CN108304441B - Network resource recommendation method and device, electronic equipment, server and storage medium - Google Patents

Network resource recommendation method and device, electronic equipment, server and storage medium Download PDF

Info

Publication number
CN108304441B
CN108304441B CN201711120052.2A CN201711120052A CN108304441B CN 108304441 B CN108304441 B CN 108304441B CN 201711120052 A CN201711120052 A CN 201711120052A CN 108304441 B CN108304441 B CN 108304441B
Authority
CN
China
Prior art keywords
information
target user
network resource
user
resource
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
CN201711120052.2A
Other languages
Chinese (zh)
Other versions
CN108304441A (en
Inventor
邓亚平
连凤宗
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.)
Tencent Technology Shenzhen Co Ltd
Original Assignee
Tencent Technology Shenzhen 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 Tencent Technology Shenzhen Co Ltd filed Critical Tencent Technology Shenzhen Co Ltd
Priority to CN201711120052.2A priority Critical patent/CN108304441B/en
Publication of CN108304441A publication Critical patent/CN108304441A/en
Application granted granted Critical
Publication of CN108304441B publication Critical patent/CN108304441B/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
    • G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
    • G06F16/43—Querying
    • G06F16/435—Filtering based on additional data, e.g. user or group profiles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
    • G06F16/63—Querying
    • G06F16/635—Filtering based on additional data, e.g. user or group profiles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/73—Querying
    • G06F16/735—Filtering based on additional data, e.g. user or group profiles
    • 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

Abstract

The invention discloses a network resource recommendation method and device, electronic equipment, a server and a storage medium, and belongs to the technical field of networks. According to the resource recommendation method and the resource recommendation system, preference prediction and recommendation are performed on the user to be recommended by utilizing the resource recommendation model obtained by training the network resource information, the user function use behavior information and the user historical use information, and as the sources of sample data acquired during model training are more diversified, the resource recommendation model can describe the relationship between the user and the favorite network resources from different angles, so that the user's preference degree on the network resources is accurately represented, the accuracy of recommending the favorite network resources to the user is high, and a strong basis is provided for commercial operation and the like.

Description

Network resource recommendation method and device, electronic equipment, server and storage medium
Technical Field
The present invention relates to the field of network technologies, and in particular, to a method and an apparatus for recommending network resources, an electronic device, a server, and a storage medium.
Background
With the development of network technology, the presentation forms that network products can provide are more and more abundant. When providing services for users, the network products can provide online or offline services of network resources in various forms such as pictures, songs, videos and the like.
For a user, the amount of data provided by the network resource server is too large, and it takes much time and traffic to find out a network resource interested by the user from the mass data, so at present, the network resource server provides a network resource recommendation function, and taking the network resource as a song and the network resource server as a song server as an example, the song server can play songs played in an album according to the history of the user to find out songs similar to the songs played in the album and recommend the songs to the user.
However, the recommendation method refers to a single piece of information, and does not necessarily represent the actual interest of the user, so that the recommendation accuracy is low.
Disclosure of Invention
In order to solve the problems in the prior art, embodiments of the present invention provide a network resource recommendation method and apparatus, an electronic device, a server, and a storage medium. The technical scheme is as follows:
in a first aspect, a method for recommending network resources is provided, where the method includes:
acquiring function use behavior information of a target user to be recommended, wherein the function use behavior information is used for expressing the use behavior of the target user on at least one function of a network resource server;
acquiring preference characteristics of the target user according to the function use behavior information of the target user;
determining the preference degree of the target user to the network resources according to the preference characteristics of the target user, the network resources to be recommended and a resource recommendation model;
when the preference degree of the target user for the network resource meets a preset recommendation condition, recommending the network resource to the target user;
the resource recommendation model is obtained by training network resource information provided by the network resource server, function use behavior information of a plurality of sample users and historical use information of the plurality of sample users.
