CN112750012A

CN112750012A - Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium

Info

Publication number: CN112750012A
Application number: CN202110045280.8A
Authority: CN
Inventors: 马翼; 王继云; 朱战伟
Original assignee: Dingdang Fast Medicine Technology Group Co ltd
Current assignee: Dingdang Fast Medicine Technology Group Co ltd
Priority date: 2021-01-13
Filing date: 2021-01-13
Publication date: 2021-05-04

Abstract

The application discloses a commodity recommendation method, a commodity recommendation device, commodity recommendation equipment and a storage medium. A method of merchandise recommendation, comprising: receiving login information sent by a client of a user; the login information comprises a user identifier; searching corresponding historical purchased commodity information according to the user identification; if the historical purchased commodity information can be searched, determining the user portrait according to the historical purchased commodity information; wherein the user representation comprises: a user identification and a user tag; determining a corresponding commodity recommendation list according to the user portrait; and sending the commodity recommendation list to a client of the user. According to the method and the device, the commodities are automatically recommended to the user, and the shopping experience of the user is improved.

Description

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

Technical Field

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

Background

With the rapid development of the internet and electronic commerce, online shopping has become a normal state. The products on the internet are various and thousands of products, but the mobile phone screen display of the user is limited, and all the commodities are difficult to be completely displayed to the user. Affecting the sale effect of the commodity.

Disclosure of Invention

The application mainly aims to provide a commodity recommendation method, a commodity recommendation device, commodity recommendation equipment and a storage medium, so as to solve the problem that shopping of a user is influenced because a mobile phone cannot display all commodities in the prior art.

In order to achieve the above object, according to one aspect of the present application, there is provided a commodity recommendation method including:

receiving login information sent by a client of a user; the login information comprises a user identifier;

searching corresponding historical purchased commodity information according to the user identification;

if the historical purchased commodity information can be searched, determining the user portrait according to the historical purchased commodity information; wherein the user representation comprises: a user identification and a user tag;

determining a corresponding commodity recommendation list according to the user portrait;

and sending the commodity recommendation list to a client of the user.

In one embodiment, if the historical purchased merchandise information is not found, determining a user representation based on the merchandise review information of the user;

and determining to send a commodity recommendation list to the user client according to the user portrait.

In one embodiment, the generation of the tag comprises:

determining a commodity set purchased by each user according to the historical purchase record of each user;

processing any commodity in the commodity set by adopting a vector conversion model to obtain a corresponding commodity vector;

calculating the average vector of the commodity set of each user according to each commodity vector;

according to the average vector of the commodity set of each user, clustering calculation is carried out on a user group based on a pre-trained clustering model to obtain a plurality of different groups;

each group is provided with a corresponding label;

and determining the label corresponding to the user according to the average vector and the grouping relation of the user.

In one embodiment, the method further comprises: determining a user group corresponding to each label;

determining a commodity set of the user group according to the commodity set of each user in the user group;

and in the commodity set of the user group, sorting according to the order of sales volume from high to low to obtain a commodity recommendation list corresponding to the label.

In one embodiment, the method further comprises periodically updating the user representation.

In one embodiment, periodically updating the user representation includes:

periodically acquiring a record of commodity purchase of a user; and counting a set of purchased commodities of each user;

determining an average vector of purchased goods of each user;

calculating the set of the users by adopting a clustering algorithm according to the average vector of the purchased commodities of each user to obtain a plurality of different groups;

determining a group corresponding to the user according to the average vector of the purchased commodities of the user; and a label.

In one embodiment, determining a user representation from merchandise browsing information of a user includes:

determining a commodity set browsed by a user;

querying each commodity vector in the commodity set;

calculating a vector mean value of the commodity set according to each commodity vector in the commodity set;

calculating the vector mean value by adopting a pre-trained clustering model to obtain a label corresponding to the commodity set;

determining a representation of the user from the tag.

