CN114119106A - Information recommendation method and device, server and storage medium - Google Patents

Information recommendation method and device, server and storage medium Download PDF

Info

Publication number
CN114119106A
CN114119106A CN202111499713.3A CN202111499713A CN114119106A CN 114119106 A CN114119106 A CN 114119106A CN 202111499713 A CN202111499713 A CN 202111499713A CN 114119106 A CN114119106 A CN 114119106A
Authority
CN
China
Prior art keywords
information
recommendation
recommendation information
clustering
historical
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202111499713.3A
Other languages
Chinese (zh)
Inventor
肖严
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Dajia Internet Information Technology Co Ltd
Original Assignee
Beijing Dajia Internet Information Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Dajia Internet Information Technology Co Ltd filed Critical Beijing Dajia Internet Information Technology Co Ltd
Priority to CN202111499713.3A priority Critical patent/CN114119106A/en
Publication of CN114119106A publication Critical patent/CN114119106A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0251—Targeted advertisements
    • G06Q30/0255—Targeted advertisements based on user history
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Accounting & Taxation (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Engineering & Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The disclosure relates to an information recommendation method, an information recommendation device, a server and a storage medium. The method comprises the following steps: responding to an information recommendation request of a target account, and acquiring first recall information from a preset recommendation information base; acquiring second recall information from historical recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of a target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request; acquiring candidate recommendation information from the first recall information and the second recall information; and determining target recommendation information from the candidate recommendation information, and sending the target recommendation information to a target account. According to the method and the device, the number of the recalled information can be increased on the basis of not increasing the number of additional machines, so that the accuracy of information recommendation can be improved.

