CN111127155A

CN111127155A - Commodity recommendation method, commodity recommendation device, server and storage medium

Info

Publication number: CN111127155A
Application number: CN201911345564.8A
Authority: CN
Inventors: 盛宇佳
Original assignee: Beijing Missfresh Ecommerce Co Ltd
Current assignee: Beijing Daily Youxian Technology Co.,Ltd.
Priority date: 2019-12-24
Filing date: 2019-12-24
Publication date: 2020-05-08

Abstract

The application discloses a commodity recommendation method, a commodity recommendation device, a server and a storage medium, and belongs to the technical field of the Internet. The commodity recommendation method provided by the embodiment of the application responds to a browsing request of a terminal and determines a plurality of target commodities; the method comprises the steps of obtaining commodity characteristics of a target commodity, obtaining user characteristics of a target user, and obtaining purchasing characteristics of the target user on the target commodity, wherein the target user is a user using a terminal; determining the recommendation probability of the target commodity according to the commodity characteristics, the user characteristics and the purchasing characteristics; determining the display position of the target commodity in the commodity interface to be requested according to the recommendation probability; and returning the commodity information and the display position of the target commodity to the terminal. According to the method, the recommendation probabilities of different target commodities are different, the more the recommendation probability of the current commodity is, the more the display position of the current commodity is forward, the more the display position is obvious, so that the exposure rate of the current commodity is improved, the promotion effect of the current commodity is improved, and the processing efficiency is high.

Description

Commodity recommendation method, commodity recommendation device, server and storage medium

Technical Field

The present application relates to the field of internet technologies, and in particular, to a method, an apparatus, a server, and a storage medium for recommending a commodity.

Background

With the development of internet technology, more and more fresh electric power suppliers emerge. The fresh commodity and the fresh commodity are sold on the Internet by means of electronic commerce, such as fruits, vegetables or meat. The user can purchase the required fresh goods through the client installed on the terminal. In order to reduce loss, merchants can deal with the temporary commodities through various promotion means.

In the related technology, an operator sets the display position of a fresh commodity on a client page in advance, and when the commodity becomes a temporary commodity, price reduction processing is performed on the commodity, and the commodity is recommended to a user at the original display position of the commodity.

But the user may not notice the promotion information of the temporary merchandise, resulting in poor promotion effect of the temporary merchandise and low processing efficiency.

Disclosure of Invention

The embodiment of the application provides a commodity recommendation method, a commodity recommendation device, a server and a storage medium, and can solve the problems of poor promotion effect and low processing efficiency of a temporary commodity. The technical scheme is as follows:

in one aspect, a method for recommending goods is provided, the method comprising:

responding to a browsing request of a terminal, and determining a plurality of target commodities, wherein the target commodities are current commodities close to the quality guarantee period;

acquiring the commodity characteristics of the target commodity, acquiring the user characteristics of a target user, and acquiring the purchase characteristics of the target user on the target commodity, wherein the target user is a user using the terminal;

determining the recommendation probability of the target commodity according to the commodity characteristics, the user characteristics and the purchase characteristics;

determining the display position of the target commodity in a commodity interface to be requested according to the recommendation probability;

and returning the commodity information and the display position of the target commodity to the terminal.

In a possible implementation manner, the obtaining the commodity characteristics of the target commodity includes:

acquiring the first number of times that the target commodity is clicked, the sales number and the first number of times that the target commodity is added into a shopping cart in at least one first historical time range;

and one or more of the first number, the sales number and the first number are combined into the commodity characteristics of the target commodity.

In another possible implementation manner, the obtaining the user characteristics of the target user includes:

acquiring a second frequency, a commodity purchasing quantity, an average consumption level and terminal information of the terminal, wherein the second frequency, the commodity purchasing quantity and the average consumption level are used by the target user to access the current server by using the terminal within at least one first historical time range;

and forming one or more of the second times, the times of purchasing commodities, the quantity of purchased commodities, the average consumption level and the terminal information into the user characteristics of the target user.

In another possible implementation manner, the obtaining of the purchase characteristics of the target user for the target product includes:

determining a commodity category to which the target commodity belongs;

acquiring a first purchase record, a first vehicle adding record and a first click record of the target user for the target commodity, and a second purchase record, a second vehicle adding record and a second click record of the target user for the commodity category within at least one first historical time range;

and forming the purchase characteristics of the target user on the target commodity by one or more of the first purchase record, the first car adding record, the first click record, the second purchase record, the second car adding record and the second click record.

In another possible implementation manner, the determining the recommendation probability of the target product according to the product feature, the user feature, and the purchase feature includes:

composing the merchandise feature, the user feature, and the purchase feature into a recommendation feature;

inputting the recommendation characteristics into a first recommendation model to obtain a recommendation score of the target commodity;

and taking the recommendation score as the recommendation probability.

In another possible implementation manner, the combining the item feature, the user feature, and the purchase feature into a recommendation feature includes:

determining a front bin corresponding to the area where the terminal is located;

acquiring the characteristics of the front bin;

acquiring the association characteristics of the associated commodities related to the target commodity;

composing the merchandise feature, the user feature, the purchase feature, the pre-bin feature, and the association feature into the recommendation feature.

In another possible implementation manner, the method further includes:

determining the association probability of the associated commodity and the target commodity according to the association characteristics;

determining the display position of the associated commodity in the commodity interface according to the association probability;

and returning the display position and the commodity information of the associated commodity to the terminal.

In another possible implementation, the determining a plurality of target commodities includes:

acquiring user information of the target user according to the user identification of the target user carried in the browsing request;

according to the user information, selecting a temporary commodity matched with the user information from a plurality of temporary commodities in a front bin corresponding to the terminal;

and taking the selected temporary commodity as the target commodity.

In another possible implementation manner, the method further includes:

obtaining sample recommendation characteristics of sample commodities and sample recommendation scores of the sample commodities;

inputting the sample recommendation characteristics of the sample commodities into a second recommendation model to obtain click rate, conversion rate, overdue probability, selling price, commodity price entering and overdue weight of the sample commodities;

determining a training score of the sample commodity according to the click rate, the conversion rate, the expiration probability, the sales price, the commodity intake price and the expiration weight;

determining a score difference between the training score and the sample recommendation score;

and adjusting the model parameters of the second recommendation model according to the grading difference to obtain the first recommendation model.