In order to achieve the above object, according to a second aspect of the present application, there is provided an article recommendation device; the device includes:

the receiving module is used for receiving login information sent by a client of a user; the login information comprises a user identifier;

the portrait determining module is used for searching corresponding historical purchased commodity information according to the user identification;

the commodity list determining module is used for determining a corresponding commodity recommendation list according to the user portrait;

and the sending module is used for sending the commodity recommendation list to the client of the user.

In one embodiment, the representation determining module is further configured to determine a user representation based on the user's product review information if historical purchased product information is not retrieved;

and the commodity list determining module is also used for determining to send a commodity recommendation list to the client of the user according to the user portrait.

In one embodiment, the representation determination module is further configured to determine a set of items purchased by each user based on the historical purchase records of each user;

each group is provided with a corresponding label;

In one embodiment, the item list determining module is further configured to determine a user group corresponding to each tag;

In one embodiment, the representation determination module is further configured to periodically update the user representation.

In one embodiment, the image determination module is further configured to periodically obtain a record of the user purchasing the merchandise; and counting a set of purchased commodities of each user;

determining an average vector of purchased goods of each user;

In one embodiment, the representation determination module is further configured to determine a set of items viewed by the user;

querying each commodity vector in the commodity set;

determining a representation of the user from the tag.

In order to achieve the above object, according to a third aspect of the present application, there is provided an article recommendation control apparatus; comprising at least one processor and at least one memory; the memory is to store one or more program instructions; the processor, configured to execute one or more program instructions, is configured to perform the following steps:

and sending the commodity recommendation list to a client of the user.

In one embodiment, the processor is further configured to determine a user representation based on the user's merchandise review information if historical purchased merchandise information is not retrieved;

In one embodiment, the processor is further configured to determine a set of items purchased by each user based on the historical purchase records of each user;

each group is provided with a corresponding label;

In one embodiment, the processor is further configured to determine a user group corresponding to each tag;

In one embodiment, the processor is further configured to periodically update the user representation.

In one embodiment, the processor is further configured to periodically update the user representation, including:

determining an average vector of purchased goods of each user;

In one embodiment, a set of items viewed by a user is determined;

querying each commodity vector in the commodity set;

determining a representation of the user from the tag.

To achieve the above object, according to a fourth aspect of the present application, there is provided a computer-readable storage medium having one or more program instructions embodied therein, the one or more program instructions being for performing the steps of:

and sending the commodity recommendation list to a client of the user.

In one embodiment, the generation of the tag comprises:

each group is provided with a corresponding label;

In one embodiment, the method further comprises:

determining a user group corresponding to each label;

In one embodiment, periodically updating the user representation includes:

determining an average vector of purchased goods of each user;

determining a commodity set browsed by a user;

querying each commodity vector in the commodity set;

determining a representation of the user from the tag.

According to the technical scheme, the login information sent by the client of the user is received; determining a user portrait according to historical purchased goods information; determining a corresponding commodity recommendation list according to the user portrait; and the commodity recommendation list is sent to the client of the user, so that the shopping experience of the user is improved.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this application, serve to provide a further understanding of the application and to enable other features, objects, and advantages of the application to be more apparent. The drawings and their description illustrate the embodiments of the invention and do not limit it. In the drawings:

FIG. 1 is a schematic diagram of an illustration of a user image according to an embodiment of the application;

FIG. 2 is a flow chart of a method for recommending merchandise according to an embodiment of the present application;

FIG. 3 is a flow chart of a method of determining a user tag according to an embodiment of the present application;

FIG. 4 is a flow chart of another merchandise recommendation method according to an embodiment of the present application;

FIG. 5 is a schematic structural diagram of an article recommendation device according to an embodiment of the present application;

fig. 6 is a schematic structural diagram of an article recommendation device according to an embodiment of the present application.

Detailed Description

In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only partial embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.

First, the terminology used in the present application will be described.