Description

Information recommendation method and device, server and storage medium
Technical Field
The present disclosure relates to the field of information recommendation technologies, and in particular, to an information recommendation method, an information recommendation apparatus, a server, and a storage medium.
Background
With the development of information recommendation technology, a technology for recommending information to a user appears, wherein recommendation information is recalled from a recommendation information base, namely, recommendation information suitable for being recommended to the user is screened out, and then the screened recommendation information can be ranked and bid according to the information value of the screened recommendation information, so that the recommendation information with high information value can be recommended to the user.
However, in the current information recommendation method, the process of recalling the recommendation information generally needs to increase the number of the recall information to ensure the recall effect, however, if the number of the recall information needs to be increased, the number of machines needed for executing the information recall needs to be increased at the same time, so that the accuracy of information recommendation in the current information recommendation method is low under the condition that the number of machines is limited.
Disclosure of Invention
The present disclosure provides an information recommendation method, apparatus, server and storage medium to at least solve the problem of low accuracy of information recommendation in related art. The technical scheme of the disclosure is as follows:
according to a first aspect of the embodiments of the present disclosure, there is provided an information recommendation method, including:
responding to an information recommendation request of a target account, and acquiring first recall information from a preset recommendation information base;
acquiring second recall information according to the historical recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of the target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request;
acquiring candidate recommendation information from the first recall information and the second recall information;
and determining target recommendation information from the candidate recommendation information, and sending the target recommendation information to the target account.
In an exemplary embodiment, the historical recommendation information is candidate recommendation information obtained from the historical recall information; the number of the historical recommendation information is multiple; the obtaining of second recall information from the historical recommendation information of the target account includes: acquiring a clustering center corresponding to each historical recommendation information, and determining a recommendation information set corresponding to each clustering center; and screening out a first amount of recommendation information from the recommendation information sets corresponding to the clustering centers to serve as the second recall information.
In an exemplary embodiment, the screening out a first amount of recommendation information from the recommendation information sets corresponding to the respective cluster centers as the second recall information includes: determining a current clustering center and a recommendation information set corresponding to the current clustering center to obtain a current recommendation information set; acquiring historical recommendation information contained in the current recommendation information set and the information number of the historical recommendation information contained in the current recommendation information set; screening out a second quantity of other recommendation information except the historical recommendation information contained in the current recommendation information set from the current recommendation information set, and taking the second quantity of other recommendation information and the historical recommendation information contained in the current recommendation information set as second recall information; and the second quantity is the difference value between the first quantity and the information quantity.
In an exemplary embodiment, the filtering out a second amount of recommendation information from the current recommendation information set, except for the historical recommendation information included in the current recommendation information set, includes: obtaining bid information corresponding to each piece of other recommendation information; and acquiring a second amount of other recommendation information from the other recommendation information according to the ranking of the bid information.
In an exemplary embodiment, before obtaining the cluster center corresponding to each piece of historical recommendation information, the method further includes: acquiring information characteristics corresponding to each piece of recommendation information in the recommendation information base and bid information corresponding to each piece of recommendation information in the recommendation information base; and clustering the recommendation information in the recommendation information base by using the information characteristics and the bid information to obtain a clustering center corresponding to the recommendation information in the recommendation information base.
In an exemplary embodiment, the clustering, by using the information features and the bid information, the recommendation information in the recommendation information base to obtain a cluster center corresponding to each recommendation information in the recommendation information base includes: acquiring a plurality of initial clustering centers, calculating a first distance between each piece of recommended information and each initial clustering center according to the information characteristics and the bid information, and screening out initial clustering information corresponding to each initial clustering center from each piece of recommended information according to the first distance to form each clustering information set; acquiring a target clustering center corresponding to each clustering information set according to the information characteristics of the initial clustering information contained in each clustering information set; and calculating a second distance between the initial clustering information contained in each clustering information set and a target clustering center corresponding to each clustering information set according to the information characteristics of the initial clustering information contained in each clustering information set and the bid information of the initial clustering information contained in each clustering information set, and determining the clustering center corresponding to each piece of recommendation information in the recommendation information base according to the second distance.
In an exemplary embodiment, the determining, according to the second distance, a clustering center corresponding to each piece of recommendation information in the recommendation information base includes: acquiring the sum of the distances of the second distances; if the sum of the distances is larger than or equal to a preset distance threshold value, taking the target clustering center as an initial clustering center, and returning to the step of calculating the first distance between each piece of recommendation information and each initial clustering center according to the information characteristics and the bid information; and/or if the sum of the distances is smaller than the distance threshold, taking the target clustering center as a clustering center corresponding to each piece of recommended information in the recommended information base.
According to a second aspect of the embodiments of the present disclosure, there is provided an information recommendation apparatus including:
a first recall acquisition unit configured to execute acquisition of first recall information from a preset recommendation information base in response to an information recommendation request of a target account;
a second recall acquisition unit configured to execute acquiring second recall information according to the historical recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of the target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request;
a candidate recommendation obtaining unit configured to perform obtaining candidate recommendation information from the first recall information and the second recall information;
and the target recommendation sending unit is configured to determine target recommendation information from the candidate recommendation information and send the target recommendation information to the target account.
In an exemplary embodiment, the historical recommendation information is candidate recommendation information obtained from the historical recall information; the number of the historical recommendation information is multiple; the second recall acquisition unit is further configured to execute acquisition of a clustering center corresponding to each historical recommendation information, and determine a recommendation information set corresponding to each clustering center; and screening out a first amount of recommendation information from the recommendation information sets corresponding to the clustering centers to serve as the second recall information.
In an exemplary embodiment, the second recall acquisition unit is further configured to determine a current clustering center and a recommended information set corresponding to the current clustering center, so as to obtain a current recommended information set; acquiring historical recommendation information contained in the current recommendation information set and the information number of the historical recommendation information contained in the current recommendation information set; screening out a second quantity of other recommendation information except the historical recommendation information contained in the current recommendation information set from the current recommendation information set, and taking the second quantity of other recommendation information and the historical recommendation information contained in the current recommendation information set as second recall information; and the second quantity is the difference value between the first quantity and the information quantity.
In an exemplary embodiment, the second recall acquisition unit is further configured to perform acquisition of bid information corresponding to each of the other recommendation information; and acquiring a second amount of other recommendation information from the other recommendation information according to the ranking of the bid information.
In an exemplary embodiment, the information recommendation apparatus further includes: the recommendation information clustering unit is configured to execute the steps of obtaining information characteristics corresponding to each piece of recommendation information in the recommendation information base and bid information corresponding to each piece of recommendation information in the recommendation information base; and clustering the recommendation information in the recommendation information base by using the information characteristics and the bid information to obtain a clustering center corresponding to the recommendation information in the recommendation information base.
In an exemplary embodiment, the recommendation information clustering unit is further configured to perform obtaining a plurality of initial clustering centers, calculate a first distance between each recommendation information and each initial clustering center according to the information features and the bid information, and screen out initial clustering information corresponding to each initial clustering center from each recommendation information according to the first distance to form each clustering information set; acquiring a target clustering center corresponding to each clustering information set according to the information characteristics of the initial clustering information contained in each clustering information set; and calculating a second distance between the initial clustering information contained in each clustering information set and a target clustering center corresponding to each clustering information set according to the information characteristics of the initial clustering information contained in each clustering information set and the bid information of the initial clustering information contained in each clustering information set, and determining the clustering center corresponding to each piece of recommendation information in the recommendation information base according to the second distance.
