CN117035893A

CN117035893A - Information recommendation method, device, equipment and storage medium

Info

Publication number: CN117035893A
Application number: CN202210476090.6A
Authority: CN
Inventors: 马豪杰; 吴小舟; 周浩; 隋冬; 池涛; 孙严; 叶宸志
Original assignee: Beijing Dajia Internet Information Technology Co Ltd
Current assignee: Beijing Dajia Internet Information Technology Co Ltd
Priority date: 2022-04-29
Filing date: 2022-04-29
Publication date: 2023-11-10

Abstract

The disclosure relates to an information recommendation method, an information recommendation device, information recommendation equipment and a storage medium, and belongs to the technical field of computers. The method comprises the following steps: responding to an item recommendation instruction of a first account to acquire a plurality of alternative items; determining the shelf rate and the acquisition rate of a plurality of candidate articles; determining recommendation probabilities of the plurality of candidate items based on the stocking rates and the acquisition rates of the plurality of candidate items; based on the recommendation probabilities of the plurality of candidate items, the plurality of candidate items are recommended for the first account. Because the loading rate indicates the probability of loading the candidate item on the item supply platform of the first account when the candidate item is recommended for the first account, and the acquisition rate indicates the probability of detecting the acquisition operation of the loaded candidate item when the candidate item is loaded on the item supply platform, the probability of loading the candidate item on the item supply platform and being acquired by other users can be improved based on the loading rate and the acquisition rate.

Description

Information recommendation method, device, equipment and storage medium

Technical Field

The disclosure relates to the field of computer technology, and in particular, to an information recommendation method, an information recommendation device and a storage medium.

Background

With the development of computer technology, live shopping has rapidly become a mainstream shopping method. In the related art, a server recommends a plurality of articles for a merchant according to the preference of the merchant, the merchant selects the articles from the plurality of articles, the selected articles are put on an article supply platform, and then a user can acquire the articles in the article supply platform. However, since the preference of the merchant is different from the preference of the user, the item recommended to the merchant may not be the item that the user likes, and thus the accuracy of item recommendation is low.

Disclosure of Invention

The disclosure provides an information recommendation method, an information recommendation device, information recommendation equipment and a storage medium, which can improve the accuracy of article recommendation. The technical scheme of the present disclosure is as follows:

according to a first aspect of embodiments of the present disclosure, there is provided an information recommendation method, the method including:

responding to an item recommendation instruction of a first account to acquire a plurality of alternative items;

determining a shelf availability and an acquisition availability of the plurality of candidate items, the shelf availability indicating a probability of a shelf of the candidate item on an item supply platform of the first account if the candidate item is recommended for the first account, the acquisition availability indicating a probability of an acquisition operation of the shelf of the candidate item being detected if the candidate item is shelf on the item supply platform;

Determining recommendation probabilities of the plurality of candidate articles based on the loading rates and the acquisition rates of the plurality of candidate articles, wherein the recommendation probabilities are in positive correlation with the loading rates and the acquisition rates;

recommending the plurality of candidate items for the first account based on the recommendation probabilities of the plurality of candidate items.

In some embodiments, determining the acquisition rate of the plurality of candidate items comprises:

acquiring first account information, second account information and article information of the plurality of alternative articles, wherein the second account is an account focusing on the first account;

for any alternative item, calling an acquisition rate prediction model, and determining the acquisition rate of the alternative item based on item information of the alternative item, the first account information and the second account information.

In some embodiments, the training process of the acquisition rate prediction model comprises:

acquiring first sample information, wherein the first sample information comprises first sample account information, second sample account information, sample article information and a first sample label, the second sample account is an account focusing on the first sample account, the first sample label indicates that after the sample article is put on a shelf on an article supply platform of the first sample account, the acquisition operation of the sample article put on the shelf by other accounts except the first sample account is detected, or the acquisition operation of the sample article is not detected;

Invoking the acquisition rate prediction model, and determining a first prediction label of the sample article based on the first sample account information, the second sample account information and the sample article information;

and training the acquisition rate prediction model based on the first sample tag and the first prediction tag so as to reduce the difference between the first prediction tag and the first sample tag obtained by the trained acquisition rate prediction model.

In some embodiments, obtaining the first sample tag comprises:

determining a first sample label corresponding to the sample article as a first numerical value when the acquisition operation of the sample article is detected within a first time period after a loading time point of the sample article, wherein the loading time point is a time point when the sample article is loaded on an article supply platform of the first sample account;

and determining a first sample label corresponding to the sample article as a second value when the acquisition operation of the sample article is not detected within the first time period after the loading time point.

In some embodiments, the determining the recommended probabilities for the plurality of candidate items based on the pick rates and the acquisition rates for the plurality of candidate items comprises:

And multiplying the loading rate of any alternative item by the acquisition rate to obtain the recommendation probability of the alternative item.

determining a first article sequence and a second article sequence based on the shelf rate and the acquisition rate of the plurality of alternative articles, wherein the first article sequence is an article sequence obtained by sequencing the plurality of alternative articles according to the order of the shelf rate from big to small, and the second article sequence is an article sequence obtained by sequencing the plurality of alternative articles according to the order of the acquisition rate from big to small;

for any alternative item, determining a first recommendation probability for the alternative item based on the order of the alternative item in the first item sequence, and determining a second recommendation probability for the alternative item based on the order of the alternative item in the second item sequence;

and weighting the first recommendation probability and the second recommendation probability to obtain the recommendation probability of the candidate object.

In some embodiments, the determining the pick up rate of the plurality of candidate items comprises:

acquiring first account information and article information of the plurality of alternative articles;

and for any alternative article, calling an inventory rating prediction model, and determining the inventory rating of the alternative article based on the article information of the alternative article and the first account information.

In some embodiments, the training process of the shelf rate prediction model includes:

acquiring second sample information, wherein the second sample information comprises third sample account information, sample article information and a second sample label, and the second sample label indicates that after recommending the sample article for the third sample account, the sample article is put on an article supply platform of the third sample account or the sample article is not put on the article supply platform;

invoking the shelf rate prediction model, and determining a second prediction label of the sample article based on the third sample account information and the sample article information;

and training the shelf rate prediction model based on the second sample label and the second prediction label so as to reduce the difference value between the second prediction label and the second sample label obtained by the trained shelf rate prediction model.

In some embodiments, obtaining the second sample tag comprises:

after recommending the sample article for the third sample account, determining a second sample label corresponding to the sample article as a third numerical value under the condition that the sample article is put on a shelf on an article supply platform of the third sample account;

and after recommending the sample object for the third sample account, determining a second sample label corresponding to the sample object as a fourth value under the condition that the sample object is not put on a shelf on an object supply platform of the third sample account.

In some embodiments, the obtaining a plurality of candidate items in response to the item recommendation instructions for the first account includes:

responding to the article recommendation instruction, determining a second account, wherein the second account is an account focusing on the first account;

for any one of a plurality of item categories, determining, from the determined second account, a number of the second accounts from which items within the item category were obtained;

determining a first number of target item categories from the plurality of item categories, wherein the number of the second accounts corresponding to the target item categories is the largest;

And selecting the plurality of candidate articles from the target article categories.

In some embodiments, before the determining the loading rate and the acquisition rate of the plurality of candidate items, the information recommendation method further includes:

for any alternative article, acquiring a third account number corresponding to the alternative article, wherein the server detects the acquisition operation of the alternative article in an article supply platform of the third account in a second time period before the current time point;

sorting the plurality of candidate articles according to the sequence from the large number to the small number of the third accounts corresponding to the plurality of candidate articles;

a second number of the candidate items ordered first is obtained.

According to a second aspect of embodiments of the present disclosure, there is provided a model training method, the method comprising:

acquiring first sample information, wherein the first sample information comprises first sample account information, second sample account information, sample article information and a first sample label, the second sample account is an account focusing on the first sample account, and the first sample label indicates that after the sample article is put on a shelf on an article supply platform of the first sample account, the acquisition operation of the sample article put on the shelf is detected or the acquisition operation of the sample article is not detected;

Invoking an acquisition rate prediction model, and determining a first prediction label of the sample article based on the first sample account information, the second sample account information and the sample article information;

According to a third aspect of embodiments of the present disclosure, there is provided an information recommendation apparatus, the apparatus including:

an item acquisition unit configured to execute an item recommendation instruction in response to a first account to acquire a plurality of candidate items;

a first determination unit configured to perform determination of a pickup rate indicating a probability of pickup of the alternative item by an item supply platform of the first account in a case where the alternative item is recommended for the first account, and an acquisition rate indicating a probability of detection of an acquisition operation of the pickup of the alternative item by other accounts than the first account in a case where the alternative item is picked up by the item supply platform;

A second determining unit configured to perform determining a recommendation probability of the plurality of candidate items based on the stocking rate and the acquisition rate of the plurality of candidate items, the recommendation probability having a positive correlation with the stocking rate and the acquisition rate;

an item recommending unit configured to execute recommending the plurality of candidate items for the first account based on the recommendation probabilities of the plurality of candidate items.

In some embodiments, the first determining unit is configured to perform obtaining first account information, second account information, and item information of the plurality of candidate items, the second account being an account focused on the first account; for any alternative item, calling an acquisition rate prediction model, and determining the acquisition rate of the alternative item based on item information of the alternative item, the first account information and the second account information.

In some embodiments, obtaining the first sample tag comprises:

In some embodiments, the second determining unit is configured to perform multiplying the pick up rate of the candidate item by the acquisition rate for any candidate item, resulting in the recommended probability of the candidate item.

In some embodiments, the second determining unit is configured to perform determining a first article sequence and a second article sequence based on the pickup rates and the acquisition rates of the plurality of candidate articles, the first article sequence being an article sequence obtained by sorting the plurality of candidate articles in order of the pickup rates from large to small, the second article sequence being an article sequence obtained by sorting the plurality of candidate articles in order of the acquisition rates from large to small; for any alternative item, determining a first recommendation probability for the alternative item based on the order of the alternative item in the first item sequence, and determining a second recommendation probability for the alternative item based on the order of the alternative item in the second item sequence; and weighting the first recommendation probability and the second recommendation probability to obtain the recommendation probability of the candidate object.