In another possible implementation manner, the determining a training score of the sample good according to the click rate, the conversion rate, the expiration probability, the sale price, the good entrance price, and the expiration weight includes:

determining a profit score of the sample commodity according to the click rate, the conversion rate and the sales price;

determining a loss score of the sample commodity according to the click rate, the conversion rate, the expiration probability, the commodity intake price and the expiration weight;

taking the sum of the profit score and the loss score as the training score.

In another possible implementation manner, the returning the commodity information and the display position of the target commodity to the terminal includes:

acquiring the inventory quantity of the target commodity;

and when the inventory quantity of the target commodity is zero, filtering the target commodity, and returning the commodity information and the display position of the remaining target commodity to the terminal.

In another aspect, there is provided an article recommendation apparatus, the apparatus including:

the system comprises a first determining module, a second determining module and a display module, wherein the first determining module is used for responding to a browsing request of a terminal and determining a plurality of target commodities, and the target commodities are current commodities close to the quality guarantee period;

the first acquisition module is used for acquiring the commodity characteristics of the target commodity, acquiring the user characteristics of a target user and acquiring the purchase characteristics of the target user on the target commodity, wherein the target user is a user using the terminal;

the second determination module is used for determining the recommendation probability of the target commodity according to the commodity characteristics, the user characteristics and the purchase characteristics;

the third determining module is used for determining the display position of the target commodity in a commodity interface to be requested according to the recommendation probability;

and the return module is used for returning the commodity information and the display position of the target commodity to the terminal.

In a possible implementation manner, the first obtaining module is further configured to obtain a first number of times the target product is clicked, a sales number, and a first number of times the target product is added to a shopping cart in at least one first historical time range; and one or more of the first number, the sales number and the first number are combined into the commodity characteristics of the target commodity.

In another possible implementation manner, the first obtaining module is further configured to obtain a second number of times that the target user accesses the current server using the terminal, a number of times of purchasing commodities, a number of purchased commodities, an average consumption level, and terminal information of the terminal within at least one first historical time range; and forming one or more of the second times, the times of purchasing commodities, the quantity of purchased commodities, the average consumption level and the terminal information into the user characteristics of the target user.

In another possible implementation manner, the first obtaining module is further configured to determine a commodity category to which the target commodity belongs; acquiring a first purchase record, a first vehicle adding record and a first click record of the target user for the target commodity, and a second purchase record, a second vehicle adding record and a second click record of the target user for the commodity category within at least one first historical time range; and forming the purchase characteristics of the target user on the target commodity by one or more of the first purchase record, the first car adding record, the first click record, the second purchase record, the second car adding record and the second click record.

In another possible implementation manner, the second determining module is further configured to combine the commodity feature, the user feature and the purchase feature into a recommendation feature; inputting the recommendation characteristics into a first recommendation model to obtain a recommendation score of the target commodity; and taking the recommendation score as the recommendation probability.

In another possible implementation manner, the second determining module is further configured to determine a front bin corresponding to an area where the terminal is located; acquiring the characteristics of the front bin; acquiring the association characteristics of the associated commodities related to the target commodity; composing the merchandise feature, the user feature, the purchase feature, the pre-bin feature, and the association feature into the recommendation feature.

In another possible implementation manner, the apparatus further includes:

a third determining module, configured to determine, according to the association feature, an association probability between the associated product and the target product;

the fourth determining module is used for determining the display position of the associated commodity in the commodity interface according to the association probability;

the return module is further configured to return the display position of the associated commodity and the commodity information to the terminal.

In another possible implementation manner, the first determining module is further configured to obtain user information of the target user according to the user identifier of the target user carried in the browsing request; according to the user information, selecting a temporary commodity matched with the user information from a plurality of temporary commodities in a front bin corresponding to the terminal; and taking the selected temporary commodity as the target commodity.

In another possible implementation manner, the apparatus further includes:

the second acquisition module is used for acquiring sample recommendation characteristics of sample commodities and sample recommendation scores of the sample commodities;

the input module is used for inputting the sample recommendation characteristics of the sample commodities into a second recommendation model to obtain the click rate, the conversion rate, the overdue probability, the sale price, the commodity entrance price and the overdue weight of the sample commodities;

a fifth determining module, configured to determine a training score of the sample commodity according to the click rate, the conversion rate, the expiration probability, the sales price, the commodity intake price, and the expiration weight;

a sixth determining module for determining a score difference between the training score and the sample recommendation score;

and the adjusting module is used for adjusting the model parameters of the second recommendation model according to the grading difference value to obtain the first recommendation model.

In another possible implementation manner, the fifth determining module is further configured to determine a revenue score of the sample commodity according to the click rate, the conversion rate, and the sales price; determining a loss score of the sample commodity according to the click rate, the conversion rate, the expiration probability, the commodity intake price and the expiration weight; taking the sum of the profit score and the loss score as the training score.

In another possible implementation manner, the return module is further configured to obtain the inventory quantity of the target product; and when the inventory quantity of the target commodity is zero, filtering the target commodity, and returning the commodity information and the display position of the remaining target commodity to the terminal.

In another aspect, a server is provided, which includes a processor and a memory, where at least one program code is stored in the memory, and the at least one program code is loaded and executed by the processor to implement the operation of any one of the above commodity recommendation methods.

In another aspect, a computer-readable storage medium is provided, in which at least one program code is stored, and the at least one program code is loaded and executed by a processor to implement the operations of any one of the above-mentioned article recommendation methods.

The technical scheme provided by the embodiment of the application has the following beneficial effects:

the commodity recommendation method provided by the embodiment of the application responds to a browsing request of a terminal, and determines a plurality of target commodities, wherein the target commodities are imminent commodities close to the quality guarantee period; the method comprises the steps of obtaining commodity characteristics of a target commodity, obtaining user characteristics of a target user, and obtaining purchasing characteristics of the target user on the target commodity, wherein the target user is a user using a terminal; determining the recommendation probability of the target commodity according to the commodity characteristics, the user characteristics and the purchasing characteristics; determining the display position of the target commodity in the commodity interface to be requested according to the recommendation probability; and returning the commodity information and the display position of the target commodity to the terminal. According to the method, the recommendation probabilities of different target commodities are different, the more the recommendation probability is, the more the display position of the temporary commodity is forward, the more the display position is obvious, so that the exposure rate of the temporary commodity is improved, a user can more visually notice the promotion information of the temporary commodity, the promotion effect of the temporary commodity is improved, and the processing efficiency is high.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

FIG. 1 is a schematic diagram of an implementation environment of merchandise recommendation provided by an embodiment of the present application;

fig. 2 is a flowchart of a commodity recommendation method according to an embodiment of the present application;

fig. 3 is a flowchart of a commodity recommendation method according to an embodiment of the present application;

FIG. 4 is a schematic diagram illustrating a server training to obtain a first recommendation model according to an embodiment of the present application;

fig. 5 is a schematic diagram illustrating a server sending an ordered list to a terminal through a first recommendation model according to an embodiment of the present application;

fig. 6 is a schematic structural diagram of a commodity recommending device according to an embodiment of the present application;

fig. 7 is a block diagram of a server according to an embodiment of the present disclosure.