In an exemplary embodiment, the recommendation information clustering unit is further configured to perform obtaining a distance sum of the respective second distances; if the sum of the distances is larger than or equal to a preset distance threshold value, taking the target clustering center as an initial clustering center, and returning to the step of calculating the first distance between each piece of recommendation information and each initial clustering center according to the information characteristics and the bid information; and the cluster center is configured to be used as the cluster center corresponding to each piece of recommended information in the recommended information base if the distance sum is smaller than the distance threshold.
According to a third aspect of the embodiments of the present disclosure, there is provided a server, 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 as defined in any one of the embodiments of the first aspect.
According to a fourth aspect of embodiments of the present disclosure, there is provided a computer-readable storage medium, wherein instructions, when executed by a processor of a server, enable the server to perform the information recommendation method according to any one of the embodiments of the first aspect.
According to a fifth aspect of embodiments of the present disclosure, there is provided a computer program product, which includes instructions that, when executed by a processor of a server, enable the server to perform the information recommendation method according to any one of the embodiments of the first aspect.
The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects:
acquiring first recall information from a preset recommendation information base by responding to an information recommendation request of a target account; acquiring second recall information according to the historical recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of a target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request; acquiring candidate recommendation information from the first recall information and the second recall information; and determining target recommendation information from the candidate recommendation information, and sending the target recommendation information to a target account. According to the method and the device, on the basis of obtaining the first recall information from the recommendation information base, the second recall information can be obtained according to the historical recommendation information of the target account, and the final target recommendation information is obtained and recommended to the target account on the basis of the first recall information and the second recall information.
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 present disclosure and, together with the description, serve to explain the principles of the disclosure and are not to be construed as limiting the disclosure.
Fig. 1 is a diagram illustrating an application environment of an information recommendation method according to an exemplary embodiment.
Fig. 2 is a flow chart illustrating an information recommendation method according to an example embodiment.
FIG. 3 is a flowchart illustrating obtaining second recall information according to an exemplary embodiment.
Fig. 4 is a flowchart illustrating obtaining second recall information according to another example embodiment.
FIG. 5 is a flow diagram illustrating a process for clustering recommendation information according to an example embodiment.
Fig. 6 is a flowchart illustrating obtaining a cluster center corresponding to recommended information according to an exemplary embodiment.
FIG. 7 is a schematic diagram illustrating a structure of a residual neural network, according to an example embodiment.
FIG. 8 is a schematic view of an ad cache of a bid queue shown in accordance with an exemplary embodiment.
FIG. 9 is a schematic diagram of ad clustering for a bid queue shown in accordance with an exemplary embodiment.
Fig. 10 is a block diagram illustrating an information recommendation apparatus according to an example embodiment.
FIG. 11 is a block diagram illustrating a server in accordance with an example embodiment.
Detailed Description
In order to make the technical solutions of the present disclosure better understood by those of ordinary skill in the art, the technical solutions in 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 and claims of the present disclosure and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the disclosure described herein are capable of operation in sequences other than those illustrated or otherwise described herein. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
It should also be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for presentation, analyzed data, etc.) referred to in the present disclosure are both information and data that are authorized by the user or sufficiently authorized by various parties.
The information recommendation method provided by the present disclosure may be applied to an application environment as shown in fig. 1. Wherein the terminal 101 interacts with the server 102 via a network. Specifically, the target account may initiate an information recommendation request for obtaining recommendation information to the server 102 through the terminal 101, the server 102 may respond to the request, obtain first recall information from a recommendation information base in which the recommendation information is stored, obtain second recall information according to history recommendation information obtained from history recall information corresponding to the history recommendation request of the target account, sort the first recall information and the second recall information, screen out candidate recommendation information, and finally determine the target recommendation information from the candidate recommendation information and return the target recommendation information to the terminal 101 of the target account. The terminal 101 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, and the server 102 may be implemented by an independent server or a server cluster composed of a plurality of servers.
Fig. 2 is a flowchart illustrating an information recommendation method according to an exemplary embodiment, where the information recommendation method is used in the server 102, as shown in fig. 2, and includes the following steps.
In step S201, in response to an information recommendation request of a target account, first recall information is acquired from a preset recommendation information base.
The target account refers to an account for obtaining recommendation information, the account may trigger a relevant recommendation information obtaining operation through the terminal 101, and initiate an information recommendation request for obtaining recommendation information sent by the server 102 to the server 102, for example, the target account may perform a refresh operation on a certain display page in the terminal 101 to trigger the information recommendation request to the server 102, and the server 102 may respond to the request, and find recall information corresponding to the information recommendation request from a pre-established recommendation information base, as the first recall information.
For example, the recommendation information that may be interested in the target account may be screened out as the first recall information according to the interest tag of the target account, or the browsing topic, or the like, or the hit recommendation information in the recommendation information base may be used as the first recall information, or the recommendation information liked by other accounts similar to the target account may be cooperatively filtered and recalled as the first recall information, or the like, and in this step, the server 102 may screen out the first recall information corresponding to different screening rules from the recommendation information base in which the recommendation information is stored according to different screening rules.
In step S202, second recall information is acquired according to the historical recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of the target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request.
The history recommendation information refers to recommendation information obtained from the history recall information, and may be, for example, recommendation information that is sorted by sorting the history recall information and is screened as history recommendation information. The history recall information refers to recall information corresponding to a history recommendation request obtained by the server 102 when a target account triggers the history information recommendation request, where the history information recommendation request refers to an information recommendation request triggered by the target account within a period of time before the information recommendation request is triggered, an information recommendation request triggered by the current triggered information recommendation request a preset number of times may be used as the history information recommendation request, and for example, an information recommendation request triggered by the target account 10 times before the current triggered information recommendation request may be used as the history information recommendation request.
Specifically, the server 102 may store historical recommendation information of the target account, after each time the target account triggers an information recommendation request, the server 102 may respond to the request to obtain corresponding recall information, and process the recall information to determine corresponding recommendation information, and then the server 102 may store the recommendation information as historical recommendation information of an information recommendation request triggered after the target account. When the information recommendation request is triggered after the target account, the stored historical recommendation information may be read, and corresponding recall information may be obtained according to the historical recommendation information, and may be used as the second recall information, for example, the historical recommendation information may be directly used as newly-added recall information, that is, the second recall information, or recommendation information related to the historical recommendation information may be screened out based on the historical recommendation information, and used as the second recall information.
In step S203, candidate recommendation information is acquired from the first recall information and the second recall information;
in step S204, target recommendation information is determined from the candidate recommendation information, and the target recommendation information is sent to the target account.
The candidate recommendation information refers to recommendation information obtained by sorting the recall information through a sorting stage, and the candidate recommendation information can be used as bidding queue information to execute a final bidding stage so as to determine recommendation information finally recommended to a target account, namely target recommendation information.
Specifically, after the server 102 obtains the first recall information and the second recall information, the first recall information and the second recall information may be used as a candidate set for ranking, and the first recall information and the second recall information are ranked, for example, the first recall information and the second recall information may be ranked according to earnings brought by the recall information according to pre-estimation values of various models, and after the ranking is completed, corresponding candidate recommendation information may be obtained, and then, the server 102 may further use the candidate recommendation information as recommendation information in a bidding queue, perform a series of filtering on the candidate recommendation information by a threshold and the like, and then calculate a value corresponding to each candidate recommendation information, thereby selecting the candidate recommendation information with the highest value as target recommendation information to be sent to the terminal 101 of the target account.
In the information recommendation method, first recall information is acquired from a preset recommendation information base by responding to an information recommendation request of a target account; acquiring second recall information according to the historical recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of a target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request; acquiring candidate recommendation information from the first recall information and the second recall information; and determining target recommendation information from the candidate recommendation information, and sending the target recommendation information to a target account. According to the method and the device, on the basis of obtaining the first recall information from the recommendation information base, the second recall information can be obtained according to the historical recommendation information of the target account, and the final target recommendation information is obtained and recommended to the target account on the basis of the first recall information and the second recall information.