In some embodiments, the first determining unit is configured to perform acquiring first account information and item information of the plurality of candidate items; and for any alternative article, calling an inventory rating prediction model, and determining the inventory rating of the alternative article based on the article information of the alternative article and the first account information.

In some embodiments, obtaining the second sample tag comprises:

In some embodiments, the item acquisition unit is configured to execute a determination of a second account in response to the item recommendation instruction, the second account being an account focused on the first account; for any one of a plurality of item categories, determining, from the determined second account, a number of the second accounts from which items within the item category were obtained; determining a first number of target item categories from the plurality of item categories, wherein the number of the second accounts corresponding to the target item categories is the largest; and selecting the plurality of candidate articles from the target article categories.

In some embodiments, the article obtaining unit is further configured to perform obtaining, for any alternative article, a third account number corresponding to the alternative article, where the server detects, in a second period of time before the current time point, an obtaining operation for the alternative article in an article supply platform of the third account number; sorting the plurality of candidate articles according to the sequence from the large number to the small number of the third accounts corresponding to the plurality of candidate articles; a second number of the candidate items ordered first is obtained.

According to a fourth aspect of embodiments of the present disclosure, there is provided a model training apparatus, the apparatus comprising:

an information acquisition unit configured to perform acquisition of first sample information including first sample account information, second sample account information, sample article information, and a first sample tag, the second sample account being an account focusing on the first sample account, the first sample tag indicating that an acquisition operation of the sample article on the shelf is detected or an acquisition operation of the sample article is not detected after the sample article is set up on an article supply platform of the first sample account;

a model calling unit configured to execute a call acquisition rate prediction model, determining a first prediction tag of the sample item based on the first sample account information, the second sample account information, and the sample item information;

and a model training unit configured to perform training of the acquisition rate prediction model based on the first sample label and the first prediction label so that a difference between the first prediction label and the first sample label obtained by the trained acquisition rate prediction model is reduced.

According to a fifth aspect of embodiments of the present disclosure, there is provided an electronic device including:

a processor;

a memory for storing the processor-executable instructions;

wherein the processor is configured to execute the instructions to implement the information recommendation method or the model training method as described in the above aspects.

According to a sixth aspect provided by embodiments of the present disclosure, there is provided a computer readable storage medium, which when executed by a processor of an electronic device, causes the electronic device to perform the information recommendation method or the model training method as described in the above aspects.

According to a seventh aspect of embodiments of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the information recommendation method or model training method of the above aspects.

In the embodiment of the disclosure, since the loading rate of the candidate item indicates the probability of loading the candidate item on the item supply platform of the first account in the case of recommending the candidate item for the first account, and the acquisition rate of the candidate item indicates the probability of detecting the acquisition operation of the loaded candidate item in the case of loading the candidate item on the item supply platform of the first account, the recommendation probability of the candidate item is determined based on the loading rate and the acquisition rate of the candidate item, and the candidate item is recommended for the first account based on the recommendation probability, so that the probability of loading the candidate item on the item supply platform and being acquired by other users can be improved, and the accuracy of item recommendation can be improved.

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

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure and do not constitute an undue limitation on the disclosure.

FIG. 1 is a schematic diagram illustrating an implementation environment according to an example embodiment.

Fig. 2 is a flowchart illustrating an information recommendation method according to an exemplary embodiment.

Fig. 3 is a flowchart illustrating another information recommendation method according to an exemplary embodiment.

FIG. 4 is a flowchart illustrating a model training method according to an exemplary embodiment.

FIG. 5 is a flow chart illustrating another model training method according to an exemplary embodiment.

Fig. 6 is a flowchart illustrating yet another information recommendation method according to an exemplary embodiment.

Fig. 7 is a flowchart illustrating yet another information recommendation method according to an exemplary embodiment.

FIG. 8 is a schematic diagram illustrating an item recommendation process, according to an example embodiment.

Fig. 9 is a block diagram illustrating an information recommendation apparatus according to an exemplary embodiment.

FIG. 10 is a block diagram illustrating a model training apparatus, according to an example embodiment.

Fig. 11 is a block diagram of a terminal according to an exemplary embodiment.

Fig. 12 is a block diagram of a server, according to an example embodiment.

Detailed Description

In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

It should be noted that the terms "first," "second," and the like in the description of the present disclosure and the claims and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present disclosure as detailed in the accompanying claims.

It should be noted that, the user information (including, but not limited to, user equipment information, user personal information, etc.) related to the present disclosure may be information authorized by the user or sufficiently authorized by each party.

The terms "at least one," "a plurality," "each," "any" as used herein, at least one includes one, two or more, a plurality includes two or more, and each refers to each of a corresponding plurality, any of which refers to any of the plurality. For example, the plurality of accounts includes 3 accounts, and each account refers to each account of the 3 accounts, and any account of the 3 accounts may be a first account, a second account, or a third account.

FIG. 1 is a schematic illustration of an implementation environment provided by embodiments of the present disclosure. Referring to fig. 1, the implementation environment includes a terminal 101 and a server 102. The terminal 101 and the server 102 are connected by a wireless or wired network. The terminal 101 is, for example, a notebook computer, a mobile phone, a tablet computer, or other terminals. The server 102 is illustratively a background server 102 of an application or a cloud server 102 providing services such as cloud computing and cloud storage.

The terminal 101 is configured to log in an account, and send an item recommendation request to the server 102, where the account is an account registered in the server 102, and the item recommendation request is configured to request that an item be recommended for the account. The server 102 is configured to receive the item recommendation request, generate an item recommendation instruction for the account, determine, by using the information recommendation method provided by the present disclosure, at least one candidate item recommended by the first account, and send the at least one candidate item to the terminal 101. The terminal 101 is configured to receive the at least one alternative item and display the at least one alternative item. In this way, the user corresponding to the terminal can select an article from at least one alternative article, the selected article is put on the article supply platform of the account, and then other users can acquire the articles in the article supply platform. For example, after the user selects an alternative item from the displayed at least one alternative item, sending an item shelving request to the server 102 through the terminal 101, where the item shelving request carries the alternative item selected by the user, the server 102 adds the alternative item selected by the user to an item shelving set of the first account, where the item in the item shelving set is used for being displayed in an item supply platform of the first account.

Illustratively, an application is installed in the terminal 101, the terminal 101 logs in an account in the application, sends an item recommendation request to the server 102 through the application, receives at least one alternative item sent by the server 102, and presents the at least one alternative item. The terminal 101 also mounts the item selected by the user to the item supply platform through the and application. The application is, for example, an application in the operating system of the terminal 101 or a third party application. Illustratively, the application is a live application, shopping application, gaming application, etc., to which embodiments of the present disclosure are not limited.

The method provided by the embodiment of the disclosure can be applied to scenes of recommending food to the user. For example, a user sends a food recommendation request to a server through a terminal, the food recommendation request carries a first account number which is currently logged in by the terminal, the server determines a plurality of alternative foods after receiving the recommendation request, determines recommendation probabilities of the plurality of alternative foods through the method provided by the disclosure, then determines a plurality of alternative foods with larger recommendation probabilities based on the recommendation probabilities of the plurality of alternative foods, and sends the plurality of alternative foods to the terminal. The terminal receives the multiple alternative delicious foods and displays the multiple alternative delicious foods. Then, the user can select food from the food, put the selected food on the goods supply platform of the account, and then, other users can acquire the food in the goods supply platform.

The method provided by the embodiment of the disclosure can be applied to a scene of recommending cosmetics to a user. For example, a user sends a cosmetic recommendation request to a server through a terminal, the cosmetic recommendation request carries a first account number which is currently logged in by the terminal, the server determines a plurality of candidate cosmetics after receiving the recommendation request, determines recommendation probabilities of the plurality of candidate cosmetics through the method provided by the disclosure, then determines a plurality of candidate cosmetics with larger recommendation probabilities based on the recommendation probabilities of the plurality of candidate cosmetics, and sends the plurality of candidate cosmetics to the terminal. The terminal receives the plurality of candidate cosmetics and displays the plurality of candidate cosmetics. Thereafter, the user can select cosmetics from among them, put the selected cosmetics on the goods supply platform of the account, and thereafter, other users can acquire the cosmetics in the goods supply platform.

The method provided by the embodiment of the disclosure can be applied to the scene of recommending the electronic product to the user. For example, a user sends an electronic product recommendation request to a server through a terminal, the electronic product recommendation request carries a first account number which is currently logged in by the terminal, the server determines a plurality of alternative electronic products after receiving the recommendation request, determines recommendation probabilities of the plurality of alternative electronic products through the method provided by the disclosure, then determines a plurality of alternative electronic products with larger recommendation probabilities based on the recommendation probabilities of the plurality of alternative electronic products, and sends the plurality of alternative electronic products to the terminal. And the terminal receives the plurality of candidate electronic products and displays the plurality of candidate electronic products. Then, the user can select the electronic products from the account, put the selected electronic products on the object supply platform of the account, and then, other users can acquire the electronic products in the object supply platform.

The information recommendation method provided by the embodiment of the disclosure can also be applied to a scene of recommending other types of articles to a user, such as sports goods, learning articles and the like, and the embodiment of the disclosure is not limited thereto.

Fig. 2 is a flowchart illustrating an information recommendation method according to an exemplary embodiment, including the following steps, as shown in fig. 2.

201. And the server responds to the item recommendation instruction of the first account to acquire a plurality of candidate items.

The article recommending instruction of the first account indicates that the article is recommended by the first account. The server generates an article recommendation instruction for the first account after receiving an article recommendation request carrying the first account sent by the terminal. The terminal sends the item recommendation request to an interface corresponding to the item recommendation service, and the server receives the item recommendation request through the interface. The first account is any account registered in the server, for example, an account registered by a merchant.

By way of example, the type of alternative item is any type, e.g., cosmetic, food product, electronic product, office product, etc. Illustratively, the server has stored therein a set of candidate items for storing a plurality of candidate items. Accordingly, the server obtains a plurality of candidate items from the candidate item set.

