and a first traffic segmentation method;

generating a product recommendation list based on a first product recommendation strategy distributed by a user and user access historical data, wherein the user access historical data comprises a user historical transaction product list and a user historical browsing product list;

acquiring user access data, wherein the user access data is associated with a product recommendation list; and

evaluating the m first product recommendation strategies based on a user access data set, wherein the user access data set comprises a set of all user access data in a test period.

2. The method of claim 1, wherein the first product recommendation policy is generated based on a product base policy and a first dynamic addition factor.

3. The method of claim 2, wherein the product base policy comprises:

and calculating a basic product recommendation score for the to-be-recommended products contained in the to-be-recommended product list, wherein the basic product recommendation score is calculated based on the user transaction time factor, the user browsing time factor, the transaction product similarity and the browsing product similarity.

4. The method of claim 3, wherein the user transaction time factor and the user browsing time factor are calculated based on a time factor algorithm, the time factor algorithm comprising:

calculating a transaction time interval between a historical product transaction date and a current visit date, wherein the historical product transaction date is obtained based on the user historical transaction product list;

calculating the browsing time interval between the browsing date of the historical products and the current visiting date, wherein the browsing date of the historical products is obtained based on the historical browsing product list of the user;

acquiring the user transaction time factor based on the transaction time interval and the product deadline attribute; and

and acquiring the user browsing time factor based on the browsing time interval and the product term attribute.

5. The method of claim 3, wherein the method of generating the deal product similarity and the browse product similarity comprises:

calculating the similarity between the product to be recommended and the historical trading product of the user based on the product similarity index and the preset similarity index configuration, and acquiring the similarity of the trading product; and

and calculating the similarity between the product to be recommended and the historical browsing product of the user based on the product similarity index and the preset similarity index configuration, and acquiring the similarity of the browsing product.

6. The method of claim 5, wherein the product similarity index comprises two or more of product term, sales channel, risk level, purchase amount, product type, transaction currency, and credit status.

7. The method of claim 2, wherein the first dynamic addition factor comprises at least one trending factor, the trending factor comprising an asterisk trending factor, an age trending factor, or a risk assessment trending factor.

8. The method of claim 7, wherein the star popularity factor is determined based on historical access data of same-star users; the age trending factor is determined based on historical access data of users in the same age group; the risk evaluation popularity factor is determined based on historical access data of users with the same risk evaluation grade, wherein the user star grade, the user age group and the user risk evaluation grade are determined based on preset evaluation rules.

9. The method of claim 2, wherein the first product recommendation policy being generated based on a product base policy and a first dynamic addition factor comprises:

obtaining a first strategy weight vector, wherein the first strategy weight vector comprises a product basic strategy weight and a first dynamic addition factor weight; and

generating the first product recommendation policy based on the product base policy, the first dynamic addition factor, and the first policy weight vector.

10. The method of claim 1, wherein said evaluating the m first product recommendation policies based on a set of user access data further comprises:

calculating a strategy evaluation index value based on the user access data set, wherein the strategy evaluation index value comprises a product basic strategy evaluation index value and a first product recommendation strategy evaluation index value;

calculating a first policy validity index value based on the product base policy evaluation index value and the first product recommendation policy evaluation index value; and

when a first strategy effectiveness index value corresponding to the kth first product recommendation strategy is larger than or equal to a preset threshold value, marking the kth first product recommendation strategy as an effective strategy, wherein k is larger than or equal to 1 and is smaller than or equal to m, and k is an integer.

11. The method of claim 10, wherein when the first policy effectiveness index values corresponding to all of the first product recommendation policies are less than a preset threshold, the method further comprises:

executing the product recommendation strategy updating method for j times, marking the second product recommendation strategy as an effective strategy when a second strategy effectiveness index value corresponding to the second product recommendation strategy is larger than or equal to a preset threshold value, wherein j is an integer larger than or equal to 1,

the product recommendation strategy updating method comprises the following steps:

obtaining at least one of a second strategy weight vector or a second dynamic addition factor, wherein the second strategy weight vector is obtained by adjusting the relative proportion of the product basic strategy weight and the first dynamic addition factor weight in the first strategy weight vector, and the second dynamic addition factor is obtained by adjusting the type of the hot degree factor contained in the first dynamic addition factor;

generating a second product recommendation strategy based on the second strategy weight vector and/or a second dynamic addition factor and a product basic strategy; and

and evaluating the second product recommendation strategy based on the same evaluation method as the first product recommendation strategy.

12. The method of claim 10 or 11, wherein when n valid policies are included, where n satisfies 2 ≦ n ≦ m and n is an integer, the method further comprising:

and sequencing the n effective strategies from large to small according to the first strategy effectiveness index value, and taking the sequenced first effective strategy as an optimal strategy.

13. The method of claim 12, wherein when there are q active policies that are first ordered in parallel, where q satisfies 2 ≦ q ≦ n and q is an integer, the method further comprising:

executing the flow segmentation updating method for i times until a first and only effective strategy is sorted, marking the effective strategy as an optimal strategy, wherein i is an integer greater than or equal to 1,

the flow segmentation updating method comprises the following steps:

adjusting the flow distribution proportion in the first flow segmentation method to obtain a second flow segmentation method;

allocating user polling to the q effective strategies based on the second traffic segmentation method;

acquiring second strategy effectiveness index values of the q effective strategies, wherein the second strategy effectiveness index values are calculated based on the same method as the first strategy effectiveness index values;

and sequencing the q effective strategies from large to small according to a second strategy effectiveness index value.

14. The method of claim 12, wherein after obtaining the optimal policy, the method further comprises:

distributing a full amount of users to the optimal strategy;

calculating an optimal policy effectiveness index value based on the same method as the first policy effectiveness index value;

and evaluating the optimal strategy based on the optimal strategy effectiveness index value and a preset evaluation period.

15. The method of claim 10, wherein the user access data comprises page statistics and access behavior data, the page statistics including at least contact identification, product recommendation policy, policy invocation amount, and product recommendation number; the access behavior data is obtained based on at least one of a user transaction behavior and a user browsing behavior.

16. The method of claim 15, wherein the policy evaluation metrics include a recommended click rate and a purchase conversion rate.

17. A product recommendation device, comprising:

the system comprises a first obtaining module, a second obtaining module and a third obtaining module, wherein the first obtaining module is configured to obtain a product recommendation strategy pool, the product recommendation strategy pool comprises m first product recommendation strategies, and m is an integer greater than or equal to 2;

the first processing module is configured to distribute user polling to the m first product recommendation strategies based on the product recommendation strategy pool and a first traffic segmentation method;

the second processing module is configured to generate a product recommendation list based on the first product recommendation strategy distributed by the user and user access historical data, wherein the user access historical data comprises a user historical transaction product list and a user historical browsing product list;

the second acquisition module is configured to acquire user access data, and the user access data is associated with the product recommendation list; and

a third processing module configured to evaluate the m first product recommendation policies based on a user access data set, wherein the user access data set includes a set of all user access data in a test period.

18. An electronic device, comprising:

one or more processors;

a storage device for storing one or more programs,

wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method of any of claims 1-16.

19. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to perform the method of any one of claims 1 to 16.

20. A computer program product comprising a computer program which, when executed by a processor, implements a method according to any one of claims 1 to 16.

Images