CN115033782B - Object recommendation method, training method, device and equipment of machine learning model - Google Patents

Object recommendation method, training method, device and equipment of machine learning model Download PDF

Info

Publication number
CN115033782B
CN115033782B CN202210551859.6A CN202210551859A CN115033782B CN 115033782 B CN115033782 B CN 115033782B CN 202210551859 A CN202210551859 A CN 202210551859A CN 115033782 B CN115033782 B CN 115033782B
Authority
CN
China
Prior art keywords
recall
path
paths
objects
score
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.)
Active
Application number
CN202210551859.6A
Other languages
Chinese (zh)
Other versions
CN115033782A (en
Inventor
肖涛
石瑾
何晓辉
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 Baidu Netcom Science and Technology Co Ltd
Original Assignee
Beijing Baidu Netcom Science and 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 Baidu Netcom Science and Technology Co Ltd filed Critical Beijing Baidu Netcom Science and Technology Co Ltd
Priority to CN202210551859.6A priority Critical patent/CN115033782B/en
Publication of CN115033782A publication Critical patent/CN115033782A/en
Application granted granted Critical
Publication of CN115033782B publication Critical patent/CN115033782B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The disclosure provides a method for recommending objects, a training method of a machine learning model, a device and equipment, and relates to the field of artificial intelligence, in particular to an intelligent recommendation technology, a machine learning technology and a deep learning technology. The method comprises the following steps: determining a plurality of objects, each object recalled by at least one of a plurality of recall paths; determining a recall path score for each object for each recall path, the recall path score for each object for the corresponding at least one recall path being determined based on the recall path, the recall path score for the object for the other recall path being a predetermined value corresponding to the other recall path; determining a recall total score for each object based on the respective recall path score for the object for each recall path in the plurality of recall paths and the respective weights for the plurality of recall paths; and determining a plurality of first objects to be recommended based on the recall total scores of the plurality of objects.