The article recommendation request also carries article recommendation conditions, and accordingly, the server selects a plurality of candidate articles meeting the article recommendation conditions from the candidate article set. Exemplary item recommendation conditions include a resource range in which the number of resources corresponding to an item is located, an acquisition amount range that an item acquisition amount needs to satisfy, a type of an item, a manner in which an alternative item is selected from an alternative item set, and the like.

202. The server determines a stocking rate and an acquisition rate for the plurality of candidate items.

The loading rate indicates a probability of loading the alternative item on the item supply platform of the first account with the alternative item recommended for the first account. The higher the loading rate, the greater the probability that the user to which the first account belongs likes the alternative item. After recommending the alternative articles to the first account, the user to which the first account belongs can determine whether to put the alternative articles on the article supply platform of the user, so that the server determines the put-on rate of the alternative articles when recommending the articles for the first account, recommends the articles for the first account based on the put-on rate, and the probability that the recommended articles are put on the article supply platform by the user can be improved, thereby improving the accuracy of article recommendation.

The acquisition rate indicates a probability of detecting an acquisition operation of the on-shelf candidate item by other accounts than the first account in a case where the on-shelf candidate item is on the item supply platform. The higher the acquisition rate, the greater the probability that other users like the alternative item. After the user to which the first account belongs puts the article on the article supply platform, other users can determine whether to acquire the article from the article supply platform, so that when recommending the article for the first account, the server determines the acquisition rate of the alternative article, recommends the article for the first account based on the acquisition rate, the probability that the recommended article is acquired by other users after the article is put on the article supply platform of the first account can be improved, and the accuracy of article recommendation is improved.

The article supply platform can display articles, and after the articles are put on the article supply platform, a user can see the articles put on the article supply platform, so that the articles put on the article supply platform are obtained. Illustratively, the item supply platform includes a living room, an online store, and the like.

203. The server determines a recommendation probability for the plurality of candidate items based on the stocking rate and the acquisition rate of the plurality of candidate items.

The recommendation probability and the loading rate are in positive correlation. For example, the server multiplies the loading rate and the obtaining rate of the candidate item to obtain the recommendation probability of the candidate item, so that the greater the recommendation probability, the greater the probability that the candidate item is successfully loaded to the item supply platform of the first account and successfully obtained by other users, that is, the greater the probability that the candidate item is liked by the user. Of course, the recommendation probability of the candidate item can also be determined in other manners, for example, the loading rate and the obtaining rate of the candidate item are weighted and summed to obtain the recommendation probability of the candidate item, which is not limited by the embodiment of the disclosure.

204. The server recommends a plurality of candidate items for the first account based on the recommendation probabilities of the plurality of candidate items.

Illustratively, the server sorts the plurality of candidate items in order of the recommendation probability from the higher to the lower, and recommends the first account with the first, target number of candidate items sorted in the first order. Wherein the target number is any number, e.g. 20. Or recommending the corresponding candidate items with the recommendation probability larger than the probability threshold value to the first account by the server. Wherein the probability threshold is any value, for example, 0.8. The method can ensure that the recommendation probability of the candidate object recommended to the first account is larger, and the larger the recommendation probability is, the larger the candidate object is successfully put on the object supply platform of the first account, and the probability of being successfully acquired by other users is larger, namely, the probability of the candidate object being liked by the users is larger, so that the accuracy of object recommendation is improved.

It should be noted that, if the candidate item is recommended for the first account based on the loading rate only, after the user to which the first account belongs loads the candidate item onto the item supply platform, the acquisition amount of the candidate item is often low, so that the GMV (Gross Merchandise Volume, total amount of commodity transaction) corresponding to the first account is low. In the embodiment of the disclosure, the alternative articles are recommended for the first account by integrating the loading rate and the obtaining rate, so that the probability that the recommended alternative articles are loaded to the article supply platform by the user to which the first account belongs is improved, the probability that the alternative articles are obtained by other users is also improved, and therefore the GMV corresponding to the first account can be improved, and the GMV corresponding to the live broadcast application program is further improved.

The other point to be described is that after the information recommendation method provided by the embodiment of the present disclosure is used for recommending the article for the account, the article exposure times of the article supply platform of the account on the day of article recommendation, the times of acquiring the articles in the article supply platform on the day of article recommendation, the number of accounts of acquiring the articles in the article supply platform on the day of article recommendation, and the GMV of the article supply platform on the day of article recommendation all have jump-type growth.

Fig. 3 is a flowchart illustrating an information recommendation method in which an acquisition rate and a stocking rate of an alternative item are determined through an acquisition rate prediction model and a stocking rate prediction model according to an exemplary embodiment. As shown in fig. 3, the method includes the following steps.

301. And the server responds to the item recommendation instruction of the first account to acquire a plurality of candidate items.

The implementation manner of this step refers to step 201 above, and will not be described here again.

302. The server invokes the acquisition rate prediction model to determine the acquisition rates of the plurality of candidate items.

In some embodiments, this step is implemented as: the server acquires the first account information, the second account information and the article information of a plurality of alternative articles, and for any alternative article, the server invokes an acquisition rate prediction model to determine the acquisition rate of the alternative article based on the article information of the alternative article, the first account information and the second account information. The second account is an account focusing on the first account.

The first account information includes attribute information of a user to which the first account belongs, information of an article on which the first account is put on shelf before a current time point, information of an article acquired by other users among articles on which the first account is put on shelf, and the like. The second account information comprises attribute information of a user to which the second account belongs, information of an article acquired by the second account and the like. Wherein, the attribute information of the user includes age, sex, region of the user, etc. The information of the article includes the type of the article, the number of resources corresponding to the article, and the like.

The server inputs the first account information, the second account information and the article information into an acquisition rate prediction model, predicts the acquisition rate prediction model based on the input first account information, second account information and article information, obtains a prediction tag of the article, and then outputs the prediction tag. Illustratively, the predictive label is a number between 0 and 1, representing the acquisition rate of the candidate item predicted by the acquisition rate prediction model. The training process of the acquisition rate prediction model is described in the embodiment shown in fig. 4 below, and will not be described herein.

In the embodiment of the disclosure, since the acquisition rate prediction model predicts that the accuracy of the acquisition rate is high and the prediction efficiency is high, the acquisition rate of the candidate object is determined by the acquisition rate prediction model, the prediction efficiency can be improved, the accuracy of the acquisition rate of the candidate object is ensured, and the object recommendation is performed by using the acquisition rate, so that the efficiency and accuracy of the object recommendation can be improved.

303. The server invokes the shelf rate prediction model to determine the shelf rates of the plurality of candidate items.

In some embodiments, the server obtains first account information and item information for a plurality of alternative items; and for any alternative article, the server calls an inventory rating prediction model, and determines the inventory rating of the alternative article based on the article information and the first account information of the alternative article.

The first account information includes attribute information of a user to which the first account belongs, information of an article on which the first account is put on shelf before a current time point, and the like. Wherein, the attribute information of the user includes age, sex, region of the user, etc. The information of the article includes the type of the article, the number of resources corresponding to the article, and the like.

The server inputs the first account information and the article information into an article loading rate prediction model, predicts the article loading rate prediction model based on the input first account information and article information, obtains a prediction label of the candidate article, and then outputs the prediction label. Illustratively, the predictive label is a number between 0 and 1, representing the shelf-up rate of the candidate item predicted by the shelf-up rate prediction model. The training process of the loading rate prediction model is described in the embodiment shown in fig. 5 below, and will not be described herein.

In the embodiment of the disclosure, since the shelf rate prediction model predicts that the shelf rate is high in accuracy and high in prediction efficiency, the shelf rate of the candidate articles is determined through the shelf rate prediction model, the prediction efficiency can be improved, the accuracy of the shelf rate of the candidate articles is ensured, and the articles are recommended by using the shelf rate, so that the efficiency and accuracy of article recommendation can be improved.

It should be noted that, the sequence of steps 302 and 303 is not limited in the embodiment of the disclosure, that is, step 303 may be performed first and then step 302 may be performed.

304. The server determines a recommendation probability for the plurality of candidate items based on the stocking rate and the acquisition rate of the plurality of candidate items.

The recommendation probability and the loading rate are in positive correlation.

In some embodiments, the server determines a recommendation probability for the plurality of candidate items based on the pick rate and the acquisition rate for the plurality of candidate items, comprising: for any alternative item, the server multiplies the shelf rate and the acquisition rate of the alternative item to obtain the recommendation probability of the alternative item, so that the recommendation probability can indicate the probability that the alternative item is successfully shelf on the item supply platform of the first account and successfully acquired by other users, and the greater the recommendation probability is, the greater the probability that the alternative item is successfully shelf on the item supply platform of the first account and successfully acquired by other users is, namely, the greater the probability that the alternative item is liked by the users is. Based on the recommendation probability, recommending the articles to the user can improve the probability that the recommended articles are liked by the user, so that the recommendation accuracy is improved.

In some embodiments, the server determines a recommendation probability for the plurality of candidate items based on the pick rate and the acquisition rate for the plurality of candidate items, including steps (1) - (3) below.

(1) The server determines a first article sequence and a second article sequence based on the shelf rate and the acquisition rate of the plurality of alternative articles, wherein the first article sequence is an article sequence obtained by sequencing the plurality of alternative articles according to the order of the shelf rate from big to small, and the second article sequence is an article sequence obtained by sequencing the plurality of alternative articles according to the order of the acquisition rate from big to small.

(2) For any of the candidate items, the server determines a first recommendation probability for the candidate item based on the order of the candidate item in the first sequence of items, and determines a second recommendation probability for the candidate item based on the order of the candidate item in the second sequence of items.

Wherein the first recommended probability of the candidate item is greater the earlier the order of the candidate item in the first sequence of items. The further back the order of the candidate items in the first sequence of items, the less the first recommendation probability for the candidate item. The earlier the order of the candidate items in the second item sequence, the greater the second recommendation probability for the candidate item. The further back the order of the candidate items in the second item sequence, the less the second recommended probability of the candidate item.

For example, the server determines a ratio of an order of the candidate items in the first item sequence to a number of candidate items in the first item sequence, and determines a difference of 1 and the ratio as the first recommendation probability.

For example, the server determines a ratio of the order of the candidate items in the second item sequence to the number of candidate items in the second item sequence, and determines a difference of 1 and the ratio as the second recommendation probability.