In an exemplary embodiment, the historical recommendation information is candidate recommendation information obtained from the historical recall information; the number of the historical recommendation information is multiple; as shown in fig. 3, step S202 may specifically include the following steps:
in step S301, a cluster center corresponding to each piece of historical recommendation information is obtained, and a recommendation information set corresponding to each cluster center is determined.
In this embodiment, the history recommendation information may be recommendation information obtained by the server 102 after the history recall information passes through the sorting phase after the target account triggers the history information recommendation request, that is, candidate recommendation information corresponding to the history information recommendation request. In this embodiment, after obtaining candidate recommendation information as a bid queue, the server 102 may store the candidate recommendation information as historical recommendation information corresponding to the next target account trigger information recommendation request, and since the historical recommendation information is composed of candidate recommendation information that is information of the bid queue, the number of the historical recommendation information may be multiple. The cluster center refers to a cluster center of the historical recommendation information, and the recommendation information set refers to a set formed by recommendation information corresponding to the cluster center.
Specifically, the server 102 may perform clustering processing on recommendation information included in the recommendation information base in advance to form a plurality of cluster centers, and a recommendation information set formed by recommendation information included in each cluster center. After the server 102 determines the historical recommendation information, it may determine a cluster center corresponding to each piece of historical recommendation information and a recommendation information set corresponding to the cluster center.
For example, the recommendation information base includes recommendation information a, recommendation information B, recommendation information C, and recommendation information D, where the recommendation information a and the recommendation information B correspond to a cluster center a, a recommendation information set corresponding to the cluster center a includes the recommendation information a and the recommendation information B, and the recommendation information C and the recommendation information D correspond to a cluster center B, and a recommendation information set corresponding to the cluster center B includes the recommendation information C and the recommendation information D, so when the recommendation information a is history recommendation information, the server 102 may use the cluster center a as a cluster center corresponding to the recommendation information a, and use the recommendation information set corresponding to the recommendation information a as a recommendation information set corresponding to the cluster center.
In step S302, a first amount of recommendation information is screened from the recommendation information sets corresponding to the respective clustering centers as second recall information.
The first quantity may refer to the quantity of the recommendation information screened from the recommendation information set, and the quantity may be set in advance, because the historical recommendation information in this embodiment is bid queue information obtained by screening the historical recall information through a sorting stage, the data volume of the historical recommendation information is smaller than that of the first recall information obtained directly from the recommendation information base, and therefore, in order to improve the recall effect, in this embodiment, the first quantity of recommendation information may also be screened from the recommendation information set corresponding to the clustering center of each piece of historical recommendation information, and used as the second recall information to expand the data volume of the second recall information.
For example, if the first number is 10, for a certain historical recommendation information a, 10 recommendation information may be screened from the recommendation information set corresponding to the certain historical recommendation information a, and the certain historical recommendation information a is used as the second recall information.
In this embodiment, the second recall information may be obtained by screening a recommendation information set corresponding to the historical recommendation information, so that the data volume of the second recall information may be increased, and the reliability of information recall may be improved.
Further, as shown in fig. 4, the step S302 may further include:
in step S401, a current clustering center and a recommendation information set corresponding to the current clustering center are determined, and a current recommendation information set is obtained.
The current clustering center refers to any one of a plurality of clustering centers corresponding to the plurality of historical recommendation information, different historical recommendation information can also correspond to different clustering centers due to the plurality of historical recommendation information, and the current clustering center refers to any one of the plurality of clustering centers. For example, the history recommendation information may include history recommendation information a and history recommendation information B, and the history recommendation information a and the history recommendation information B respectively correspond to different cluster centers, that is, respectively correspond to a cluster center a and a cluster center B, then the current cluster center may be any one of the cluster center a or the cluster center B, the current recommendation information set refers to a recommendation information set corresponding to the current cluster center, if the cluster center a is the current cluster center, then the recommendation information set corresponding to the cluster center a may be the current recommendation information set, and if the cluster center B is the current cluster center, then the recommendation information set corresponding to the cluster center B may be the current recommendation information set.
In step S402, the historical recommendation information included in the current recommendation information set and the information number of the historical recommendation information included in the current recommendation information set are obtained.
After determining the current recommendation information set, the server 102 may further determine the information number of the historical recommendation information included in the set, for example, for both the historical recommendation information a and the historical recommendation information C, which correspond to the same cluster center a, so that when the cluster center a serves as the current cluster center, the current recommendation information set may include two pieces of historical recommendation information, and if the historical recommendation information B corresponds to the cluster center B, when the cluster center B serves as the current cluster center, the current recommendation information set may include one piece of historical recommendation information.
In step S403, a second amount of other recommendation information except the historical recommendation information included in the current recommendation information set is screened from the current recommendation information set, and the second amount of other recommendation information and the historical recommendation information included in the current recommendation information set are used as second recall information; wherein, the second quantity is the difference value between the first quantity and the information quantity.
The other recommendation information refers to recommendation information other than the historical recommendation information in the current recommendation information set, after determining the historical recommendation information included in the current recommendation information set, the server 102 may use the recommendation information other than the historical recommendation information in the current recommendation information set as the other recommendation information, screen out a second number of other recommendation information from the other recommendation information, and use a second number of other recommendation information and the historical recommendation information included in the current recommendation information set as second recall information, where the second number is a difference between the first number and the information number.
For example, the first number may be set to 10, if a certain current recommendation information set includes 2 pieces of historical recommendation information, the second number may be 8, that is, 8 pieces of other recommendation information are screened from the current recommendation information set, and 8 pieces of other recommendation information and 2 pieces of historical recommendation information are taken as the second recall information, and if a certain current recommendation information set includes only 1 piece of historical recommendation information, the second number may be 9, that is, 9 pieces of other recommendation information are screened from the current recommendation information set, and 9 pieces of other recommendation information and 1 piece of historical recommendation information are taken as the second recall information.
In this embodiment, the second recall information may be obtained by screening a second number of other pieces of recommendation information in the current recommendation information set and historical recommendation information in the current recommendation information set, so that it is ensured that the number of the second recall information screened in each recommendation information set satisfies the first number, and it is ensured that the second recall information includes the historical recommendation information, and diversity and accuracy of the second recall information can be improved.
Further, step S403 may further include: obtaining bid information corresponding to each piece of other recommendation information; and acquiring a second amount of other recommendation information from the other recommendation information according to the ranking of the bid information.
In this embodiment, after determining the other recommendation information, the server 102 may further obtain a bid of the advertiser corresponding to each other recommendation information as bid information corresponding to each other recommendation information, and then the server 102 may further sort the other recommendation information according to the bid information, for example, sort the other recommendation information according to a size order of the bid information, and select a second quantity of other recommendation information according to the sort, that is, select a second quantity of other recommendation information with a higher bid.
For example, the other recommendation information is sorted according to the size order of the bid information and is other recommendation information a, other recommendation information B, other recommendation information C, other recommendation information D, and other recommendation information E, if the determined second number is 3, the server 102 may use the other recommendation information a, the other recommendation information B, and the other recommendation information C as the screened second number of other recommendation information, and if the determined second number is 2, the server 102 may use the other recommendation information a and the other recommendation information B as the screened second number of other recommendation information.
In this embodiment, the second amount of other selected recommendation information may be selected according to bid information corresponding to the other recommendation information, so that the other selected recommendation information has higher bid information, thereby improving the information value of the second recall information.
In an exemplary embodiment, as shown in fig. 5, before step S202, the method may further include:
in step S501, information characteristics corresponding to each piece of recommendation information in the recommendation information base and bid information corresponding to each piece of recommendation information in the recommendation information base are obtained.
In this embodiment, the recommendation information may be a video content feature of the advertisement video, the information feature may be extracted through a pre-trained residual neural network, for example, the information feature is extracted through ResNet-34, and the bid information refers to a bid of an advertiser for the advertisement video. Specifically, the server 102 may extract video content features of each piece of recommendation information in the recommendation information base, that is, each advertisement video, through a residual neural network, as information features corresponding to each piece of recommendation information, and may also obtain an advertiser bid corresponding to each advertisement video, as bid information corresponding to each piece of recommendation information.
In step S502, clustering is performed on each piece of recommendation information in the recommendation information base by using the information features and the bid information, so as to obtain a clustering center corresponding to each piece of recommendation information in the recommendation information base.
Then, the server 102 may perform clustering processing on the recommendation information in the recommendation information base by using the information features and the bid information determined in step S501, so that recommendation information with similar information features and bid information may be clustered into the same recommendation information set, and a cluster center of the recommendation information set may also be determined, thereby obtaining a cluster center corresponding to each recommendation information in the recommendation information base.
In this embodiment, the server 102 may further perform clustering processing on the recommendation information in the recommendation information base by using the information features and the bid information, so that the clustered recommendation information satisfies both the similarity of the information features and the similarity of the bid information, thereby further improving the clustering accuracy of the recommendation information clustering.
Further, as shown in fig. 6, step S502 may further include:
in step S601, a plurality of initial clustering centers are obtained, a first distance between each piece of recommended information and each initial clustering center is calculated according to the information characteristics and the bid information, and initial clustering information corresponding to each initial clustering center is screened from each piece of recommended information according to the first distance to form each clustering information set.
The first distance may be used to represent a similarity between each piece of recommendation information in the recommendation information base and an initial clustering center, where the initial clustering center may be randomly generated by the server 102, and the initial clustering information refers to recommendation information included in a clustering information set corresponding to the initial clustering center.
Specifically, the server 102 may randomly generate a plurality of cluster centers as initial cluster centers, and then the server 102 may calculate a first distance between each piece of recommendation information and each initial cluster center by using the information feature and the bid information of each piece of recommendation information, so as to determine the initial cluster center corresponding to each piece of recommendation information, for example, the initial cluster center with the smallest first distance corresponding to each piece of recommendation information may be used as the initial cluster center corresponding to the piece of recommendation information, so that the server 102 may determine the piece of recommendation information corresponding to each initial cluster center, that is, the piece of initial cluster information, and combine the pieces of initial cluster information into a cluster information set matching the initial cluster centers.
For example, the recommendation information may include: the server 102 may calculate first distances to the initial clustering centers a and the initial clustering centers B respectively by using information features and bid information of the recommended information, so that the initial clustering centers with smaller first distances are used as the initial clustering centers corresponding to each recommended information, the recommended information is used as the initial clustering information corresponding to the initial clustering centers, the initial clustering information corresponding to the initial clustering centers a may be the recommended information a, the recommended information B and the recommended information C, and the initial clustering information corresponding to the initial clustering centers B may be the recommended information D and the recommended information E, so as to form a clustering information set of the initial clustering centers a, i.e. including recommendation information a, recommendation information B and recommendation information C, and a set of cluster information forming an initial cluster center B, i.e. including recommendation information D and recommendation information E. The calculation formula of the first distance may be as follows:
Figure BDA0003401025510000131
wherein x isiInformation characteristic representing the ith recommendation information, CenterkThen the k-th initial cluster center, dist (x) is representedi,Centerk) It indicates a first distance of the ith recommendation information from the kth initial cluster center,
Figure BDA0003401025510000132
bid information indicating the ith recommendation information,
Figure BDA0003401025510000133
then representNormalizing the bid information, and
Figure BDA0003401025510000134
the similarity between the ith recommendation information and the kth initial clustering center is shown.
In step S602, a target clustering center corresponding to each clustering information set is obtained according to the information characteristics of the initial clustering information included in each clustering information set.
The target cluster center is a cluster center for each cluster information set that is determined again by the server 102 after each cluster information set is determined, and the target cluster center may be calculated by the server 102 according to information characteristics of initial cluster information included in each cluster information set, for example, by the following formula:
Figure BDA0003401025510000135
wherein, the CenterkRepresenting the re-determined target cluster center, CkThe number of information, x, of the initial clustering information included in the clustering information set corresponding to the target clustering centeriThe information characteristics of the ith initial clustering information are represented, and the average value of the initial clustering information included in the clustering information set is obtained, so that the corresponding target clustering center can be determined.
In step S603, according to the information features of the initial clustering information included in each clustering information set and the bid information of the initial clustering information included in each clustering information set, a second distance between the initial clustering information included in each clustering information set and a target clustering center corresponding to each clustering information set is calculated, and a clustering center corresponding to each piece of recommendation information in the recommendation information base is determined according to the second distance.
The second distance may represent a similarity between each initial cluster information in the cluster information sets and the re-determined target cluster center for the cluster information sets, and after determining the target cluster center for each cluster information set by the server 102, a second distance between each initial cluster information and the target cluster center can be obtained again according to the way of calculating the first distance, so that the server can determine the clustering center corresponding to each initial clustering information according to the second distance, for example, if the second distance corresponding to a certain cluster information set is smaller than a set threshold, the target cluster center may be used as the cluster center of all the initial cluster information in the cluster information set, therefore, the clustering center of each initial clustering information can be determined, and the clustering center corresponding to each recommended information is obtained.
In this embodiment, the clustering center of each piece of recommendation information may be determined by calculating the distance from the clustering center to the recommendation information and the bid information, so that the accuracy of the determined clustering center may be improved.
Further, step S603 may further include: acquiring the sum of the distances of the second distances; if the sum of the distances is greater than or equal to the preset distance threshold, taking the target clustering center as an initial clustering center, and returning to the step S601; and/or if the sum of the distances is smaller than the distance threshold, taking the target clustering center as the clustering center corresponding to each piece of recommended information in the recommended information base.
Wherein, the distance sum refers to the sum of the second distances between all the initial clustering information and the corresponding target clustering centers, after the server 102 calculates the second distances between all the initial cluster information and the corresponding target cluster centers, all the obtained second distances may be added to obtain a distance sum, and the server 102 may compare the sum of the distances with a predetermined distance threshold, and if the sum of the distances is greater than or equal to the predetermined distance threshold, it indicates that the clustering effect is still not ideal, and at this time, the server 102 may use the determined target clustering center as the initial clustering center of the new iteration, and returns to step S601 until the sum of the distances is less than the distance threshold, which indicates that the clustering effect is ideal enough, then the target clustering center determined by the last iteration can be used as the clustering center corresponding to each piece of recommendation information in the recommendation information base.
In this embodiment, iteration of the clustering process may also be implemented by setting the distance threshold, and the iteration process is stopped until the sum of the distances is smaller than the distance threshold, so that the clustering effect and the accuracy of the determined clustering center may be further improved.
In an exemplary embodiment, a video retrieval method based on a cluster center is further provided, and the profit is maximized while the result relevance is ensured. The scheme can be realized by an offline part and an online part, and can be specifically realized by the following steps:
(1) the off-line part is used for carrying out the off-line part,
the main function of the off-line part is to extract embedding and clustering videos of a full library, the full-library video embedding is extracted by utilizing a deep neural network ResNet-34, the ResNet-34 mainly structurally comprises a residual error design part, the structure of the residual error part is shown in fig. 7, and through the connection setting of 'skip level' of f (x) + x, shaving loss caused by back propagation can be effectively relieved, and the neural network can better extract embedding of the videos.
After the embedding of all videos is extracted, clustering is carried out through a method of kmeans added with bid information, and compared with the traditional kmeans, the bid information is introduced into the clustering.
And for the newly created video of the advertiser, the new advertisement can be estimated by regularly utilizing the clustered model and then belongs to the corresponding class, so that the rapid starting of the new advertisement can be ensured.
(2) On-line part
The online portion is mainly a cluster center that is deployed and recalled as detailed below.