Detailed Description

In order to make the technical solutions and advantages of the present application more clear, the following describes the embodiments of the present application in further detail.

An embodiment of the present application provides an implementation environment for commodity recommendation, and referring to fig. 1, the implementation environment includes: a terminal 101 and a server 102. The terminal 101 and the server 102 may be connected by wireless or wired connection. The terminal 101 is installed with a target client, and the server 102 is a server 102 corresponding to the target client. The target client may be a client for selling any commodity, for example, the target client may be a client for selling clothing, a client for selling fresh food, a client for selling cosmetics, and the like. Some fresh commodities may have temporary commodities, and in order to reduce loss, merchants deal with temporary commodities through various sales promotion means.

However, in the related art, the temporary commodities are subjected to price reduction processing at the original display positions of the temporary commodities, the temporary commodities cannot obtain enough exposure, and users may not obtain promotion information of the temporary commodities, so that the promotion effect of the temporary commodities is poor, and the processing efficiency is low.

In the commodity recommendation method provided by the embodiment of the application, the server 102 determines the recommendation probability of the target commodity according to the commodity characteristics of the target commodity, the user characteristics of the target user and the purchase characteristics of the target user on the target commodity. The target product is a product whose expiration date is close to the expiration date, and the target user is a user using the terminal 101. The server 102 recommends the display position of the target commodity in the commodity interface to the terminal 101 according to the recommendation probability of the target commodity, the recommendation probabilities of different target commodities are different, and the display position of the current commodity with the higher recommendation probability is more obvious as the display position is more forward, so that the exposure rate of the current commodity is improved, a user can more intuitively notice the promotion information of the current commodity, the promotion effect of the current commodity is improved, and the processing efficiency is high.

In addition, in the related art, price reduction processing is performed at the original display position of the current commodity, the personalized requirements of the user cannot be met, thousands of people are not recommended, the user preference is not considered, and therefore the promotion effect of the current commodity is poor to a certain extent. In the commodity recommendation method provided by the embodiment of the application, the recommendation probability of the target commodity is determined by the server 102 based on the commodity characteristics of the target commodity, the user characteristics of the target user and the purchase characteristics of the target commodity by the target user, and the method considers the preferences of different users to different commodities and the preferences of different categories, so that the thousands of people and thousands of faces of personalized recommendation can be realized to the preferences of the temporary commodity, the sale speed of the temporary commodity can be increased, and the overall profit of the temporary commodity can be increased.

An embodiment of the present application provides a method for recommending a commodity, and with reference to fig. 2, the method includes:

step 201: and responding to a browsing request of the terminal, and determining a plurality of target commodities, wherein the target commodities are current commodities with the shelf life close to.

Step 202: the method comprises the steps of obtaining commodity characteristics of a target commodity, obtaining user characteristics of a target user, and obtaining purchasing characteristics of the target user on the target commodity, wherein the target user is a user using a terminal.

Step 203 determines the recommendation probability of the target commodity according to the commodity characteristics, the user characteristics and the purchasing characteristics.

Step 204: and determining the display position of the target commodity in the commodity interface to be requested according to the recommendation probability.

Step 205: and returning the commodity information and the display position of the target commodity to the terminal.

In one possible implementation manner, the acquiring the commodity characteristics of the target commodity includes:

acquiring the first number of times that a target commodity is clicked, the sales number and the first number of times that the target commodity is added into a shopping cart in at least one first historical time range;

one or more of the first number, the sales number, and the first number are combined into the commodity characteristics of the target commodity.

acquiring a second frequency, commodity purchasing quantity, average consumption level and terminal information of a terminal, wherein the second frequency, the commodity purchasing quantity and the average consumption level are used by a target user to access a current server by using the terminal within at least one first historical time range;

and forming one or more of the second time, the commodity purchasing quantity, the average consumption level and the terminal information into the user characteristics of the target user.

In another possible implementation manner, the obtaining of the purchase characteristics of the target user on the target product includes:

determining a commodity category to which the target commodity belongs;

acquiring a first purchase record of a target user on a target commodity, a first car adding record and a first click record of adding into a shopping car, and a second purchase record of the target user on a commodity category, a second car adding record and a second click record of adding into the shopping car within at least one first historical time range;

In another possible implementation manner, determining the recommendation probability of the target product according to the product feature, the user feature and the purchase feature includes:

the commodity characteristics, the user characteristics and the purchase characteristics form recommendation characteristics;

inputting the recommendation characteristics into a recommendation model to obtain a recommendation score of the target commodity;

and taking the recommendation score as the recommendation probability.

In another possible implementation, combining the goods feature, the user feature, and the purchase feature into a recommendation feature includes:

determining a front bin corresponding to an area where a terminal is located;

acquiring the characteristics of a front bin;

and combining the commodity characteristics, the user characteristics, the purchase characteristics, the pre-bin characteristics and the association characteristics into recommendation characteristics.

In another possible implementation manner, the method further includes:

and returning the display position of the associated commodity and the commodity information to the terminal.

In another possible implementation, determining a plurality of target goods includes:

acquiring user information of a target user according to a user identifier of the target user carried in a browsing request;

selecting a temporary commodity matched with the user information from a plurality of temporary commodities in a front bin corresponding to the terminal according to the user information;

and taking the selected temporary commodity as a target commodity.