Description

Object recommendation method, training method, device and equipment of machine learning model
Technical Field
The present disclosure relates to the field of artificial intelligence, and in particular, to an intelligent recommendation technique, a machine learning technique, and a deep learning technique, and more particularly, to a method for recommending an object, a training method for a machine learning model, an apparatus for recommending an object, a training apparatus for a machine learning model, an electronic device, a computer-readable storage medium, and a computer program product.
Background
Artificial intelligence is the subject of research that causes computers to simulate certain human mental processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.), both at the hardware level and at the software level. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing, and the like; the artificial intelligence software technology mainly comprises a computer vision technology, a voice recognition technology, a natural language processing technology, machine learning/deep learning, a big data processing technology, a knowledge map technology and the like.
The appearance and popularization of the internet bring a great deal of information to users, and the requirements of the users on the information in the information age are met, but the quantity of the information on the internet is greatly increased along with the rapid development of the network, so that the users cannot obtain the part of information which is really useful for the users when facing a great amount of information, and the use efficiency of the information is reduced on the contrary. In such a context, recommendation systems have come to mind.
The approaches described in this section are not necessarily approaches that have been previously conceived or pursued. Unless otherwise indicated, it should not be assumed that any of the approaches described in this section qualify as prior art merely by virtue of their inclusion in this section. Similarly, unless otherwise indicated, the problems mentioned in this section should not be considered as having been acknowledged in any prior art.
Disclosure of Invention
The present disclosure provides a method of recommending an object, a training method of a machine learning model, an apparatus of recommending an object, a training apparatus of a machine learning model, an electronic device, a computer-readable storage medium, and a computer program product.
According to an aspect of the present disclosure, there is provided a method of recommending an object, including: determining a plurality of objects recalled by a plurality of recall paths, wherein each object of the plurality of objects is recalled by at least one of the plurality of recall paths; determining a respective recall path score for each of the plurality of objects for each of the plurality of recall paths, wherein the respective recall path score for each of the plurality of objects for each of the corresponding at least one recall path is determined based on the recall path, and the respective recall path scores for other recall paths of the object other than the corresponding at least one recall path are at preset values corresponding to the other recall paths; for each object in the plurality of objects, determining a recall score for the object based on the respective recall path score for the object in each recall path in the plurality of recall paths and the respective weights for the plurality of recall paths; and determining a plurality of first objects to be recommended in the plurality of objects based on the recall total scores of the plurality of objects.
According to another aspect of the present disclosure, there is provided a training method of a machine learning model, including: determining a plurality of sample objects recalled through a plurality of recall paths and respective real click results of the plurality of sample objects, wherein each sample object in the plurality of sample objects is recalled through at least one recall path in the plurality of recall paths; determining a respective recall path score for each of the plurality of sample objects for each of the plurality of recall paths, wherein the respective recall path score for each of the plurality of sample objects for each of the corresponding at least one recall path is determined based on the recall path, and the respective other recall path score for the other recall path for the sample object outside of the corresponding at least one recall path is a preset value corresponding to the other recall path; for each sample object in the plurality of sample objects, inputting a respective recall path score for the sample object in each recall path in the plurality of recall paths into a machine learning model, wherein the machine learning model is configured to output a recall total score for the corresponding sample object based on the respective weights for the plurality of recall paths and the respective recall path score for the corresponding sample object in each recall path in the plurality of recall paths; calculating a loss value based on the recall total score of the sample object and the real click result of the sample object; and adjusting parameters of the machine learning model based on the loss values.
According to another aspect of the present disclosure, there is provided an apparatus for recommending an object, including: a first determining unit configured to determine a plurality of objects recalled through a plurality of recall paths, wherein each of the plurality of objects is recalled through at least one of the plurality of recall paths; a second determining unit configured to determine a respective recall path score of each of the plurality of objects on each of a plurality of recall paths, wherein the respective recall path score of each of the plurality of objects on each of a corresponding at least one recall path is determined based on the recall path, the respective recall path scores of the objects on other recall paths than the corresponding at least one recall path being a preset value corresponding to the other recall paths; a third determining unit configured to determine, for each of the plurality of objects, a recall total score of the object based on the corresponding recall path score of the object in each of the plurality of recall paths and the respective weights of the plurality of recall paths; and a fourth determination unit configured to determine a plurality of first objects to be recommended among the plurality of objects based on the recall total score of each of the plurality of objects.
According to another aspect of the present disclosure, there is provided a training apparatus for a machine learning model, including: a seventh determining unit configured to determine a plurality of sample objects recalled through a plurality of recall paths and respective real click results of the plurality of sample objects, wherein each of the plurality of sample objects is recalled through at least one of the plurality of recall paths; an eighth determining unit configured to determine a respective recall path score of each of the plurality of sample objects in each of the plurality of recall paths, wherein the respective recall path score of each of the plurality of sample objects in each of the corresponding at least one recall path is determined based on the recall path, and the respective recall path scores of the sample objects in other recall paths than the corresponding at least one recall path are preset values corresponding to the other recall paths; a machine learning model configured to output, for each of a plurality of sample objects, a recall total score for the sample object based on respective weights of the plurality of recall paths and a respective recall path score for the sample object at each of the plurality of recall paths; a calculating unit configured to calculate a loss value based on the recall total score of the sample object and the real click result of the sample object; and a parameter tuning unit configured to train the machine learning model based on the loss value.
According to another aspect of the present disclosure, there is provided an electronic device including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform the method described above.
According to another aspect of the present disclosure, there is provided a non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the above method.
According to another aspect of the disclosure, a computer program product is provided, comprising a computer program, wherein the computer program realizes the above method when executed by a processor.
According to one or more embodiments of the disclosure, by acquiring the recall path score of each object on each recall path and determining the recall total score of each object based on the recall path scores and the weight corresponding to each recall path, objects obtained from different recall paths can be compared according to the recall total score to obtain a globally optimal object set.
It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present disclosure, nor do they limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the embodiments and, together with the description, serve to explain the exemplary implementations of the embodiments. The illustrated embodiments are for purposes of illustration only and do not limit the scope of the claims. Throughout the drawings, identical reference numbers designate similar, but not necessarily identical, elements.
Fig. 1 illustrates a schematic diagram of an exemplary system in which various methods described herein may be implemented, in accordance with embodiments of the present disclosure;
FIG. 2 illustrates a flow chart of a method of recommending objects according to an exemplary embodiment of the present disclosure;
FIG. 3 illustrates a flowchart of a method of recommending objects according to an exemplary embodiment of the present disclosure;
FIG. 4 shows a flow diagram of a method of training a machine learning model according to an example embodiment of the present disclosure;
fig. 5 is a block diagram illustrating a structure of an apparatus for recommending an object according to an exemplary embodiment of the present disclosure;
fig. 6 is a block diagram illustrating a structure of an apparatus for recommending an object according to an exemplary embodiment of the present disclosure;
FIG. 7 shows a block diagram of a training apparatus for a machine learning model according to an example embodiment of the present disclosure; and
FIG. 8 sets forth a block diagram of exemplary electronic devices that can be used to implement embodiments of the present disclosure.
Detailed Description
Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of the embodiments of the disclosure are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, it will be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