In an advertisement system, a plurality of personalized models are usually deployed for personalized recall, in order to ensure the effect, a large number of machines are often used for recalling thousands of advertisements, but in the advertisement system, one advertisement can be called out at a time pv, most of the recalling results are not wined, if the number of recalls is reduced, the recalling results can not be wined, the large-disk effect cannot be ensured, and if the number of recalls is increased, a large number of machines are used. From the above, it can be seen that in this case, the effect and the machine need to be balanced, and in order to solve this problem, a method for recalling based on clustering of bidding advertisement queues is proposed, so that a great amount of machines are saved while the effect is ensured.
The clustering deployment recalling method based on the bidding advertisement queue mainly comprises two steps:
in a first step, shown in FIG. 8, advertisements in a bid queue are cached, which has the benefits of:
1. the recall efficiency is high, the recalled advertisements are the advertisements which are targeted once, most of the recalled advertisements also meet the targeting conditions, the delivery system can be used for delivering the recalled advertisements again, and the recall result is prevented from being filtered by the targeting information
2. The system is sensitive to response, any change of the system can be reflected in the bidding queue, the recalling result can be changed immediately, and the change of the system can be sensed very quickly
3. The bidding queue is a stage that all the recalling methods need to go through by integrating the results of other various recalling methods, and the recalling in the bidding queue is equivalent to re-screening better results of the results of each recall, so that the screened results can be called back and participate in bidding, in other words, even if the capacity of other recalling methods is improved, the bidding method can also be fed back to the bidding advertisement recalling method in a standing horse, and thus the recalling method can also be improved.
4. A large number of machines are saved, and because the cached bidding advertisements of the user dimensions are recalled, a large number of recall models do not need to be deployed for estimation, so that personalization is considered, and a large number of machines are saved.
And secondly, because the advertisement data volume of the bidding queue is generally dozens of times, the number of recalls is small compared with the model number, and in order to supplement the richness of the recalls, a recalling strategy of clustering expansion is provided. As shown in fig. 9, items 1 to itemn are cached bidding queues, and cluster centers 1 and … cluster center n are offline partially determined cluster centers, and for each bidding queue, a cluster center is found, and all the next advertisements are found through the cluster center, and are ranked by bid, and K advertisements ranked in the front are taken for recall, so that the following advantages are mainly achieved:
1. increase the size of the candidate set of the recall and improve the rate of the recall request
2. Through a clustering method, photo related information is introduced, and the accuracy of the advertisement is increased
3. Clustering expansion truncation is performed through the bid, advertiser bidding information is considered, bidding smoothness is improved, and platform income is increased.
In the embodiment, videos can be clustered through a deep neural network, so that the generalization capability of the whole recommendation library is improved, the newly created objects can be rapidly started, bid information, namely bid information of an advertiser, is introduced when similar videos are searched, the search effect is guaranteed, the platform profit is maximized, a recalling special mode can be developed through a clustering center, the effect is guaranteed, and the problem that a large number of machines are needed by an online model is solved.
It should be understood that, although the steps in the flowcharts of the present disclosure are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in the figures may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of execution of the steps or stages is not necessarily sequential, but may be performed alternately or in alternation with other steps or at least some of the other steps or stages.
It is understood that the same/similar parts between the embodiments of the method described above in this specification can be referred to each other, and each embodiment focuses on the differences from the other embodiments, and it is sufficient that the relevant points are referred to the descriptions of the other method embodiments.
Fig. 10 is a block diagram illustrating an information recommendation apparatus according to an example embodiment. Referring to fig. 10, the apparatus includes a first recall acquisition unit 1001, a second recall acquisition unit 1002, a candidate recommendation acquisition unit 1003, and a target recommendation transmitting unit 1004.
A first recall acquisition unit 1001 configured to perform acquisition of first recall information from a preset recommendation information base in response to an information recommendation request of a target account;
a second recall acquisition unit 1002 configured to perform acquisition of second recall information according to the history recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of a target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request;
a candidate recommendation acquisition unit 1003 configured to perform acquisition of candidate recommendation information from the first recall information and the second recall information;
and a target recommendation sending unit 1004 configured to perform determining target recommendation information from the candidate recommendation information, and sending the target recommendation information to the target account.
In an exemplary embodiment, the historical recommendation information is candidate recommendation information obtained from the historical recall information; the number of the historical recommendation information is multiple; the second recall acquisition unit 1002 is further configured to perform acquiring a clustering center corresponding to each historical recommendation information, and determine a recommendation information set corresponding to each clustering center; and screening a first amount of recommendation information from the recommendation information sets corresponding to the clustering centers to serve as second recall information.
In an exemplary embodiment, the second recall acquisition unit 1002 is further configured to determine a current clustering center and a recommendation information set corresponding to the current clustering center, so as to obtain a current recommendation information set; acquiring historical recommendation information contained in a current recommendation information set and the information number of the historical recommendation information contained in the current recommendation information set; screening out a second amount of other recommendation information except the historical recommendation information contained in the current recommendation information set from the current recommendation information set, and taking the second amount of other recommendation information and the historical recommendation information contained in the current recommendation information set as second recall information; wherein, the second quantity is the difference value between the first quantity and the information quantity.
In an exemplary embodiment, the second recall acquisition unit 1002 is further configured to perform acquiring bid information corresponding to each of the other recommendation information; and acquiring a second amount of other recommendation information from the other recommendation information according to the ranking of the bid information.
In an exemplary embodiment, the information recommendation apparatus further includes: the recommendation information clustering unit is configured to acquire information characteristics corresponding to each piece of recommendation information in the recommendation information base and bid information corresponding to each piece of recommendation information in the recommendation information base; and clustering all recommendation information in the recommendation information base by using the information characteristics and the bid information to obtain a clustering center corresponding to all recommendation information in the recommendation information base.
In an exemplary embodiment, the recommendation information clustering unit is further configured to perform obtaining a plurality of initial clustering centers, calculate a first distance between each recommendation information and each initial clustering center according to the information characteristics and the bid information, and screen out initial clustering information respectively corresponding to each initial clustering center from each recommendation information according to the first distance to form each clustering information set; acquiring a target clustering center corresponding to each clustering information set according to the information characteristics of the initial clustering information contained in each clustering information set; and calculating a second distance between the initial clustering information contained in each clustering information set and a target clustering center corresponding to each clustering information set according to the information characteristics of the initial clustering information contained in each clustering information set and the bid information of the initial clustering information contained in each clustering information set, and determining the clustering center corresponding to each piece of recommendation information in the recommendation information base according to the second distance.
In an exemplary embodiment, the recommendation information clustering unit is further configured to perform obtaining a distance sum of the respective second distances; if the sum of the distances is larger than or equal to a preset distance threshold value, taking the target clustering center as an initial clustering center, and returning to the step of calculating the first distance between each piece of recommendation information and each initial clustering center according to the information characteristics and the bid information; and the cluster center is configured to be used as the cluster center corresponding to each piece of recommended information in the recommended information base if the distance sum is smaller than the distance threshold.
With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
FIG. 11 is a block diagram illustrating an electronic device 1100 for information recommendation, according to an example embodiment. For example, the electronic device 1100 may be a server. Referring to fig. 11, electronic device 1100 includes a processing component 1120 that further includes one or more processors, and memory resources, represented by memory 1122, for storing instructions, such as application programs, that are executable by processing component 1120. The application programs stored in memory 1122 may include one or more modules that each correspond to a set of instructions. Further, the processing component 1120 is configured to execute instructions to perform the above-described methods.
The electronic device 1100 may further include: the power component 1124 is configured to perform power management of the electronic device 1100, the wired or wireless network interface 1126 is configured to connect the electronic device 1100 to a network, and an input-output (I/O) interface S28. The electronic device 1100 may operate based on an operating system, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, or the like, stored in the memory 1122.
In an exemplary embodiment, a computer-readable storage medium comprising instructions, such as memory 1122 comprising instructions, executable by a processor of electronic device 1100 to perform the above-described method is also provided. The storage medium may be a computer-readable storage medium, which may be, for example, a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
In an exemplary embodiment, a computer program product is also provided, which includes instructions executable by a processor of the electronic device 1100 to perform the above-described method.
It should be noted that the descriptions of the above-mentioned apparatus, the electronic device, the computer-readable storage medium, the computer program product, and the like according to the method embodiments may also include other embodiments, and specific implementations may refer to the descriptions of the related method embodiments, which are not described in detail herein.
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 variations, uses, or adaptations of the disclosure following, in general, the 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 will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims (10)

