Multi-path recall algorithm-integrating information recommendation method, apparatus and device, and medium

Abstract

Disclosed in the present application are a multi-path recall algorithm-integrating information recommendation method, apparatus and device, and a medium. The method comprises: acquiring an object set to be recommended, a user set and behavior data of the user set within a historical time period; dividing the behavior data within the historical time period into behavior data in a first time period and behavior data in a second time period, the first time period being earlier than the second time period; on the basis of the behavior data of the user set in the first time period, using a plurality of modes to determine in each mode pushing users corresponding to candidate objects in said object set; on the basis of the behavior data of the user set in the second time period, integrating the pushing users corresponding to the candidate objects in different modes, so as to obtain the pushing users corresponding to the candidate objects; and acquiring pushing information of the candidate objects and sending same to the pushing users corresponding to the candidate objects. The technical solution of the embodiments of the present application can improve information pushing accuracy.

Description

Information recommendation method, device, equipment and medium integrating multi-path recall algorithm

[0001] This application claims priority to the Chinese patent application filed with the China Patent Office on December 21, 2023, with application number 202311775509.9. The entire contents of the above application are incorporated by reference into this application. Technical Field

[0002] The present application relates to the field of computer technology, and in particular to an information recommendation method, apparatus, device, and medium integrating a multi-path recall algorithm. Background Art

[0003] With the development of domestic science and technology, various technological products have been applied to all aspects of people's lives, and people can obtain all kinds of information through technological products.

[0004] If you want to send messages to users, you can use traditional information push methods. Operations personnel or product managers can push information to target user groups based on their experience.

[0005] However, the traditional method of pushing information based on experience may push invalid or erroneous information to users.

[0006] Summary of the Invention

[0007] The present application provides an information recommendation method, apparatus, device, and medium that integrate a multi-path recall algorithm. The technical solutions of the embodiments of the present application can improve the accuracy of information push.

[0008] In a first aspect, an embodiment of the present application provides a method for pushing information based on behavior data, the method comprising:

[0009] Obtain the set of recommended objects, the set of users, and the behavioral data of the set of users in the historical time period;

[0010] Dividing the behavior data of the historical time period into behavior data of a first time period and behavior data of a second time period; wherein the first time period is prior to the second time period;

[0011] Determine, based on the behavioral data of the user set in the first time period, in multiple ways, the push users corresponding to the candidate objects in the set of objects to be recommended under each way;

[0012] Based on the behavioral data of the user set in the second time period, the push users corresponding to each candidate object in different ways are merged to obtain the push users corresponding to each candidate object;

[0013] Get the push information of the candidate object and send it to the push user corresponding to the candidate object.

[0014] In a second aspect, an embodiment of the present application further provides an information recommendation device integrating a multi-path recall algorithm, the device comprising:

[0015] A data acquisition module is configured to acquire a set of objects to be recommended, a set of users, and behavioral data of the set of users in a historical time period;

[0016] a time division module configured to divide the behavior data of a historical time period into behavior data of a first time period and behavior data of a second time period; wherein the first time period precedes the second time period;

[0017] a push user determination module configured to determine, based on the behavioral data of the user set in the first time period, a push user corresponding to the candidate object in the set of recommended objects in each of the methods using multiple methods;

[0018] A user fusion module is configured to fuse the push users corresponding to each candidate object in different ways based on the behavioral data of the user set in the second time period to obtain the push users corresponding to each candidate object;

[0019] The information sending module is configured to obtain push information of the candidate object and send it to the push user corresponding to the candidate object.

[0020] In a third aspect, an embodiment of the present application further provides an information recommendation device integrating a multi-path recall algorithm, the device comprising:

[0021] at least one processor; and

[0022] a memory communicatively connected to at least one processor; wherein,

[0023] The memory stores a computer program that can be executed by at least one processor. The computer program is executed by at least one processor so that the at least one processor can execute the information recommendation method integrating the multi-way recall algorithm of any embodiment of the present application.

[0024] According to another aspect of the present application, a computer-readable medium is provided, wherein the computer storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the information recommendation method of the fusion multi-path recall algorithm of any embodiment of the present application when executed.

[0025] The technical solution of the embodiment of the present application obtains push users corresponding to multiple methods by using multiple methods for the behavioral data of the first time period, and obtains push users in a multi-dimensional manner, which is conducive to improving the comprehensiveness of obtaining push users. The push users obtained in the first time period are integrated with the behavioral data of the second time period, and the push users are further refined to obtain the final push users, which is conducive to reducing the amount of data and improving the efficiency of information push.

[0026] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0028] FIG1 is a flow chart of an information recommendation method integrating a multi-path recall algorithm according to a first embodiment of the present application;

[0029] FIG2 is a flow chart of an information recommendation method integrating a multi-path recall algorithm according to the second embodiment of the present application;

[0030] FIG3 is a flow chart of an information recommendation method integrating a multi-path recall algorithm according to the second embodiment of the present application;

[0031] FIG4 is a structural diagram of an information recommendation device integrating a multi-path recall algorithm according to an embodiment of the present application;

[0032] FIG5 is a schematic diagram of the structure of an information recommendation device that implements the multi-path recall algorithm provided in an embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0034] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] In the technical solution of the embodiment of the present application, the acquisition, storage and application of the behavioral data of the user set involved in the historical time period are triggered by the user and obtained with the user's authorization, which complies with the provisions of relevant laws and regulations and does not violate public order and good morals.

[0036] Example 1

[0037] Figure 1 is a flow chart of an information recommendation method that incorporates a multi-path recall algorithm, as provided in Example 1 of the present application. This embodiment of the present application is applicable to information push scenarios, and the method can be performed by an information recommendation device based on the multi-path recall algorithm, which can be implemented in hardware and / or software.