In another possible implementation manner, the method further includes:

acquiring sample recommendation characteristics of sample commodities and sample recommendation scores of the sample commodities;

determining training scores of the sample commodities according to the click rate, the conversion rate, the overdue probability, the sale price, the commodity entrance price and the overdue weight;

In another possible implementation manner, determining a training score of a sample commodity according to a click rate, a conversion rate, an expiration probability, a sales price, a commodity entry price and an expiration weight includes:

determining the income score of the sample commodity according to the click rate, the conversion rate and the sale price;

determining a loss score of the sample commodity according to the click rate, the conversion rate, the overdue probability, the commodity incoming price and the overdue weight;

the sum of the profit and loss scores is taken as the training score.

In another possible implementation manner, returning the commodity information and the display position of the target commodity to the terminal includes:

acquiring the inventory quantity of target commodities;

and when the stock quantity of the target commodity is zero, filtering the target commodity, and returning the commodity information and the display position of the remaining target commodity to the terminal.

An embodiment of the present application provides a commodity recommendation method, see fig. 3, applied to a terminal and a server, where the method includes:

step 301: and the terminal sends a browsing request to the server, wherein the browsing request carries the user identification of the target user.

When a target user accesses a target client installed on a terminal, the terminal sends a browsing request to a server, wherein the browsing request carries a user identifier. The server is a server corresponding to the target client. The target client may be a fresh sales client, a clothing sales client, a cosmetics sales client, and the like, in this embodiment, the target client is only taken as the fresh sales client for example to explain, and correspondingly, the server is a server corresponding to the fresh sales client.

Step 302: the server receives a browsing request from the terminal and determines a plurality of target products.

The target commodity is a current commodity approaching shelf life.

In one possible implementation, the server may determine the plurality of target goods by: the server receives a browsing request sent by the terminal, and acquires user information of a target user according to a user identifier in the browsing request; according to the user information, selecting a temporary commodity matched with the user information from a plurality of temporary commodities in a front bin corresponding to the terminal; and taking the selected temporary commodity as a target commodity.

In the implementation mode, the front warehouse is a warehouse corresponding to an area where the terminal is located, and multiple types of temporary commodities are stored in the front warehouse. The server selects the temporary commodity matched with the user information from the preposed bin according to the user information, the temporary commodity matched with the user information is a commodity purchased by a target user once or a commodity intentionally purchased by the target user, and the user can be recommended in a personalized manner subsequently, so that the processing efficiency of the temporary commodity is improved.

The server can acquire one or more of a record of historical purchased commodities, a record of currently browsed commodities, a record of currently clicked commodities and a record of commodities currently added into the shopping cart of the target user according to the user identification, so that the user information of the target user is acquired.

In another possible implementation manner, the server may take a plurality of temporary commodities in a front bin corresponding to the terminal as a plurality of target commodities. In another possible implementation, the server may also take a plurality of temporary commodities in the central bin as a plurality of target commodities. In the embodiment of the present application, the manner in which the server determines the plurality of target commodities is not particularly limited.

Step 303: the server acquires the commodity characteristics of the target commodity, acquires the user characteristics of the target user and acquires the purchase characteristics of the target user on the target commodity.

In one possible implementation manner, the server may obtain the commodity characteristics of the target commodity by the following steps: the server acquires the first number of times that the target commodity is clicked, the sales number and the first number of times that the target commodity is added into a shopping cart in at least one first historical time range; one or more of the first number, the sales number, and the first number are combined into the commodity characteristics of the target commodity.

In this implementation manner, the first historical time range may be set and changed as needed, and the number of the first historical time ranges may also be set and changed as needed, which is not specifically limited in this embodiment of the application. For example, the first history time range may be one or more of 1 day, 3 days, 7 days, 14 days, and 30 days, when the first history time range is 1 day, 3 days, 7 days, 14 days, and 30 days, the server composes the first number of times, the sales number, and the first number into the commodity feature, the server acquires the first number of times, the sales number, and the first number added to the shopping cart of the target commodity clicked within the past 1 day, the server acquires the first number of times, the sales number, and the first number added to the shopping cart of the target commodity clicked within the past 3 days, the server acquires the first number of times, the sales number, and the first number added to the shopping cart of the target commodity clicked within the past 7 days, the server acquires the first number of times, the sales number, and the first number added to the shopping cart of the target commodity clicked within the past 14 days, and the server acquires the first number of times, the sales number, and the first number of times, the target commodity clicked within the past 30 days, The sales amount and the first amount added to the shopping cart, and the server combines the first number of times the target commodity is clicked, the sales amount and the first amount in each first historical time range into the commodity characteristics of the target commodity.

In one possible implementation manner, the server may obtain the user characteristics of the target user by the following steps: the server acquires second times, commodity purchasing quantity, average consumption level and terminal information of the terminal, wherein the second times, the commodity purchasing quantity and the average consumption level are used by a target user to access the current server by using the terminal in at least one first historical time range; and forming one or more of the second time, the commodity purchasing quantity, the average consumption level and the terminal information into the user characteristics of the target user.

In this implementation manner, the terminal information of the terminal may be a type of an operating system, that is, the terminal is an IOS operating system or an Android operating system. When the first history time range is 1 day, 3 days, 7 days, 14 days and 30 days, the server composes the second times, the commodity purchasing quantity, the average consumption level and the terminal information into the user characteristics of the target user, the server obtains the second times, the commodity purchasing level, the average consumption level and the terminal information of the target user accessing the current server by using the terminal in the past 1 day, the server obtains the second times, the commodity purchasing quantity, the average consumption level and the terminal information of the target user accessing the current server by using the terminal in the past 3 days, the server obtains the second times, the commodity purchasing quantity, the average consumption level and the terminal information of the target user accessing the current server by using the terminal in the past 7 days, and the server obtains the second times, the commodity purchasing level, the average consumption level and the terminal information of the target user accessing the current server by using the terminal in, The number of times of purchasing commodities, the number of purchased commodities, the average consumption level and the terminal information are purchased, the server acquires a second number of times of accessing the current server by the target user through the terminal in the last 30 days, the number of times of purchasing commodities, the number of purchased commodities, the average consumption level and the terminal information, and the server combines the second number of times of accessing the current server by the target user through the terminal, the number of times of purchasing commodities, the number of purchased commodities, the average consumption level and the terminal information in each first historical time range into the user characteristics of the target user.

In addition, in this implementation, the number of times of purchasing commodities may be the number of times of purchasing commodities within a historical time range from the time when the target user registers the server to the time when the target user currently accesses the server, the number of purchased commodities may be the number of purchased commodities within a historical time range from the time when the target user registers the server to the time when the target user currently accesses the server, and the average consumption level may be the consumption level of the target user within a unit time range from the time when the target user registers the server to the time when the target user currently accesses the server. In the embodiments of the present application, this is not particularly limited.