User portrait: a user representation is a tagged user model abstracted according to information such as user consumption habits, living habits and the like. Referring to FIG. 1, there is shown a schematic illustration of a user representation; including but not limited to: natural characteristics, social characteristics, medicine purchasing characteristics, medical history characteristics and the like. Natural characteristics include, but are not limited to, gender, age, education, etc.; social features include, but are not limited to, marital, family, occupation, etc.; medical history characteristics include, but are not limited to, past medical history, and the like. And the medicine purchasing characteristics comprise historical medicine purchasing records and the like.

Based on this, the present application proposes a commodity recommendation method, see the flow chart of a commodity recommendation method shown in fig. 2; the method comprises the following steps:

step S202, receiving login information sent by a client of a user; the login information comprises a user identifier; the user identification can be a user name or a mobile phone number of the user; the login information also includes the password of the user.

Specifically, other relevant information of the user can be determined according to the registration information of the user; including but not limited to age, occupation, gender, cell phone number, etc.

Illustratively, after Zhang three login, a user name zhangsan is input, and after a password 111111, the login is successful.

Step S204, searching corresponding historical purchased commodity information according to the user identification;

illustratively, according to the user name zhangsan, commodity information of a historical period purchased by the user can be determined; the historical period may be the last week or the last day; the length of the historical period can be flexibly set.

Of course, in order to avoid confusion caused by the duplicate name of the user name, other additional authentication information, such as a mobile phone number, can be used to ensure the uniqueness of the user.

Step S206, if the historical purchased commodity information can be searched, determining the user portrait according to the historical purchased commodity information; wherein the user representation comprises: a user identification and a user tag;

the user tags are group identifications grouped for all users; each user belongs to a group. Each group corresponds to a product recommendation list.

Step S208, determining a corresponding commodity recommendation list according to the user portrait;

specifically, a product recommendation list of the corresponding group is determined according to the user tag in the user representation.

The commodity recommendation list is sorted according to the order of sales volume from high to low or from low to high.

Step S210, sending the commodity recommendation list to a client of the user; so that the user's client displays the item recommendation list.

The method of the invention determines the user portrait through the information of the purchased goods of the user, and determines the corresponding goods recommendation list according to the user portrait. The method and the device can accurately determine the commodities preferred by the user and actively push the commodity list, thereby improving the commodity selling effect.

In one embodiment, when the user tag is generated, referring to a flowchart of a method for determining the user tag shown in fig. 3: the method comprises the following steps:

step S301, determining a commodity set purchased by each user according to the historical purchase record of each user;

step S302, for any commodity in the commodity set, processing by adopting a vector conversion model to obtain a corresponding commodity vector;

specifically, each commodity is processed by adopting a vector conversion model to obtain an embedded vector of each commodity, wherein the vector conversion model can adopt a word2vec model.

Step S303, calculating the average vector of the commodity set of each user according to each commodity vector;

illustratively, the vector of product 1 purchased by Zhang Sanqi is (w)₁₁，w₁₂，w₁₃，w₁₄…w_1n)；

The vector of the product 2 is (w)₂₁，w₂₂，w₂₃，w₂₄…w_2n)；

The vector of item 3 is (w)₂₁，w₂₂，w₂₃，w₂₄…w_2n)；

…

The vector of the commodity m is (w)_m1，w_m2，w_m3，w_m4…w_mn)；

Calculating the average vector of m commodities of Zhang III as X1:

step S304, according to the average vector of the commodity set of each user, clustering calculation is carried out on a user group based on a pre-trained clustering model to obtain a plurality of different groups;

in specific implementation, the clustering model adopts kmeans or Mini Batch K-Means, and if the data volume is too large, a kmeans algorithm using spark platform can be considered.

Determining the number of user clusters in the model training process;

for an arbitrary group of quantities, calculating the calculation error of the clustering model of the group of quantities;

wherein, the calculation error is the square sum of the errors in the clusters;

determining the optimal user clustering number by adopting an elbow method according to the error corresponding to each clustering number;

specifically, the smallest sum of squares of the errors in the clusters is determined; and the corresponding number of clusters.