In the present disclosure, unless otherwise specified, the use of the terms "first", "second", etc. to describe various elements is not intended to limit the positional relationship, the timing relationship, or the importance relationship of the elements, and such terms are used only to distinguish one element from another. In some examples, a first element and a second element may refer to the same instance of the element, and in some cases, based on the context, they may also refer to different instances.
The terminology used in the description of the various described examples in this disclosure is for the purpose of describing the particular examples only and is not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the elements may be one or more. Furthermore, the term "and/or" as used in this disclosure is intended to encompass any and all possible combinations of the listed items.
In the related art, in the existing multi-recall recommendation system, a large number of objects to be recommended can be obtained through a plurality of recall paths, but the number of objects which can be actually recommended to a user is very limited. The prior art can only determine a part of objects to be recommended from each recall path to be sent to the next stage of the recommendation system, or select the objects to be recommended of a part of recall paths and discard the objects to be recommended of other recall paths.
To solve the above problem, the present disclosure enables comparing objects obtained from different recall paths according to a recall score to obtain a globally optimal set of objects by obtaining a corresponding recall path score for each object in each recall path and determining a recall score for each object based on the recall path scores and a weight corresponding to each recall path.
Embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.
Fig. 1 illustrates a schematic diagram of an exemplary system 100 in which various methods and apparatus described herein may be implemented in accordance with embodiments of the present disclosure. Referring to fig. 1, the system 100 includes one or more client devices 101, 102, 103, 104, 105, and 106, a server 120, and one or more communication networks 110 coupling the one or more client devices to the server 120. Client devices 101, 102, 103, 104, 105, and 106 may be configured to execute one or more applications.
In embodiments of the present disclosure, the server 120 may run one or more services or software applications that enable the method of recommending objects to be performed.
In some embodiments, the server 120 may also provide other services or software applications that may include non-virtual environments and virtual environments. In certain embodiments, these services may be provided as web-based services or cloud services, such as provided to users of client devices 101, 102, 103, 104, 105, and/or 106 under a software as a service (SaaS) network.
In the configuration shown in fig. 1, server 120 may include one or more components that implement the functions performed by server 120. These components may include software components, hardware components, or a combination thereof, which may be executed by one or more processors. A user operating a client device 101, 102, 103, 104, 105, and/or 106 may, in turn, utilize one or more client applications to interact with the server 120 to take advantage of the services provided by these components. It should be understood that a variety of different system configurations are possible, which may differ from system 100. Accordingly, fig. 1 is one example of a system for implementing the various methods described herein and is not intended to be limiting.
The user may use client devices 101, 102, 103, 104, 105, and/or 106 to perform operations of the recommendation system front end. The client device may provide an interface that enables a user of the client device to interact with the client device, e.g., the user may utilize various inputs of the client to select pages, goods, multimedia, links, etc. objects of interest to the user. The client device may also output information to the user via the interface, e.g., the client may output to the user an object recommended by the recommendation system according to the user's information. Although fig. 1 depicts only six client devices, those skilled in the art will appreciate that any number of client devices may be supported by the present disclosure.
Client devices 101, 102, 103, 104, 105, and/or 106 may include various types of computer devices, such as portable handheld devices, general purpose computers (such as personal computers and laptops), workstation computers, wearable devices, smart screen devices, self-service terminal devices, service robots, gaming systems, thin clients, various messaging devices, sensors or other sensing devices, and so forth. These computer devices may run various types and versions of software applications and operating systems, such as MICROSOFT Windows, APPLE iOS, UNIX-like operating systems, linux, or Linux-like operating systems (e.g., GOOGLE Chrome OS); or include various Mobile operating systems such as MICROSOFT Windows Mobile OS, iOS, windows Phone, android. Portable handheld devices may include cellular telephones, smart phones, tablets, personal Digital Assistants (PDAs), and the like. Wearable devices may include head-mounted displays (such as smart glasses) and other devices. The gaming system may include a variety of handheld gaming devices, internet-enabled gaming devices, and the like. The client device is capable of executing a variety of different applications, such as various Internet-related applications, communication applications (e.g., email applications), short Message Service (SMS) applications, and may use a variety of communication protocols.
Network 110 may be any type of network known to those skilled in the art that may support data communications using any of a variety of available protocols, including but not limited to TCP/IP, SNA, IPX, etc. By way of example only, one or more networks 110 may be a Local Area Network (LAN), an ethernet-based network, a token ring, a Wide Area Network (WAN), the internet, a virtual network, a Virtual Private Network (VPN), an intranet, an extranet, a Public Switched Telephone Network (PSTN), an infrared network, a wireless network (e.g., bluetooth, WIFI), and/or any combination of these and/or other networks.
The server 120 may include one or more general purpose computers, special purpose server computers (e.g., PC (personal computer) servers, UNIX servers, mid-end servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and/or combination. The server 120 may include one or more virtual machines running a virtual operating system, or other computing architecture involving virtualization (e.g., one or more flexible pools of logical storage that may be virtualized to maintain virtual storage for the server). In various embodiments, the server 120 may run one or more services or software applications that provide the functionality described below.
The computing units in server 120 may run one or more operating systems including any of the operating systems described above, as well as any commercially available server operating systems. The server 120 may also run any of a variety of additional server applications and/or middle tier applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, and the like.
In some implementations, the server 120 can include one or more applications to analyze and consolidate data feeds and/or event updates received from users of the client devices 101, 102, 103, 104, 105, and 106. Server 120 may also include one or more applications to display data feeds and/or real-time events via one or more display devices of client devices 101, 102, 103, 104, 105, and 106.
In some embodiments, the server 120 may be a server of a distributed system, or a server incorporating a blockchain. The server 120 may also be a cloud server, or a smart cloud computing server or a smart cloud host with artificial intelligence technology. The cloud Server is a host product in a cloud computing service system, and is used for solving the defects of high management difficulty and weak service expansibility in the conventional physical host and Virtual Private Server (VPS) service.
The system 100 may also include one or more databases 130. In some embodiments, these databases may be used to store data and other information. For example, one or more of the databases 130 may be used to store information such as audio files and video files. The data store 130 may reside in various locations. For example, the data store used by the server 120 may be local to the server 120, or may be remote from the server 120 and may communicate with the server 120 via a network-based or dedicated connection. The data store 130 may be of different types. In certain embodiments, the data store used by the server 120 may be a database, such as a relational database. One or more of these databases may store, update, and retrieve data to and from the databases in response to the commands.
In some embodiments, one or more of the databases 130 may also be used by applications to store application data. The databases used by the application may be different types of databases, such as key-value stores, object stores, or conventional stores supported by a file system.
The system 100 of fig. 1 may be configured and operated in various ways to enable application of the various methods and apparatus described in accordance with the present disclosure.
According to an aspect of the present disclosure, a method of recommending an object is provided. As shown in fig. 2, the method of recommending an object includes: step S201, determining a plurality of objects recalled through a plurality of recall paths, wherein each object in the plurality of objects is recalled through at least one recall path in the plurality of recall paths; step S202, determining a recall path score of each object in a plurality of recall paths, wherein the recall path score of each object in the plurality of objects in each recall path in the corresponding at least one recall path is determined based on the recall path, and the recall path score of the object in other recall paths except the corresponding at least one recall path is a preset value corresponding to the other recall paths; step S203, aiming at each object in the plurality of objects, determining a recall total score of the object based on the corresponding recall path score of the object in each recall path in the plurality of recall paths and the respective weight of the plurality of recall paths; and step S204, determining a plurality of first objects to be recommended in the plurality of objects based on the recall total scores of the plurality of objects.
Therefore, by acquiring the corresponding recall path score of each object in each recall path and determining the recall total score of each object based on the recall path scores and the weight corresponding to each recall path, the objects obtained from different recall paths can be compared according to the recall total score to obtain the globally optimal object set.