[0038] Referring to FIG1 , the information recommendation method integrating the multi-path recall algorithm includes:

[0039] S101: Obtain a set of objects to be recommended, a set of users, and behavioral data of the user set within a historical time period.

[0040] Among them, the set of objects to be recommended can be a set of objects pushed to the user, which can be represented by I. The objects to be recommended can be represented by ITEMID, ITEMID∈I. Each object to be recommended has at least one attribute information, and the attribute information of the objects to be recommended in different business scenarios is different. For example, if the business scenario is news, the attribute information of the objects to be recommended includes but is not limited to: news title, time of news event, location of news event, and people involved in news event, etc. If the business scenario is a commodity, the attribute information of the objects to be recommended includes but is not limited to: unique identity document (ID), commodity name, commodity label, commodity description, commodity brand and commodity price, etc.

[0041] The user set can be the set of users to be screened for information push, represented by U. Each user has at least one attribute, including a unique user ID (userid), age, region, and gender. The historical time period can be a preset past period, perhaps a year or several months, depending on the business scenario. Behavioral data can be information about a user and an object performing a specific action at a specific point in time. Behavioral data includes attributes such as the user's unique ID, the item's unique ID, the type of action, and the time of the action. Behavioral data can be represented by T. Behavioral types vary depending on the business scenario, such as clicks, likes, or favorites. User interest in pushed information is often determined by whether the user has behavioral data related to the pushed information. It should be noted that user information and behavioral data, including user-related data, are obtained only with user authorization.

[0042] Specifically, acquisition methods include, but are not limited to, crawler acquisition, user input, or query data logs. By acquiring the set of recommended objects, the set of users, and the behavioral data of the user set within a historical time period, it is convenient to determine the objects to be recommended and the number of users to whom information is pushed, represented as ITEMID and N, respectively.

[0043] In one example, the preset historical time period is from 00:00:00 on January 1, XX, to 24:00:00 on August 31, XX. A set of objects to be recommended, a set of users, and behavioral data of the user sets within the historical time period are obtained through user input; the set of objects to be recommended includes short-sleeved shirts, pants, and shoes; the set of users includes user 1, user 2, and user 3; the behavioral data of the user sets within the period from 00:00:00 on January 1, XX, to 24:00:00 on August 31, XX, includes: user 1 favorited pants at 01:01:01 on January 1, XX, user 2 favorited short-sleeved shirts at 05:05:05 on May 5, XX, and user 3 favorited shoes at 08:08:08 on August 31, XX.

[0044] In a specific example, the preset historical time period is from 00:00:00 on February 1, XX to 24:00:00 on July 31, XX. A set of recommended objects, a set of users, and behavioral data of the user sets within the historical time period are obtained through user input; the set of recommended objects includes current affairs news and entertainment news; the user set includes user 1 and user 2; and the behavioral data of the user set between 00:00:00 on February 1, XX and 24:00:00 on July 31, XX includes: user 1 favorited current affairs news at 02:02:02 on February 2, XX, and user 2 favorited entertainment news at 03:03:03 on March 3, XX.

[0045] S102: Divide the behavior data of the historical time period into behavior data of a first time period and behavior data of a second time period; wherein the first time period is prior to the second time period.

[0046] The historical time period may be a time period before the current time, and the historical time period includes a first time period and a second time period. The length of the first time period may be the same as or different from the length of the second time period, and this embodiment is not limited thereto. The sum of the lengths of the first time period and the second time period equals the preset historical time period.

[0047] Specifically, clean the data in I, U, and T, including checking for null values, determining the type and range of field values, and deleting any non-compliant items. Split T into data at a certain point in time. For example, the data D for the past period (e.g., half a year) and the data D for the most recent period (e.g., one month) are used. A The length of D is the number of behavioral data, denoted as |D|.

[0048] In one example, the preset historical time period is from 00:00:00 on January 1, XX to 24:00:00 on August 31, XX. The historical time period is divided into a first time period and a second time period. The first time period is from 00:00:00 on January 1, XX to 24:00:00 on June 30, XX, and the second time period is from 00:00:00 on July 1, XX to 24:00:00 on August 31, XX. The behavior data from 00:00:00 on January 1, XX to 24:00:00 on June 30, XX is used as the behavior data for the first time period; and the behavior data from 00:00:00 on July 1, XX to 24:00:00 on August 31, XX is used as the behavior data for the second time period.

[0049] S103: Based on the behavioral data of the user set in the first time period, multiple methods are used to determine the push users corresponding to the candidate objects in the set of objects to be recommended under each method.

[0050] The candidate objects may be objects to be pushed, and the set of objects to be recommended is composed of the candidate objects. The pushing user may be the user to whom the candidate objects are pushed.

[0051] Specifically, the behavioral data of each user in the user set in the first time period is obtained, and the push users of each candidate object in the set of recommended objects obtained by multiple methods are obtained, and a candidate user list is generated. The push users corresponding to the candidate objects obtained by each method may be the same or different. For the ITEMID of the object to be pushed and the number of users N for the required information recommendation, multiple methods are used to determine the push users corresponding to the candidate objects in the set of objects to be recommended under each method.

[0052] In one example, the user set includes User 1 and User 2; the candidate objects in the set of recommended objects include T-shirts and pants. Based on the user set's behavioral data during the first time period, the user for whom the recommended T-shirt is obtained through Method 1 is User 3; the user for whom the recommended T-shirt is obtained through Method 2 is User 4; the user for whom the recommended T-shirt is obtained through Method 1 is User 5; and the user for whom the recommended T-shirt is obtained through Method 2 is User 6.