In one possible implementation manner, the server may obtain the purchase characteristics of the target commodity by the target user through the following steps: the server determines the commodity category to which the target commodity belongs; acquiring a first purchase record, a first vehicle adding record and a first click record of a target user for a target commodity, and a second purchase record, a second vehicle adding record and a second click record of the target user for the commodity category in at least one first historical time range; and forming the purchase characteristics of the target user on the target commodity by one or more of the first purchase record, the first car adding record, the first click record, the second purchase record, the second car adding record and the second click record.

In the implementation mode, the server not only obtains the cross characteristics of the target user and the target commodity, namely the first purchase record, the first car adding record and the first click record of the target user on the target commodity, but also obtains the cross characteristics of the categories of the target user and the target commodity, namely the second purchase record, the second car adding record and the second click record of the target user on the commodity category, so that the obtained purchase characteristics are more comprehensive, and the temporary commodity can be specifically and individually recommended to the user subsequently according to the purchase characteristics.

When the first historical time range is 1 day, 3 days, 7 days, 14 days and 30 days, the server combines the first purchase record, the first car adding record, the first click record, the second purchase record, the second car adding record and the second click record into a purchase characteristic, the server acquires the first purchase record, the first car adding record and the first click record of the target user on the target commodity in the past 1 day, and the second purchase record, the second car adding record and the second click record of the target user on the commodity category, the server acquires the first purchase record, the first car adding record and the first click record of the target user on the target commodity in the past 3 days, and the second purchase record, the second car adding record and the second click record of the target user on the commodity category, the server acquires the first purchase record, the first car adding record and the first click record of the target user on the target commodity in the past 7 days, and a second purchase record, a second car-adding record and a second click record of the target user on the commodity category, the server acquires a first purchase record, a first car-adding record and a first click record of the target user on the target commodity within the last 14 days, and a second purchase record, a second car-adding record and a second click record of the target user on the commodity category, and the server acquires a first purchase record, a first car adding record and a first click record of the target user on the target commodity within the last 30 days, and a second purchase record, a second car adding record and a second click record of the target user to the commodity category, wherein the server records the first purchase record, the first car adding record and the first click record of the target user to the target commodity in each first historical time range, and the second purchase record, the second car adding record and the second click record of the target user on the commodity category form the purchase characteristics of the target user on the target commodity.

Step 304: the server combines the merchandise feature, the user feature, and the purchase feature into a recommendation feature.

In this step, the recommendation feature includes: the commodity characteristics of the target commodity, the user characteristics of the target user and the purchase characteristics of the target user on the target commodity.

It should be noted that the recommendation feature may include a pre-bin feature, an association feature, and a pre-bin feature and an association feature in addition to the merchandise feature, the user feature, and the purchase feature.

In one possible implementation, when the recommendation feature further includes a pre-bin feature, the step may be: the server determines a front-end bin corresponding to the area where the terminal is located; acquiring the characteristics of a front bin; and combining the commodity characteristics, the user characteristics, the purchase characteristics and the pre-bin characteristics into recommendation characteristics.

One region comprises a plurality of regions, and each region is provided with a front bin, so that the target commodity can be timely and quickly delivered to the target user after the target user purchases the target commodity, and the freshness of the target commodity is ensured.

In this implementation manner, the server may determine the front-end bin corresponding to the area where the terminal is located according to the location where the terminal is located. The step of the server acquiring the characteristics of the front-end bin may be: the server acquires the heat of the front bin, the sales quantity of the front bin, the average delivery time of the front bin and the times of bin explosion of the front bin in at least one first historical time range; the heat of the front bin, the sales quantity of the front bin, the average delivery time of the front bin and the number of times of bin explosion of the front bin are combined into the front bin characteristic.

The heat degree of the front bin is the fire heat degree of the front bin corresponding to the area where the terminal is located in the plurality of front bins in the area. The number of times of bin explosion of the front bin is the number of times that orders received by the front bin are more, so that commodities in the front bin cannot meet the requirements of users.

When the first historical time ranges are 1 day, 3 days, 7 days, 14 days and 30 days, the server acquires the heat of the front bin, the sales quantity of the front bin, the average delivery time of the front bin and the times of explosion of the front bin within the past 1 day, the server acquires the heat of the front bin, the sales quantity of the front bin, the average delivery time of the front bin and the times of explosion of the front bin within the past 3 days, the server acquires the heat of the front bin, the sales quantity of the front bin, the average delivery time of the front bin and the times of explosion of the front bin within the past 7 days, the server acquires the heat of the front bin, the sales quantity of the front bin, the average delivery time of the front bin and the times of explosion of the front bin within the past 14 days, and the server acquires the heat of the front bin, the sales quantity of the front bin, the average delivery time of the front bin and the times of explosion of the front bin within the past 30 days, the server connects the front bin heat, the front bin sales quantity, the average delivery time and the times of explosion of the front bin explosion, The sale number of the front bins, the average delivery time of the front bins and the number of times of bin explosion of the front bins form the characteristics of the front bins.

In this implementation, the server loads the cross feature according to the relationship between the commodity feature, the user feature, the purchase feature, and the micro-bin feature, thereby forming the recommendation feature.

It should be noted that, when the area has no front bin but only a central bin, the front bin feature may be a central bin feature, and the server obtains the central bin feature and combines the commodity feature, the user feature, the purchase feature and the central bin feature into the recommendation feature. The manner of the server acquiring the central bin features is similar to the manner of the server acquiring the front bin features, and is not described herein again.

In another possible implementation, when the recommended features further include an associated feature, the step may be: the server acquires the association characteristics of the associated commodities related to the target commodity; and combining the commodity characteristics, the user characteristics, the purchase characteristics and the associated characteristics into recommendation characteristics.

In this implementation, the server may obtain the association feature by: the server determines whether the commodity clicked by the target user is related to the target commodity, whether the commodity added into the shopping cart by the target user is related to the target commodity, whether the commodity clicked by the target user or the commodity added into the shopping cart is a commodity in the same menu or not, and whether the commodity clicked by the target user or the commodity added into the shopping cart and the target commodity can be mined out through a frequent item set or not; the server combines the features into associated features.