The method of the present disclosure may be used in various recommendation scenarios, for example, a search engine, information flow, online video browsing, online shopping, and the like, which are not limited herein.
In step S201, a plurality of objects recalled through a plurality of recall paths are determined.
According to some embodiments, the plurality of recall paths may include recall paths based on various recall algorithms, such as, for example, user-based collaborative filtering, item-based collaborative filtering, matching networks (MatchNet), business rule-based recalls, or other recall paths, not limited herein. Each recall path may recall a certain number of objects, and the recalled objects may come from multiple recall paths simultaneously. It is to be appreciated that the plurality of recall paths are recalled in accordance with the target user or a target object associated with the target user (e.g., a video, an item, etc. being viewed by the target user) to obtain a plurality of objects.
In step S202, a respective recall path score for each of the plurality of objects in each of the plurality of recall paths is determined.
According to some embodiments, a respective recall path score for each object of the plurality of objects for each recall path of the corresponding at least one recall path may be determined based on the recall path. During the recall process, each recall path can generate a recall score, i.e., a recall path score, for the object recalled for that path. This score can reflect the degree of association between the recalled object and the target user/target object, or the "quality" of the recalled object itself.
According to some embodiments, the plurality of recall paths may include at least one of: utilizing a first recall path based on collaborative filtering of users, a respective recall path score of a first object recalled by the first recall path at the first recall path may indicate a degree of similarity between the target user and a user associated with the first object; a second recall path utilizing collaborative filtering based on the items, a respective recall path score for a second object recalled by the second recall path at the second recall path may be indicative of a similarity between the target object and the second object; and a third recall path based on the business rules, a respective recall path score for a third object recalled by the third recall path at the third recall path may indicate at least one of a click volume, timeliness, and author rating of the third object. It will be appreciated that the plurality of recall paths may also include other recall paths that each determine a respective recall path score for objects recalled in the path.
And recommending the object preferred by the user similar to the target user by calculating the similarity between the target user and other users based on the collaborative filtering of the user. The recall path score may be, for example, a similarity between the target user and the user associated with the recalled first object, or other values derived based on the similarity.
And recommending the object similar to the target object to a recommendation list, page or user corresponding to the target object by calculating the similarity of the target object and other objects based on the collaborative filtering of the article. The recall path score may be, for example, a similarity between the target object and the recalled second object, or another value obtained based on the similarity.
The recalling based on the business rule can be based on the attribute of the object to be recommended or the attribute of the target object/target user. In one exemplary embodiment, the recall path score may be determined based on at least one of click through volume, timeliness, and author rating of the object to be recommended. The higher the click quantity of the object to be recommended, the stronger the timeliness, or the higher the author grade, the higher the recall path score. It is understood that the object to be recommended may also include other attributes that can be used to determine the recall path score, which are not limited herein. In an exemplary embodiment, the recall path score may also be determined according to relevant information of the target user, or information related to both the object to be recommended and the target user, such as a spatial distance between the target user and the object to be recommended. It can be understood that the corresponding recall path score determination mode may be set based on the business rule according to the requirement, and is not limited herein.
According to some embodiments, the respective recall path score of each object of the plurality of objects for other recall paths other than the corresponding at least one recall path may be a preset value corresponding to the other recall paths. For each object in the plurality of objects, a default recall path score may be set for recall paths that do not recall the object, such that each object has a score on each recall path. In some embodiments, the preset value of the corresponding recall path score of the other recall path may be zero or other values, and may also be set according to the recall path score of the object recalled by the recall path (e.g., an average value, a median value, etc. of the recall path scores of the object recalled by the recall path). It can be understood that different recall routes may have the same preset value of the recall route score, or may have different preset values of the recall route score, which is not limited herein.
In step S203, for each of the plurality of objects, a recall total score of the object is determined based on the corresponding recall path score of the object in each of the plurality of recall paths and the respective weights of the plurality of recall paths.
According to some embodiments, there may be a mapping between the respective recall path scores of different recall paths. In one exemplary embodiment, the posterior results of the corresponding recall path scores (user similarity) for the first recall path based on the collaborative filtering by the user between 0.7-0.8 and the corresponding recall path scores (number of clicks on object) for the third recall path based on the business rules between 20-50 are substantially consistent. That is, from the user behavior data, it can be derived that the probability of the user clicking on the first object between the respective recall path scores of the first recall path and 0.7-0.8 is close to the probability between the respective recall path scores of the third recall path and 20-50. Thus, the first recall path and the third recall path may have respective weights such that the recall score corresponding to the segment having the respective recall path score of 0.7-0.8 for the first recall path is able to coincide as much as possible with the recall score corresponding to the segment having the respective recall path score of 20-50 for the third recall path. In this way, a way of comparing recall scores of different recall paths is provided, so that a globally optimal set of objects can be determined based on the recall path scores of the objects recalled by the plurality of recall paths at the respective recall paths.
According to some embodiments, the step S203, for each of the plurality of objects, determining the recall total score of the object based on the corresponding recall path score of the object in each of the plurality of recall paths and the respective weights of the plurality of recall paths, may include: and inputting the corresponding recall path score of each object in the plurality of objects in each recall path in the plurality of recall paths into the trained machine learning model to obtain the recall total score of the object output by the machine learning model. The machine learning model may be configured to output a recall score for the corresponding object based on the respective weights of the plurality of recall paths and the corresponding recall path score for the corresponding object at each of the plurality of recall paths. The weights of the plurality of recall paths are obtained by training a machine learning model using the plurality of sample objects recalled through the plurality of recall paths and the true click results of the plurality of sample objects.
Thus, by using the machine learning model to obtain the weight of each recall path, the mapping relationship between scores of different recall paths can be learned based on a large amount of user data, so that the recall total score output by the machine learning model can evaluate whether the object is globally optimal.
According to some embodiments, inputting the respective recall path score for each of the plurality of objects in each of the plurality of recall paths into the trained machine learning model may comprise: normalizing the respective recall path score for each object of the plurality of objects in each recall path of the plurality of recall paths; and inputting the normalized respective recall path score for the subject at each of the plurality of recall paths into the trained machine learning model. The machine learning model may be further configured to output a recall total score for the corresponding object based on the respective weights for the plurality of recall paths and the normalized respective recall path score for the corresponding object for each of the plurality of recall paths. Therefore, the learning difficulty of the model can be reduced and the learned weight value is more accurate by performing normalization before inputting the machine learning model.
According to some embodiments, the machine learning model may be a linear regression model. The linear regression model is one of the simplest machine learning models, and thus by using the linear regression model as a model for calculating the recall total score, accurate weights corresponding to each recall path can be obtained with less cost (training cost, inference cost).
According to some embodiments, the machine learning model may output a number between 0-1, with higher values indicating better relevance of the object to the target user/target object.
According to some embodiments, the machine learning model may be updated in response to determining that at least one recall route of the plurality of recall routes satisfies a preset condition. The preset condition may be, for example, an update (e.g., increase, decrease, change of a feature, etc.) of object information in the database or an update of user information, or may be another condition. Due to the high updating frequency of the data in the recommendation system, the relationship between the score of each recall path and the final recommendation result is changed, so that the real-time performance of the weight of the corresponding recall path score of each recall path can be ensured by updating the machine learning model after determining that at least one recall path meets the preset condition.