(3) And the server performs weighting processing on the first recommendation probability and the second recommendation probability to obtain the recommendation probability of the candidate object.

Illustratively, the server sums the first recommendation probability and the second recommendation probability in a weighted manner to obtain the recommendation probability of the candidate item. The weights of the first recommendation probability and the second recommendation probability can be set to any value, for example, the weight of the first recommendation probability is 0.6, and the weight of the second recommendation probability is 0.4, which is not limited by the embodiments of the present disclosure.

In the embodiment of the disclosure, the first recommendation probability and the second recommendation probability are determined by using the sequence of the candidate articles in the first article sequence and the sequence of the second article sequence, and then the final recommendation probability of the candidate articles is obtained by weighting the first recommendation probability and the second recommendation probability instead of directly determining the recommendation probability of the candidate articles by using the absolute values of the acquisition rate and the loading rate of the candidate articles, so that the determined values of the recommendation probabilities of the candidate articles have smaller difference, that is, the distribution situation of the recommendation probabilities of the plurality of candidate articles is not excessively dispersed, the data processing efficiency of the subsequent recommendation probabilities can be improved, and the article recommendation efficiency is further improved.

305. The server recommends a plurality of candidate items for the first account based on the recommendation probabilities of the plurality of candidate items.

The implementation of this step is referred to above in step 204, and will not be described here again.

In the embodiment of the disclosure, the acquisition rate and the loading rate of the candidate object are determined through the acquisition rate prediction model and the loading rate prediction model, so that the acquisition rate and the acquisition efficiency and the accuracy of the loading rate of the candidate object are ensured, and the accuracy of the recommendation probability of the candidate object and the recommendation efficiency of the candidate object are ensured.

FIG. 4 is a flow chart illustrating a method of information recommendation describing a training process for an acquisition rate prediction model for determining the acquisition rate of an alternative item, according to an exemplary embodiment. As shown in fig. 4, the method includes the following steps.

401. The server acquires first sample information, wherein the first sample information comprises first sample account information, second sample account information, sample article information and a first sample label, the second sample account is an account focusing on the first sample account, the first sample label indicates that after the sample articles are put on a shelf on an article supply platform of the first sample account, the acquisition operation of other accounts except the first sample account on the put-on sample articles is detected, or the acquisition operation of the sample articles is not detected.

The server may randomly select a plurality of accounts from the registered accounts as a first sample account, then take other accounts focusing on the first sample account as a second sample account, and determine that the first sample account was the article on the shelf in the article supply platform as a sample article. The first sample account information includes attribute information of a user to whom the first sample account belongs, information of an article on which the first sample account is put on shelf before a current time point, information of an article acquired by other users in the articles on which the first sample account is put on shelf, and the like. The second sample account information comprises attribute information of a user to which the second sample account belongs, information of an article acquired by the second sample account and the like. Wherein, the attribute information of the user includes age, sex, region of the user, etc. The information of the article includes the type of the article, the number of resources corresponding to the article, and the like.

In some embodiments, the server obtains a first sample tag comprising: the server determines a first sample label corresponding to the sample article as a first value when the acquisition operation of the sample article is detected in a first period of time after the sample article is put on shelf, and determines a first sample label corresponding to the sample article as a second value when the acquisition operation of the sample article is not detected in the first period of time after the sample article is put on shelf. Wherein the racking time point is a time point when the sample article is racking on the article supply platform of the first sample account. The first value represents a first time period after the sample article is put on shelf and is acquired by other users. The second value indicates that the sample item was not acquired by other users for a second period of time after being put on shelf. Illustratively, the first value is 1 and the second value is 0. Illustratively, the first duration is any duration, for example, 3 days. For example, if the first sample tag is a first value, the sample corresponding to the first sample tag is referred to as a positive sample. If the first sample label is the second value, the sample corresponding to the first sample label is called a negative sample.

In the embodiment of the disclosure, in the case that the acquisition operation of the sample article is detected in a first period after the time point of the sample article being put on shelf, a first sample tag corresponding to the sample article is determined to be a first value, and in the case that the acquisition operation of the sample article is not detected in the first period after the time point of the sample article being put on shelf, a first sample tag corresponding to the sample article is determined to be a second value by the server, then the acquisition rate prediction model is trained based on the first sample tag, so that the predicted tag obtained by the acquisition rate prediction model can reflect the acquisition rate of the article, namely, the probability that the article is acquired by other users in the first period after the time point of the sample article being put on shelf, thereby recommending the article based on the acquisition rate and improving the accuracy of article recommendation.

402. The server invokes an acquisition rate prediction model to determine a first prediction tag for the sample item based on the first sample account information, the second sample account information, and the sample item information.

The server inputs the first sample account information, the second sample account information and the sample article information into an acquisition rate prediction model, predicts the acquisition rate prediction model based on the input first sample account information, second sample account information and sample article information, obtains a first prediction tag of the sample article, and then outputs the first prediction tag.

Illustratively, the first predictive label is a number between 0 and 1, representing the acquisition rate of the sample item predicted by the acquisition rate prediction model. For example, a first predictive label of 0.8 indicates that the sample item is 80% likely to be acquired by other users within a first time period after being set up on the item supply platform of the first sample account.

403. The server trains an acquisition rate prediction model based on the first sample tag and the first prediction tag.

The training targets of the acquisition rate prediction model are as follows: and the difference value between the first prediction label and the first sample label obtained by the trained acquisition rate prediction model is reduced. Correspondingly, the server trains the acquisition rate prediction model based on the first sample tag and the first prediction tag, so that the difference between the first prediction tag and the first sample tag obtained by the trained acquisition rate prediction model is reduced.

In an exemplary embodiment, the server determines a loss value for the acquisition rate prediction model based on the first sample tag and the first prediction tag, the loss value indicating a difference between the first prediction tag and the first sample tag, and adjusts the model parameters based on the loss value such that the adjusted loss value for the acquisition rate prediction model is reduced until the loss value is less than a loss threshold.

In the embodiment of the disclosure, considering that the first sample account information, the second sample account information and the sample article information are all important factors capable of affecting that the sample article in the article supply platform of the first sample account is successfully acquired by other users, the acquisition rate prediction model is trained based on the three information, so that the acquisition rate prediction model can learn the relationship between the three information and the first sample tag, namely, the relationship between the three information and whether the sample article in the article supply platform of the first sample account is successfully acquired by other users or not, thereby having the information based on any account, the information of other accounts focusing on the account and the information of any article, and accurately determining the probability that the article in the article supply platform of the account is acquired by other users. The recommendation probability of the articles is determined based on the probability, and the articles are recommended based on the recommendation probability, so that the probability that the recommended articles are liked by users can be improved, and the accuracy of article recommendation is improved.

By way of example, the acquisition rate prediction model can also be trained using information other than the three information described above, which is not limiting in accordance with the disclosed embodiments. The acquisition rate prediction model can also be trained by electronic devices other than a server, for example, nor is the embodiment of the disclosure limited.

FIG. 5 is a flow chart illustrating a method of information recommendation describing a training process for a shelf rate prediction model for determining the shelf rate of an alternative item, according to an exemplary embodiment. As shown in fig. 5, the method includes the following steps.

501. The server obtains second sample information, the second sample information including third sample account information, sample item information, and a second sample tag, the second sample tag indicating whether sample items are or are not being shelved on the item supply platform of the third sample account after recommending sample items for the third sample account.

The server may select a plurality of accounts from the registered accounts as the third sample account, and then determine the recommended item for the third sample account as the sample item. The third sample account information includes attribute information of a user to whom the third sample account belongs, information of an article on which the third sample account is put on shelf before the current time point, and the like. Wherein, the attribute information of the user includes age, sex, region of the user, etc. The information of the article includes the type of the article, the number of resources corresponding to the article, and the like.

In some embodiments, the server obtains a second sample tag comprising: after recommending the sample articles for the third sample account, the server determines a second sample label corresponding to the sample articles as a third numerical value under the condition that the sample articles are put on the shelf on an article supply platform of the third sample account; after recommending the sample object for the third sample account, the server determines a second sample label corresponding to the sample object as a fourth value under the condition that the sample object is not put on the object supply platform of the third sample account. And after the third numerical value is expressed as that the third sample account recommends the sample article, the user to which the third sample account belongs puts the sample article on the article supply platform of the third sample account. And after the fourth numerical value indicates that the third sample account recommends the sample article, the user to which the third sample account belongs does not put the sample article on the article supply platform of the third sample account. Illustratively, the third value is 1 and the fourth value is 0. For example, if the second sample label is a third value, the sample corresponding to the second sample label is referred to as a positive sample. If the second sample label is a fourth value, the sample corresponding to the first sample label is called a negative sample.

For example, considering that after recommending the sample item for the third sample account, the user to which the third sample account belongs needs a certain time to decide whether to put the sample item on the item supply platform, in order to avoid that the delay of the user on the putting on the sample item leads to the marking error of the second sample label corresponding to the sample item, the server delays the time of marking the second sample label corresponding to the sample item, that is, the server does not mark the second sample label corresponding to the sample item within the target time period after recommending the sample item for the third sample account, but determines the second sample label corresponding to the sample item as the third numerical value when the sample item is put on the item supply platform of the third sample account, and determines the second sample label corresponding to the sample item as the fourth numerical value when the sample item is not put on the item supply platform of the third sample account after the target time period is spaced from the time point when the sample item is recommended for the third sample account. Wherein the target time period is any time period, such as 30 minutes, 1 hour, etc.

In the embodiment of the disclosure, after recommending a sample item for a third sample account, determining a second sample label corresponding to the sample item as a third value when the sample item is on the item supply platform of the third sample account, and determining a second sample label corresponding to the sample item as a fourth value when the sample item is not on the item supply platform of the third sample account, training an item loading rate prediction model based on the second sample label, so that the predicted label obtained by the item loading rate prediction model can reflect the item loading rate, namely, the probability that the item is loaded to the item supply platform by the user after recommending the item to the user, thereby recommending the item based on the loading rate and improving the item recommendation accuracy.

502. The server calls an overhead rate prediction model, and determines a second prediction label of the sample item based on the third sample account information and the sample item information.