In another possible implementation, when the recommended features further include a pre-bin and an associated feature, the step may be: the server determines a front-end bin corresponding to the area where the terminal is located; acquiring the characteristics of a front bin; acquiring the association characteristics of the associated commodities related to the target commodity; and combining the commodity characteristics, the user characteristics, the purchase characteristics, the pre-bin characteristics and the association characteristics into recommendation characteristics.

In this implementation manner, the manner in which the server obtains the pre-bin feature is the same as the manner in which the pre-bin feature is obtained, and the manner in which the server obtains the association feature is the same as the manner in which the association feature is obtained, which are not described herein again.

Step 305: and the server inputs the recommendation characteristics into the first recommendation model to obtain a recommendation score of the target commodity, and the recommendation score is used as a recommendation probability.

In the step, the server inputs the recommendation characteristics into the first recommendation model, and finally the recommendation probability of the target commodity is obtained. The recommendation probabilities obtained by different target commodities are different, the larger the recommendation probability is, the more obvious the display position of the target commodity corresponding to the target commodity in the commodity interface is, and the higher the probability of purchasing by the target user is.

It should be noted that, before this step, the server trains to obtain the first recommendation model. In a possible implementation manner, the step of training the server to obtain the first recommendation model may be implemented by the following steps (1) to (5), including:

(1) the server obtains the sample recommendation characteristics of the sample commodities and the sample recommendation scores of the sample commodities.

When the first recommendation model is obtained through training, the sample commodities displayed in the commodity interface of the server are far larger than the clicked sample commodities, the clicked sample commodities are far larger than the sample commodities added into the shopping cart, and the sample commodities added into the shopping cart are far larger than the purchased sample commodities. Therefore, when the server acquires the sample commodity, the sample commodity is acquired according to the corresponding proportion by taking the click rate, the car-adding rate and the purchase rate as references.

The click rate is the quotient of the number of the clicked sample commodities and the number of the displayed sample commodities, the car-adding rate is the quotient of the number of the clicked sample commodities and the number of the sample commodities added into the shopping car, and the purchase rate is the quotient of the number of the purchased sample commodities and the number of the displayed sample commodities.

The server can obtain the sample recommendation characteristics and the sample recommendation scores of the sample commodities through the server log and the front end buried point log corresponding to the server.

(2) And the server inputs the sample recommendation characteristics of the sample commodities into the second recommendation model to obtain click rate, conversion rate, overdue probability, selling price, commodity price entering and overdue weight of the sample commodities.

And the server inputs the sample recommendation characteristics into a second recommendation model, and the click rate, the conversion rate, the overdue probability, the sale price, the commodity entrance price and the overdue weight of the sample commodity are obtained through the model parameters of the second recommendation model.

The second recommended model may be an initial model or a model obtained by training of other servers, which is not specifically limited in the application embodiment. When the second recommended model is an initial model, the model parameters are initial model parameters.

(3) And the server determines the training scores of the sample commodities according to the click rate, the conversion rate, the overdue probability, the sale price, the commodity entrance price and the overdue weight.

In one possible implementation, the server determines the profit score of the sample commodity according to the click rate, the conversion rate and the selling price, see the following formula one; determining a loss score of the sample commodity according to the click rate, the conversion rate, the overdue probability, the commodity incoming price and the overdue weight, and referring to a formula II; the sum of the yield score and the loss score is taken as the training score, see equation three below.

The formula I is as follows: score₁＝ctr×cvr×p₁

Wherein, Score₁For revenue scoring, ctr is click-through rate, cvr is conversion rate, p₁Is the selling price; the conversion rate is the quotient of the number of the purchased sample commodities and the number of the clicked sample commodities, the expiration weight is a parameter for adjusting the time between the non-expiration commodity and the expiration commodity, and the expiration probability is the probability that the expiration is caused because the expiration is not caused by sale of the expiration commodity.

The formula II is as follows: score₂＝r×p×(1-ctr×cvr)×p₂

Wherein, Score₂For loss score, r is the expiration weight, p is the expiration probability, ctr is the click rate, cvr is the conversion rate, p₂And (5) feeding the commodity.

The formula III is as follows: score ═ Score₁+Score₂＝ctr×cvr×p₁+r×p×(1-ctr×cvr)×p₂

Wherein Score is the training Score, Score₁Score for revenue scoring₂The loss was scored.

From equation three, it can be seen that: when it is overdueScore at a rate of 0, i.e., p is 0₂Is 0, Score ═ Score₁Only the revenue scores of the sample goods are considered at this time. When p is not 0, the larger p, the Score₂The larger the Score, the higher the training Score, the higher the ranking, and the easier the goods will be sold.

(4) The server determines a score difference between the training score and the sample recommendation score.

And (4) the sample recommendation score is a known score, and after the training score is obtained by the server through the step (3), the training score is compared with the sample recommendation score to determine a score difference value between the training score and the sample recommendation score.

(5) And the server adjusts the model parameters of the second recommendation model according to the grading difference value to obtain the first recommendation model.

The server adjusts model parameters of the second recommendation model according to the size of the score difference, inputs the sample recommendation characteristics into the second recommendation model after model parameter adjustment, obtains click rate, conversion rate, expiration probability, sale price, commodity price and expiration weight of the sample commodity again, determines training scores of the sample commodity again according to the click rate, the conversion rate, the expiration probability, the sale price, the commodity price and the expiration weight, and repeats iterative training until the difference between the training scores and the sample recommendation scores is within a preset difference range or the iteration times reach preset iteration times, so that the first recommendation model is obtained.

From the above formula three, it can be seen that: the recommendation model considers the profit score and the loss score at the same time, so that the recommendation score of the recommendation model for the target commodity is scored based on the comprehensive performance of the profit and the loss, and the obtained recommendation score is more objective and comprehensive.

In one possible implementation, the first recommendation model may employ a gradient boosting tree. When the first recommendation model obtains the recommendation score according to each feature in the recommendation features, an optimal division point with the best feature is found, the whole data set is divided into two subsets which are larger than and smaller than the division point, then the optimal division point is respectively found for the two divided subsets until the termination condition is met, and finally the construction of the gradient lifting tree is completed to obtain the recommendation score.