In step S204, a plurality of first objects to be recommended are determined among the plurality of objects based on the recall total score of each of the plurality of objects.
After the machine learning model, a plurality of objects recalled by a plurality of recall paths all have a comparable score, and a globally optimal resource set, namely at least one first object to be recommended, can be obtained by taking from high to low according to the score. These objects may be recommended directly to the user or may be further processed for preference.
In some embodiments, the plurality of first objects to be recommended may be determined based on the recall total score of each of the plurality of objects in various manners, for example, a preset number of objects with the highest recall total score may be used as the first objects to be recommended, a preset proportion of objects with the highest recall total score may be used as the first objects to be recommended, or the first objects to be recommended may be determined in other manners, which is not limited herein.
According to some embodiments, as shown in fig. 3, the method of recommending an object may further include: step S305, inputting at least one characteristic of each first object to be recommended in the plurality of first objects to be recommended into a recommended ranking model to obtain a ranking score of each first object to be recommended; and step S306, determining at least one second object to be recommended in the plurality of first objects to be recommended based on the respective ranking scores of the plurality of first objects to be recommended. The operations of steps S301 to S304 in fig. 3 are similar to the operations of steps S201 to S204 in fig. 2, and are not limited herein.
Although the objects to be recommended can be sorted and filtered by the recall total score, the machine learning model described above uses only the corresponding recall path score of the object as the basis for reasoning, and such sorting and filtering is very rough. However, the number of objects to be recommended can be reduced quickly and efficiently using this method.
We now consider the ranking model in the recommendation system again. The recommendation ranking model is used to refine a large number of objects recalled in a recommendation system, and inferences are typically made based on the specific characteristics of these objects. Such a model can provide accurate scores for objects, thereby determining a globally optimal few objects for recommendation to a user, but cannot rank an excessive number of objects to be recommended. Therefore, in a multi-recall recommendation system, it may happen that too many objects are recalled by multiple recall paths, and the crowd ordering model does not have enough computing resources to order the objects.
Therefore, after ranking is performed based on the recall total score of each of the plurality of objects, the plurality of first objects to be recommended indicated by the ranking result can be input into the recommendation ranking model for fine ranking. By the method, the problem that the machine learning model is in a coarse sorting mode is solved, the problem that the recommending sorting model cannot process objects to be recommended in an overlarge quantity is solved, the recommending result which is more fit with the target object or the target user can be output, and the user experience is improved.
After recommending the ranking model, at least one second object to be recommended may be exposed to present the recommendation to the user.
According to another aspect of the present disclosure, a method of training a machine learning model is provided. As shown in fig. 4, the method includes: step S401, determining a plurality of sample objects recalled through a plurality of recall paths and respective real click results of the plurality of sample objects, wherein each sample object in the plurality of sample objects is recalled through at least one recall path in the plurality of recall paths; step S402, determining a recall path score of each sample object in a plurality of recall paths, wherein the recall path score of each sample object in the plurality of sample objects in each recall path in the corresponding at least one recall path is determined based on the recall path, and the recall path scores of the sample objects in other recall paths than the corresponding at least one recall path are preset values corresponding to the other recall paths; step S403, inputting, for each of the plurality of sample objects, a corresponding recall path score of the sample object in each of the plurality of recall paths into a machine learning model, wherein the machine learning model is configured to output a recall total score of the corresponding sample object based on respective weights of the plurality of recall paths and the corresponding recall path score of the corresponding sample object in each of the plurality of recall paths; step S404, calculating a loss value based on the recall total score of the sample object and the real click result of the sample object; and step S405, training the machine learning model based on the loss value. The operations of steps S401 to S403 in fig. 4 are similar to the operations of steps S201 to S203 in fig. 2, and are not repeated herein.
Therefore, the machine learning model is trained by using real sample data of a user, so that the machine learning model can learn the weight of each recall path, the capability of determining the recall total score of an object based on the recall path score of the object in each recall path is provided, and objects obtained from different recall paths can be compared according to the recall total score to obtain a globally optimal object set.
In some embodiments, the plurality of sample objects and the real click results of each of the plurality of sample objects may be, for example, history objects previously recommended to the target user by the recommendation system and real click results of whether the target user clicked on the history objects. The recommendation system may record this information during the recommendation process, along with the recall path from which each object came and the corresponding recall path score.
According to some embodiments, the plurality of recall paths may include at least one of: utilizing a first recall path based on collaborative filtering of users, a respective recall path score of a first sample object recalled by the first recall path at the first recall path may indicate a similarity between a target user and a target user associated with the first sample object; a second recall path utilizing collaborative filtering based on the items, a respective recall path score for a second sample object recalled by the second recall path at the second recall path may be indicative of a similarity between the target object and the second sample object; and utilizing a third recall path based on the business rules, a respective recall path score for a third sample object recalled by the third recall path at the third recall path may indicate at least one of an amount of clicks, a timeliness, and an author rating of the third sample object.
According to some embodiments, the step S403, for each of the plurality of sample objects, inputting the corresponding recall path score of the sample object in each of the plurality of recall paths into the machine learning model, may include: normalizing the respective recall path score for each of the plurality of sample objects for each of the plurality of recall paths; and inputting the normalized respective recall path score for the sample object at each of the plurality of recall paths into the trained machine learning model. The machine learning model may be further configured to output a recall total score for the corresponding sample object based on the respective weights for the plurality of recall paths and the normalized respective recall path score for the corresponding sample object for each of the plurality of recall paths.
It is understood that, when performing steps S404-S405, an appropriate loss function may be determined according to the requirement, so as to calculate a loss value based on the recall total score of the sample object and the real click result of the sample object, and adjust the parameters of the model based on the loss value, which is not limited herein. In some embodiments, training or adjusting parameters of the machine learning model may include: the weights of each of the plurality of recall paths are adjusted.
According to some embodiments, the machine learning model may be a linear regression model.
According to another aspect of the present disclosure, an apparatus for recommending an object is provided. As shown in fig. 5, the apparatus 500 includes: a first determining unit 510 configured to determine a plurality of objects recalled through a plurality of recall paths, wherein each object of the plurality of objects is recalled through at least one recall path of the plurality of recall paths; a second determining unit 520 configured to determine a respective recall path score of each object of the plurality of objects in each recall path of the plurality of recall paths, wherein the respective recall path score of each object of the plurality of objects in each recall path of the corresponding at least one recall path is determined based on the recall path, and the respective recall path scores of the objects in other recall paths than the corresponding at least one recall path are preset values corresponding to the other recall paths; a third determining unit 530 configured to determine, for each object of the plurality of objects, a recall total score of the object based on the corresponding recall path score of the object in each recall path of the plurality of recall paths and the respective weights of the plurality of recall paths; and a fourth determining unit 540 configured to determine a plurality of first objects to be recommended among the plurality of objects based on the recall total score of each of the plurality of objects. The operations of the units 510-530 in the apparatus 500 are similar to the operations of the steps S201-S203 in fig. 2, and are not described herein again.
According to some embodiments, the plurality of recall paths may include at least one of: utilizing a first recall path based on collaborative filtering of users, a respective recall path score of a first object recalled by the first recall path at the first recall path may indicate a degree of similarity between the target user and a user associated with the first object; a second recall path utilizing collaborative filtering based on the items, a respective recall path score for a second object recalled by the second recall path at the second recall path may be indicative of a similarity between the target object and the second object; and a third recall path based on the business rules, a respective recall path score for a third object recalled by the third recall path at the third recall path may indicate at least one of a click volume, timeliness, and author rating of the third object.