[0053] For example, in a news scenario, the user set includes user A and user B, and the candidate objects in the set of recommended objects include current affairs news and entertainment news. Based on the user set's behavioral data during the first time period, the user to whom current affairs news is pushed, obtained through method 1, is user C; the user to whom current affairs news is pushed, obtained through method 2, is user D; the user to whom entertainment news is pushed, obtained through method 1, is user E; and the user to whom entertainment news is pushed, obtained through method 2, is user F.

[0054] S104: Based on the behavioral data of the user set in the second time period, the push users corresponding to each candidate object in different modes are integrated to obtain the push users corresponding to each candidate object.

[0055] The fusion is used to clean the pushed users, for example, it can be used to remove duplicates of the same pushed user.

[0056] Specifically, set the pre-fetch quantity K. If each of the K user-to-be-pushed pairs generated by each method <userid i ,ITEMID> in the second time period D AIf there is corresponding behavior data, such as collection, the count is increased by 1, otherwise the count is not increased by 1. The final count is P, then P≤K, and the push users corresponding to each candidate object are obtained.

[0057] In an example, the pre-fetch quantity is set to 1, and ITEMID represents short sleeves. Based on the behavioral data of the user set in the first time period, the push user 1 corresponding to the short sleeves is obtained through method 1. The generated user-to-be-pushed object pair is<userid1,ITEMID> ; Get the push user 2 corresponding to the short sleeve through method 2, and the generated user-to-be-pushed object pair is<userid2,ITEMID> The behavior data of the second time period includes user 1 collecting short-sleeved shirts at 08:08:08 on August XX.<userid1,ITEMID> The count is increased by 1.

[0058] S105: Obtain push information of the candidate object and send it to the push user corresponding to the candidate object.

[0059] The push information may be information of a candidate object to be pushed, such as attribute information or description information of the candidate object.

[0060] Specifically, after obtaining the push user, the attribute information of the candidate object is sent to the push user according to a preset method. The preset method includes but is not limited to: sending at a certain time point, sending at a specified frequency, or sending according to the number of push users.

[0061] In an example, the preset number of push users is 100. After the push users are acquired, the attribute information of the pushed candidate objects is sent to the push users according to the preset number of push users.

[0062] The technical solution of the embodiment of the present application obtains push users corresponding to multiple methods by using multiple methods for the behavioral data of the first time period, and obtains push users in a multi-dimensional manner, which is conducive to improving the comprehensiveness of obtaining push users. The push users obtained in the first time period are integrated with the behavioral data of the second time period, and the push users are further refined to obtain the final push users, which is conducive to reducing the amount of data and improving the efficiency of information push.

[0063] Optional, alternative objects include: news, articles, web pages, videos, live broadcast rooms or products.

[0064] Specifically, news, articles, web pages, videos, live broadcast rooms or product images, videos or documents can be sent to push users.

[0065] In one example, the origin, color, and length information of a short-sleeved shirt are sent to the push user; the time of an entertainment news event, the people involved in the news event, and the location of the news event are sent to the push user; the title, content, or author information of a novel are sent to the push user; the webpage address and title information of a webpage are sent to the push user; the video title, video content introduction, and author information of a video are sent to the push user; the live broadcast room title and live broadcast room link information are sent to the push user; and the material and origin information of a toy car are sent to the push user.

[0066] By sending news, articles, web pages, videos, live broadcast rooms or product information to push users, it is helpful to push information to users who are interested in it, thereby improving the scope and efficiency of information push.

[0067] Example 2

[0068] Figure 2 is a flow chart of the information recommendation method of the fusion multi-path recall algorithm provided in Example 2 of the present application. Based on the above embodiments, the present embodiment optimizes and improves the information recommendation operation of the fusion multi-path recall algorithm.

[0069] Furthermore, "based on the behavioral data of the user set in the first time period, use multiple methods to determine the push users corresponding to the alternative objects in the set of objects to be recommended under each method" is refined into "based on the behavioral data of the alternative users in the user set in the first time period, determine the alternative users who have interactive behaviors with the alternative objects in the set of objects to be recommended, and the number of behaviors of each alternative user with respect to each alternative object; based on the behavioral data of the alternative users in the user set in the first time period, and the number of behaviors of each alternative user with respect to each alternative object, determine the interaction weight of each alternative user with respect to each alternative object; based on the alternative users who have interactive behaviors with each alternative object, and the interaction weight of each alternative user with respect to each alternative object, determine the push users corresponding to the alternative objects in the set of objects to be recommended under the behavioral data method", so as to improve the operation of information push.

[0070] It should be noted that for parts not described in detail in the embodiments of this application, reference can be made to the descriptions in other embodiments.

[0071] Referring to FIG2 , the information recommendation method integrating the multi-path recall algorithm includes:

[0072] S201: Obtain a set of objects to be recommended, a set of users, and behavioral data of the user set within a historical time period.

[0073] S202: Divide the behavior data of the historical time period into behavior data of a first time period and behavior data of a second time period; wherein the first time period is prior to the second time period.

[0074] S203: Determine candidate users who have interactive behaviors with candidate objects in the set of recommended objects, and the number of behaviors of each candidate user with respect to each candidate object, based on the behavioral data of the candidate users in the user set in the first time period.

[0075] Specifically, the candidate user is a user who has interactive operations on the candidate object. The number of behaviors is the number of interactive behaviors of the candidate user on the candidate object. According to the behavioral data of the candidate users in the user set in the first time period, all the data of D are traversed. i , count any corresponding candidate userid p The number of behaviors count p .

[0076] In one example, for a candidate item with the ITEMID short-sleeved shirt, user set D includes userid1 and userid2. Based on the behavior data of the candidate users in the user set during the first time period, all data in D are traversed. The value of count1 for userid1's behavior is 2, and the value of count2 for userid2's behavior is 1.