The server inputs the third sample account information and the sample article information into an overhead rate prediction model, predicts the overhead rate prediction model based on the input third sample account information and sample article information, obtains a second prediction tag of the sample article, and then outputs the second prediction tag.

Illustratively, the second predictive label is a value between 0 and 1 representing the pick-up rate of the sample item predicted by the pick-up rate prediction model. For example, the second predictive label is 0.8, indicating that after recommending the sample item for the third sample account, the third sample account has an 80% likelihood of selecting the sample item to be shelved into the item supply platform.

503. The server trains an up-rate prediction model based on the second sample tag and the second prediction tag.

The training targets of the shelf rate prediction model are as follows: and the difference value between the second prediction label and the second sample label obtained by the trained shelf rate prediction model is reduced. Correspondingly, the server trains the loading rate prediction model based on the second sample label and the second prediction label, so that the difference between the second prediction label and the second sample label obtained by the trained loading rate prediction model is reduced.

In an exemplary embodiment, the server determines a loss value for the put-on-rate prediction model based on the second sample tag and the second prediction tag, the loss value indicating a difference between the second prediction tag and the second sample tag, and adjusts the model parameters based on the loss value such that the loss value obtained by the adjusted put-on-rate prediction model decreases until the loss value is less than a loss threshold.

In the embodiment of the disclosure, considering that the third sample account information and the sample article information are both important factors that can affect the sample article being put on the article supply platform by the third sample account after recommending the sample article for the third sample account, the put-on-shelf rate prediction model is trained based on the two information, so that the put-on-shelf rate prediction model can learn the relationship between the two information and the second sample label, that is, whether the third sample account selects the relationship between the sample article to be put on the article supply platform after recommending the sample article for the third sample account, thereby having the information based on any account and the information of any article, and accurately determining the probability of putting the article on the article supply platform by the account after recommending the article by the account. The recommendation probability of the articles is determined based on the probability, and the articles are recommended based on the recommendation probability, so that the probability that the recommended articles are liked by users can be improved, and the accuracy of article recommendation is improved.

For example, the loading rate prediction model can also be trained using other information than the two information described above, e.g., information concerning other accounts of the third sample account, which is not limiting in the disclosed embodiments. Illustratively, the shelf rate prediction model can also be trained by other electronic devices besides the server, which is not limited by the disclosed embodiments.

FIG. 6 is a flow chart illustrating a method of information recommendation in accordance with an exemplary embodiment describing a method of selecting an alternative item. As shown in fig. 6, the method includes the following steps.

601. The server responds to the article recommending instruction to determine a second account, wherein the second account is the account focusing on the first account.

602. The server determines, for any one of a plurality of item categories, the number of second accounts from which items within the item category were obtained from the determined second accounts.

Illustratively, the plurality of item categories are tertiary item categories. The item categories include multiple levels of item categories, and lower item categories belong to sub-categories of upper item categories. For example, item categories include: sports related articles-outdoor sports related articles-wading outdoor sports related articles. Wherein, related articles of wading outdoor exercises are three-level articles. In the embodiment of the disclosure, the candidate articles are selected from a plurality of three-level article categories, so that the candidate articles with more diversified types can be selected.

603. The server determines a first number of target item categories from the plurality of item categories, the number of second accounts corresponding to the target item categories being the greatest.

For example, the server ranks the plurality of item categories in order of the number of corresponding second accounts from greater than lesser than the first number of item categories ranked first as the target item category. Where the first number is any number, e.g., 3, 4, etc., and embodiments of the present disclosure are not limited in this regard.

604. The server selects a plurality of candidate items from the target item categories.

In the embodiment of the disclosure, the articles obtained by the user who pays attention to the second account of the first account are utilized to determine the favorite target article categories of the user, and the alternative articles are selected from the target article categories, so that the possibility that the alternative articles recommended for the first account are favored by the user is higher, and the accuracy of article recommendation is improved. In addition, the article information of the first account is not needed to be used for being historically put on shelf in the article supply platform, so that article recommendation can be accurately performed for the first account even if articles are never put on shelf before the first account. The influence of the cold start problem caused by the lack of the historical article information of the first account when recommending articles for the first account is avoided.

Illustratively, the server randomly selects a target number of items from each target item category, the target number being any number, e.g., 5, 6, etc. The server selects a plurality of candidate items from the target item categories based on the information such as the acquired amount of the items, the amount of allocated resources corresponding to the items, the amount of third accounts corresponding to the items, and the like. Wherein the acquired amount of the article refers to the total data amount of the article acquired in all article supply platforms. The number of allocated resources corresponding to the article refers to the number of resources allocated to the account number to which the article supply platform belongs by the server after other users acquire the article from the article supply platform. In addition, for any article, if the server detects the operation of acquiring the article in any article supply platform within the second time period before the current time point, the account to which the article supply platform belongs is the third account corresponding to the article.

Illustratively, the server selects a plurality of candidate items from the target item category based on the information of the acquired amount of the items, the amount of the allocated resources corresponding to the items, the amount of the third account corresponding to the items, and the like, and the method comprises the following steps (A) - (E):

(A) For any target item category, the server sorts the items in the target item category according to the order of the obtained quantity from large to small to obtain a third item sequence, and the first selection probability of each item is determined based on the order of each item in the third item sequence.

Wherein the first selection probability of an item is greater the further forward the item is in the third sequence of items. The closer the order of the items in the third sequence of items, the less the first selection probability of the items.

For example, the server determines a ratio of an order of items in the third sequence of items to a number of items in the third sequence of items, and determines a difference of 1 to the ratio as the first selection probability.

(B) The server sorts the articles in the target article category according to the order of the number of the allocated resources from large to small to obtain a fourth article sequence, and determines a second selection probability of each article based on the order of each article in the fourth article sequence.

The earlier the order of the items in the fourth sequence of items, the greater the second selection probability of the items. The further back the order of the items in the fourth sequence of items, the less the second selection probability of the items.

Illustratively, the server determines a ratio of the order of the items in the fourth sequence of items to the number of candidate items in the fourth sequence of items, and determines a difference of 1 from the ratio as the second selected probability.

(C) The server sorts the articles in the target article category according to the sequence from the large number to the small number of the corresponding third accounts to obtain a fifth article sequence, and determines a third selection probability of each article based on the sequence of each article in the fifth article sequence.

The earlier the order of the items in the fifth sequence of items, the greater the third picking probability of the item. The further back the order of the items in the fifth sequence of items, the less the third picking probability of the item.

Illustratively, the server determines a ratio of the order of the items in the fifth sequence of items to the number of alternative items in the fifth sequence of items, and determines a difference of 1 from the ratio as the third selected probability.

(D) The server carries out weighting processing on the first selection probability, the second selection probability and the third selection probability of each article to obtain the final selection probability of each article.

The server performs weighted summation on the first selection probability, the second selection probability and the third selection probability of each article to obtain a final selection probability of each article. The first selection probability, the second selection probability, and the third selection probability are weighted arbitrarily, for example, the first selection probability is weighted 0.3, the second selection probability is weighted 0.3, and the third selection probability is weighted 0.4.

(E) The server selects an alternative item from the target item category according to the selection probabilities of the plurality of items in the target item category.

Illustratively, the server sorts the plurality of items in order of the probability of choosing from a greater to a lesser, and chooses the top-ranked, target number of items as the candidate items. Wherein the target number is any number, e.g., 5, 6, etc. Or the server selects the corresponding item with the selection probability larger than the probability threshold value as the candidate item. Wherein the probability threshold is any value, for example, 0.8. The method can ensure that the selection probability of the selected candidate object is larger, and the larger the selection probability is, the larger the distribution resource quantity of the candidate object is and the larger the probability of the candidate object being acquired by a user is, namely, the larger the probability of the candidate object being liked by the user is, so that the accuracy of object recommendation is improved.

It should be noted that, the server may also determine the selection probability of the article according to any one or two of the acquired amount of the article, the amount of allocated resources corresponding to the article, and the amount of the third account corresponding to the article, for example, determine the third selection probability determined based on the amount of the third account corresponding to the article as the final selection probability of the article, which is not limited in the embodiments of the present disclosure.

605. The server determines a stocking rate and an acquisition rate for the plurality of candidate items.

The loading rate indicates a probability of loading the candidate item on the item supply platform of the first account in a case where the candidate item is recommended for the first account, and the acquisition rate indicates a probability of detecting an acquisition operation of the candidate item loaded on the other accounts than the first account in a case where the candidate item is loaded on the item supply platform.

606. The server determines a recommendation probability for the plurality of candidate items based on the stocking rate and the acquisition rate of the plurality of candidate items.

607. The server recommends a plurality of candidate items for the first account based on the recommendation probabilities of the plurality of candidate items.

For the implementation of steps 605-607, please refer to steps 502-505, and the description thereof is omitted.

The server determines, according to the recommendation probability of the plurality of candidate articles, the plurality of candidate articles to be recommended, and then reorders the plurality of candidate articles to be recommended according to the article categories to which the plurality of candidate articles to be recommended belong, so as to obtain a candidate article sequence, wherein adjacent candidate articles do not belong to the same article category in the candidate article sequence, and therefore the subsequent terminal displays the candidate articles according to the candidate article sequence, and the display of the articles belonging to the same category can be prevented from being too concentrated. The server reorders the plurality of candidate items to be recommended according to the three-level item category to which the candidate items to be recommended belong, to obtain a candidate item sequence, wherein adjacent candidate items do not belong to the same three-level item category. The above-mentioned manner of reordering the plurality of candidate articles to be recommended according to the category of the tertiary article to which the plurality of candidate articles to be recommended belong is called diversity break-up based on the category of the tertiary article.