Preferably, the number of groups of user clusters is 50;

for example, the average vector of each user is calculated in step S303, which sequentially includes: x₁，X₂，X₃，X₄…X_k(ii) a Wherein k is the number of users; dividing the obtained mixture into L groups after passing through a clustering model; l is a positive integer greater than zero.

Wherein, (X1, X4, X6,) … (X2, X)_k，X_K-1)；

Wherein (X1, X4, X6) is the first group; (X2, X)_k，X_K-1) Is the Lth group;

step S305, setting corresponding labels for each group; the label may be a number, a letter, a character string, etc., and the specific form is not limited.

Step S306, determining the label corresponding to the user according to the average vector and the grouping of the user.

Exemplarily, the group of the mean vector X1 of zhangsan is the first group; the labels are of a first group, the arabic numerals 1.

To determine the item list for each group, in one embodiment, a user group for each tag is determined;

illustratively, the set of items of the first group corresponding to tag 1 is: item set of Zhao + item set of second user + item set of fourth user.

It is worth emphasizing that the same kind of goods may appear in the set of goods of Zhang III as in the set of goods of the second user; for example, Zhang III purchased Lihua qingwen capsules, and Li IV purchased Lihua qingwen capsules; in one embodiment, therefore, before sorting, a deduplication operation is performed according to the names of the commodities.

Considering that new products are always put on shelves; and new users join, old users may logout, or old users 'consumption habits also change, so the user's merchandise representation should change dynamically and randomly. Therefore, in an embodiment, the method further includes updating the user portrait periodically, where the period may be set to 24 hours or 12 hours, and may be flexibly set.

Periodically acquiring a record of commodity purchase of a user;

counting a set of purchased commodities of each user according to the purchased commodity record;

determining an average vector of purchased goods of each user;

illustratively, Zhang three has re-purchased some products within the last day, and the average vector for this group of products is X11;

the number of users is z; z is a positive integer greater than zero. Each user vector is as follows

X₁₁，X₁₂，X₁₃，X₁₄…X₁z；

exemplarily, the clustering is performed again to obtain h groups, and the group of vector X11 of Zhang III is the h group (X11, X14, X)₁z); the label is h; h is a positive integer greater than zero.

Determining a group corresponding to the user according to the average vector of the user and the corresponding grouping relation; and a label.

If the user is a newly registered user, the consumption record of the user does not exist in the database. In one embodiment, if the historical purchased merchandise information is not found, determining a user representation based on the merchandise review information of the user;

Wherein, confirm user portrait according to user's commodity browsing information, include:

determining a commodity set browsed by a user;

querying each commodity vector in the commodity set;

the embedded vector of each commodity can be stored in a database in advance; the embedded vector of the commodity is obtained and stored by the commodity through a vector conversion model in advance.

determining a representation of the user from the tag.

Another method of merchandise recommendation is described below, with reference to the flow chart of another method of merchandise recommendation shown in fig. 4; the method comprises the following steps:

step S401, obtaining sample data: and cleaning the commodities purchased by each user, and sequencing according to the time sequence to obtain a user commodity table in a format of 'user id, purchased commodity set'.

Step S402, converting the sample data by adopting a word2vec model, and deriving an embedded embedding vector of each commodity.

Step S403, calculating the embedded embedding vector mean value of all commodities purchased by each user; and making a user vector table in a format of 'user id, user vector'.

Step S404, carrying out clustering calculation on the data of the user vector table, selecting the kmeans of python or Mini Batch K-Means according to different data volumes, and training by considering the kmeans algorithm of spark platform to obtain a clustering model if the data volume is too large.

Step S405, determining the number of user clusters in the model training process; different intra-cluster error square sums can be obtained by training different numbers of clustering models; and obtaining the optimal user clustering number by adopting an elbow method.

Step S406, calling a model to calculate vector data of the user, determining a user tag according to a group where the vector data is located, and writing the user tag into a user commodity table in a format of 'user id, purchased commodity set and user tag'.

Step S407, grouping all commodities according to user tags; and counting the sales volume of the commodities corresponding to each label, and sequencing according to the sales volume to obtain a commodity sequencing list of user preferences under different labels.