According to some embodiments, the third determining unit may include: a machine learning model configured to output a recall score for a corresponding object based on respective weights of the plurality of recall paths obtained by training the machine learning model with real click results of a plurality of sample objects and a plurality of sample objects recalled through the plurality of recall paths and a corresponding recall path score for the corresponding object at each of the plurality of recall paths.
According to some embodiments, the machine learning model may be further configured to: normalizing the respective recall path score of the corresponding object for each of the plurality of recall paths; and outputting a recall score for the corresponding object based on the respective weights of the plurality of recall paths and the normalized respective recall path score for the corresponding object for each of the plurality of recall paths.
According to some embodiments, the machine learning model may be a linear regression model.
According to some embodiments, the machine learning model may be updated in response to determining that at least one recall route of the plurality of recall routes satisfies a preset condition.
According to some embodiments, as shown in fig. 6, the apparatus 600 may further include: a fifth determining unit 650, configured to input at least one feature of each of the plurality of first objects to be recommended into the recommendation ranking model to obtain a ranking score of each of the plurality of first objects to be recommended; and a sixth determining unit 660 configured to determine at least one second object to be recommended among the plurality of first objects to be recommended based on the respective ranking scores of the plurality of first objects to be recommended.
According to another aspect of the present disclosure, a training apparatus for a machine learning model is provided. As shown in fig. 7, the apparatus 700 includes: a seventh determining unit 710 configured to determine a plurality of sample objects recalled through a plurality of recall paths and a real click result of each of the plurality of sample objects, wherein each of the plurality of sample objects is recalled through at least one of the plurality of recall paths; an eighth determining unit 720, configured to determine a respective recall path score of each of the plurality of sample objects in each of the plurality of recall paths, wherein the respective recall path score of each of the plurality of sample objects in each of the corresponding at least one recall path is determined based on the recall path, and the respective recall path scores of the sample objects in other recall paths than the corresponding at least one recall path are preset values corresponding to the other recall paths; a machine learning model 730 configured to, for each of a plurality of sample objects, output a recall total score for the sample object based on a respective weight of the plurality of recall paths and a respective recall path score for the sample object for each of the plurality of recall paths; a calculating unit 740 configured to calculate a loss value based on the recall total score of the sample object and the real click result of the sample object; and a parameter tuning unit 750 configured to train the machine learning model based on the loss value. The operations of the units 710 to 750 in the apparatus 700 are similar to those of the steps S401 to S405 in fig. 4, and are not described herein again.
According to some embodiments, a respective recall path score for each recall path of each sample object of the plurality of sample objects in the corresponding at least one recall path may be determined based on the recall path, and the respective recall path score for the recall path of the sample object outside the corresponding at least one recall path may be a preset value.
According to some embodiments, the plurality of recall paths may include at least one of: utilizing a first recall path based on collaborative filtering of users, a respective recall path score of a first sample object recalled by the first recall path at the first recall path may indicate a similarity between a target user and a target user associated with the first sample object; a second recall path utilizing collaborative filtering based on the items, a respective recall path score for a second sample object recalled by the second recall path at the second recall path may be indicative of a similarity between the target object and the second sample object; and utilizing a third recall path based on the business rules, a respective recall path score for a third sample object recalled by the third recall path at the third recall path may indicate at least one of an amount of clicks, a timeliness, and an author rating of the third sample object.
According to some embodiments, the machine learning model may be further configured to: normalizing the respective recall path score for each of the plurality of sample objects for each of the plurality of recall paths; and outputting a recall total score for the sample object based on the respective weights of the plurality of recall paths and the normalized corresponding recall path score for the sample object for each of the plurality of recall paths
According to some embodiments, the machine learning model may be a linear regression model.
In the technical scheme of the disclosure, the collection, storage, use, processing, transmission, provision, disclosure and other processing of the personal information of the related user are all in accordance with the regulations of related laws and regulations and do not violate the good customs of the public order.
According to an embodiment of the present disclosure, there is also provided an electronic device, a readable storage medium, and a computer program product.
Referring to fig. 8, a block diagram of a structure of an electronic device 800, which may be a server or a client of the present disclosure, which is an example of a hardware device that may be applied to aspects of the present disclosure, will now be described. Electronic device is intended to represent various forms of digital electronic computer devices, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 8, the apparatus 800 includes a computing unit 801 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 802 or a computer program loaded from a storage unit 808 into a Random Access Memory (RAM) 803. In the RAM803, various programs and data required for the operation of the device 800 can also be stored. The calculation unit 801, the ROM 802, and the RAM803 are connected to each other by a bus 804. An input/output (I/O) interface 805 is also connected to bus 804.
A number of components in the device 800 are connected to the I/O interface 805, including: an input unit 806, an output unit 807, a storage unit 808, and a communication unit 809. The input unit 806 may be any type of device capable of inputting information to the device 800, and the input unit 806 may receive input numeric or character information and generate key signal inputs related to user settings and/or function controls of the electronic device, and may include, but is not limited to, a mouse, a keyboard, a touch screen, a track pad, a track ball, a joystick, a microphone, and/or a remote control. Output unit 807 can be any type of device capable of presenting information and can include, but is not limited to, a display, speakers, a video/audio output terminal, a vibrator, and/or a printer. The storage unit 808 may include, but is not limited to, a magnetic disk, an optical disk. The communication unit 809 allows the device 800 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers and/or chipsets, such as bluetooth (TM) devices, 802.11 devices, wiFi devices, wiMax devices, cellular communication devices, and/or the like.
Computing unit 801 may be a variety of general and/or special purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various dedicated Artificial Intelligence (AI) computing chips, various computing units running machine learning network algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, and so forth. The calculation unit 801 executes the respective methods and processes described above, such as the method of recommending an object and the training method of a machine learning model. For example, in some embodiments, the method of recommending objects and the method of training a machine learning model may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program can be loaded and/or installed onto device 800 via ROM 802 and/or communications unit 809. When loaded into RAM803 and executed by computing unit 801, a computer program may perform one or more steps of the method of recommending objects and the method of training a machine learning model described above. Alternatively, in other embodiments, the computing unit 801 may be configured in any other suitable way (e.g., by means of firmware) to perform the method of recommending objects and the training method of the machine learning model.
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), system on a chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, 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 compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, 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: local Area Networks (LANs), wide Area Networks (WANs), and the Internet.
The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The Server can be a cloud Server, also called a cloud computing Server or a cloud host, and is a host product in a cloud computing service system, so as to solve the defects of high management difficulty and weak service expansibility in the traditional physical host and VPS service ("Virtual Private Server", or simply "VPS"). The server may also be a server of a distributed system, or a server incorporating a blockchain.
It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present disclosure may be performed in parallel, sequentially or in different orders, and are not limited herein as long as the desired results of the technical solutions disclosed in the present disclosure can be achieved.
While embodiments or examples of the present disclosure have been described with reference to the accompanying drawings, it is to be understood that the above-described methods, systems and apparatus are merely illustrative embodiments or examples and that the scope of the invention is not to be limited by these embodiments or examples, but only by the claims as issued and their equivalents. Various elements in the embodiments or examples may be omitted or may be replaced with equivalents thereof. Further, the steps may be performed in an order different from that described in the present disclosure. Further, various elements in the embodiments or examples may be combined in various ways. It is important that as technology evolves, many of the elements described herein may be replaced with equivalent elements that appear after the present disclosure.