It should be noted that, the steps 601-604 are actually a way to select the candidate item by using the fan preference focusing on the first account, and of course, the candidate item can be selected by other ways, for example, the following ways:

first, historical item recalls. For any one of a plurality of item categories, the server determines an item acquisition amount corresponding to the item category, where the item acquisition amount is an acquisition amount of an item in the item category in an item supply platform of the first account. For example, items a and B in the item category are set up to the item supply platform of the first account, and the item a is acquired in the item supply platform by 50, and the item B is acquired in the item supply platform by 40, then the item corresponding to the item category is acquired by 90. The server then determines a first number of target item categories from the plurality of item categories, the target item categories corresponding to the most available items, and then selects an alternative item from the target item categories. The implementation manner of the server to determine the first number of target item categories from the multiple item categories and select the candidate item from the target item categories is the same as the implementation manner of the steps 603-604, which is not described herein. The above historical item recall mode determines the most preferred target item of the items on the first account from the item belonging items of the items on the first account based on the acquired amount of the items in the item supply platform of the first account, and selects the alternative item from the target item items, so that the more likely the alternative item recommended for the first account is preferred by the user, and the accuracy of item recommendation is improved.

Second, the graph recalls. Behavior data of a plurality of accounts registered in the server is represented by a bipartite graph, for example, the behavior data comprises a plurality of doublets, and each doublet (u, i) represents that the account u produces behavior on the article i. In the bipartite graph, the node corresponding to the account belonging to one bipartite group is directly connected with the node corresponding to the article. And then, based on the correlation degree of the nodes of the articles which are not directly connected with the nodes of the first account and the nodes of the first account in the bipartite graph, selecting a plurality of article nodes with the highest correlation degree, wherein the articles corresponding to the article nodes are candidate articles. Illustratively, the correlation between the node of the first account and the item node is determined by at least one of: the number of paths between the two nodes, the path length between the two nodes, i.e. the number of nodes between the two nodes through which the path passes, and the degree of egress of the nodes through which the path passes between the two nodes, wherein the number of edge stripes of the nodes is called the degree of egress of the nodes. According to the mode of the graph recall, the correlation degree between the account and the article can be accurately determined by utilizing the structural relation between the account node and the article node in the graph, so that the article with the larger correlation degree with the first account can be selected as the candidate article, and the accuracy rate of article recommendation can be improved.

Third, list recall. And selecting an alternative article from the search quantity ranking list. The search amount ranking list comprises a plurality of objects with larger search amount, and the objects are ranked in the order from large to small. To rack an item in the item supply platform, or to acquire an item from the item rack platform, a user can search for the item from a server. The server counts the times of searching each article, so as to determine the searching amount of each article, and a searching amount ranking list is obtained according to the searching amount of each article. The method for selecting the candidate item from the search ranking list by the server is similar to the method for selecting the candidate item from the target item category, and will not be described herein. Illustratively, the server selects a top-ranked target number of items in the search ranking list as candidate items. Because the search quantity of the articles can indicate the interested degree of the user on the articles, the alternative articles with higher search quantity are recommended to the user based on the search quantity ranking list, the articles recommended to the user are highly likely to be liked by the user, and the accuracy of article recommendation is improved.

It should be noted that the server may also be able to select an alternative item based on other sheets, for example, from an acquisition ranking list. The acquisition quantity ranking list comprises a plurality of objects with larger acquisition quantity, and the objects are ranked according to the order of the acquisition quantity from large to small. Or, the server selects the alternative item from the ranking list based on the number of allocated resources. The ranking list of the allocated resource quantity comprises a plurality of articles with higher allocated resource quantity, and the articles are ranked according to the order of the allocated resource quantity from high to low. The method for selecting the candidate item from the other list by the server is the same as the method for selecting the candidate item from the search ranking list, and is not described here again.

Fourth, collaborative filtering. Wherein collaborative filtering includes collaborative filtering based on items, that is, in the case of recommending items for a first account, determining items liked by a user to whom the first account belongs, and determining alternative items similar to the items. The collaborative filtering further comprises collaborative filtering based on the user, that is, in the case of recommending the items for the first account, determining the similar accounts of the first account, and determining the preferred alternative items of the user to which the similar accounts belong. The candidate articles liked by the user to which the first account belongs can be determined through collaborative filtering, so that the accuracy of article recommendation is improved.

Fifth, a double tower model recall. The double-tower model comprises a user feature extraction network and an article feature extraction network, which are respectively used for acquiring user features and article features; for any user, selecting an alternative item that is closer to the user feature based on the distance between the user feature and each item feature. For example, the server invokes a user feature extraction network in the dual-tower model, performs feature extraction based on the first account information to obtain a first account feature, invokes an item feature extraction network, and performs feature extraction based on a plurality of item information to obtain a plurality of item features. And then determining the similarity between the first account feature and each item feature, wherein the target number of items with higher similarity are used as candidate items.

It should be noted that the above ways of selecting the alternative articles can be combined in any manner, and the embodiments of the present disclosure are not limited thereto.

FIG. 7 is a flow chart illustrating a method of information recommendation, according to an exemplary embodiment describing a process of screening candidate items based on the embodiment shown in FIG. 6. As shown in fig. 7, the method includes the following steps.

701. And the server responds to the item recommendation instruction of the first account to acquire a plurality of candidate items.

The implementation of this step is referred to the embodiment shown in fig. 6 and will not be described here again.

In some embodiments, the server obtains the plurality of candidate items and then filters the plurality of candidate items. For example, the server filters out repeated candidate items in the plurality of candidate items, filters out candidate items which do not accord with the item recommendation conditions in the item recommendation request, filters out candidate items which have been recommended to the first account within a certain period of time before the current time point, and the like, so that the accuracy of item recommendation can be improved.

702. And the server acquires the number of the third accounts corresponding to the alternative items for any alternative item, wherein the server detects the acquisition operation of the alternative items in the item supply platform of the third account in a second time period before the current time point.

For any alternative article, if the server detects the acquisition operation of the alternative article in any article supply platform within a second time period before the current time point, the account corresponding to the article supply platform is the third account corresponding to the alternative article. Wherein the second time period is any time period, for example, 3 days.

703. And the server sorts the plurality of alternative articles according to the sequence from the large number to the small number of the third accounts corresponding to the plurality of alternative articles.

704. The server obtains a second number of alternate items ordered first.

Wherein the second number is any number, e.g., 20.

In the embodiment of the disclosure, the plurality of candidate articles are sequenced according to the sequence from large to small of the number of the third accounts corresponding to the plurality of candidate articles, and the second number of the candidate articles sequenced in front is obtained, so that the number of the third accounts corresponding to the obtained candidate articles is ensured to be large, that is, the number of the article supply platforms which are successfully obtained by other users after the articles are put on the shelf is large in the second time before the current time point, which means that the possibility that the candidate articles are liked by the users is larger, and therefore the accuracy of article recommendation is improved.

It should be noted that, steps 702-704 are just one method for screening the candidate articles, and the candidate articles can be screened in other manners, for example, the server ranks the plurality of candidate articles acquired in 701 according to the order of the corresponding acquisition amounts from large to small, so as to obtain a sixth article sequence, determines a fourth selection probability of each article based on the order of each candidate article in the sixth article sequence, and then acquires a second number of candidate articles with the maximum fourth selection probability. Or the server sorts the plurality of candidate articles obtained in 701 according to the sequence from large to small of the corresponding allocated resource quantity to obtain a seventh article sequence, determines a fifth selection probability of each article based on the sequence of each candidate article in the seventh article sequence, and then obtains a second number of candidate articles with the largest fifth selection probability. Or the server screens the candidate articles by combining the number, the acquired amount and the allocated resource number of the corresponding third accounts. For example, the server sorts the multiple candidate articles obtained in 701 according to the order from large to small of the number of the corresponding third accounts to obtain an eighth article sequence, determines a sixth selection probability of each article based on the order of each candidate article in the eighth article sequence, weights the fifth selection probability, the sixth selection probability and the seventh selection probability of each candidate article to obtain a final selection probability of each candidate article, and then selects a second number of candidate articles with the largest final selection probability from the multiple candidate articles.

705. The server determines a stocking rate and an acquisition rate for the plurality of candidate items.

706. The server determines the recommendation probability of the plurality of candidate articles based on the loading rate and the acquisition rate of the plurality of candidate articles, wherein the recommendation probability is in positive correlation with the loading rate and the acquisition rate.

707. The server recommends a plurality of candidate items for the first account based on the recommendation probabilities of the plurality of candidate items.

For the implementation of steps 705-707, please refer to steps 502-505 above, and further description is omitted here.

It should be noted that the above embodiments may be combined in any manner, and the present disclosure is not limited thereto.

Fig. 8 is a schematic diagram of an item recommendation process provided by an embodiment of the present disclosure. Referring to fig. 8, after receiving the item recommendation request, the server recalls the items based on a plurality of recall modes through the recall module, that is, a plurality of alternative items are determined through the recall module. The plurality of candidate items is then filtered. After filtering, the coarse ranking module ranks the plurality of candidate articles in the manner shown in fig. 7, and obtains a part of candidate articles ranked in front. The truncation in fig. 8 refers to that a part of the candidate articles ordered in the front in the ordered article sequence are reserved, and the rest of the candidate articles in the article sequence are discarded. And then, the server determines the loading rate of the candidate articles by a precision arranging module, invokes the loading rate prediction model, invokes the acquisition rate prediction model to determine the acquisition rate of the candidate articles, fuses the loading rate and the acquisition rate to obtain recommended probabilities, sorts the sorted article sequences according to the order of the recommended probabilities of the plurality of candidate articles from large to small, namely, keeps partial candidate articles sorted in the front, and discards the rest candidate articles in the article sequences. The server then reorders the remaining portion of the alternatives by the reordering module in the manner described above in step 607, i.e., based on the diversity of the three-level object class. And then returning the rearranged article sequence to the terminal sending the article recommendation request.

Fig. 9 is a block diagram illustrating an information recommendation apparatus according to an exemplary embodiment. Referring to fig. 9, the apparatus includes:

an item acquisition unit 901 configured to execute an item recommendation instruction in response to a first account to acquire a plurality of candidate items;

a first determining unit 902 configured to perform determining a loading rate and an acquisition rate of a plurality of candidate articles, the loading rate indicating a probability of loading the candidate articles on an article supply platform of the first account in a case where the candidate articles are recommended for the first account, the acquisition rate indicating a probability of detecting an acquisition operation of the loaded candidate articles by other accounts than the first account in a case where the candidate articles are loaded on the article supply platform;

a second determining unit 903 configured to determine a recommendation probability of the plurality of candidate items based on the stocking rate and the acquisition rate of the plurality of candidate items, the recommendation probability having a positive correlation with the stocking rate and the acquisition rate;

an item recommending unit 904 configured to execute recommending a plurality of candidate items for the first account based on the recommending probabilities of the plurality of candidate items.