In a second aspect, an apparatus for recommending network resources is provided, the apparatus comprising:
the system comprises a first acquisition module, a second acquisition module and a recommendation module, wherein the first acquisition module is used for acquiring function use behavior information of a target user to be recommended, and the function use behavior information is used for expressing the use behavior of the target user on at least one function of a network resource server;
the second acquisition module is used for acquiring the preference characteristics of the target user according to the function use behavior information of the target user;
the determining module is used for determining the preference degree of the target user on the network resources according to the preference characteristics of the target user, the network resources to be recommended and a resource recommendation model;
the recommending module is used for recommending the network resources to the target user when the preference degree of the target user for the network resources meets a preset recommending condition;
the resource recommendation model is obtained by training network resource information provided by the network resource server, function use behavior information of a plurality of sample users and historical use information of the plurality of sample users.
In a third aspect, an electronic device is provided, which includes a processor and a memory, where at least one instruction, at least one program, a set of codes, or a set of instructions is stored in the memory, and the at least one instruction, the at least one program, the set of codes, or the set of instructions is loaded and executed by the processor to implement the network resource recommendation method according to the first aspect.
In a fourth aspect, a server is provided, which includes a processor and a memory, where at least one instruction, at least one program, a set of codes, or a set of instructions is stored in the memory, and the at least one instruction, the at least one program, the set of codes, or the set of instructions is loaded and executed by the processor to implement the network resource recommendation method according to the first aspect.
In a fifth aspect, a computer readable storage medium has stored therein at least one instruction, at least one program, a set of codes, or a set of instructions that is loaded and executed by a processor to implement the network resource recommendation method according to the first aspect.
The technical scheme provided by the embodiment of the invention has the following beneficial effects:
the method comprises the steps of utilizing a resource recommendation model obtained by training network resource information, function use behavior information of a target user and historical use information of the target user to predict and recommend the preference of the target user to be recommended, wherein the resource recommendation model is more diversified in sources of relevant characteristics of model use, so that the resource recommendation model can describe the relationship between the user and the network resources which are liked from different angles, the liked degree of the user to the network resources is accurately represented, the accuracy of recommending the liked network resources to the user is higher, and a strong basis is provided for commercial operation and the like.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a schematic diagram of an implementation environment of a network resource recommendation method according to an embodiment of the present invention;
fig. 2 is a flowchart of a network resource recommendation method according to an embodiment of the present invention;
fig. 3 is a flowchart of a network resource recommendation method, which is provided by an embodiment of the present invention, for example, a multimedia resource is a song;
FIG. 4 is a schematic diagram of an interface for recommending network resources according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a network resource recommendation device according to an embodiment of the present invention;
fig. 6 is a block diagram of an electronic device according to an embodiment of the present invention;
fig. 7 is a block diagram of a server according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
Fig. 1 is a schematic diagram of an implementation environment of a network resource recommendation method according to an embodiment of the present invention. In this embodiment, at least one terminal 101 and a network resource server 102 are included.
The user of the terminal 101 can access the network resource server 102 through the client to use the network service provided by the network resource server 102. For example, the terminal 101 may access the web resource server 102 through a video application client, and may also access a web portal of the web resource server 102 through a browser client. When the target user accesses the client by the method, functions of publishing, searching and the like of the network resource can be used by the client, and information generated by using the functions, such as record information generated by searching the network resource and the like, is the function use behavior information. After the target user uses the network resources, the information of the network resources is recorded to form the historical use information of the target user, for example, when a certain target user plays a certain song, the information of the song is recorded in the historical use information of the target user.
The network resource may be a multimedia resource, and the multimedia resource may be a video or an audio. When the multimedia resource is a video, the video may be a movie, a television show, a self-portrait small video, etc., and the multimedia feature information extracted based on the multimedia resource information may be a video popularity, a language type of the video, a video genre, etc. The video popularity can be the playing times of the video, the praise times of the video, the comment times of the video, and the like, and the video type can be an action type, a comedy type, a horror type, and the like.
When the multimedia asset is audio, the audio may be a song or the like, and the multimedia feature information extracted based on the multimedia asset information may be audio heat, language of the audio, audio type, or the like. The audio popularity may be the number of times the audio is played, the number of times the audio is liked, the number of times the audio is shared, etc., and the audio type may be balladry, electronic music, popular music, etc.