Claims (24)

1. A method of recommending an object, comprising:
determining a plurality of objects recalled by a plurality of recall paths, wherein each object of the plurality of objects was recalled by at least one recall path of the plurality of recall paths;
determining a respective recall path score for each of the plurality of objects for each of the plurality of recall paths, wherein the respective recall path score for each of the plurality of objects for each of the corresponding at least one recall path is determined based on the recall path, and the respective recall path scores for other recall paths of the object outside of the corresponding at least one recall path are at preset values corresponding to the other recall paths;
for each of the plurality of objects, determining a recall score for the object based on the object's respective recall path score for each of the plurality of recall paths and the respective weights of the plurality of recall paths; and
determining a plurality of first objects to be recommended in the plurality of objects based on the recall total scores of the plurality of objects respectively.
2. The method of claim 1, wherein determining, for each of the plurality of objects, a recall score for the object based on the object's respective recall path score for each of the plurality of recall paths and the respective weights of the plurality of recall paths comprises:
inputting a respective recall path score for each of the plurality of objects in each of the plurality of recall paths into a trained machine learning model, wherein the machine learning model is configured to output a recall total score for a corresponding object based on respective weights of the plurality of recall paths obtained by training the machine learning model using a plurality of sample objects recalled through a plurality of recall paths and real click results of the plurality of sample objects and corresponding recall path scores of the corresponding object in each of the plurality of recall paths.
3. The method of claim 2, wherein inputting the respective recall path score for each of the plurality of objects in each of the plurality of recall paths into a trained machine learning model comprises:
normalizing a respective recall path score for each object of the plurality of objects at each recall path of the plurality of recall paths; and
inputting the normalized respective recall pathway score of the object for each of the plurality of recall pathways into the trained machine learning model, wherein the machine learning model is further configured to output a recall total score for the corresponding object based on the respective weights for the plurality of recall pathways and the normalized respective recall pathway score for the corresponding object for each of the plurality of recall pathways.
4. The method of any of claims 1-3, further comprising:
inputting at least one characteristic of each first object to be recommended in the plurality of first objects to be recommended into a recommendation ranking model to obtain a ranking score of each first object to be recommended; and
and determining at least one second object to be recommended in the plurality of first objects to be recommended based on the respective ranking scores of the plurality of first objects to be recommended.
5. The method of any of claims 1-3, wherein the plurality of recall paths includes at least one of:
a first recall path utilizing user-based collaborative filtering, a first object recalled by the first recall path having a respective recall path score at the first recall path indicative of a degree of similarity between a target user and a user associated with the first object;
a second recall path utilizing collaborative filtering based on items, a respective recall path score of a second object recalled by the second recall path at the second recall path indicating a degree of similarity between a target object and the second object; and
a third recall path based on business rules, a respective recall path score of a third object recalled by the third recall path at the third recall path indicating at least one of an amount of clicks, a timeliness, and an author rating of the third object.
6. The method of claim 2, wherein the machine learning model is a linear regression model.
7. The method of claim 2, wherein the machine learning model is updated in response to determining that at least one of the plurality of recall paths satisfies a preset condition.
8. A method of training a machine learning model, comprising:
determining a plurality of sample objects recalled through a plurality of recall paths and respective real click results of the plurality of sample objects, wherein each sample object of the plurality of sample objects is recalled through at least one recall path of the plurality of recall paths;
determining a respective recall path score for each of the plurality of sample objects for each of the plurality of recall paths, wherein the respective recall path score for each of the plurality of sample objects for each of the corresponding at least one recall path is determined based on the recall path, and the respective recall path scores for other recall paths of the sample objects outside the corresponding at least one recall path are at preset values corresponding to the other recall paths;
for each of the plurality of sample objects, inputting a respective recall path score for the sample object in each of the plurality of recall paths into a machine learning model, wherein the machine learning model is configured to output a recall score for the corresponding sample object based on the respective weights of the plurality of recall paths and the respective recall path score for the corresponding sample object in each of the plurality of recall paths;
calculating a loss value based on the recall total score of the sample object and the real click result of the sample object; and
training the machine learning model based on the loss value.
9. The method of claim 8, wherein inputting, for each of the plurality of sample objects, the respective recall path score for that sample object in each of the plurality of recall paths into a machine learning model comprises:
normalizing a respective recall path score for each of the plurality of sample objects at each of the plurality of recall paths; and
inputting the normalized respective recall path score for the sample object for each of the plurality of recall paths into the machine learning model, wherein the machine learning model is further configured to output a recall total score for the corresponding sample object based on the respective weights for the plurality of recall paths and the normalized respective recall path score for the corresponding sample object for each of the plurality of recall paths.
10. The method of claim 8, wherein the plurality of recall paths comprises at least one of:
a first recall path utilizing user-based collaborative filtering, a first sample object recalled by the first recall path indicating a degree of similarity between a target user and a target user associated with the first sample object at a respective recall path score of the first recall path;
a second recall path utilizing collaborative filtering based on items, a respective recall path score of a second sample object recalled by the second recall path at the second recall path indicating a degree of similarity between a target object and the second sample object; and
utilizing a third recall path based on business rules, a respective recall path score of a third sample object recalled by the third recall path at the third recall path indicating at least one of an amount of clicks, a timeliness, and an author rating of the third sample object.
11. The method of any of claims 8-10, wherein the machine learning model is a linear regression model.
12. An apparatus for recommending an object, comprising:
a first determining unit configured to determine a plurality of objects recalled through a plurality of recall paths, wherein each object of the plurality of objects is recalled through at least one recall path of the plurality of recall paths;
a second determining unit configured to determine a respective recall route score of each of the plurality of objects on each of the plurality of recall routes, wherein the respective recall route score of each of the plurality of objects on each of the corresponding at least one recall route is determined based on the recall route, and the respective recall route scores of the objects on other recall routes than the corresponding at least one recall route are preset values corresponding to the other recall routes;
a third determining unit configured to determine, for each of the plurality of objects, a recall total score for the object based on the respective recall path score of the object in each of the plurality of recall paths and the respective weights of the plurality of recall paths; and
a fourth determining unit configured to determine a plurality of first objects to be recommended among the plurality of objects based on the recall total score of each of the plurality of objects.
13. The apparatus of claim 12, wherein the third determining unit comprises:
a machine learning model configured to output a recall score for a corresponding object based on respective weights of the plurality of recall paths and the corresponding recall path score for the corresponding object for each of the plurality of recall paths, wherein the respective weights of the plurality of recall paths are derived by training the machine learning model with real click results for a plurality of sample objects and the plurality of sample objects recalled through a plurality of recall paths.
14. The apparatus of claim 13, wherein the machine learning model is further configured to:
normalizing a respective recall path score of the corresponding object for each of the plurality of recall paths; and
outputting a recall score for the corresponding object based on the respective weights for the plurality of recall paths and the normalized respective recall path score for the corresponding object for each of the plurality of recall paths.
15. The apparatus of claim 12, further comprising:
a fifth determining unit, configured to input at least one characteristic of each of the plurality of first objects to be recommended into the recommendation ranking model to obtain a ranking score of each of the plurality of first objects to be recommended; and
a sixth determining unit configured to determine at least one second object to be recommended among the plurality of first objects to be recommended based on the respective ranking scores of the plurality of first objects to be recommended.
16. The apparatus of claim 12, wherein the plurality of recall paths comprise at least one of:
a first recall path utilizing user-based collaborative filtering, a first object recalled by the first recall path having a respective recall path score at the first recall path indicative of a degree of similarity between a target user and a user associated with the first object;
a second recall path utilizing collaborative filtering based on items, a respective recall path score of a second object recalled by the second recall path at the second recall path indicating a degree of similarity between a target object and the second object; and
a third recall path based on business rules, a respective recall path score of a third object recalled by the third recall path at the third recall path indicating at least one of an amount of clicks, a timeliness, and an author rating of the third object.
17. The apparatus of claim 13, wherein the machine learning model is a linear regression model.
18. The apparatus of claim 13, wherein the machine learning model is updated in response to determining that at least one of the plurality of recall paths satisfies a preset condition.
19. A training apparatus for a machine learning model, comprising:
a seventh determining unit configured to determine a plurality of sample objects recalled through a plurality of recall paths and real click results of the respective plurality of sample objects, wherein each of the plurality of sample objects is recalled through at least one of the plurality of recall paths;
an eighth determining unit configured to determine a recall path score for each of the plurality of sample objects in each of the plurality of recall paths, wherein the respective recall path score for each of the plurality of sample objects in each of the corresponding at least one recall path is determined based on the recall path, and the respective recall path scores for other sample objects in the recall paths other than the corresponding at least one recall path are preset values corresponding to the other recall paths;
a machine learning model configured to output, for each of the plurality of sample objects, a recall total score for the sample object based on the respective weights of the plurality of recall paths and the respective recall path score for the sample object at each of the plurality of recall paths;
a calculating unit configured to calculate a loss value based on the recall total score of the sample object and the real click result of the sample object; and
a parameter tuning unit configured to train the machine learning model based on the loss value.
20. The apparatus of claim 19, wherein the machine learning model is further configured to:
normalizing a respective recall path score for each of the plurality of sample objects at each of the plurality of recall paths; and
and outputting the recall total score of the sample object based on the respective weights of the plurality of recall paths and the normalized corresponding recall path score of the sample object in each recall path of the plurality of recall paths.
21. The apparatus of claim 19, wherein the plurality of recall paths comprise at least one of:
a first recall path utilizing user-based collaborative filtering, a first sample object recalled by the first recall path having a respective recall path score at the first recall path indicative of a similarity between a target user and a target user associated with the first sample object;
a second recall path utilizing collaborative filtering based on items, a respective recall path score of a second sample object recalled by the second recall path at the second recall path indicating a degree of similarity between a target object and the second sample object; and
utilizing a third recall path based on business rules, a respective recall path score of a third sample object recalled by the third recall path at the third recall path indicating at least one of an amount of clicks, a timeliness, and an author rating of the third sample object.
22. The apparatus of any of claims 19-21, wherein the machine learning model is a linear regression model.
23. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein
The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-11.
24. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1-11.
CN202210551859.6A 2022-05-18 2022-05-18 Object recommendation method, training method, device and equipment of machine learning model Active CN115033782B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210551859.6A CN115033782B (en) 2022-05-18 2022-05-18 Object recommendation method, training method, device and equipment of machine learning model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210551859.6A CN115033782B (en) 2022-05-18 2022-05-18 Object recommendation method, training method, device and equipment of machine learning model

Publications (2)

Publication Number Publication Date
CN115033782A CN115033782A (en) 2022-09-09
CN115033782B true CN115033782B (en) 2023-03-28

Family

ID=83121178

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210551859.6A Active CN115033782B (en) 2022-05-18 2022-05-18 Object recommendation method, training method, device and equipment of machine learning model

Country Status (1)

Country Link
CN (1) CN115033782B (en)

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111859126B (en) * 2020-07-09 2024-05-14 有半岛(北京)信息科技有限公司 Recommended item determining method, device, equipment and storage medium
CN111931055B (en) * 2020-08-14 2023-09-12 中国工商银行股份有限公司 Object recommendation method, object recommendation device and electronic equipment
CN112948678B (en) * 2021-02-26 2023-07-21 北京房江湖科技有限公司 Article recall method and system and article recommendation method and system
CN114443989B (en) * 2022-01-30 2023-02-03 北京百度网讯科技有限公司 Ranking method, training method and device of ranking model, electronic equipment and medium

Also Published As

Publication number Publication date
CN115033782A (en) 2022-09-09

Similar Documents

Publication Publication Date Title
CN112579909A (en) Object recommendation method and device, computer equipment and medium
US20230112576A1 (en) Techniques for data processing predictions
CN114791982B (en) Object recommendation method and device
CN116541536B (en) Knowledge-enhanced content generation system, data generation method, device, and medium
CN115114424A (en) Response method and device for query request
WO2023142406A1 (en) Ranking method and apparatus, ranking model training method and apparatus, and electronic device and medium
CN116883181B (en) Financial service pushing method based on user portrait, storage medium and server
CN116450944A (en) Resource recommendation method and device based on recommendation model, electronic equipment and medium
CN115033782B (en) Object recommendation method, training method, device and equipment of machine learning model
CN115269989A (en) Object recommendation method and device, electronic equipment and storage medium
CN114238745A (en) Method and device for providing search result, electronic equipment and medium
CN114676062A (en) Method and device for testing difference data of interface, electronic equipment and medium
CN114445147A (en) Electronic ticket issuing method, electronic ticket issuing device, electronic ticket issuing apparatus, and electronic ticket issuing medium
CN114219046A (en) Model training method, matching method, device, system, electronic device and medium
CN114443896A (en) Data processing method and method for training a predictive model
CN114281990A (en) Document classification method and device, electronic equipment and medium
CN113868453A (en) Object recommendation method and device
CN113312554A (en) Method and device for evaluating recommendation system, electronic equipment and medium
CN115809364B (en) Object recommendation method and model training method
CN114861071B (en) Object recommendation method and device
US20230044508A1 (en) Data labeling processing
CN115292608A (en) Content resource recommendation method, neural network training method, device and equipment
CN118113934A (en) Data processing method, apparatus, electronic device, medium, and program product
CN115829653A (en) Method, device, equipment and medium for determining relevancy of advertisement text
CN113420227A (en) Training method of click rate estimation model, and method and device for estimating click rate

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
GR01 Patent grant
GR01 Patent grant