The process of the server training through the above steps (1) to (5) to obtain the first recommendation model can be seen in fig. 4, and as can be seen from fig. 4: the method comprises the steps that a server obtains sample recommendation characteristics of sample commodities, obtains training scores of the sample commodities through forward propagation of initial model parameters, determines score difference values between the training scores and the sample recommendation scores, and obtains a first recommendation model when the score difference values are within a preset difference value range. And when the score difference is not within the preset difference range, the model parameters are adjusted in a gradient reverse mode, the training scores of the sample commodities are obtained again through the adjusted model parameters, and the training is iterated in the mode until the score difference is within the preset difference range or the iteration times reach the preset iteration times, so that the first recommendation model is obtained.

Step 306: and the server determines the display position of the target commodity in the commodity interface to be requested according to the recommendation probability.

In this step, the server may sort the recommendation probabilities corresponding to each target commodity from large to small, and the target commodity with the larger recommendation probability has a more obvious display position in the commodity interface.

Referring to fig. 5, it can be seen from fig. 5 that: and the server trains the second recommendation model through the sample recommendation characteristics and the sample recommendation scores of the sample commodities to obtain the first recommendation model. The server responds to the browsing request of the terminal and then determines a plurality of target commodities. Wherein the plurality of target products may be located in a list. The server inputs the recommendation characteristics corresponding to each target commodity into the first recommendation model, obtains the recommendation score, namely the recommendation probability, corresponding to the target commodity through the first recommendation model, and then sorts the target commodities according to the recommendation probability to obtain a sorted list, wherein the display position of the target commodity sorted in the front in the commodity interface is more obvious. And the server sends an ordered list to the terminal, wherein the ordered list comprises commodity information and display positions of the target commodities.

Step 307: the server returns the commodity information and the display position of the target commodity to the terminal.

And after determining the display position of the target commodity in the commodity interface, the server sends the commodity information and the display position of the target commodity to the terminal.

In a possible implementation manner, the server may recommend not only the provisional commodity to the target user, but also recommend the associated commodity related to the provisional commodity to the target user. The related commodity may be a provisional commodity or a non-provisional commodity. For example, the server may recommend not only fruits such as strawberries and hami melons to the target user, but also yogurt and salad dressing to the target user, and the yogurt may be matched with fruits such as strawberries and hami melons to produce fruit salad. For another example, the server may recommend not only chicken wings but also cola to the target user, and the chicken wings may be matched with the cola to make cola chicken wings.

In the implementation mode, the server can promote the sales volume of the temporary commodities and the sales volume of the related commodities related to the temporary commodities through the strategy of taking a sale, so that the overall profit is improved.

Accordingly, the server returns not only the commodity information and the display position of the target commodity to the terminal, but also the commodity information and the display position of the associated commodity to the terminal. In one possible implementation, the method further includes: the server determines the association probability of the associated commodity and the target commodity according to the association characteristics; determining the display position of the associated commodity in the commodity interface according to the association probability; and returning the display position of the associated commodity and the commodity information to the terminal.

The display area of the associated commodity in the commodity interface and the display area of the target commodity in the commodity interface can be the same or different. When the two are the same, the server transmits not only the commodity information and the display position of the target commodity located in the display area but also the commodity information and the display position of the associated commodity located in the display area to the terminal. The terminal displays the commodity information of the target commodity at the display position corresponding to the target commodity in the display area and displays the commodity information of the related commodity at the display position corresponding to the related commodity in the display area. When the two are different, for the sake of convenience of distinction, the display region in which the target product is located is referred to as a first display region, and the display region in which the associated product is located is referred to as a second display region. The server sends the commodity information and the display position of the target commodity in the first display area to the terminal, and sends the commodity information and the display position of the related commodity in the second display area to the terminal. The terminal displays the commodity information of the target commodity at the display position corresponding to the target commodity in the first display area, and displays the commodity information of the related commodity at the display position corresponding to the related commodity in the second display area.

In this implementation manner, the manner in which the server determines the association probability according to the association feature is similar to the manner in which the server determines the recommendation probability according to the recommendation feature, and the manner in which the server determines the display position of the associated commodity in the commodity interface according to the association probability is similar to the manner in which the server determines the display position of the target commodity in the commodity interface according to the recommendation probability, which is not described herein again.

In a possible implementation mode, the server can also obtain the inventory quantity of the target commodities, filter out sold-out commodities according to the inventory quantity of the target commodities, and improve the exposure rate of other target commodities, so that the overall yield of the temporary commodities is improved. Correspondingly, the steps can be as follows: the server acquires the inventory quantity of the target commodity; and when the stock quantity of the target commodity is zero, filtering the target commodity, and returning the commodity information and the display position of the remaining target commodity to the terminal.

In the implementation mode, the server can obtain the stock quantity of each target commodity, and when the stock quantity of any target commodity is zero, the server filters the target commodity and returns the commodity information and the display position of the remaining target commodity to the terminal. The server may obtain the inventory quantity of the target product in real time, or may obtain the inventory quantity of the target product on the level periodically, which is not specifically limited in this embodiment of the application.

It should be noted that, in the embodiment of the present application, the server may not only adjust the display position of the temporary product according to the user characteristics, the product characteristics, and the purchase characteristics of the target user for the target product, and through the first recommendation model, optimize the exposure position of the temporary product, but also filter out products sold out according to the inventory information of the target product, improve the exposure rate of other products, and improve the sales volume of related products through a strategy of selling.

Step 308: and the terminal receives the commodity information and the display position of the target commodity sent by the server, and displays the commodity information at the display position corresponding to the target commodity in the commodity interface.

The terminal receives the commodity information and the display position of the target commodity, and the commodity information is rendered at the display position corresponding to the target commodity in the commodity interface, so that the target user can select the target commodity required by the user.

An embodiment of the present application provides a commodity recommending apparatus, referring to fig. 6, the apparatus includes:

a first determining module 601, configured to determine, in response to a browsing request from a terminal, a plurality of target commodities, where a target commodity is a current commodity close to a shelf life;

a first obtaining module 602, configured to obtain a commodity feature of a target commodity, obtain a user feature of a target user, and obtain a purchase feature of the target user on the target commodity, where the target user is a user using a terminal;

a second determining module 603, configured to determine, according to the commodity feature, the user feature, and the purchase feature, a recommendation probability of the target commodity;

a third determining module 604, configured to determine, according to the recommendation probability, a display position of the target product in the product interface to be requested;

a returning module 605, configured to return the commodity information and the display position of the target commodity to the terminal.