1. An information recommendation method, comprising:
responding to an information recommendation request of a target account, and acquiring first recall information from a preset recommendation information base;
acquiring second recall information according to the historical recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of the target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request;
acquiring candidate recommendation information from the first recall information and the second recall information;
and determining target recommendation information from the candidate recommendation information, and sending the target recommendation information to the target account.
2. The method of claim 1, wherein the historical recommendation information is candidate recommendation information derived from the historical recall information; the number of the historical recommendation information is multiple;
the obtaining of second recall information according to the historical recommendation information of the target account includes:
acquiring a clustering center corresponding to each historical recommendation information, and determining a recommendation information set corresponding to each clustering center;
and screening out a first amount of recommendation information from the recommendation information sets corresponding to the clustering centers to serve as the second recall information.
3. The method according to claim 2, wherein the screening out a first amount of recommendation information from the recommendation information sets corresponding to the respective clustering centers as the second recall information includes:
determining a current clustering center and a recommendation information set corresponding to the current clustering center to obtain a current recommendation information set;
acquiring historical recommendation information contained in the current recommendation information set and the information number of the historical recommendation information contained in the current recommendation information set;
screening out a second quantity of other recommendation information except the historical recommendation information contained in the current recommendation information set from the current recommendation information set, and taking the second quantity of other recommendation information and the historical recommendation information contained in the current recommendation information set as second recall information; and the second quantity is the difference value between the first quantity and the information quantity.
4. The method of claim 3, wherein the filtering out a second amount of recommendation information from the current set of recommendation information other than the historical recommendation information included in the current set of recommendation information comprises:
obtaining bid information corresponding to each piece of other recommendation information;
and acquiring a second amount of other recommendation information from the other recommendation information according to the ranking of the bid information.
5. The method according to claim 2, wherein before the obtaining of the clustering center corresponding to each historical recommendation information, the method further comprises:
acquiring information characteristics corresponding to each piece of recommendation information in the recommendation information base and bid information corresponding to each piece of recommendation information in the recommendation information base;
and clustering the recommendation information in the recommendation information base by using the information characteristics and the bid information to obtain a clustering center corresponding to the recommendation information in the recommendation information base.
6. The method according to claim 5, wherein the clustering the recommendation information in the recommendation information base by using the information features and the bid information to obtain a clustering center corresponding to each recommendation information in the recommendation information base comprises:
acquiring a plurality of initial clustering centers, calculating a first distance between each piece of recommended information and each initial clustering center according to the information characteristics and the bid information, and screening out initial clustering information corresponding to each initial clustering center from each piece of recommended information according to the first distance to form each clustering information set;
acquiring a target clustering center corresponding to each clustering information set according to the information characteristics of the initial clustering information contained in each clustering information set;
and calculating a second distance between the initial clustering information contained in each clustering information set and a target clustering center corresponding to each clustering information set according to the information characteristics of the initial clustering information contained in each clustering information set and the bid information of the initial clustering information contained in each clustering information set, and determining the clustering center corresponding to each piece of recommendation information in the recommendation information base according to the second distance.
7. An information recommendation apparatus, comprising:
a first recall acquisition unit configured to execute acquisition of first recall information from a preset recommendation information base in response to an information recommendation request of a target account;
a second recall acquisition unit configured to execute acquiring second recall information according to the historical recommendation information of the target account; the historical recommendation information is obtained according to historical recall information corresponding to a historical information recommendation request of the target account, and the historical information recommendation request is an information recommendation request triggered by the target account within a preset time period before the information recommendation request;
a candidate recommendation obtaining unit configured to perform obtaining candidate recommendation information from the first recall information and the second recall information;
and the target recommendation sending unit is configured to determine target recommendation information from the candidate recommendation information and send the target recommendation information to the target account.
8. A server, 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 one of claims 1 to 6.
9. A computer-readable storage medium, wherein instructions in the computer-readable storage medium, when executed by a processor of a server, enable the server to perform the information recommendation method of any one of claims 1 to 6.
10. A computer program product comprising instructions, which, when executed by a processor of a server, enable the server to carry out the information recommendation method according to any one of claims 1 to 6.
CN202111499713.3A 2021-12-09 2021-12-09 Information recommendation method and device, server and storage medium Pending CN114119106A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111499713.3A CN114119106A (en) 2021-12-09 2021-12-09 Information recommendation method and device, server and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111499713.3A CN114119106A (en) 2021-12-09 2021-12-09 Information recommendation method and device, server and storage medium