The network resource server 102 is used for providing network services, which may refer to video services, audio services, picture services, reading services, question answering services, and the like. Taking a network resource server as an example of a video server, the video services provided by the network resource server may include services such as live video, online video playing, video downloading, and the like, and for the network resource server, the services provided by the network resource server may not be a single service, for example, for the video server, the video server may not be limited to only a video service, but also provide other types of network services such as an audio service, and for the audio server, the audio server may not be limited to only an audio service, but also provide more types of network services such as a video service, such as a karaoke service, audio downloading, and the like.
Of course, the network resource server may also provide functions such as forwarding and commenting, and this is not specifically limited in the embodiment of the present invention. The online video playing service may refer to converting a certain movie into a video data stream, and providing the video data stream to the terminal 101 through a video client or a web portal for online playing or offline downloading.
For the network resource server 102, the network resource server 102 may further have at least one database for storing network resource characteristic information, historical usage information of sample users, and functional usage behavior information of sample users.
The network resource feature information may include a heat degree of the network resource, a language of the network resource, a type of the network resource, and the like. For example, when the network resource is a multimedia resource and the multimedia resource is a video, the multimedia feature information may be a video heat, a language and a video type of the video, and when the multimedia resource is an audio, the multimedia feature information may be an audio heat, a language and an audio type of the audio, and the like.
The sample user may refer to a user served by the network resource server, and the historical usage information of the sample user may include information such as the number of times the sample user uses the network resource.
The function usage behavior information of the sample user may include network resource list information, network resource publishing information, network resource search click information, social network resource sharing information, interaction information of the sample user with the network resource, and the like.
The function usage behavior information of the sample users may be different for different network resource types, for example, when the network resource is a multimedia resource and the multimedia resource is a video, the function usage behavior information of the plurality of sample users includes at least one of the following information:
1. user registration information such as user name, age, gender, favorite video type, etc.
2. Video list information, such as video favorite list information of the sample user, or video list information created by the sample user, etc.
3. And uploading information of the video, such as a small video shot by a user of the sample user.
4. And the video search click information is used for indicating that a sample user wants to search a certain video, and searching the information of the video recorded in the record database when inputting other related information such as a video name or a video author name and the like in a search box for searching through a video application client or a browser client.
5. The social network sharing information of the video may be the number of times of sharing in the video application client, or the number of times of sharing to other clients, such as the social application client.
6. And interaction information of any video, such as comments, praise and the like of a certain video.
When the network resource is a multimedia resource and the multimedia resource is audio, taking the multimedia resource as a song as an example, the function usage behavior information of the plurality of sample users includes at least one of the following information:
1. user registration information such as user name, age, gender, favorite music type, etc.
2. Song list information such as song collection list information of the sample user, song list information created by the sample user himself or song list information collected by the sample user, and the like.
3. And the song information represents song information uploaded to the multimedia resource server when the sample user uses the song function through the music client.
4. The song search click information is used for indicating that a sample user wants to search for a certain song, and searching the information of the song recorded in the record database when inputting the song name or the name of the singer and other related information in the search box for searching through the music client or the browser client.
5. The social network sharing information of the song may be the number of times that a certain song is shared in the music client, or the number of times that the certain song is shared to other clients, such as a social application client.
6. Interaction information for any song, such as comments, praise, etc. for a song.
It should be noted that the function usage behavior information of the sample user may be obtained from the function usage behavior information database by the network resource server when model training is required.
Fig. 2 is a flowchart of a network resource recommendation method according to an embodiment of the present invention, and referring to fig. 2, the flowchart shown in fig. 2 specifically includes two parts, a first part is a model training process based on function usage behavior information of sample users and historical usage information of the sample users, and a second part is a prediction process based on a recommendation model. After the resource recommendation model is obtained through the training in steps 201 to 202, the resource recommendation model can be directly used when prediction is needed, such as steps 203 to 206, without performing the training again or complex calculation. In this embodiment, taking the example that the network resource is a multimedia resource, the method execution main body is a multimedia resource server, and details of each step are described with reference to fig. 2:
201. and screening and extracting the characteristics of the original sample data to obtain the characteristic information of a plurality of positive samples and the characteristic information of a plurality of negative samples, wherein the original sample data comprises the function use behavior information of the sample users and the historical use information of the plurality of sample users.
The positive sample refers to a set of multimedia resources and sample users with the stay time being longer than or equal to a preset time, and the negative sample refers to a set of multimedia resources and sample users with the stay time being shorter than the preset time. For example, for audio or video, the dwell time may refer to a play time, and for web page resources and the like, the dwell time may be a browse time.
The inventor realizes that when the multimedia resources preferred by the user are predicted by combining the function use behavior information of the user and the multimedia characteristic information of the multimedia resources, the prediction accuracy of the preference degree of the multimedia resources of the user is higher because the reference information is more comprehensive and the preference of the user to the multimedia resources is analyzed from multiple angles.
Based on the multimedia resource information provided by the multimedia resource server, the multimedia resource server can extract multimedia feature information, user function use behavior information and user historical use information, and store the extracted information in a corresponding database of the multimedia resource server for subsequent feature extraction. Taking the multimedia resource as a song as an example, the multimedia resource server can extract the song characteristic information, the function use behavior information of the user and the song listening history information of the user based on the song information in the song library, and store the extracted information in a corresponding database of the multimedia resource server.
Before training the model, the sample information needs to be processed, and the preprocessing process includes: a sample screening process, an evaluation reference preference degree process, a positive and negative sample classification process, a feature source screening process, and a feature extraction process, each of which is described in detail below.
Sample screening process
After the multimedia resource server obtains the function use behavior information and the historical use information of a plurality of sample users to be screened, the sample users to be screened can be screened, after the sample users are determined, appropriate multimedia resource characteristic information is selected for each sample user.
When a sample user to be screened is screened, the social network sharing information, the interaction information of the multimedia resource, the accumulated time of playing the multimedia resource and the like in the function use behavior information of the sample user to be screened can be screened, a function use behavior information threshold value is preset, when the number of times of sharing the multimedia resource in the social network sharing information of the sample user to be screened is larger than the function use behavior information threshold value, or the number of times of interacting the multimedia resource in the interaction information of the sample user to be screened is larger than the function use behavior information threshold value, the sample user is selected and determined to be the sample user. Or, presetting a multimedia resource playing threshold, when the accumulated time for playing the multimedia resources of the sample users to be screened is greater than the multimedia resource playing threshold, selecting the sample users, and determining the sample users as the sample users. Therefore, the sample users are screened, the sample users obtained after screening can be more representative, and the model obtained by training is more representative.
Process for evaluating reference preference degree
After the screened sample user is obtained, the sample comprises the sample user and the multimedia resource, so that the multimedia resource provided by the multimedia resource server needs to be evaluated for the reference preference degree. The multimedia resource server can evaluate the historical staying time of the multimedia resources in the historical use information of the sample user to obtain the reference preference degree of the multimedia resources, and the reference preference degree can be expressed in a score form.
Taking the example that the multimedia resource is a song, if the range of the score is set as [0,1], firstly determining the song to be scored, then determining the historical staying time of the song corresponding to the sample user, and dividing the historical staying time by the whole song time to obtain the result, namely the score of the song corresponding to the sample user. In addition, the score may also be jointly scored according to the historical usage information and the functional usage behavior information of the sample user, for example, according to the processing method, a sub-score is obtained according to the historical usage information, then according to the similar processing method, a sub-score is obtained according to the song K song duration, and an average value of the two sub-scores is calculated as the score of the song corresponding to the sample user.
In the above process of evaluating the reference preference degree, the multimedia resource evaluating the reference preference degree may be an original multimedia resource or a multimedia resource after being filtered, and this embodiment is not limited herein.
Positive and negative sample classification process
After the reference preference degree of the multimedia resource is obtained through the reference preference degree evaluation process, the sample is divided into a positive sample and a negative sample according to the reference preference degree of the multimedia resource. When classifying the sample, the multimedia resource server can obtain the reference preference degree of the multimedia resource in the sample data, and then according to a preset classification method, the sample is classified into two types, namely a positive sample and a negative sample.
The preset classification method may be that the reference preference degree is represented in a score form, a sample with a score larger than a preset threshold is a positive sample, and a sample with a score smaller than or equal to the preset threshold is a negative sample. For example, samples with a score greater than 0.5 become positive samples, and samples with a score less than or equal to 0.5 become negative samples.
Characteristic source screening process
Because one data source of the multimedia resource characteristic information is the song list, the song list of the sample user can be screened, so that the selected sample user and the multimedia resource characteristic information can be guaranteed to be more representative, and the trained model has more universal applicability. When the multimedia resource characteristic information is screened, when the source of the multimedia resource characteristic information is a song list, the multimedia resource characteristic information can be screened according to the heat degree of the song list, and can also be screened according to a singer or an album of songs in the song list.
Taking the example that the multimedia resource is a song, when the multimedia resource server performs screening according to the popularity of the song, the multimedia resource server performs screening according to the playing times, the praise times and the shared times of the song, a popularity threshold is preset, and when the popularity of the song is greater than the popularity threshold, the song is screened out to be determined as a characteristic source. Taking the praise times as an example, the preset heat threshold is 100, when the praise times of a certain song list is greater than 100, the song list is screened out to determine that the song list is a feature source, and other heat types can be processed in the same way, which is not described herein again. The source of the multimedia resource characteristic information is screened in this way, more representative data providing characteristics can be left, and noise can be prevented.
When the multimedia resource server performs screening according to the singer or the album of the songs in the song list, the multimedia resource server may perform screening according to the same degree of the singer or the album of the songs in the song list, and the same degree threshold value is preset. Taking the singer of the songs in the singing list as an example, the preset threshold value of the same degree is 0.6, when 30 songs exist in the singing list and 20 songs are all performed by the same singer, the numerical value of the same degree is calculated to be 0.66, and the same degree is smaller than the preset threshold value of the same degree, the singing list can be screened out and determined to be a characteristic source. Therefore, the singing list is screened, and the selected singing list is prevented from being too single, so that the extracted song characteristic information is inaccurate when the song characteristic information is extracted from the singing list.
The specific process of the above screening is only described by taking a song as an example, and other types of multimedia resources can be processed in the same way, which is not described herein again.
Feature extraction process
After the screened feature source data is obtained, the preference features of the sample user can be obtained according to the historical use information and the function use behavior information of the sample user. The specific process can be as follows from step 1 to step 3:
step 1: and performing vector feature extraction on a plurality of multimedia resource information included in the historical use information and the function use behavior information of the sample user to obtain a user vector based on the multimedia resource, a user vector based on the label and feature information of the multimedia resource.
In the implementation, for example, multimedia resources are songs, all the song list information obtained after the screening and all the song identifiers in the song list information are determined, the song identifiers are extracted, each song identifier is used as a word, then all the song identifiers in one song list form a sentence, a plurality of song lists form a plurality of sentences, the sentences are used as linguistic data, and then the song vectors of the sentences are extracted by adopting a vector extraction algorithm, such as word2vector and the like. Then, according to the song listening history information of the sample user and the song vector, a song-based user vector can be obtained.
Then, from the song vector and the song-based user vector, a distance between the vectors may be calculated, which may be used to indicate how liked the sample user is for the song.
Determining all the song list information obtained after screening and all the label information in the song list information, extracting the label information, taking each label as a word, then forming a sentence by all the labels in one song list, forming a plurality of sentences by a plurality of song lists, taking the sentences as linguistic data, and then extracting the label vectors by adopting a vector extraction algorithm, such as word2vector and the like. And then, according to the song listening history information of the sample user and the label vector, obtaining a label-based user vector.
Then, from the tag vectors and the tag-based user vectors, a distance between the vectors is calculated, which can be used to indicate how liked the sample user likes songs of different tags.
Step 2: and performing collaborative filtering feature extraction on the historical use information of the sample user to obtain the collaborative filtering feature of the sample user based on the multimedia resource and the collaborative filtering feature of the sample user based on the sample user.
In an implementation, the process of obtaining the song-based collaborative filtering feature may be as follows:
and calculating the similarity between the songs according to the historical use information of all sample users, wherein the greater the similarity is, the greater the possibility that two songs are liked by the same sample user is. Acquiring all songs from historical song listening information of a sample user, generating a song similarity list similar to the song in the range of all songs in a song library aiming at one song, and generating a song-based collaborative filtering feature vector for the sample user according to the song similarity list and the song listening list of the sample user, wherein each digit in the collaborative filtering feature vector can represent the possibility that the sample user corresponding to the digit likes the song.
The process of obtaining collaborative filtering features based on sample users may be as follows: and calculating the similarity between the sample users according to the historical use information of all the sample users, wherein the two sample users with higher similarity have higher probability of liking the same song. Then, for each sample user, a list of similarity degrees with other sample users is generated, and according to the similarity degrees with other sample users in the similarity degree list and songs that other sample users may like, a collaborative filtering feature vector based on the sample user is generated for the sample user, wherein each bit in the collaborative filtering feature vector represents the possibility that the sample user likes the song corresponding to the bit in the song list.
And step 3: and extracting the characteristics of the sample user according to the historical use information and the function use behavior information of the sample user to obtain the user portrait characteristics of the sample user.
When the function use behavior information of the sample user contains user registration information, user feature extraction can be performed on the user registration information, the extracted user name, gender, age and other features form a sample user identity feature sub-vector, and the sample user identity feature sub-vector can be used for indicating the identity information of the sample user.
According to the historical use information of the sample user and other function use behavior information, such as song type information liked by the sample user in the user registration information, and the like, the characteristics of the song type, singer, song language and the like possibly liked by the sample user are extracted, and the characteristics are formed into a sample user preference sub-vector which can be used for indicating the song characteristics liked by the sample user.
And sequentially synthesizing the sample user identity characteristic sub-vector and the sample user preference sub-vector into a user portrait characteristic vector of the sample user, wherein the user portrait characteristic vector of the sample user can be used for indicating identity information and favorite song characteristics of the sample user.
It should be noted that, which steps of the above steps 1 to 3 are included in the process of obtaining the preference characteristics of the sample user may be determined according to information specifically included in the obtained function usage behavior information. For example, when the acquired function usage behavior information includes all information in the function usage behavior information, that is, the user registration information, the song list information, the song search click information, the social network sharing information of the song, and the interaction information for any song are acquired, the above steps 1 to 3 may be performed at the same time. When the acquired function usage behavior information only includes song list information and interaction information for any song, only step 1 above may be performed.
202. And performing model training based on the feature information of the positive samples, the feature information of the negative samples, the reference preference degrees of the positive samples and the negative samples and the multimedia feature information of each multimedia resource to obtain a resource recommendation model.
Inputting the feature information of a plurality of positive samples, the feature information of a plurality of negative samples, the reference preference degrees of the positive and negative samples, and the multimedia feature information of each multimedia resource into a resource recommendation model to be trained, wherein step 202 may include the following steps 1 to 2:
step 1: and inputting the characteristic information of the positive samples, the characteristic information of the negative samples, the reference preference degrees of the positive samples and the negative samples and the multimedia characteristic information of each multimedia resource into the current recommendation model, and outputting the prediction preference degrees of the sample users to the sample data.
In an implementation, the recommendation model may be an FTRL (logistic regression algorithm) online learning algorithm model, and feature information of a plurality of positive samples, feature information of a plurality of negative samples, reference preference degrees of the positive and negative samples, and multimedia feature information of each multimedia resource are input into the current recommendation model, where the feature information of the samples includes a preference feature of each sample user, multimedia feature information of the multimedia resource, and a corresponding multimedia resource reference preference degree, and the preference feature of the sample user may include a preference feature calculated in a plurality of vector forms. By giving initial weight to each preference feature, and then calculating a prediction preference degree, the prediction preference degree can be expressed in the form of a component value, the range of the prediction score can be [0,1], and is used for expressing that the current recommendation model predicts the likeness degree of the sample user to the multimedia resource, and the higher the prediction score is, the higher the likeness degree of the sample user to the recommended multimedia resource is.
Step 2: and adjusting the weight of each feature in the current recommendation model according to the prediction preference degree and the reference preference degree of the sample data to obtain a preset recommendation model.
In implementation, after the prediction preference degrees output by the multiple models are obtained, each prediction preference degree is compared with the reference preference degree corresponding to the multimedia resource in the input sample, and the error between the prediction preference degrees and the reference preference degrees is calculated.
And continuously adjusting the weight of each feature in the current recommendation model according to the obtained error, and taking the finally obtained weight of each feature of the model as the weight of the resource recommendation model, thereby obtaining the trained resource recommendation model. The trained resource recommendation model can accurately predict the multimedia resource like degree of the user, and can improve the accuracy of recommending the multimedia resource like to the user.
It should be noted that the above process is only an example of the resource recommendation model training process, and in the case of performing the same function, the models and algorithms in the above process may be replaced by other models or algorithms, such as a matrix decomposition algorithm model, a deep learning algorithm model, and the like, which is not limited in this disclosure.
It should be noted that the above-mentioned model training process and the process of predicting the user's likeability of the multimedia resource may be implemented on the same device, for example, both may be implemented by the server, and the model training is performed on the server side, and the resource recommendation model obtained by the model training is identified, or after the model training is performed on the server side, the resource recommendation model obtained by the training is issued to other electronic devices, and the other electronic devices identify based on the resource recommendation model.
Fig. 3 is a flowchart of a network resource recommendation method, which is provided by an embodiment of the present invention, for example, a multimedia resource is a song. With reference to fig. 3, taking the multimedia resource as a song as an example, the above steps are briefly summarized:
the multimedia resource server obtains song information of the song library, function use behavior information of the sample user and song listening history information of the sample user from a corresponding database, and then screens the sample user. And scoring the songs according to the song listening history information corresponding to the screened sample user to obtain the score of each song. The samples are then divided into positive and negative samples based on the score. And then, the multimedia resource server performs feature extraction on the function use information of the sample user and the song listening history information of the sample user to obtain the preference feature of the sample user. Sample user preference features include feature-based user vectors, collaborative filtering features, and user profile features. The feature-based user vector comprises a song-based user vector and a tag-based user vector, the collaborative filtering feature comprises a song-based collaborative filtering feature and a user-based collaborative filtering feature, and the user portrait feature comprises a user identity feature and a user preference feature.
Then, acquiring the characteristic information of the song, wherein the characteristic information of the song comprises song popularity, song language and song type. And finally, the multimedia resource server inputs the obtained feature information of the songs, the preference features of the sample users and the scores of the songs into a model for training. After the model is trained, the model can be used for predicting the likeness of the user to the song.
The following describes a process of predicting the user's likeness to a multimedia resource based on a resource recommendation model.
203. And acquiring the function use behavior information of the target user to be recommended.
In implementation, according to the function usage behavior information and the multimedia resource information of the target user to be recommended, the favorite degree of the target user on the multimedia resource can be predicted, and the history usage information of the target user is not necessary information.
204. And acquiring the preference characteristics of the target user according to the function use behavior information of the target user.
In implementation, after acquiring the function usage behavior information of the target user, according to the processing manner of step 201, feature extraction is performed on the function usage behavior information of the target user to form feature information in a vector form, and the obtained feature information in the vector form constitutes preference features of the target user.
205. And determining the preference degree of the target user to the multimedia resources according to the preference characteristics of the target user, the multimedia resources to be recommended and the resource recommendation model.
It should be noted that, if the multimedia resource to be recommended is pre-stored in the multimedia resource server, the multimedia resource server may obtain the multimedia resource feature information from the multimedia feature information database; and if the multimedia resource to be recommended is not pre-stored in the multimedia resource database, the multimedia resource server performs feature extraction on the multimedia resource to be recommended to obtain multimedia resource feature information.