In a possible implementation manner, the first obtaining module 602 is further configured to obtain a first number of times that the target product is clicked, a sales number, and a first number of times that the target product is added to the shopping cart within at least one first historical time range; one or more of the first number, the sales number, and the first number are combined into the commodity characteristics of the target commodity.

In another possible implementation manner, the first obtaining module 602 is further configured to obtain a second number of times that the target user accesses the current server using the terminal, a number of times of purchasing commodities, a number of purchased commodities, an average consumption level, and terminal information of the terminal within at least one first historical time range; and forming one or more of the second time, the commodity purchasing quantity, the average consumption level and the terminal information into the user characteristics of the target user.

In another possible implementation manner, the first obtaining module 602 is further configured to determine a category of a target product; acquiring a first purchase record of a target user on a target commodity, a first car adding record and a first click record of adding into a shopping car, and a second purchase record of the target user on a commodity category, a second car adding record and a second click record of adding into the shopping car within at least one first historical time range; and forming the purchase characteristics of the target user on the target commodity by one or more of the first purchase record, the first car adding record, the first click record, the second purchase record, the second car adding record and the second click record.

In another possible implementation manner, the second determining module 603 is further configured to combine the commodity feature, the user feature, and the purchase feature into a recommendation feature; inputting the recommendation characteristics into a first recommendation model to obtain a recommendation score of the target commodity; and taking the recommendation score as the recommendation probability.

In another possible implementation manner, the second determining module 603 is further configured to determine a front bin corresponding to an area where the terminal is located; acquiring the characteristics of a front bin; acquiring the association characteristics of the associated commodities related to the target commodity; and combining the commodity characteristics, the user characteristics, the purchase characteristics, the pre-bin characteristics and the association characteristics into recommendation characteristics.

In another possible implementation manner, the apparatus further includes:

a third determining module 604, configured to determine, according to the association characteristic, an association probability between the associated item and the target item;

the returning module 605 is further configured to return the display position of the associated product and the product information to the terminal.

In another possible implementation manner, the first determining module 601 is further configured to obtain user information of the target user according to a user identifier of the target user carried in the browsing request; selecting a temporary commodity matched with the user information from a plurality of temporary commodities in a front bin corresponding to the terminal according to the user information; and taking the selected temporary commodity as a target commodity.

In another possible implementation manner, the apparatus further includes:

the second acquisition module is used for acquiring the sample recommendation characteristics of the sample commodities and the sample recommendation scores of the sample commodities;

the input module is used for inputting the sample recommendation characteristics of the sample commodities into the second recommendation model to obtain the click rate, the conversion rate, the overdue probability, the sale price, the commodity price and the overdue weight of the sample commodities;

the fifth determining module is used for determining the training scores of the sample commodities according to the click rate, the conversion rate, the overdue probability, the selling price, the commodity price and the overdue weight;

In another possible implementation manner, the fifth determining module is further configured to determine a profit score of the sample commodity according to the click rate, the conversion rate and the sales price; determining a loss score of the sample commodity according to the click rate, the conversion rate, the overdue probability, the commodity incoming price and the overdue weight; the sum of the profit and loss scores is taken as the training score.

In another possible implementation manner, the returning module 605 is further configured to obtain the inventory quantity of the target product; and when the stock quantity of the target commodity is zero, filtering the target commodity, and returning the commodity information and the display position of the remaining target commodity to the terminal.

The commodity recommending device provided by the embodiment of the application responds to a browsing request of a terminal, and determines a plurality of target commodities, wherein the target commodities are imminent commodities close to the quality guarantee period; the method comprises the steps of obtaining commodity characteristics of a target commodity, obtaining user characteristics of a target user, and obtaining purchasing characteristics of the target user on the target commodity, wherein the target user is a user using a terminal; determining the recommendation probability of the target commodity according to the commodity characteristics, the user characteristics and the purchasing characteristics; determining the display position of the target commodity in the commodity interface to be requested according to the recommendation probability; and returning the commodity information and the display position of the target commodity to the terminal. The device has the advantages that the recommendation probabilities of different target commodities are different, the more the recommendation probability is, the more the display position of the temporary commodity is forward, the more the temporary commodity is obvious, so that the exposure rate of the temporary commodity is improved, a user can more intuitively notice the promotion information of the temporary commodity, the promotion effect of the temporary commodity is improved, and the processing efficiency is high.

Fig. 7 is a block diagram of a server 700 according to an embodiment of the present disclosure. The server 700 may have a relatively large difference due to different configurations or performances, and may include one or more processors (CPUs) 701 and one or more memories 702, where the memory 702 stores at least one program code, and the at least one program code is loaded and executed by the processors 701 to implement the methods provided by the above-mentioned method embodiments. Of course, the server 700 may also have components such as a wired or wireless network interface, a keyboard, and an input/output interface, so as to perform input and output, and the server 700 may also include other components for implementing the functions of the device, which are not described herein again.

The embodiment of the present application further provides a computer-readable storage medium, where the computer-readable storage medium is applied to a server, and at least one program code is stored in the computer-readable storage medium, and the at least one program code is loaded and executed by a processor, so as to implement the operations performed by the server in the commodity recommendation method according to the above embodiment.

The above description is only for facilitating the understanding of the technical solutions of the present application by those skilled in the art, and is not intended to limit the present application. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for recommending an article, the method comprising:

2. The method of claim 1, wherein the obtaining the product characteristics of the target product comprises:

3. The method of claim 1, wherein the obtaining the user characteristics of the target user comprises:

4. The method of claim 1, wherein the obtaining the purchase characteristics of the target user for the target product comprises:

determining a commodity category to which the target commodity belongs;

5. The method of claim 1, wherein said determining a recommendation probability for the target item based on the item characteristics, the user characteristics, and the purchase characteristics comprises:

and taking the recommendation score as the recommendation probability.

6. The method of claim 5, wherein said combining said merchandise characteristic, said user characteristic, and said purchase characteristic into a recommendation characteristic comprises:

acquiring the characteristics of the front bin;

7. The method of claim 5, further comprising:

8. An article recommendation device, the device comprising:

9. A server, characterized in that the server comprises a processor and a memory, wherein at least one program code is stored in the memory, and the at least one program code is loaded and executed by the processor to implement the merchandise recommendation method according to any one of claims 1 to 7.

10. A computer-readable storage medium having at least one program code stored therein, the at least one program code being loaded and executed by a processor to implement the merchandise recommendation method of any one of claims 1 to 7.