Publications (1)

Publication Number Publication Date
CN114119106A true CN114119106A (en) 2022-03-01

Family

ID=80364059

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111499713.3A Pending CN114119106A (en) 2021-12-09 2021-12-09 Information recommendation method and device, server and storage medium

Country Status (1)

Country Link
CN (1) CN114119106A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114785852A (en) * 2022-04-07 2022-07-22 北京达佳互联信息技术有限公司 Push content determining method and device, electronic equipment and storage medium
CN115689616A (en) * 2022-12-20 2023-02-03 陕西长锦网络科技有限公司 Cloud content pushing method and system based on big data characteristic analysis

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114785852A (en) * 2022-04-07 2022-07-22 北京达佳互联信息技术有限公司 Push content determining method and device, electronic equipment and storage medium
CN114785852B (en) * 2022-04-07 2024-04-09 北京达佳互联信息技术有限公司 Push content determining method and device, electronic equipment and storage medium
CN115689616A (en) * 2022-12-20 2023-02-03 陕西长锦网络科技有限公司 Cloud content pushing method and system based on big data characteristic analysis
CN115689616B (en) * 2022-12-20 2023-11-17 北京国联视讯信息技术股份有限公司 Cloud content pushing method and system based on big data feature analysis

Similar Documents

Publication Publication Date Title
US20190018900A1 (en) Method and Apparatus for Displaying Search Results
US10977322B2 (en) Systems and methods for recommending temporally relevant news content using implicit feedback data
US9706008B2 (en) Method and system for efficient matching of user profiles with audience segments
US10102307B2 (en) Method and system for multi-phase ranking for content personalization
US9270767B2 (en) Method and system for discovery of user unknown interests based on supplemental content
US10380649B2 (en) System and method for logistic matrix factorization of implicit feedback data, and application to media environments
US8874586B1 (en) Authority management for electronic searches
CN110909182B (en) Multimedia resource searching method, device, computer equipment and storage medium
US20150242750A1 (en) Asymmetric Rankers for Vector-Based Recommendation
CN112052387B (en) Content recommendation method, device and computer readable storage medium
US11100538B1 (en) Image recognition based content item selection
CN108153792B (en) Data processing method and related device
CN109409928A (en) A kind of material recommended method, device, storage medium, terminal
CN114119106A (en) Information recommendation method and device, server and storage medium
Symeonidis et al. Feature-weighted user model for recommender systems
CN110955831B (en) Article recommendation method and device, computer equipment and storage medium
Margaris et al. Improving collaborative filtering's rating prediction quality in dense datasets, by pruning old ratings
CN105718566A (en) Intelligent music recommendation system
CN111159563A (en) Method, device and equipment for determining user interest point information and storage medium
US9430572B2 (en) Method and system for user profiling via mapping third party interests to a universal interest space
CN113362109B (en) Recommendation method and device for media resources, electronic equipment and storage medium
CN112115354A (en) Information processing method, information processing apparatus, server, and storage medium
Zhang et al. An efficient collaborative filtering approach using smoothing and fusing
WO2012166958A1 (en) Determining matching degrees between information categories and displayed information
CN115391589A (en) Training method and device for content recall model, electronic equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination