CN104599160B

CN104599160B - Commodity recommendation method and device

Info

Publication number: CN104599160B
Application number: CN201510064107.7A
Authority: CN
Inventors: 吴鹏; 刘淑艳; 向守兵; 杨震; 张蓉; 陈澄; 云培研
Original assignee: Tencent Technology Shenzhen Co Ltd; Sichuan Engineering Technical College
Current assignee: Tencent Technology Shenzhen Co Ltd; Sichuan Engineering Technical College
Priority date: 2015-02-06
Filing date: 2015-02-06
Publication date: 2020-09-08
Anticipated expiration: 2035-02-06
Also published as: CN104599160A

Abstract

A merchandise recommendation method, comprising the steps of: extracting the interested commodities of the user from the interested commodity library of the user; acquiring a universal purchase cycle of the interested commodity; extracting a historical record of the same type of goods purchased by the user from the goods purchase record of the user, and counting the latest time for the user to purchase the same type of goods according to the historical record; setting a recommended time of the commodity of interest according to the universal purchase period and the latest time; and recommending the interested commodity to the user according to the recommending time. The method can improve the hit rate of the commodity push information and effectively improve the utilization rate of network and computer resources. In addition, a commodity recommending device corresponding to the commodity recommending method is further provided, and another commodity recommending method and device for recommending commodities in an interested commodity package are further provided.

Description

Commodity recommendation method and device

Technical Field

The invention belongs to the technical field of network information mining, and particularly relates to a commodity recommendation method and device.

Background

With the development of internet technology and e-commerce technology, online shopping is becoming more and more popular. In order to guide the online shopping behavior of the internet user, the e-commerce operator pushes various commodity information to the internet user through various channels, for example, through various internet application clients and various popular websites and the like.

The existing commodity information pushing scheme generally obtains commodities which users are interested in by collecting shopping records of internet users and recommends similar commodities to the users.

Although the commodities recommended by the existing scheme have a certain hit rate, namely the quantity of the commodity push information successfully guiding the user to generate corresponding shopping behaviors accounts for a certain proportion of the total quantity of the commodity push information, the existing scheme has the inherent defects that: the prior scheme recommends the same type of commodities to the user, wherein the commodities are purchased by the user, and the user probably does not consider the recommended commodities any more due to the purchased commodities. Therefore, the existing scheme can generate a large amount of invalid push information, and wastes network and computer resources.

Disclosure of Invention

In view of the above, it is desirable to provide a commodity recommendation method and apparatus that can improve the hit rate of commodity recommendation and thus effectively utilize network and computing resources.

A merchandise recommendation method, comprising the steps of:

extracting the interested commodities of the user from the interested commodity library of the user;

acquiring a universal purchase cycle of the interested commodity;

extracting a historical record of the same type of goods purchased by the user from the goods purchase record of the user, and counting the latest time for the user to purchase the same type of goods according to the historical record;

setting a recommended time of the commodity of interest according to the universal purchase period and the latest time;

and recommending the interested commodity to the user according to the recommending time.

A merchandise recommendation method, comprising the steps of:

extracting an interested commodity package of a user from an interested commodity library of the user, wherein the interested commodity package comprises a plurality of commodities belonging to a plurality of stages, the plurality of stages have sequential time sequence, and the purchase time sequence of the commodities of the previous stage is prior to that of the commodities of the next stage;

acquiring the universal purchase interval duration of two-stage commodities adjacent in time sequence in the interested commodity package;

extracting the historical records of the same type of commodities of the commodities in the interested commodity package purchased by the user according to the commodity purchase records of the user, and determining the latest time for the user to purchase the same type of commodities and the latest stage of the same type of commodities in the plurality of stages according to the historical records;

setting the recommended time of the commodities in each stage after the latest stage in the interested commodity package according to the universal purchasing interval duration by taking the latest time as a starting point;

and recommending commodities in each stage after the latest stage to the user according to the recommending time.

An article recommendation device comprising:

the commodity extraction module is used for extracting the commodity of interest of the user from the commodity library of interest of the user;

the purchase cycle acquisition module is used for acquiring the universal purchase cycle of the interested commodity;

the latest time counting module is used for extracting the historical records of the similar commodities of the interested commodities purchased by the user according to the commodity purchase records of the user and counting the latest time for the user to purchase the similar commodities according to the historical records;

the first recommendation time setting module is used for setting recommendation time of the interested commodity according to the universal purchase cycle and the latest time;

and the first recommending module is used for recommending the interested commodity to the user according to the recommending time.

An article recommendation device comprising:

the commodity package extraction module is used for extracting the commodity package of interest of the user from the commodity library of interest of the user, wherein the commodity package of interest comprises a plurality of commodities belonging to a plurality of stages, the plurality of stages have time sequence, and the purchase time sequence of the commodities in the previous stage is prior to that of the commodities in the next stage;

the purchase interval duration acquisition module is used for acquiring the universal purchase interval duration of the two-stage commodities adjacent in time sequence in the interested commodity package;

the latest time and stage acquisition module is used for extracting the historical records of the same type of commodities of the commodities in the interested commodity package purchased by the user according to the commodity purchase records of the user, and determining the latest time of purchasing the same type of commodities by the user and the latest stage of the same type of commodities in the plurality of stages according to the historical records;

the second recommendation time setting module is used for setting the recommendation time of the commodities in each stage after the latest stage in the interested commodity package by taking the latest time as a starting point according to the universal purchasing interval duration;

and the second recommending module is used for recommending commodities in each stage after the latest stage to the user according to the recommending time.

The first commodity recommending method and device for recommending the interested commodities extract the interested commodities of the user from the interested commodity library of the user, acquire the universal purchase cycle of the interested commodities, set the recommending time of the interested commodities according to the universal purchase cycle and the latest time of purchasing the same kind of commodities by the user, and recommend the interested commodities to the user according to the recommending time.

The second commodity recommendation method and device for recommending commodities in an interested commodity package extract the interested commodity package of a user from an interested commodity library of the user, determine which stages of similar commodities of the interested commodity package are purchased by the user, determine the latest stage and latest purchase time corresponding to the purchased similar commodities, set the recommendation time of commodities of each stage after the latest stage in the interested commodity package according to the universal purchase interval duration of the commodities of two stages adjacent to each other in time sequence in the interested commodity package and the latest purchase time, and recommend commodities of each stage after the latest stage to the user according to the recommendation time, wherein the recommendation time is matched with the time of purchasing the commodities of each stage after the latest stage by the user, rather than randomly recommending the commodities to the user at any time, so that the hit rate of pushing information of the commodities is improved, the utilization rate of network and computer resources is effectively improved.

Drawings

FIG. 1 is a block diagram of a portion of an embodiment of an apparatus operable to perform a merchandise recommendation method according to the present application;

FIG. 2 is a flowchart illustrating a method for recommending merchandise according to an embodiment;

FIG. 3 is a flowchart illustrating the steps of counting the universal purchase cycles of merchandise in one embodiment;

FIG. 4 is a flowchart illustrating a method for recommending merchandise according to an embodiment;

FIG. 5 is a flowchart illustrating a method for recommending merchandise according to an embodiment;

FIG. 6 is a flowchart illustrating the steps of counting the time intervals between the ordinary purchases of two stages of merchandise in the time sequence of the merchandise package according to an embodiment;

FIG. 7 is a flowchart illustrating a method for recommending merchandise according to an embodiment;

FIG. 8 is a flowchart illustrating steps for building a commodity package model in one embodiment;

FIG. 9 is a schematic diagram of a configuration of an article recommendation device in one embodiment;

FIG. 10 is a schematic diagram of a configuration of an article recommendation device in one embodiment;

FIG. 11 is a schematic diagram of the structure of an article recommendation device in one embodiment;

FIG. 12 is a schematic structural diagram of an article recommendation device in one embodiment;

FIG. 13 is a schematic diagram of a configuration of an article recommendation device in one embodiment;

FIG. 14 is a schematic structural diagram of an article recommendation device in one embodiment;

FIG. 15 is a schematic diagram of the structure of an article recommendation device in one embodiment;

fig. 16 is a schematic structural diagram of an article recommendation device in one embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

Fig. 1 is a partial block diagram of an apparatus that can execute the product recommendation method of the present application in one embodiment. As shown in FIG. 1, in one embodiment, the server includes a processor, a storage medium, a memory, and a network interface connected by a system bus. Wherein, the network interface is used for network communication; the storage medium is stored with an operating system, a database and software instructions for realizing the commodity recommendation method, wherein the database is used for storing commodity information of interest of a user, commodity general purchase cycle information, commodity purchase records and the like; the memory is used for caching data; the processor coordinates the work among the various components and executes the software instructions described above to implement the merchandise recommendation method described herein. The structure shown in fig. 1 is a block diagram of only a part of the structure related to the present application, and does not constitute a limitation of the device to which the present application is applied, and a specific device may include more or less components than those shown in the drawings, or combine some components, or have a different arrangement of components.

As shown in fig. 2, in one embodiment, a method for recommending an article includes the steps of:

in step S202, the product of interest of the user is extracted from the product library of interest of the user.

And step S204, acquiring the universal purchase cycle of the interested commodity.

In one embodiment, the universal purchase period of the commodity is a comprehensive index obtained by summarizing and counting the purchase periods of the commodities of a plurality of users. In one embodiment, the universal purchase period of the product is an average or weighted average calculated according to the purchase periods of the products by the multiple users, wherein the weight corresponding to the user who purchases the product with a large number may be higher than the weight corresponding to the user who purchases the product with a small number.

In one embodiment, the general purchase cycle data of each type of goods may be stored in advance, and the step S204 may extract the general purchase cycle of the goods of interest from the pre-stored purchase cycle data.

In one embodiment, the method for recommending commodities further comprises the step of counting the universal purchase period of the commodities, as shown in fig. 3, and in one embodiment, the step comprises the following steps:

step S302, calculating the average purchasing period of various commodities purchased by a plurality of users from the purchasing record database, wherein the step of calculating the average purchasing period of various commodities purchased by a certain user comprises the following steps: extracting name keywords of the commodities in the user purchase record, matching the name keywords with preset commodity class names, marking the commodities as commodity classes corresponding to the commodity class names matched with the name keywords, and calculating the average purchase period of the commodities purchased by the user according to the purchase time of the same class of commodities purchased by the user.

In one embodiment, part or all of the purchase records in the purchase record database may be extracted as sample data, and the average purchase period of each user for purchasing each type of goods may be counted according to the extracted sample data.

In one embodiment, the step of extracting the name keyword of the commodity in a certain purchase record comprises: and segmenting the commodity name in the purchase record to obtain a plurality of words, extracting nouns and noun phrases in the words, scoring the nouns or noun phrases if the nouns or noun phrases are extracted, and taking the noun or noun phrase with the highest score as the name keyword of the commodity. In one embodiment, the step of scoring a plurality of nouns or noun phrases comprises: the nouns or noun phrases can be matched with the pictures of the commodities and the commodity attribute information (such as the commodity size, the material and the like), and the scores can be carried out according to the matching degree of the nouns or noun phrases and the pictures of the commodities and the commodity attribute information. For example, the "child vest" is extracted from the commodity name "spring and autumn pure cotton knitwear children's garment child vest" as the name keyword of the commodity.

In one embodiment, the matching relationship between nouns can be preset, and a noun relationship matching library is stored. For example, semantically approximating words may be set to have matching relationships between them. In one embodiment, a matching noun corresponding to the name keyword of the product may be searched in the noun relation matching library, and if a certain product class name is included in the matching noun, it is determined that the product class name matches the name keyword. In one embodiment, if the name key and the item class name are the same, it may also be determined that the name key and the item class name match.

In one embodiment, the step of calculating the average purchase period of the same type of goods purchased by the user according to the purchase time of the same type of goods purchased by the user comprises the following steps: the method comprises the steps of sequencing purchase records of the user for purchasing the same type of commodities according to purchase time, obtaining a preset number of purchase records adjacent to the sequencing position, obtaining purchase time spans of the preset number of purchase records (namely the interval duration of the first purchase time and the last purchase time in the preset number of purchase records), and calculating the ratio of the purchase time spans to the preset number to serve as an average purchase period of the user for purchasing the type of commodities. In another embodiment, the step of calculating the average purchase period of the same type of goods purchased by the user according to the purchase time of the same type of goods purchased by the user comprises the following steps: the method comprises the steps of sequencing purchase records of the user for purchasing the same type of commodities according to purchase time, intercepting a section of subsequence of which the purchase time span exceeds a threshold value in a sequenced purchase record sequence, counting the number of the purchase records and the purchase time span contained in the subsequence, and calculating the ratio of the purchase time span to the number of the purchase records to serve as the average purchase period of the user for purchasing the commodity.

Step S304, calculating the universal purchase cycle of the same type of commodity according to the average purchase cycle of each user of the same type of commodity.

In one embodiment, an average or weighted average of the average purchase periods of the various users of the same type of merchandise may be calculated as the universal purchase period for that type of merchandise.

Step S206, extracting the historical record of the same type of goods purchased by the user from the goods purchase record of the user, and counting the latest time for the user to purchase the same type of goods according to the historical record.

In one embodiment, the step of extracting the history of the similar commodities of the interested commodities purchased by the user according to the commodity purchase record of the user comprises the following steps: and if the name key words of the commodities are matched with the class names of preset commodity classes to which the interested commodities belong, extracting the corresponding commodity purchase records as the historical records of the same type of commodities the user purchases the interested commodities.

And step S208, setting the recommendation time of the interested commodity according to the universal purchase cycle of the interested commodity and the latest time of the user for purchasing the same kind of interested commodity.

In one embodiment, the latest time when the user purchases the same type of interested commodity is taken as a starting point, the product of the universal purchase cycle of the interested commodity and a preset weight is taken as a time increment, and the recommended time of the interested commodity is set as a corresponding time point after the time increment is added to the starting point. For example, the preset weight is 1 or 0.8, and so on.

And step S210, recommending interested commodities to the user according to the recommending time.

The user can be pushed the related information of the interested goods, such as the goods purchasing web page link, the goods name, the goods price and the goods picture, etc.

As shown in fig. 4, in an embodiment, the method for recommending a product further includes the following steps:

in step S402, a personal purchase cycle of the user for the interested goods is acquired.

In one embodiment, the personal purchase period of the same type of goods for which the user purchases the goods of interest can be calculated and taken as the personal purchase period of the user for the goods of interest.

In step S404, if the personal purchase cycle of the user for the interested commodity is shorter than the universal purchase cycle of the interested commodity, the recommendation time of the interested commodity is advanced according to the personal purchase cycle.

In one embodiment, the recommended time of the interested product is advanced by a length of time which is a product of a difference between the general purchase period and the individual purchase period and a preset weight.

In this embodiment, if the personal purchase cycle of the user for the interested commodity is shorter than the general purchase cycle of the interested commodity, the recommendation time of the interested commodity is advanced according to the personal purchase cycle, so that the recommendation time is more consistent with the time when the user needs to purchase the interested commodity, and the hit rate of the commodity push information is improved.

As shown in fig. 5, a commodity recommendation method includes the following steps:

step S502, extracting the interested commodity package of the user from the interested commodity library of the user, wherein the interested commodity package comprises a plurality of commodities belonging to a plurality of stages, the plurality of stages have sequential time sequence, and the purchase time sequence of the commodities of the previous stage is prior to that of the commodities of the next stage.

In one embodiment, a package of items corresponds to items corresponding to a plurality of phases included in a complete event process. For example, a kit of items containing items for home decoration includes items that are needed for a number of stages included in the process of being assigned to a home device. For another example, the fitness calculation includes a plurality of phases, and the merchandise package includes merchandise needed for the plurality of phases of the fitness plan.

Step S504, the universal purchase interval duration of the two-stage commodities adjacent to the time sequence in the interested commodity package is obtained.

For example, if the package of goods includes three stages of goods, the three stages are sorted from first to last in time sequence as follows: the first stage, the second stage and the third stage are two stages with adjacent time sequences, and the second stage and the third stage are two stages with adjacent time sequences.

In one embodiment, the universal purchase interval duration of the two-stage commodities adjacent to each other in time sequence is a comprehensive index obtained by summarizing and counting the purchase interval durations of the two-stage commodities purchased by a plurality of users. In one embodiment, the universal purchase interval duration of the two-stage commodities is an average value or a weighted average value calculated according to the purchase interval duration of the two-stage commodities purchased by a plurality of users, wherein the weight corresponding to the user purchasing the two-stage commodities with a large number may be higher than the weight corresponding to the user purchasing the commodities with a small number.

In one embodiment, the universal purchase interval duration of the two-stage commodities adjacent to the time sequence of each commodity package may be stored in advance, and the step S504 may extract the universal purchase interval duration of the two-stage commodities adjacent to the time sequence of the commodity package of interest from the pre-stored universal purchase interval duration data of the two-stage commodities adjacent to the time sequence of each commodity package.

In an embodiment, the method for recommending commodities further includes a step of counting the time length of the ordinary purchasing interval of the two-stage commodities adjacent to each other in the time sequence in the commodity package, as shown in fig. 6, the step includes the following steps:

in step S602, purchase records classified by the user are extracted from the purchase record database.

Step S604, extracting the name key words of the commodities in the purchase records of the users, marking the stage of the commodity package to which the commodities belong, and marking the stage of the commodity package to which the commodities belong if the name key words of the commodities are matched with the commodities at the stage in the commodity package.

In one embodiment, the matching relationship between nouns can be preset, and a noun relationship matching library is stored. For example, semantically approximating words may be set to have matching relationships between them. In one embodiment, if the name keyword of a certain product and the class name of the preset product class to which a certain product at a certain stage in the product package belong have a matching relationship in the noun relationship matching library, that is, if the name keyword of a certain product matches the class name of the preset product class to which a certain product at a certain stage in the product package belongs, it can be determined that the name keyword of the certain product matches the product at the certain stage in the product package.

Step S606, the average time length of the purchasing interval of the commodities in two adjacent stages of the time sequence of the commodity package purchased by each user is counted.

In one embodiment, the average value of the time-sequence adjacent two-stage commodities of a commodity package purchased by a user for multiple times can be calculated as the average purchase interval duration of the time-sequence adjacent two-stage commodities purchased by the user. For example, the records of the commodities of a certain commodity package purchased by a certain user are sorted from first to last according to the purchase time to obtain a purchase record sequence: (t1, first stage), (t2, second stage), (t3, third stage), (t4, first stage), (t5, second stage), (t6, third stage), where each purchase record is abbreviated as a data pair of the purchase time and the stage to which the commodity belongs, then the average purchase duration of the commodities in the first stage and the second stage for the user to purchase the commodity package can be calculated as: [ (t 2-t 1) + (t 5-t 4) ]/2.

Step S608, calculating the universal purchase interval duration of the two-stage commodities adjacent to the time sequence according to the average purchase interval duration of each user of the two-stage commodities adjacent to the time sequence.

In one embodiment, the average value or the weighted average value of the average purchase interval duration of the time sequence adjacent two-stage commodities of a commodity package purchased by each user can be calculated as the universal purchase interval duration of the time sequence adjacent two-stage commodities of the commodity package.

Step S506, according to the commodity purchase record of the user, extracting the historical record of the similar commodities of the commodities in the commodity package of interest purchased by the user, and according to the historical record, determining the latest time when the user purchases the similar commodities and the latest stages of the similar commodities in the multiple stages of the commodity package of interest.

In one embodiment, the step of extracting the history of the purchase of the same type of goods in the interest goods package by the user according to the goods purchase record of the user comprises the following steps: and if the name key words of the commodities in one commodity purchase record are matched with the class name of a preset commodity class to which any commodity in the interested commodity package belongs, the commodity purchase record is extracted as the historical record of the similar commodities the user purchases the interested commodities. And further, the stage corresponding to the commodity purchase record can be marked as the stage where the matched commodity in the interested commodity package is located.

And further, by comparing the stages corresponding to the historical records of the similar commodities of the interested commodities purchased by the user, the latest stage of the similar commodities in the multiple stages of the interested commodity package can be obtained.

Step S508, using the latest time as a starting point, setting recommended time of the commodities in each stage after the latest stage in the interested commodity package according to the universal purchasing interval duration of the commodities in two stages adjacent to each other in time sequence in the interested commodity package.

In one embodiment, the latest time is used as a starting point, and the product of the universal purchase interval duration of the latest stage and the next stage of commodities and a preset weight is used as a time increment, so as to set the recommended time of the next stage of commodities in the latest stage; and sequentially setting the recommendation time of the commodities of the next and later stages, wherein the recommendation time of the commodities of each later stage is increased by taking the recommendation time of the commodities of the last stage of the stage as a starting point and multiplying the universal purchase interval time of the commodities of the stage and the last stage by a preset weight.

Step S510, recommending the commodities of each stage after the latest stage to the user according to the recommended time of each stage.

As shown in fig. 7, in an embodiment, the method for recommending a product further includes the following steps:

step S702, acquiring the personal purchase interval duration of the time sequence adjacent two-stage commodities in the interested commodity package.

In one embodiment, a user's purchase record may be extracted from a purchase record database; extracting name keywords of the commodities in the purchase record of the user, marking the stage of each commodity belonging to a plurality of stages of the interested commodity package, and if the name keywords of a certain commodity are matched with the commodities at a certain stage in the interested commodity package, marking the stage of the commodity belonging to the interested commodity package; and counting the average purchasing interval duration of the commodities in two adjacent stages of the time sequence of the interested commodity package purchased by the user to obtain the personal purchasing interval duration.

In step S704, if the personal purchase interval duration of the two-stage commodities adjacent to the time sequence is shorter than the universal purchase interval duration of the two-stage commodities adjacent to the time sequence, the recommended time of the commodities in each stage after the latest stage in the interested commodity package is advanced according to the personal purchase interval duration.

In one embodiment, the time length of the recommended time advance of each stage of the commodities can be the product of the difference between the time length of the universal purchase interval between the stage and the previous stage of the commodities and the time length of the corresponding personal purchase interval and a preset weight.

In this embodiment, if the personal purchase interval duration of the two-stage commodities adjacent to the time sequence is shorter than the universal purchase interval duration of the two-stage commodities adjacent to the time sequence, the recommendation time of the commodities in each stage after the latest stage in the interested commodity package is advanced according to the personal purchase interval duration, so that the recommendation time is more consistent with the time for the user to purchase the commodities in each stage in the interested commodity package, and the hit rate of the commodity push information is improved.

In one embodiment, the above commodity recommendation method further includes the steps of: extracting commodities which are not purchased by a user in the commodities of the interested commodity package at the latest stage; and recommending commodities which are not purchased by the user to the user in real time.

The step of recommending commodities which are not purchased by the user to the user in real time comprises the following steps: and recommending the commodities which are not purchased by the user to the user by taking the current time as the recommendation time.

In this embodiment, the latest stage of the commodity package of interest purchased by the user may be the commodity required by the user at the current stage, and the current time may be the recommended time, so that the commodities not purchased by the user at the latest stage are recommended to the user in real time, and the hit rate of the commodity push information can be improved.

In an embodiment, before step S502, the method for recommending a commodity further includes a step of constructing a commodity package model, as shown in fig. 8, the step includes the following steps:

in step S802, purchase records classified by the user are extracted from the purchase record database.

Step S804, extracting the commodities purchased by each user at the same time period, and dividing a plurality of commodities, which are simultaneously appeared in the commodities purchased by each user at the same time period and have a frequency exceeding a threshold, into commodities at the same stage, thereby summarizing a plurality of stages and commodities corresponding to the stages.

Step 806, counting the time interval of purchasing each stage of commodity according to the purchasing records of each user.

Step S808, if the time length of the purchase interval for each user to purchase a two-stage commodity is approximately the same and the purchase sequence for each user to purchase the two-stage commodity is consistent, dividing the two-stage commodity into two-stage commodities of the same commodity package, thereby obtaining a commodity package composed of a plurality of stage commodities, and storing the commodity package data.

In one embodiment, a commodity package containing the same kind of commodity as the commodity of interest of the user may be set as the commodity package of interest of the user, that is, if a commodity is contained in the commodity package, the commodity is the same kind of commodity as the commodity of interest of the user, the commodity package may be set as the commodity package of interest of the user. Further, the user's interest package may be stored in a user interest goods repository.

The first commodity recommendation method for recommending the interested commodity and the second commodity recommendation method for recommending the commodity in the interested commodity package can be implemented in a combined manner. A commodity recommendation method includes the steps in the first commodity recommendation method in any of the above embodiments and the steps in the second commodity recommendation method in any of the above embodiments, and then the commodity recommendation method also belongs to the protection scope of the present application.

As shown in fig. 9, in one embodiment, a product recommendation apparatus includes a product extraction module 902, a purchase period acquisition module 904, a latest time statistics module 906, a first recommendation time setting module 908, and a first recommendation module 910, wherein:

the goods extraction module 902 is used for extracting the goods of interest of the user from the goods library of interest of the user.

The purchase cycle acquiring module 904 is used for acquiring a universal purchase cycle of the interested goods.

As shown in fig. 10, in an embodiment, the article recommending apparatus further includes a general purchase period counting module 1002 and a general purchase period storing module 1004, wherein the general purchase period counting module 1002 is configured to count a general purchase period of the article; the universal purchase cycle storage module 1004 is configured to store universal purchase cycle data of various types of goods, and the purchase cycle acquiring module 904 may extract a universal purchase cycle of a good of interest from the pre-stored purchase cycle data.

In one embodiment, the universal purchase period counting module 1002 is configured to count average purchase periods of various types of commodities purchased by a plurality of users from the purchase record database, wherein the process of counting average purchase periods of various types of commodities purchased by a certain user includes: extracting name keywords of the commodities in the user purchase record, matching the name keywords with preset commodity class names, marking the commodities as commodity classes corresponding to the commodity class names matched with the name keywords, and calculating the average purchase period of the commodities purchased by the user according to the purchase time of the same class of commodities purchased by the user.

In one embodiment, the universal purchase period statistics module 1002 may extract part or all of the purchase records in the purchase record database as sample data, and count the average purchase period of each user for purchasing each type of goods according to the extracted sample data.

In one embodiment, the process of the universal purchase cycle statistics module 1002 extracting the name keyword of the item in a certain purchase record comprises: and segmenting the commodity name in the purchase record to obtain a plurality of words, extracting nouns and noun phrases in the words, scoring the nouns or noun phrases if the nouns or noun phrases are extracted, and taking the noun or noun phrase with the highest score as the name keyword of the commodity. In one embodiment, the process of the universal purchase cycle statistics module 1002 scoring a plurality of nouns or noun phrases includes: the nouns or noun phrases can be matched with the pictures of the commodities and the commodity attribute information (such as the commodity size, the material and the like), and the scores can be carried out according to the matching degree of the nouns or noun phrases and the pictures of the commodities and the commodity attribute information.

In an embodiment, the article recommendation apparatus further includes a noun matching relationship storage module (not shown in the figure) for setting a matching relationship between nouns and storing a noun relationship matching library. For example, semantically approximating words may be set to have matching relationships between them. In one embodiment, the general purchase period statistic module 1002 may search the noun relation matching library for a matching noun corresponding to the name keyword of the product, and determine that a product class name matches the name keyword if the product class name is included in the matching noun. In one embodiment, if the name key and the product class name are the same, then the universal purchase cycle statistics module 1002 may also determine that the name key and the product class name match.

In one embodiment, the process of the universal purchase period statistic module 1002 calculating the average purchase period of the same type of goods purchased by the user according to the purchase time of the same type of goods purchased by the user includes: the method comprises the steps of sequencing purchase records of the user for purchasing the same type of commodities according to purchase time, obtaining a preset number of purchase records adjacent to the sequencing position, obtaining purchase time spans of the preset number of purchase records (namely the interval duration of the first purchase time and the last purchase time in the preset number of purchase records), and calculating the ratio of the purchase time spans to the preset number to serve as an average purchase period of the user for purchasing the type of commodities. In another embodiment, the process of the universal purchase period statistic module 1002 calculating the average purchase period of the same type of goods purchased by the user according to the purchase time of the same type of goods purchased by the user includes: the method comprises the steps of sequencing purchase records of the user for purchasing the same type of commodities according to purchase time, intercepting a section of subsequence of which the purchase time span exceeds a threshold value in a sequenced purchase record sequence, counting the number of the purchase records and the purchase time span contained in the subsequence, and calculating the ratio of the purchase time span to the number of the purchase records to serve as the average purchase period of the user for purchasing the commodity.

Further, the universal purchase period counting module 1002 is further configured to calculate a universal purchase period of the same type of goods according to the average purchase period of each user of the same type of goods.

In one embodiment, the universal purchase cycle statistics module 1002 may calculate an average or weighted average of the average purchase cycles of the various users of the same type of item as the universal purchase cycle for that type of item.

The latest time counting module 906 is configured to extract a history of the similar product that the user purchased the interested product according to the product purchase record of the user, and count the latest time that the user purchased the similar product according to the history.

In one embodiment, the process of the latest time statistic module 906 extracting the historical records of the same type of goods purchased by the user for the interested goods according to the goods purchase records of the user includes: and if the name key words of the commodities are matched with the class names of preset commodity classes to which the interested commodities belong, extracting the corresponding commodity purchase records as the historical records of the same type of commodities the user purchases the interested commodities.

The first recommendation time setting module 908 is used for setting the recommendation time of the interested commodity according to the universal purchase cycle of the interested commodity and the latest time when the user purchases the same commodity of the interested commodity.

In one embodiment, the first recommended time setting module 908 may use the latest time when the user purchased the same type of interested product as a starting point, use a product of the general purchase period of the interested product and a preset weight as a time increment, and set the recommended time of the interested product as a corresponding time point after the starting point is increased by the time increment.

The first recommending module 910 is configured to recommend the interested item to the user according to the recommending time.

In one embodiment, the first recommending module 910 may push the related information of the interested goods to the user, such as the goods purchase web page link, the goods name, the goods price and the goods picture, etc.

As shown in fig. 11, in an embodiment, the article recommendation apparatus further includes a personal purchase period obtaining module 1102 and a first recommendation time adjusting module 1104, wherein:

the personal purchase cycle acquiring module 1102 is used for acquiring a personal purchase cycle of a user for a product of interest.

In one embodiment, the personal purchase cycle acquiring module 1102 may calculate a personal purchase cycle of the same type of goods in which the user purchased the goods of interest, and take the personal purchase cycle as the personal purchase cycle of the user for the goods of interest.

The first recommendation time adjustment module 1104 is configured to advance the recommendation time of the item of interest according to the personal purchase cycle if the personal purchase cycle of the user for the item of interest is shorter than the universal purchase cycle of the item of interest.

As shown in fig. 12, a commodity recommending apparatus includes a commodity package extracting module 1202, a purchase interval duration obtaining module 1204, a latest time and stage obtaining module 1206, a second recommendation time setting module 1208, and a second recommending module 1210, wherein:

the commodity package extracting module 1202 is configured to extract a commodity package of interest of a user from a commodity library of interest of the user, where the commodity package of interest includes a plurality of commodities belonging to a plurality of stages, and the plurality of stages have a chronological order, and a purchase timing sequence of a commodity of a previous stage is prior to a purchase timing sequence of a commodity of a next stage.

The purchase interval duration obtaining module 1204 is configured to obtain a universal purchase interval duration of two-stage commodities adjacent to each other in time sequence in the interested commodity package.

As shown in fig. 13, in an embodiment, the article recommendation apparatus further includes a general purchase interval duration counting module 1302 and a general purchase interval duration storage module 1304, wherein: the ordinary purchasing interval duration module 1302 is configured to count the ordinary purchasing interval duration of the two-stage commodities adjacent to each other in the time sequence in the commodity package; the common purchase interval duration storage module 1304 is configured to store the common purchase interval durations of the two-stage commodities adjacent to the time sequence of each commodity package, and the purchase interval duration acquisition module 1204 may extract the common purchase interval durations of the two-stage commodities adjacent to the time sequence of the commodity package of interest from the pre-stored common purchase interval duration data of the two-stage commodities adjacent to the time sequence of each commodity package.

In one embodiment, the universal purchase interval duration module 1302 is configured to extract purchase records sorted by user from a purchase record database.

Further, the ordinary purchase interval duration module 1302 is further configured to extract the name keyword of the commodity in the purchase record of each user, mark a stage of the commodity package to which each commodity belongs, and mark a stage of the commodity package to which the commodity belongs if the name keyword of a certain commodity matches with a commodity at the stage in the commodity package.

In an embodiment, the article recommendation apparatus further includes a noun matching relationship storage module (not shown in the figure) for setting a matching relationship between nouns and storing a noun relationship matching library. For example, semantically approximating words may be set to have matching relationships between them. In one embodiment, if the name keyword of a certain product matches the class name of the preset product class to which a certain product at a certain stage in the product package, i.e. if the name keyword of a certain product matches the class name of the preset product class to which a certain product at a certain stage in the product package belongs, the universal purchase interval duration module 1302 may determine that the name keyword of the certain product matches the product at the certain stage in the product package.

Further, the ordinary purchase interval duration module 1302 is further configured to count average purchase interval durations of two adjacent stages of the time sequence of the purchase commodity package of each user.

In one embodiment, the general purchase interval duration module 1302 may calculate an average value of the purchase interval durations of the two adjacent stages of the time sequence of the multiple purchases of the commodity package by a user as the average purchase interval duration of the two adjacent stages of the time sequence of the purchase of the user.

Further, the ordinary purchasing interval duration module 1302 is further configured to calculate the ordinary purchasing interval duration of the time-sequence adjacent two-stage commodities according to the average purchasing interval duration of each user of the time-sequence adjacent two-stage commodities.

In one embodiment, the common purchase interval duration module 1302 may calculate the common purchase interval duration of the two-stage commodities adjacent to the time sequence of a commodity package as an average or weighted average of the average purchase interval durations of the two-stage commodities adjacent to the time sequence of the commodity package purchased by each user.

The latest time and stage obtaining module 1206 is configured to extract a history record of the similar commodities of the commodities in the commodity package of interest purchased by the user according to the commodity purchase record of the user, and determine the latest time when the user purchases the similar commodities and the latest stage to which the similar commodities belong in the multiple stages of the commodity package of interest according to the history record.

In one embodiment, the process of the latest time and stage obtaining module 1206 extracting the historical records of the same type of goods that the user purchased the goods in the interest package according to the goods purchase record of the user includes: and if the name key words of the commodities in one commodity purchase record are matched with the class name of a preset commodity class to which any commodity in the interested commodity package belongs, the commodity purchase record is extracted as the historical record of the similar commodities the user purchases the interested commodities. Further, the latest time and stage acquiring module 1206 may mark the stage corresponding to the commodity purchase record as the stage where the matched commodity in the interested commodity package is located.

Further, the latest time and stage obtaining module 1206 may obtain the latest stage to which the similar product belongs in the multiple stages of the interested product package by comparing the stages corresponding to the history records of the similar product purchased by the user for the interested product.

The second recommendation time setting module 1208 is configured to set, with the latest time as a starting point, the recommendation time of the commodities in each stage after the latest stage in the interested commodity package according to the universal purchase interval duration of the commodities in two stages adjacent to each other in the interested commodity package.

In an embodiment, the second recommended time setting module 1208 may use the latest time as a starting point, and use a product of a time interval between the latest stage and a universal purchase of a commodity in a next stage of the latest stage and a preset weight as a time increment, so as to set the recommended time of the commodity in the next stage of the latest stage; and sequentially setting the recommendation time of the commodities of the next and later stages, wherein the recommendation time of the commodities of each later stage is increased by taking the recommendation time of the commodities of the last stage of the stage as a starting point and multiplying the universal purchase interval time of the commodities of the stage and the last stage by a preset weight.

The second recommending module 1210 is used for recommending commodities in each stage after the latest stage to the user according to the recommending time of each stage.

As shown in fig. 14, in one embodiment, the article recommending apparatus further includes a personal purchase interval duration 1402 and a second recommended time adjustment module 1404, wherein:

the personal purchase interval duration 1402 is used for acquiring the personal purchase interval duration of the user for the time-sequence adjacent two-stage commodities in the interested commodity package.

In one embodiment, personal purchase interval duration 1402 may extract a user's purchase record from a purchase record database; extracting name keywords of the commodities in the purchase record of the user, marking the stage of each commodity belonging to a plurality of stages of the interested commodity package, and if the name keywords of a certain commodity are matched with the commodities at a certain stage in the interested commodity package, marking the stage of the commodity belonging to the interested commodity package; and counting the average purchasing interval duration of the commodities in two adjacent stages of the time sequence of the interested commodity package purchased by the user to obtain the personal purchasing interval duration.

The second recommended time adjustment module 1404 is configured to advance the recommended time of the commodities in each stage after the latest stage in the interested commodity package according to the personal purchase interval duration if the personal purchase interval duration of the commodities in two stages adjacent to the time sequence is shorter than the universal purchase interval duration of the commodities in two stages adjacent to the time sequence.

As shown in fig. 15, in an embodiment, the article recommendation apparatus further includes an unpurchased article extraction module 1502 and a third recommendation module 1504, where:

the unpurchased commodity extraction module 1502 is configured to extract commodities that have not been purchased by the user in the commodities of the latest stage of the interested commodity package; the third recommending module 1504 recommends commodities, which the user has not purchased, to the user in real time.

The process of the third recommending module 1504 recommending commodities, which have not been purchased by the user, to the user in real time includes: and recommending the commodities which are not purchased by the user to the user by taking the current time as the recommendation time.

As shown in fig. 16, in an embodiment, the above commodity recommendation device further includes a commodity package model building module 1602 and a commodity package of interest setting module 1604, where:

the commodity package model building module 1602 is used to extract the purchase records classified by the user from the purchase record database.

Further, the commodity package model building module 1602 is further configured to extract commodities purchased by each user at the same time period, and classify a plurality of commodities, which have a frequency exceeding a threshold value and appear simultaneously in the commodities purchased by each user at the same time period, as commodities at the same stage, so as to summarize a plurality of stages and commodities corresponding to the plurality of stages.

Further, the commodity package model building module 1602 is further configured to count the purchase interval duration of each stage of commodity purchased by each user according to the purchase record of each user.

Further, the commodity package model building module 1602 is further configured to divide the two-stage commodities into the two-stage commodities of the same commodity package if the duration of the purchase interval for each user to purchase a certain two-stage commodity is approximately the same and the purchase sequence for each user to purchase the two-stage commodity is consistent, so as to obtain a commodity package composed of a plurality of stage commodities, and store the commodity package data.

The interested commodity package setting module 1604 is configured to set a commodity package containing a commodity of the same type as the commodity of interest of the user as the interested commodity package of the user, that is, if the commodity package contains a commodity of the same type as the commodity of interest of the user, the commodity package can be set as the interested commodity package of the user. Further, the interested goods package setting module 1604 may store the interested goods package of the user in the interested goods library of the user.

The first commodity recommending device for recommending the interesting commodities and the second commodity recommending device for recommending the commodities in the interesting commodity package can be combined into one device. A commodity recommendation device includes a module in the first commodity recommendation device in any of the above embodiments and a module in the second commodity recommendation device in any of the above embodiments, and the commodity recommendation device also belongs to the protection scope of the present application.

The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims

1. A commodity information pushing method comprises the following steps:

the server extracts an interested commodity package of the user from an interested commodity library of the user, wherein the interested commodity package comprises a plurality of commodities in a plurality of stages contained in a complete event process, the plurality of stages have sequential time sequence, and the purchase time sequence of the commodities in the previous stage is prior to that of the commodities in the next stage; wherein, the statistics step of the commodity package comprises: extracting purchase records classified by users from a purchase record database; extracting commodities purchased by each user at the same time period, and dividing a plurality of commodities which simultaneously appear in the commodities purchased by each user at the same time period and have the times exceeding a threshold value into commodities at the same stage, thereby summarizing a plurality of stages and commodities corresponding to the stages; counting the purchase interval duration of each user for purchasing commodities at each stage according to the purchase record of each user; if the time length of the purchase interval for each user to purchase a certain two-stage commodity approaches to the same time length and the purchase sequence for each user to purchase the two-stage commodity is consistent, dividing the two-stage commodity into two-stage commodities of the same commodity package, thereby obtaining a commodity package consisting of a plurality of stage commodities;

the server acquires the universal purchase interval duration of two-stage commodities adjacent in time sequence in the interested commodity package;

the server extracts the historical records of the same type of commodities of the commodities in the interested commodity package purchased by the user according to the commodity purchase records of the user, and determines the latest time for the user to purchase the same type of commodities and the latest stage of the same type of commodities in the plurality of stages according to the historical records;

the server sets the pushing time of the commodities in each stage after the latest stage in the interested commodity package by taking the latest time as a starting point according to the universal purchasing interval duration;

and the server pushes commodity information corresponding to commodities in each stage after the latest stage to a terminal corresponding to the user according to the pushing time.