In the embodiment of the disclosure, when the loading rate of the candidate item indicates that the candidate item is recommended for the first account, the probability of loading the candidate item on the item supply platform of the first account can be described, and when the candidate item is loaded on the item supply platform of the first account, the probability of acquiring the candidate item on the loading is detected, and the probability of the candidate item being liked by other users can be described, so that the recommendation probability of the candidate item is determined based on the loading rate and the acquisition rate of the candidate item, and the candidate item is recommended for the first account based on the recommendation probability, so that the item recommended for the first account is liked not only by the user to which the first account belongs, but also by other users, and therefore, the accuracy of item recommendation can be improved.

In some embodiments, the first determining unit 902 is configured to perform acquiring first account information, second account information, and article information of a plurality of alternative articles, the second account being an account focused on the first account; and for any candidate object, calling an acquisition rate prediction model, and determining the acquisition rate of the candidate object based on the object information, the first account information and the second account information of the candidate object.

In some embodiments, the training process of the acquisition rate prediction model includes:

acquiring first sample information, wherein the first sample information comprises first sample account information, second sample account information, sample article information and a first sample label, the second sample account is an account focusing on the first sample account, the first sample label indicates that after sample articles are put on a sample article supply platform of the first sample account, the acquisition operation of other accounts except the first sample account on the put sample articles is detected, or the acquisition operation of the sample articles is not detected;

calling an acquisition rate prediction model, and determining a first prediction label of the sample article based on the first sample account information, the second sample account information and the sample article information;

In some embodiments, obtaining the first sample tag includes:

determining a first sample label corresponding to the sample article as a first numerical value when the acquisition operation of the sample article is detected within a first time period after a loading time point of the sample article, wherein the loading time point is a time point when the sample article is loaded on an article supply platform of a first sample account;

and in the first time period after the racking time point, if the acquisition operation of the sample article is not detected, determining the first sample label corresponding to the sample article as a second numerical value.

In some embodiments, the second determining unit 903 is configured to perform multiplying the pick up rate of the candidate item by the acquisition rate for any candidate item, resulting in a recommended probability of the candidate item.

In some embodiments, the second determining unit 903 is configured to determine a first article sequence and a second article sequence based on the loading rates and the acquisition rates of the plurality of candidate articles, where the first article sequence is an article sequence obtained by sorting the plurality of candidate articles in order of the loading rates from large to small, and the second article sequence is an article sequence obtained by sorting the plurality of candidate articles in order of the acquisition rates from large to small; for any candidate item, determining a first recommendation probability of the candidate item based on the order of the candidate item in the first item sequence, and determining a second recommendation probability of the candidate item based on the order of the candidate item in the second item sequence; and weighting the first recommendation probability and the second recommendation probability to obtain the recommendation probability of the candidate object.

In some embodiments, the first determining unit 902 is configured to perform acquiring first account information and item information of a plurality of alternative items; and for any alternative article, calling an inventory rating prediction model, and determining the inventory rating of the alternative article based on the article information and the first account information of the alternative article.

acquiring second sample information, wherein the second sample information comprises third sample account information, sample article information and a second sample label, and the second sample label indicates that sample articles are put on an article supply platform of the third sample account or are not put on the article supply platform after recommending sample articles for the third sample account;

calling an upper frame rate prediction model, and determining a second prediction label of the sample article based on the third sample account information and the sample article information;

and training the loading rate prediction model based on the second sample label and the second prediction label so as to reduce the difference value between the second prediction label and the second sample label obtained by the trained loading rate prediction model.

In some embodiments, obtaining the second sample tag includes:

After recommending the sample articles for the third sample account, determining a second sample label corresponding to the sample articles as a third numerical value under the condition that the sample articles are put on the shelf on an article supply platform of the third sample account;

and after recommending the sample object for the third sample account, determining a second sample label corresponding to the sample object as a fourth value under the condition that the sample object is not put on the object supply platform of the third sample account.

In some embodiments, the item obtaining unit 901 is configured to execute a determination of a second account in response to the item recommendation instruction, the second account being an account focused on the first account; for any one of a plurality of item categories, determining, from the determined second account, the number of second accounts for which items within the item category have been acquired; determining a first number of target item categories from a plurality of item categories, wherein the number of second accounts corresponding to the target item categories is the largest; a plurality of candidate items is selected from the target item category.

In some embodiments, the article obtaining unit 901 is further configured to perform obtaining, for any alternative article, a third account number corresponding to the alternative article, where the server detects, in a second period of time before the current time point, an obtaining operation for the alternative article in the article supply platform of the third account; sequencing the plurality of alternative articles according to the sequence from the large number to the small number of the third accounts corresponding to the plurality of alternative articles; a second number of the top ranked candidate items is obtained.

FIG. 10 is a block diagram illustrating a model training apparatus, according to an example embodiment. Referring to fig. 10, the apparatus includes:

an information obtaining unit 1001 configured to perform obtaining first sample information including first sample account information, second sample account information, sample article information, and a first sample tag, the second sample account being an account focusing on the first sample account, the first sample tag indicating that after a sample article is put on a rack on an article supply platform of the first sample account, an obtaining operation of the put sample article by other accounts than the first sample account is detected, or an obtaining operation of the sample article is not detected;

a model calling unit 1002 configured to execute a call acquisition rate prediction model, determining a first prediction tag of a sample item based on the first sample account information, the second sample account information, and the sample item information;

the model training unit 1003 is configured to perform training of the acquisition rate prediction model based on the first sample label and the first prediction label such that a difference between the first prediction label and the first sample label obtained by the trained acquisition rate prediction model is reduced.

It should be noted that: the information recommending device and the model training device provided in the above embodiments only exemplify the division of the above functional modules when recommending information or training a model, and in practical application, the above functional allocation may be performed by different functional modules according to needs, that is, the internal structure of the electronic device is divided into different functional modules, so as to complete all or part of the functions described above. In addition, the information recommending apparatus and the information recommending method provided in the foregoing embodiments belong to the same concept, and specific implementation processes thereof are detailed in the method embodiment, and the model training apparatus and the model training method provided in the foregoing embodiments belong to the same concept, and specific implementation processes thereof are detailed in the method embodiment, and are not described herein.

In an exemplary embodiment, there is also provided an electronic device including one or more processors and volatile or non-volatile memory for storing one or more processor-executable instructions, the one or more processors being configured to execute the instructions to implement the information recommendation method or the model training method of the above-described embodiments.

Optionally, the electronic device is provided as a terminal. Fig. 11 shows a block diagram of a terminal 1100 provided by an exemplary embodiment of the present disclosure. The terminal 1100 may be: a smart phone, a tablet computer, an MP3 player (Moving Picture Experts Group Audio Layer III, motion picture expert compression standard audio plane 3), an MP4 (Moving Picture Experts Group Audio Layer IV, motion picture expert compression standard audio plane 4) player, a notebook computer, or a desktop computer. Terminal 1100 may also be referred to by other names of user devices, portable terminals, laptop terminals, desktop terminals, and the like.

The terminal 1100 includes: a processor 1101 and a memory 1102.

The processor 1101 may include one or more processing cores, such as a 4-core processor, an 8-core processor, and the like. The processor 1101 may be implemented in at least one hardware form of DSP (Digital Signal Processing ), FPGA (Field-Programmable Gate Array, field programmable gate array), PLA (Programmable Logic Array ). The processor 1101 may also include a main processor, which is a processor for processing data in an awake state, also called a CPU (Central Processing Unit ), and a coprocessor; a coprocessor is a low-power processor for processing data in a standby state. In some embodiments, the processor 1101 may be integrated with a GPU (Graphics Processing Unit, image processor) for taking care of rendering and rendering of content that the display screen is required to display. In some embodiments, the processor 1101 may also include an AI (Artificial Intelligence ) processor for processing computing operations related to machine learning.

Memory 1102 may include one or more computer-readable storage media, which may be non-transitory. Memory 1102 may also include high-speed random access memory, as well as non-volatile memory, such as one or more magnetic disk storage devices, flash memory storage devices. In some embodiments, a non-transitory computer readable storage medium in memory 1102 is used to store at least one program code for execution by processor 1101 to implement the information recommendation method or model training method provided by the method embodiments in the present disclosure.

In some embodiments, the terminal 1100 may further optionally include: a peripheral interface 1103 and at least one peripheral. The processor 1101, memory 1102, and peripheral interface 1103 may be connected by a bus or signal lines. The individual peripheral devices may be connected to the peripheral device interface 1103 by buses, signal lines or circuit boards. Specifically, the peripheral device includes: at least one of radio frequency circuitry 1104, a display screen 1105, a camera assembly 1106, audio circuitry 1107, a positioning assembly 1108, and a power supply 1109.

A peripheral interface 1103 may be used to connect I/O (Input/Output) related at least one peripheral device to the processor 1101 and memory 1102. In some embodiments, the processor 1101, memory 1102, and peripheral interface 1103 are integrated on the same chip or circuit board; in some other embodiments, any one or both of the processor 1101, memory 1102, and peripheral interface 1103 may be implemented on a separate chip or circuit board, which is not limited in this embodiment.

The Radio Frequency circuit 1104 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 1104 communicates with a communication network and other communication devices via electromagnetic signals. The radio frequency circuit 1104 converts an electrical signal into an electromagnetic signal for transmission, or converts a received electromagnetic signal into an electrical signal. Optionally, the radio frequency circuit 1104 includes: antenna systems, RF transceivers, one or more amplifiers, tuners, oscillators, digital signal processors, codec chipsets, subscriber identity module cards, and so forth. The radio frequency circuitry 1104 may communicate with other terminals via at least one wireless communication protocol. The wireless communication protocol includes, but is not limited to: metropolitan area networks, various generations of mobile communication networks (2G, 3G, 4G, and 5G), wireless local area networks, and/or WiFi (Wireless Fidelity ) networks. In some embodiments, the radio frequency circuitry 1104 may also include NFC (Near Field Communication, short range wireless communication) related circuitry, which is not limited by this disclosure.