Step S408, when the commodity needs to be recommended, only the label of the user needs to be inquired, the commodity under the label is recalled according to the label, 30 commodities are recalled from high sales volume to low sales volume in sequence every time the commodity is called.

Step S409, for a new user without a label, acquiring a commodity list of the access behavior interaction of the user; inquiring the embedding vector of each commodity, and calculating the vector mean value of the commodities in the commodity list; calling a clustering model to calculate a vector mean value and determining a label; recalling the goods according to the label; and a temporary clustering label of the user can be obtained and temporarily stored as the data of the user portrait.

In one embodiment, the geographic location information of the user can also be acquired; converting the geographical location information into a vector;

determining users with similar geographic positions as a group by using a clustering model; recommending a pharmaceutical product to the group of users;

specifically, users around a well-known hospital may be set as a group, with a predetermined radius around the hospital as a range; a certain medication or doctor specific to the hospital is recommended to the group of users.

Illustratively, a drug for treating skin diseases, such as "compound phellodendron bark liquid", is recommended within a predetermined range of 3000 meters around a nationally known dermatology hospital.

In the range of 3000 m around a nationally well-known hospital for treating teeth, a dental floss specialized for cleaning teeth, which is a specialty of the hospital, is recommended.

When clustering is carried out by the clustering model, the geographic positions of different hospitals are taken as centers, clustering is respectively carried out, and different user groups are automatically divided.

Presetting a recommended medicine list corresponding to each group of users; the recommended medicine list is the corresponding medicine specific to the hospital; the drugs may be ordered by their degree of thermal pin.

After the user logs in, the specific medicine of the famous hospital near the user address is automatically recommended, and after the user purchases the medicine, the medicine is sent to home in an express way.

In one embodiment, clustering analysis may also be performed in the age dimension of the user; dividing the age of the user into different groups; and recommending a corresponding product recommendation list for each group of users.

Illustratively, elderly people are aged 90-60 years, and the recommended products to this group of users are nutraceuticals, such as dong a donkey-hide gelatin, Tang Ji Jian, and the like.

The age is 30-50 for middle-aged people, and the recommended product to the group of users may be a health product.

In one embodiment, clustering analysis may also be performed with diseases;

illustratively, users who have purchased cold medications on the last two days are a group; recommending a list of drugs, such as cold drugs, to the users of the group; the list includes: flos coptidis and antipyretic granules, Ganmaoling granules and compound paracetamol and amantadine hydrochloride tablets; the ordering is performed according to the order of the sales amount from high to low.

Using the users who purchased the oral ulcer medicaments in the last two days as a group; the group of users is recommended a medication of the canker sore type. For example, dexamethasone, desserts, pome and Qinghuo capsules; the ordering is performed according to the order of the sales amount from high to low.

It should be noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system such as a set of computer-executable instructions and that, although a logical order is illustrated in the flowcharts, in some cases, the steps illustrated or described may be performed in an order different than presented herein.

According to an embodiment of the present invention, there is also provided an apparatus for commodity recommendation control, as shown in fig. 5, the apparatus including:

a receiving module 51, configured to receive login information sent by a client of a user; the login information comprises a user identifier;

the portrait determining module 52 is used for searching corresponding historical purchased commodity information according to the user identification;

a commodity list determining module 53, configured to determine a corresponding commodity recommendation list according to the user portrait;

and the sending module 54 is configured to send the commodity recommendation list to the client of the user.

In one embodiment, representation determination module 52 is further configured to determine a user representation based on the user's product view information if historical purchased product information is not retrieved;

the commodity list determining module 53 is further configured to determine, according to the user image, to send a commodity recommendation list to the client of the user.

In one embodiment, representation determination module 52 is further configured to determine a set of items purchased by each user based on the historical purchase records of each user;

each group is provided with a corresponding label;

In one embodiment, the item list determining module 53 is further configured to determine a user group corresponding to each tag;

In one embodiment, representation determination module 52 is further configured to periodically update the user representation.