2. The commodity information pushing method according to claim 1, further comprising a step of counting a time interval between ordinary purchases of two stages of commodities adjacent in time sequence in the commodity package, the step comprising the steps of:

the server extracts purchase records classified by users from a purchase record database;

the server extracts the name key words of the commodities in the purchase records of the users, marks the commodity package to which the commodities belong, and marks the stage to which the commodities belong if the name key words of the commodities are matched with the commodities at a certain stage in the commodity package;

the server counts the average purchase interval duration of the commodities in two adjacent stages of the time sequence of each user purchasing the commodity package;

and the server calculates the universal purchase interval duration of the time sequence adjacent two-stage commodities according to the average purchase interval duration of each user of the time sequence adjacent two-stage commodities.

3. The commodity information pushing method according to claim 1, further comprising the steps of:

the server acquires the personal purchase interval duration of the user aiming at the time sequence adjacent two-stage commodities in the interested commodity package;

if the personal purchase interval duration of the two-stage commodities adjacent to the time sequence is shorter than the universal purchase interval duration of the two-stage commodities adjacent to the time sequence, the server advances the pushing time of the commodities in each stage after the latest stage in the interested commodity package according to the personal purchase interval duration.

4. The commodity information pushing method according to claim 1, further comprising the steps of:

the server extracts commodities which are not purchased by the user in the latest stage commodities of the interested commodity package;

and the server pushes commodity information corresponding to the commodities not purchased to the terminal corresponding to the user in real time.

5. A server for pushing merchandise information, comprising:

the commodity package model building module is used for extracting purchase records classified by users from the purchase record database; extracting commodities purchased by each user at the same time period, and dividing a plurality of commodities which simultaneously appear in the commodities purchased by each user at the same time period and have the times exceeding a threshold value into commodities at the same stage, thereby summarizing a plurality of stages and commodities corresponding to the stages; counting the purchase interval duration of each user for purchasing commodities at each stage according to the purchase record of each user; if the time length of the purchase interval for each user to purchase a certain two-stage commodity approaches to the same time length and the purchase sequence for each user to purchase the two-stage commodity is consistent, dividing the two-stage commodity into two-stage commodities of the same commodity package, thereby obtaining a commodity package consisting of a plurality of stage commodities;

the commodity package extraction module is used for extracting the commodity package of interest of the user from the commodity library of interest of the user, wherein the commodity package of interest comprises a plurality of commodities in a plurality of stages contained in a complete event process, the plurality of stages have sequential time sequence, and the purchase time sequence of the commodity in the previous stage is prior to that of the commodity in the next stage;

a second pushing time setting module, configured to set, with the latest time as a starting point, pushing times of the commodities in each stage subsequent to the latest stage in the interested commodity package according to the universal purchasing interval duration;

and the second pushing module is used for pushing the commodity information corresponding to the commodities in each stage after the latest stage to the terminal corresponding to the user according to the pushing time.

6. The server according to claim 5, further comprising a general purchase interval duration counting module and a general purchase interval duration storage module;

the ordinary purchasing interval duration counting module is used for counting the ordinary purchasing interval duration of the commodities in two adjacent stages of the time sequence in the commodity package;

the ordinary purchasing interval duration storage module is used for storing the ordinary purchasing interval duration of the commodities in two adjacent stages of the time sequence of various commodity packages;

the purchase interval duration acquisition module is further used for extracting the universal purchase interval duration of the commodities in the two stages adjacent to the time sequence of the commodity packet of interest from the pre-stored universal purchase interval duration data of the commodities in the two stages adjacent to the time sequence of various commodity packets.

7. The server according to claim 6, wherein the general purchase interval duration statistic module is further configured to extract purchase records classified by users from the purchase record database; extracting the name key words of the commodities in the purchase records of the users, marking the commodity package to which the commodities belong, and marking the stage to which the commodities belong if the name key words of the commodities are matched with the commodities at a certain stage in the commodity package; counting the average purchasing interval duration of the commodities in two adjacent stages of the time sequence of purchasing the commodity package by each user; and calculating the universal purchase interval duration of the time sequence adjacent two-stage commodities according to the average purchase interval duration of each user of the time sequence adjacent two-stage commodities.

8. The server according to claim 6, wherein the ordinary purchasing interval duration counting module is further configured to calculate the ordinary purchasing interval duration of the two-stage commodities adjacent to the time sequence of a certain commodity package as an average or weighted average of average purchasing interval durations of the two-stage commodities adjacent to the time sequence of the commodity package purchased by each user.

9. The server according to claim 5, wherein the server further comprises a personal purchase interval duration acquisition module and a second recommended time adjustment module;

the personal purchase interval duration acquisition module is used for acquiring the personal purchase interval duration of the user aiming at the time sequence adjacent two-stage commodities in the interested commodity package;

the second recommendation time adjusting module is used for advancing the pushing time of the commodities in each stage after the latest stage in the interested commodity package according to the personal purchase interval duration if the personal purchase interval duration of the commodities in two stages adjacent to the time sequence is shorter than the universal purchase interval duration of the commodities in two stages adjacent to the time sequence.

10. The server of claim 5, further comprising:

the unpurchased commodity extraction module is used for extracting commodities which are not purchased by the user in the latest stage commodities of the interested commodity package;

and the third pushing module is used for pushing the commodity information corresponding to the commodity which is not purchased to the user.

11. A computer-readable storage medium, storing a computer program which, when executed by a processor, causes the processor to carry out the steps of the method according to any one of claims 1 to 4.