The display screen 1105 is used to display a UI (User Interface). The UI may include graphics, text, icons, video, and any combination thereof. When the display 1105 is a touch display, the display 1105 also has the ability to collect touch signals at or above the surface of the display 1105. The touch signal may be input to the processor 1101 as a control signal for processing. At this time, the display screen 1105 may also be used to provide virtual buttons and/or virtual keyboards, also referred to as soft buttons and/or soft keyboards. In some embodiments, the display 1105 may be one, providing a front panel of the terminal 1100; in other embodiments, the display 1105 may be at least two, respectively disposed on different surfaces of the terminal 1100 or in a folded design; in other embodiments, the display 1105 may be a flexible display disposed on a curved surface or a folded surface of the terminal 1100. Even more, the display 1105 may be arranged in a non-rectangular irregular pattern, i.e., a shaped screen. The display 1105 may be made of LCD (Liquid Crystal Display ), OLED (Organic Light-Emitting Diode) or other materials.

The camera assembly 1106 is used to capture images or video. Optionally, the camera assembly 1106 includes a front camera and a rear camera. The front camera is arranged on the front panel of the terminal, and the rear camera is arranged on the back of the terminal. In some embodiments, the at least two rear cameras are any one of a main camera, a depth camera, a wide-angle camera and a tele camera, so as to realize that the main camera and the depth camera are fused to realize a background blurring function, and the main camera and the wide-angle camera are fused to realize a panoramic shooting and Virtual Reality (VR) shooting function or other fusion shooting functions. In some embodiments, the camera assembly 1106 may also include a flash. The flash lamp can be a single-color temperature flash lamp or a double-color temperature flash lamp. The dual-color temperature flash lamp refers to a combination of a warm light flash lamp and a cold light flash lamp, and can be used for light compensation under different color temperatures.

The audio circuit 1107 may include a microphone and a speaker. The microphone is used for collecting sound waves of users and environments, converting the sound waves into electric signals, and inputting the electric signals to the processor 1101 for processing, or inputting the electric signals to the radio frequency circuit 1104 for voice communication. For purposes of stereo acquisition or noise reduction, a plurality of microphones may be provided at different portions of the terminal 1100, respectively. The microphone may also be an array microphone or an omni-directional pickup microphone. The speaker is used to convert electrical signals from the processor 1101 or the radio frequency circuit 1104 into sound waves. The speaker may be a conventional thin film speaker or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, not only the electric signal can be converted into a sound wave audible to humans, but also the electric signal can be converted into a sound wave inaudible to humans for ranging and other purposes. In some embodiments, the audio circuit 1107 may also include a headphone jack.

The location component 1108 is used to locate the current geographic location of the terminal 1100 to enable navigation or LBS (Location Based Service, location based services). The positioning component 1108 may be a positioning component based on the United states GPS (Global Positioning System ), the Beidou system of China, the Granati system of Russia, or the Galileo system of the European Union.

A power supply 1109 is used to supply power to various components in the terminal 1100. The power source 1109 may be an alternating current, a direct current, a disposable battery, or a rechargeable battery. When the power supply 1109 includes a rechargeable battery, the rechargeable battery may support wired or wireless charging. The rechargeable battery may also be used to support fast charge technology.

In some embodiments, terminal 1100 also includes one or more sensors 1110. The one or more sensors 1110 include, but are not limited to: acceleration sensor 1111, gyroscope sensor 1112, pressure sensor 1113, fingerprint sensor 1114, optical sensor 1115, and proximity sensor 1116.

The acceleration sensor 1111 may detect the magnitudes of accelerations on three coordinate axes of a coordinate system established with the terminal 1100. For example, the acceleration sensor 1111 may be configured to detect components of gravitational acceleration in three coordinate axes. The processor 1101 may control the display screen 1105 to display the user interface in a landscape view or a portrait view according to the gravitational acceleration signal acquired by the acceleration sensor 1111. Acceleration sensor 1111 may also be used for the acquisition of motion data of a game or a user.

The gyro sensor 1112 may detect a body direction and a rotation angle of the terminal 1100, and the gyro sensor 1112 may collect a 3D motion of the user on the terminal 1100 in cooperation with the acceleration sensor 1111. The processor 1101 may implement the following functions based on the data collected by the gyro sensor 1112: motion sensing (e.g., changing UI according to a tilting operation by a user), image stabilization at shooting, game control, and inertial navigation.

The pressure sensor 1113 may be disposed at a side frame of the terminal 1100 and/or at a lower layer of the display screen 1105. When the pressure sensor 1113 is disposed at a side frame of the terminal 1100, a grip signal of the terminal 1100 by a user may be detected, and the processor 1101 performs a right-left hand recognition or a shortcut operation according to the grip signal collected by the pressure sensor 1113. When the pressure sensor 1113 is disposed at the lower layer of the display screen 1105, the processor 1101 realizes control of the operability control on the UI interface according to the pressure operation of the user on the display screen 1105. The operability controls include at least one of a button control, a scroll bar control, an icon control, and a menu control.

The fingerprint sensor 1114 is used to collect a fingerprint of the user, and the processor 1101 identifies the identity of the user based on the collected fingerprint of the fingerprint sensor 1114, or the fingerprint sensor 1114 identifies the identity of the user based on the collected fingerprint. Upon recognizing that the user's identity is a trusted identity, the user is authorized by the processor 1101 to perform relevant sensitive operations including unlocking the screen, viewing encrypted information, downloading software, paying for and changing settings, etc. Fingerprint sensor 1114 may be disposed on the front, back, or side of terminal 1100. When a physical key or vendor Logo is provided on the terminal 1100, the fingerprint sensor 1114 may be integrated with the physical key or vendor Logo.

The optical sensor 1115 is used to collect the ambient light intensity. In one embodiment, the processor 1101 may control the display brightness of the display screen 1105 based on the intensity of ambient light collected by the optical sensor 1115. Specifically, when the intensity of the ambient light is high, the display luminance of the display screen 1105 is turned up; when the ambient light intensity is low, the display luminance of the display screen 1105 is turned down. In another embodiment, the processor 1101 may also dynamically adjust the shooting parameters of the camera assembly 1106 based on the intensity of ambient light collected by the optical sensor 1115.

A proximity sensor 1116, also referred to as a distance sensor, is provided on the front panel of the terminal 1100. The proximity sensor 1116 is used to collect a distance between the user and the front surface of the terminal 1100. In one embodiment, when the proximity sensor 1116 detects that the distance between the user and the front face of the terminal 1100 gradually decreases, the processor 1101 controls the display 1105 to switch from the bright screen state to the off screen state; when the proximity sensor 1116 detects that the distance between the user and the front surface of the terminal 1100 gradually increases, the processor 1101 controls the display screen 1105 to switch from the off-screen state to the on-screen state.

Those skilled in the art will appreciate that the structure shown in fig. 11 is not limiting and that terminal 1100 may include more or fewer components than shown, or may combine certain components, or may employ a different arrangement of components.

Optionally, the electronic device is provided as a server. Fig. 12 is a schematic structural diagram of a server provided in an embodiment of the disclosure, where the server 1200 may have a relatively large difference due to different configurations or performances, and may include one or more processors (central processing units, CPU) 1201 and one or more memories 1202, where at least one program code is stored in the memories 1202, and the at least one program code is loaded and executed by the processors 1201 to implement the information recommendation method or the model training method provided in the foregoing method embodiments. Of course, the server may also have a wired or wireless network interface, a keyboard, an input/output interface, and other components for implementing the functions of the device, which are not described herein.

In an exemplary embodiment, a computer readable storage medium, such as a memory, comprising instructions executable by a processor in an electronic device to perform the information recommendation method or the model training method of the above embodiments is also provided. For example, the computer readable storage medium may be ROM, random Access Memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, etc.

In an exemplary embodiment, a computer program product is also provided, comprising a computer program, which, when executed by a processor, implements the information recommendation method or the model training method in the above-described embodiments.

Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any adaptations, uses, or adaptations of the disclosure following the general principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

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

Claims

1. An information recommendation method, comprising:

Determining a shelf availability and an acquisition availability of the plurality of candidate items, the shelf availability indicating a probability of shelf of the candidate item on an item supply platform of the first account in the case of recommending the candidate item for the first account, the acquisition availability indicating a probability of detecting an acquisition operation of the shelf of the candidate item by other accounts than the first account in the case of shelf of the candidate item on the item supply platform;

2. The information recommendation method according to claim 1, wherein determining the acquisition rate of the plurality of candidate items comprises:

3. The information recommendation method according to claim 2, wherein the training process of the acquisition rate prediction model includes:

4. The information recommendation method according to claim 3, wherein acquiring the first sample tag comprises:

5. The information recommendation method according to claim 1, wherein the determining of the recommendation probability of the plurality of candidate items based on the stocking rate and the acquisition rate of the plurality of candidate items includes:

6. The information recommendation method according to claim 1, wherein the determining of the recommendation probability of the plurality of candidate items based on the stocking rate and the acquisition rate of the plurality of candidate items includes:

7. The information recommendation method of claim 1, wherein the determining the stocking rates of the plurality of candidate items comprises:

8. The information recommendation method according to claim 7, wherein the training process of the shelf rate prediction model includes:

9. The information recommendation method of claim 8, wherein obtaining the second sample tag comprises:

10. The method according to any one of claims 1 to 9, wherein the obtaining a plurality of candidate items in response to the item recommendation command to the first account includes:

11. The information recommendation method according to any one of claims 1 to 9, wherein prior to said determining the stocking rate and the acquisition rate of said plurality of candidate items, said information recommendation method further comprises:

a second number of the candidate items ordered first is obtained.

12. An information recommendation device, characterized by comprising:

13. An electronic device, the electronic device comprising:

a processor;

a memory for storing the processor-executable instructions;

wherein the processor is configured to execute the instructions to implement the information recommendation method of any of claims 1-11.

14. A computer readable storage medium, characterized in that instructions in the computer readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the information recommendation method according to any of claims 1-11.

15. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the information recommendation method of any one of claims 1-11.