In one embodiment, the image determination module 52 is further configured to periodically obtain a record of the user's purchases of merchandise; and counting a set of purchased commodities of each user;

determining an average vector of purchased goods of each user;

In one embodiment, representation determination module 52 is further configured to determine a set of items viewed by the user;

querying each commodity vector in the commodity set;

determining a representation of the user from the tag.

According to a third aspect of the present application, there is provided an article recommendation control apparatus; referring to fig. 6, including at least one processor 61 and at least one memory 62; the memory 62 is used to store one or more program instructions; the processor 61 is configured to execute one or more program instructions to perform the following steps:

and sending the commodity recommendation list to a client of the user.

The processor 61 is further configured to determine a user representation based on the user's merchandise browsing information if historical purchased merchandise information is not found;

The processor 61 is further configured to determine a commodity set purchased by each user according to the historical purchase record of each user;

each group is provided with a corresponding label;

The processor 61 is further configured to determine a user group corresponding to each tag;

The processor 61 is also configured to periodically update the user representation.

The processor 61 is further configured to periodically obtain a record of the user purchasing the goods; and counting a set of purchased commodities of each user;

determining an average vector of purchased goods of each user;

The processor 61 is further configured to determine a commodity set browsed by the user;

querying each commodity vector in the commodity set;

determining a representation of the user from the tag.

According to a fourth aspect of the present application, there is provided a computer readable storage medium having one or more program instructions embodied therein for performing the steps of:

and sending the commodity recommendation list to a client of the user.

In one embodiment, the generation of the tag comprises:

each group is provided with a corresponding label;

In one embodiment, the method further comprises:

determining a user group corresponding to each label;

In one embodiment, periodically updating the user representation includes:

determining an average vector of purchased goods of each user;

determining a commodity set browsed by a user;

querying each commodity vector in the commodity set;

determining a representation of the user from the tag.

In an embodiment of the invention, the processor may be an integrated circuit chip having signal processing capability. The Processor may be a general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete Gate or transistor logic device, discrete hardware component.

The various methods, steps and logic blocks disclosed in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The processor reads the information in the storage medium and completes the steps of the method in combination with the hardware.

The storage medium may be a memory, for example, which may be volatile memory or nonvolatile memory, or which may include both volatile and nonvolatile memory.

The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable PROM (EEPROM), or a flash Memory.

The volatile Memory may be a Random Access Memory (RAM) which serves as an external cache. By way of example and not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), SLDRAM (SLDRAM), and Direct Rambus RAM (DRRAM).

The storage media described in connection with the embodiments of the invention are intended to comprise, without being limited to, these and any other suitable types of memory.

Those skilled in the art will appreciate that the functionality described in the present invention may be implemented in a combination of hardware and software in one or more of the examples described above. When software is applied, the corresponding functionality may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.

The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. 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, comprising:

and sending the commodity recommendation list to a client of the user.

2. The article recommendation method according to claim 1,

if the historical commodity purchasing information cannot be found, determining the user portrait according to the commodity browsing information of the user;

3. The item recommendation method of claim 1, wherein the generating of the tag comprises:

each group is provided with a corresponding label;

4. The article recommendation method of claim 3, further comprising:

determining a user group corresponding to each label;

5. The merchandise recommendation method of claim 1 further comprising periodically updating said user representation.

6. The merchandise recommendation method of claim 1, wherein periodically updating said user representation comprises:

determining an average vector of purchased goods of each user;

7. The merchandise recommendation method of claim 2, wherein determining the user representation based on the merchandise browsing information of the user comprises:

determining a commodity set browsed by a user;

querying each commodity vector in the commodity set;

determining a representation of the user from the tag.

8. An article recommendation device, comprising:

9. An article recommendation apparatus characterized by comprising: at least one processor and at least one memory; the memory is to store one or more program instructions; the processor, configured to execute one or more program instructions to perform the method of any of claims 1-7.

10. A computer-readable storage medium having one or more program instructions embodied therein for performing the method of any one of claims 1-7.