[0077] S204: Determine the interaction weight of each candidate user for each candidate object based on the behavior data of the candidate users in the user set in the first time period and the number of behaviors of each candidate user for each candidate object.

[0078] The interaction weight of the candidate object is used to determine the probability that the candidate user will interact with the candidate object.

[0079] Specifically, based on the behavior data T of the candidate users in the user set U in the first time period D, and the number of behaviors of each candidate user on each candidate object, all data in D are traversed, and for each candidate object ITEMID, the corresponding userid of any candidate user is counted. p The number of behaviors count p , and then the interaction weight of each candidate user for each candidate object is obtained. The calculation formula of interaction weight is as follows:

[0080] The user list with behavior data corresponding to ITEMID is as follows:

[0081] D ITEEID =[userid1:weight1,userid2:weight2,…,userid r :weight r ]

[0082] In one example, for a candidate item with an ITEM ID of short-sleeved shirt, there is one attribute, itemid1, which is color. A user set D includes userid1 and userid2. Based on the behavior data of the candidate users in the user set during the first time period, all data in D are traversed. The number of times userid1's behavior occurred (count1) is 2, while the number of times userid2's behavior occurred (count2) is 1.

[0083] For short sleeves, the value of userid1 and the number of times the behavior occurs count1 is 1, and the value of the number of times the behavior occurs count2 of userid2 is 2, and then the interaction weight is obtained. The interaction weight of userid1 is The interaction weight of userid2 is The list of candidate users with behavioral data corresponding to short sleeves is as follows:

[0084] S205 : Determine, based on the candidate users who have interactive behaviors with each candidate object and the interaction weight of each candidate user with respect to each candidate object, the push users corresponding to the candidate objects in the set of objects to be recommended in a behavioral data manner.

[0085] Specifically, based on the behavior data T of each candidate user in the user set U in the first time period D, and the number of behaviors of each candidate user on each candidate object, all data in D are traversed, and for each candidate object ITEMID, the corresponding userid of any candidate user is counted. p Behavior count p , and then the interaction weight of each candidate object is obtained. The candidate user list corresponding to ITEMID is as follows:

[0086] D ITEMID =[userid1:weight1,userid2:weight2,…,userid t :weight t ]

[0087] In one example, for the candidate object ITEMID of short sleeves, the user set U includes: userid1 and userid2. Based on the behavior data of each candidate user in the user set in the first time period, all data in D are traversed, and the value of the number of userid1's behavior count1 is 2, and the value of the number of userid2's behavior count2 is 1, and then the interaction weight of each candidate object is obtained. The candidate user list corresponding to ITEMID is Select the push user corresponding to the candidate object based on the candidate user list.

[0088] S206 : Based on the behavioral data of the user set in the second time period, the push users corresponding to the candidate objects in different modes are integrated to obtain the push users corresponding to the candidate objects.

[0089] S207: Obtain push information of the candidate object and send it to the push user corresponding to the candidate object.

[0090] The embodiment of the present application improves the accuracy of information push by calculating the interaction weight of the candidate users for the candidate objects, judging the probability of the candidate users generating behavior towards the candidate objects and the degree of interest.

[0091] Optionally, based on the alternative users with whom each alternative object has interactive behaviors, and the interaction weights of each alternative user with respect to each alternative object, the push users corresponding to the alternative objects in the set of objects to be recommended are determined in a behavioral data manner, including: determining the characteristic information of each alternative user based on the alternative users with whom each alternative object has interactive behaviors, and the interaction weights of each alternative user with respect to each alternative object; determining the similarity value between each alternative user and other alternative users based on the characteristic information of each alternative user; determining the push probability of each alternative object to each alternative user based on the alternative users with whom each alternative object has interactive behaviors, the interaction weights of each alternative user with respect to each alternative object, and the similarity value between each alternative user and other alternative users; determining the push users corresponding to each alternative object based on the push probability of each alternative object to each alternative user.

[0092] The feature information is used to describe the user and can be represented by a string consisting of user information in the candidate user list. The similarity value can be the degree of similarity between each candidate user and other candidate users. The push probability can be the probability that each candidate object will push to each candidate user.

[0093] Specifically, for any ITEMID, its candidate user list is further processed into STR ITEMID =“userid1userid2…userid r ”, that is, it is converted into a string form, and the value obtained by splitting it with spaces is determined as the feature information of each candidate user. The feature information of the candidate users of all ITEMIDs in I is input into the word to vector (word2vec) model in the Generate Similarity (Gensim) library for processing. Then, for any ITEMID∈U, the trained word2vec model is called to obtain a list of v similar users, which is recorded as SIM ITEMID =[userid1:sim1,userid2:sim2,…,userid v :simv ], where v is a constant that can be set in advance, and sim v For userid and userid v Furthermore, for any ITEMID∈I, we can get the push probability of ITEMID to userid, which is recorded as bias(userid,ITEMID).

[0094] Furthermore, the normalized push probability is obtained, which is recorded as norm_bias(userid, ITEMID).

[0095] Sort by norm_bias from high to low, and then get a list of push users who have a push probability for ITEMID.

[0096] B itemid =[userid1:weight1,userid2:weight2,…,userid x :weight x ]

[0097] where weight x Indicates userid x Normalized push probability for ITEMID, i.e. norm_bias(userid x ,ITEMID).

[0098] In one example, the value of the guaranteed candidate user list is D bd =[userid1:2,userid2:1]. The candidate user list corresponding to ITEMID is Further processed into STR ITEMID = "userid1userid2", which is determined as the feature information of the candidate users. The feature information of all candidate users of ITEMID in I is input into the word2vec model in the Gensim library for processing. Then, for any ITEMID∈U, the trained word2vec model is called to obtain two lists of similar users, recorded as SIM ITEMID =[userid1:0.8,userid2:0.2], and then, for any ITEMID∈I, we can get the push probability of userid for ITEMID, recorded as bias(userid,ITEMID), and further, we can get the normalized push probability, recorded as norm_bias(userid,ITEMID), and sort them from high to low according to norm_bias, and then get the push user list B with push probability for ITEMID.ITEMID .

[0099] B ITEMID =[userid1:weight1,userid2:weight2]

[0100] By determining the similarity value between each candidate user and other candidate users based on the feature information of each candidate user, it is helpful to determine the push probability of each candidate object to each candidate user, obtain push users with higher push probability, and improve the accuracy of information push.

[0101] Optionally, based on the behavioral data of the user set in the first time period, multiple methods are used to determine the push users corresponding to the alternative objects in the set of recommended objects under each method, including: obtaining the number of behaviors of each alternative user based on the behavioral data of the alternative users in the user set in the first time period; and screening the push users corresponding to each alternative object among the alternative users based on the number of behaviors of each alternative user.

[0102] The number of behaviors may be the number of behavioral data items of the user for the same candidate object.

[0103] Specifically, based on the behavior data of the candidate users in the user set in the first time period, traverse all the data of D, and for ITEMID, count the corresponding userid of any candidate user p Behavior count p , sort the candidate users according to the number of times their behaviors occur from high to low, and then generate a guaranteed candidate user list D bd , according to the guaranteed candidate user list D bd The number of behaviors of each candidate user in the guaranteed candidate user list D bd Filter the push users corresponding to each candidate object from the candidate users.

[0104] D bd =[userid1:count1,userid2:count2,…userid S :count s ]

[0105] In one example, for the candidate object ITEMID of short sleeves, the user set D includes: userid1 and userid2. Based on the behavior data of the candidate users in the user set in the first time period, all data in D are traversed, and the value of count1 of userid1's behavior is 2, and the value of count2 of userid2's behavior is 1. The users are sorted from high to low according to the number of behavior, and the value of the guaranteed candidate user list is D. bd=[userid1:2,userid2:1].

[0106] By counting the number of actions of candidate users, it is easy to determine the degree of interest of different candidate users in the same candidate object.

[0107] Optionally, based on the behavioral data of the user set in the first time period, multiple methods are used to determine the push users corresponding to the alternative objects in the set of objects to be recommended under each method, including: determining the alternative users who have interactive behaviors for each attribute information and the number of behaviors of each alternative user for each attribute information based on the behavioral data of the alternative users in the user set in the first time period and at least one attribute information of the alternative objects in the set of objects to be recommended; determining the interaction weight of each alternative user for each attribute information based on the behavioral data of the alternative users in the user set in the first time period and the number of behaviors of each alternative user for each attribute information; determining the push users corresponding to the alternative objects in the set of objects to be recommended under the object attribute information method based on at least one attribute information of each alternative object and the interaction weight of each alternative user for each attribute information.

[0108] The interaction weight of the attribute information may be the probability of the user generating an action on the attribute.

[0109] Specifically, based on the behavior data T of the candidate users in the user set U in the first time period D, and at least one attribute information of the candidate objects in the set of recommended objects, combined with different attribute information of the candidate objects, such as tags, categories, and regions, for any candidate object ITEMID, there are m attribute information in total, and its i-th attribute information is itemid i , determine the candidate users who have interactive behaviors with each attribute information. For example, the candidate object ITEMID is short-sleeved, which has two attribute information. The first attribute information itemid1 is color, and the second attribute information itemid2 is the region.

[0110] Traverse all the data of D, for itemid i , count any corresponding candidate userid p And the number of behaviors of each candidate user for each attribute information count p , and then the interaction weight of each candidate user for each attribute information is obtained. The calculation formula of the interaction weight is as follows:

[0111] Itemid i The corresponding candidate user list is

[0112] Based on m attribute information, we can obtain a list of m candidate users. If an attribute information has multiple values, we can perform the calculation for each attribute information. For example, if the attribute information of a recommended object is "attribute information 1; attribute information 2; attribute information 3", we can split it into three attribute information according to the ";" and then accumulate the counts for attribute information 1, attribute information 2, and attribute information 3 respectively.

[0113] In one example, for the candidate object ITEMID of short-sleeved shirt, there is a total of 1 attribute information, and the attribute information itemid1 is color. The user set D includes: userid1 and userid2. Based on the behavior data of the candidate users in the user set in the first time period, all data in D are traversed, and the value of count1 of userid1's behavior is 2, and the value of count2 of userid2's behavior is 1. The users are sorted from high to low according to the number of behavior, and the value of the guaranteed candidate user list can be generated. bd =[userid1:2,userid2:1].

[0114] For short sleeves, the value of userid1 and the number of behaviors count1 is 1, and the value of userid2's number of behaviors count2 is 2, and then the interaction weight is obtained. The candidate user list corresponding to itemid1 is In this way, based on one attribute, a list of candidate users can be obtained.

[0115] By calculating the interaction weight of each user for different attribute information of each candidate object, it is helpful to obtain the probability of the user's behavior on the attribute and improve the accuracy of information push.

[0116] Optionally, based on the behavioral data of the user set in the second time period, the push users corresponding to each alternative object in different modes are merged to obtain the push users corresponding to each alternative object, including: for each alternative object, according to the extraction ratio corresponding to each mode, extracting target users from the push users corresponding to each mode to form an object user group of alternative objects and target users; based on the behavioral data of the alternative users in the user set in the second time period, statistics are performed on each object user group; based on the statistical results of each object user group, the extraction ratio corresponding to each mode is adjusted; for each alternative object, according to the extraction ratio corresponding to each mode, push users are extracted from the push users corresponding to the alternative object in each mode, and merged to obtain the push users corresponding to the alternative object.

[0117] The extraction ratio may be the ratio of the number of push users selected from the push users obtained through various methods to the total number of push users. The statistical result may be a detection result of whether the user has the same behavior in the second time period.

[0118] Specifically, the behavioral data of each candidate user in the user set in the first time period is obtained, and the push users of each candidate object in the set of recommended objects obtained by multiple methods are obtained to generate a candidate user list. ITEMID Indicates. The push users corresponding to each candidate object obtained by each method may be the same or different. For the ITEMID of the object to be pushed and the number of users N for the required information recommendation, multiple methods are used to extract target users from the push users corresponding to each method according to the extraction ratio corresponding to each method, and determine the object user group that forms the candidate object and the target user under each method. Set the pre-fetch quantity K and pre-fetch in three ways. The first method can be from Take K target users from the list to form a candidate user group of candidate objects and target users; the second method is to select the corresponding candidate users from the list of candidate users according to the m attribute information of ITEMID. Take K target users from B to form a target user group of candidate objects and target users; the third method can be to select K target users from B itemid There are m+2 ways of pre-fetching target users. For each way, K target user groups of candidate objects and target users can be formed, namely:<userid1,ITEMID> ,<userid2,ITEMID> ,…, <userid K ,ITEMID>.

[0119] if <userid i ,ITEMID> in D A If there is behavioral data, such as collection, the count is increased by 1, otherwise it is not counted. The final count is P, then P≤K. Based on the statistical results, the prefetch hit rate Hit@K of each prefetch method can be calculated. The prefetch hit rate calculation formula is as follows.

[0120] Set the number of candidate results M = α * N, where α is the incremental coefficient, which may overlap and is a preset constant value greater than 1. N is the preset number of push users. Based on the m+2-way pre-fetching user method, the number of candidate results obtained in the j-th way is Q j It is the product of M and its prefetch hit rate, that is, Q j =M*Hit@K j , the corresponding user list is:

[0121] If we merge the user lists obtained from m+2 channels, the extraction ratio of userid to itemid is:

[0122] The final user list after merging is

[0123] R=[userid1:weight1,userid2:weight2,…,userid z :weight z ]

[0124] Where z is the number of users after merging. If z ≥ N, then take the first N users in R as the push user set S, otherwise go to the quantity maintenance process and start from D bd Take users from front to back, if the user is not in S, add it to S, if it exists, delete it, until the number of users in S is N or D bd Terminate when all is taken.

[0125] In one example, as shown in Figure 3, the object to be pushed is determined, and data acquisition and preprocessing are performed. The preset value of K is 3. For the candidate object ITEMID is short-sleeved, there is a total of 1 attribute information, and the attribute information itemid1 is color. The user set D includes: userid1 and userid2. According to the behavior data of the candidate users in the user set in the first time period, all data in D are traversed, and the value of the number of userid1's behavior count1 is 2, and the value of the number of userid2's behavior count2 is 1. Sort by the number of user behaviors from high to low, and generate a guaranteed candidate user list based on the behavior data. The value is D bd =[userid1:2,userid2:1]. Generate a minimum candidate user list based on behavioral data to obtain push users.<userid1,ITEMID> In D A If there is behavioral data, the value of count1 is 3, and the final count P of count1 of userid1 is 3, then P≤K. Then the prefetch hit rate Hit@K of each prefetch method can be calculated. Set the number of candidate results M = α*N, where α is a constant value greater than 1 with some overlap in the incremental coefficient. N is the preset number of push users. Based on the aforementioned m+2 prefetch user method, the number of candidate results obtained in the jth path is Q j It is the product of M and its prefetch hit rate, that is, Q j =M*Hit@K j , the corresponding user list is:

[0126] R j =[userid1:weight1,userid2:weight2]

[0127] Merge the user lists obtained from m+2 routes, and the extraction ratio of userid to itemid is:

[0128] The final user list after merging is

[0129] R = [userid1:weight1, userid2:weight2], where z is the number of users after merging. If z ≥ N, the process will be quantity-preserving.

[0130] Through the statistical results of each object user group, the extraction ratio corresponding to each method is adjusted. According to the extraction ratio corresponding to each method, push users are extracted from the push users corresponding to the alternative objects under each method. This is conducive to adjusting the number of extracted users and the extraction method, making it easier to obtain a more accurate set of push users and improve the accuracy of information push.

[0131] Example 3

[0132] Figure 4 is a schematic diagram of the structure of an information recommendation device integrating a multi-path recall algorithm provided in Example 4 of the present application. This embodiment of the present application is applicable to information push, and the device can execute an information recommendation method integrating a multi-path recall algorithm, and the device can be implemented in the form of hardware and / or software.

[0133] Referring to FIG4 , the information recommendation device integrating the multi-path recall algorithm includes: a data acquisition module 401 , a time division module 402 , a push user determination module 403 , a user integration module 404 and an information sending module 405 , wherein:

[0134] The data acquisition module 401 is used to acquire the set of recommended objects, the set of users, and the behavioral data of the set of users in a historical time period;

[0135] A time division module 402 is configured to divide the behavior data of a historical time period into behavior data of a first time period and behavior data of a second time period; wherein the first time period precedes the second time period;

[0136] The push user determination module 403 is configured to determine, based on the behavior data of the user set in the first time period, the push user corresponding to the candidate object in the set of recommended objects in each of the multiple ways;

[0137] The user fusion module 404 is configured to fuse the push users corresponding to each candidate object in different modes based on the behavior data of the user set in the second time period to obtain the push users corresponding to each candidate object;

[0138] The information sending module 405 is used to obtain push information of the candidate object and send it to the push user corresponding to the candidate object.

[0139] The technical solution of the embodiment of the present application obtains push users corresponding to multiple methods by using multiple methods for the behavioral data of the first time period, and obtains push users in a multi-dimensional manner, which is conducive to improving the comprehensiveness of obtaining push users. The push users obtained in the first time period are integrated with the behavioral data of the second time period, and the push users are further refined to obtain the final push users, which is conducive to reducing the amount of data and improving the efficiency of information push.

[0140] Optionally, the push user determination module 403 includes:

[0141] A user determination unit is configured to determine candidate users who have interactive behaviors with candidate objects in the set of recommended objects, and the number of behaviors of each candidate user with respect to each candidate object, based on the behavioral data of the candidate users in the user set in the first time period;

[0142] a weight determination unit, configured to determine an interaction weight of each candidate user with respect to each candidate object based on the behavior data of the candidate users in the user set in the first time period and the number of behaviors of each candidate user with respect to each candidate object;

[0143] The user recommendation unit is used to determine the push users corresponding to the candidate objects in the set of recommended objects in the behavioral data mode according to the candidate users who have interactive behaviors with each candidate object and the interaction weights of each candidate user for each candidate object.

[0144] Optional user recommendation unit, specifically used for:

[0145] Determine characteristic information of each candidate user based on candidate users who have interactive behaviors with each candidate object and the interaction weight of each candidate user with respect to each candidate object;

[0146] Determine the similarity between each candidate user and other candidate users based on the feature information of each candidate user;

[0147] Determine the push probability of each candidate object to each candidate user based on the candidate users who have interacted with each candidate object, the interaction weight of each candidate user for each candidate object, and the similarity value between each candidate user and other candidate users;

[0148] According to the push probability of each candidate object to each candidate user, the push user corresponding to each candidate object is determined.

[0149] Optionally, the push user determination module 403 is specifically configured to:

[0150] Determining, based on the behavioral data of candidate users in the user set during the first time period and at least one attribute information of candidate objects in the set of recommended objects, candidate users with interactive behaviors for each attribute information and the number of behaviors of each candidate user with respect to each attribute information;

[0151] Determining the interaction weight of each candidate user for each attribute information based on the behavior data of the candidate users in the user set in the first time period and the number of behaviors of each candidate user for each attribute information;

[0152] According to at least one attribute information of each candidate object and the interaction weight of each candidate user for each attribute information, the push user corresponding to the candidate object in the set of objects to be recommended is determined in the object attribute information mode.

[0153] Optionally, the push user determination module 403 is specifically configured to:

[0154] Obtaining the number of actions of each candidate user based on the behavior data of the candidate users in the user set in the first time period;

[0155] According to the number of actions of each candidate user, the push users corresponding to each candidate object are selected from each candidate user.

[0156] Optionally, the user fusion module 404 is specifically configured to:

[0157] For each candidate object, according to the extraction ratio corresponding to each method, the target user is extracted from the push users corresponding to each method to form a target user group of candidate objects and target users;

[0158] Performing statistics on each target user group based on the behavioral data of the candidate users in the user set during the second time period;

[0159] Adjust the extraction ratio corresponding to each method based on the statistical results of each target user group;

[0160] For each candidate object, according to the extraction ratio corresponding to each method, push users are extracted from the push users corresponding to the candidate object in each method, and are merged to obtain the push users corresponding to the candidate object.

[0161] Optional, alternative objects include: news, articles, web pages, videos, live broadcast rooms or products.

[0162] The information push device based on behavioral data provided in the embodiment of the present application can execute the information recommendation method of the fusion multi-channel recall algorithm provided in any embodiment of the present application, and has the corresponding functional modules and beneficial effects of executing the information recommendation method of the fusion multi-channel recall algorithm.

[0163] Example 4

[0164] FIG5 shows a schematic structural diagram of an information recommendation device 500 that can be used to implement an embodiment of the present application that incorporates a multi-path recall algorithm.

[0165] As shown in Figure 5, the information recommendation device 500 that integrates the multi-channel recall algorithm includes at least one processor 501 and a memory that is communicatively connected to the at least one processor 501, such as a read-only memory (ROM) 502, a random access memory (RAM) 503, etc., wherein the memory stores a computer program that can be executed by the at least one processor, and the processor 501 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 502 or the computer program loaded from the storage unit 508 to the random access memory (RAM) 503. Various programs and data required for the operation of the information push device 500 can also be stored in the RAM 503. The processor 501, ROM 502, and RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0166] Multiple components in the information recommendation device 500 integrating the multi-path recall algorithm are connected to an I / O interface 505, including an input unit 506, such as a keyboard and mouse; an output unit 507, such as various types of displays and speakers; a storage unit 508, such as a magnetic disk and optical disk; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the information recommendation device 500 integrating the multi-path recall algorithm to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0167] Processor 501 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of processor 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors that run machine learning model algorithms, digital signal processors (DSPs), and any appropriate processor, controller, microcontroller, etc. Processor 501 executes the various methods and processes described above, such as the information recommendation method that integrates a multi-path recall algorithm.

[0168] In some embodiments, the information recommendation method of the fused multi-way recall algorithm may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed onto the information recommendation device 500 of the fused multi-way recall algorithm via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the processor 501, one or more steps of the information recommendation method of the fused multi-way recall algorithm described above may be performed. Alternatively, in other embodiments, the processor 501 may be configured to execute the information recommendation method of the fused multi-way recall algorithm by any other appropriate means (for example, by means of firmware).

[0169] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0170] Computer programs for implementing the methods of the present application may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0171] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0172] In order to provide interaction with the user, the systems and techniques described herein can be implemented on an information recommendation device that integrates a multi-path recall algorithm, which has: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball), through which the user can provide input to the information recommendation device that integrates the multi-path recall algorithm. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0173] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0174] A computing system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers and establishing a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within a cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and VPS (Virtual Private Server) services.

[0175] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this application can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of this application can be achieved. This is not limited herein.

[0176] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. An information recommendation method integrating a multi-path recall algorithm, comprising: Obtaining a set of objects to be recommended, a set of users, and behavioral data of the set of users in a historical time period; Dividing the behavior data of the historical time period into behavior data of a first time period and behavior data of a second time period; wherein the first time period is prior to the second time period; According to the behavior data of the user set in the first time period, using multiple methods, determining the push user corresponding to the candidate object in the set of objects to be recommended under each of the methods; According to the behavior data of the user set in the second time period, the push users corresponding to each of the candidate objects in different ways are merged to obtain the push users corresponding to each of the candidate objects; The push information of the candidate object is obtained and sent to the push user corresponding to the candidate object.

2. The method according to claim 1, wherein: The method of using multiple methods to determine, based on the behavior data of the user set in the first time period, the push user corresponding to the candidate object in the set of objects to be recommended in each of the methods includes: Determine, based on the behavior data of the candidate users in the user set in the first time period, candidate users who have interactive behaviors with the candidate objects in the set of objects to be recommended, and the number of behaviors of each candidate user with respect to each candidate object; Determining the interaction weight of each candidate user with respect to each candidate object according to the behavior data of the candidate users in the user set in the first time period and the number of behaviors of each candidate user with respect to each candidate object; According to the candidate users who have interactive behaviors with each of the candidate objects and the interaction weights of each of the candidate users with respect to each of the candidate objects, the push users corresponding to the candidate objects in the set of objects to be recommended are determined in a behavior data manner.

3. The method according to claim 2, wherein: The determining, based on the candidate users with whom each candidate object has an interactive behavior and the interaction weight of each candidate user with respect to each candidate object, the push user corresponding to the candidate object in the set of objects to be recommended in a behavior data manner includes: Determining characteristic information of each candidate user according to candidate users who have interaction behaviors with each candidate object and interaction weights of each candidate user with respect to each candidate object; Determining a similarity value between each candidate user and other candidate users according to the feature information of each candidate user; Determine the push probability of each candidate object to each candidate user according to the candidate users with whom each candidate object has an interactive behavior, the interaction weight of each candidate user with respect to each candidate object, and the similarity value between each candidate user and other candidate users; According to the push probability of each candidate object to each candidate user, the push user corresponding to each candidate object is determined.

4. The method according to claim 1, wherein: The method of using multiple methods to determine, based on the behavior data of the user set in the first time period, the push user corresponding to the candidate object in the set of objects to be recommended in each of the methods includes: According to the behavior data of the candidate users in the user set in the first time period and at least one attribute information of the candidate objects in the to-be-recommended object set, determine the candidate users with which the attribute information has interactive behaviors and the number of behaviors of the candidate users with respect to the attribute information; Determining the interaction weight of each candidate user for each attribute information according to the behavior data of the candidate users in the user set in the first time period and the number of behaviors of each candidate user for each attribute information; According to at least one attribute information of each candidate object and the interaction weight of each candidate user for each attribute information, a push user corresponding to the candidate object in the set of objects to be recommended is determined in the object attribute information mode.

5. The method according to claim 1, wherein: The method of using multiple methods to determine, based on the behavior data of the user set in the first time period, the push user corresponding to the candidate object in the set of objects to be recommended in each of the methods includes: According to the behavior data of the candidate users in the user set in the first time period, obtaining the number of behaviors of each of the candidate users; According to the number of behaviors of each candidate user, the push users corresponding to each candidate object are screened from among the candidate users.

6. The method according to claim 1, wherein: The step of fusing the push users corresponding to the candidate objects in different ways according to the behavior data of the user set in the second time period to obtain the push users corresponding to the candidate objects includes: For each candidate object, according to the extraction ratio corresponding to each method, a target user is extracted from the corresponding push users under each method to form a target user group of the candidate object and the target user; Performing statistics on each of the target user groups according to the behavior data of the candidate users in the user set in the second time period; According to the statistical results of each target user group, adjusting the extraction ratio corresponding to each method; For each candidate object, according to the extraction ratio corresponding to each method, push users are extracted from the push users corresponding to the candidate object in each method, and merged to obtain the push users corresponding to the candidate object.

7. The method according to claim 1, wherein: The candidate objects include: news, articles, web pages, videos, live broadcast rooms or products.

8. An information recommendation device integrating a multi-path recall algorithm, comprising: A data acquisition module, configured to acquire a set of objects to be recommended, a set of users, and behavioral data of the set of users in a historical time period; A time division module, configured to divide the behavior data of the historical time period into behavior data of a first time period and behavior data of a second time period; wherein the first time period is prior to the second time period; A push user determination module is configured to determine, based on the behavior data of the user set in the first time period, a push user corresponding to the candidate object in the set of objects to be recommended in each of the methods in a plurality of ways; A user fusion module is configured to fuse the push users corresponding to each candidate object in different ways according to the behavior data of the user set in the second time period, so as to obtain the push users corresponding to each candidate object; The information sending module is configured to obtain push information of the candidate object and send it to a push user corresponding to the candidate object.

9. An information recommendation device integrating a multi-path recall algorithm, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the information recommendation method of the fused multi-way recall algorithm according to any one of claims 1 to 7.

10. A computer readable medium, wherein: The computer-readable medium stores computer instructions, and the computer instructions are used to enable a processor to implement the information recommendation method of any one of claims 1-7 by fusing a multi-way recall algorithm when executing the instructions.

Images