When the likeness degree of the target user to the multimedia resource needs to be predicted, the multimedia resource feature information and the preference feature of the target user can be input into the resource recommendation model together. Or processing the multimedia resource to be recommended and the function use behavior information of the target user together, generating multimedia resource feature information by the multimedia resource to be recommended, generating preference features of the target user by the function use behavior information of the target user, and then inputting the multimedia resource feature information and the preference features of the target user into the resource recommendation model.
And inputting the multimedia resource characteristic information and the preference characteristic of the target user into a resource recommendation model to obtain the preference degree of the target user to the multimedia resource, wherein the preference degree represents the favorite degree of the target user to the song. The preference degree may be expressed in the form of a score, the larger the score, the greater the degree of liking of the song by the target user.
The resource recommendation model is obtained by training multimedia resource information provided by a multimedia resource server, function use behavior information of a plurality of sample users and historical use information of the plurality of sample users.
206. And recommending the multimedia resources to the target user when the preference degree of the target user for the multimedia resources meets the preset recommendation condition.
Comparing the preference degree of the target user for the multimedia resource obtained in the step 205 with the preset recommendation condition, and if the preference degree meets the preset recommendation condition, determining the multimedia resource corresponding to the preference degree as the initial multimedia resource to be recommended. The preset recommendation condition may be that the value of the preference degree is greater than a preset score, and the preset score is a threshold value selected by a technician according to the score range of the multimedia resource, for example, if the value range of the preference degree of the multimedia resource is [0,1], the preset score may be set to 0.5.
It should be noted that after the initial multimedia resource to be recommended is determined, it is necessary to determine whether the initial multimedia resource to be recommended meets a preset supplement condition, if the preset supplement condition is met, the multimedia resource to be recommended is recommended to the target user, otherwise, the multimedia resource is not recommended to the target user. The preset replenishment conditions may be as follows:
the first condition is as follows: the historical usage information of the target user does not include multimedia resources. Or when the historical use information of the target user comprises the multimedia resources, and the time interval between the staying time of the multimedia resources and the current time is greater than or equal to the preset time length.
In implementation, whether the initial multimedia resource to be recommended is contained in the historical use information of the target user can be inquired, if not, the multimedia resource can be determined as the multimedia resource to be recommended and recommended to the target user; if the time interval between the staying time and the current time is greater than or equal to the preset time length, namely the multimedia resource is not played in the near future, the multimedia resource can be recommended to the target user, otherwise, the multimedia resource is not recommended to the target user. Therefore, the multimedia resources recently played by the target user can be prevented from being recommended to the target user, the recommendation success rate is prevented from being reduced, and the recommendation success rate is actually improved.
And a second condition: the score of the multimedia resource is greater than the preset score.
The scoring of the multimedia resource refers to scoring of the multimedia resource by all target users on the relevant client, for example, scoring of a certain movie by the target users on the application client or the browser client.
After the multimedia resource server inquires the score of the multimedia resource to be recommended, the score of the multimedia resource to be recommended is compared with a preset score, if the score of the multimedia resource is larger than the preset score, the multimedia resource can be recommended to a target user, and otherwise, the multimedia resource is not recommended to the target user. Therefore, the recommendation of badly evaluated multimedia resources to the target user can be avoided, and the experience of the target user is better. Wherein the preset score is a score selected by a technician based on a plurality of experimental results.
And (3) carrying out a third condition: and the multimedia resource name is different from the determined multimedia resource to be recommended.
The multimedia resources to be recommended are deduplicated, for example, three versions of a certain song exist in the songs to be recommended, the three versions are sung by three singers with different numbers, in this case, the version with the highest song score in the three versions is selected and determined as the song to be recommended, and the other two versions are not recommended to the target user, so that the recommendation opportunity is prevented from being wasted.
The above conditions may be used alone, two conditions may be used in combination, or three conditions may be used in combination. For example, the condition one and the condition two may be used in combination, that is, the user's historical usage information does not include the multimedia resource. Or, when the historical use information of the target user comprises the multimedia resource, the time interval between the staying time of the multimedia resource and the current time is greater than or equal to the preset time length; and the score of the multimedia resource is larger than the preset score. In this case, the processing of this condition may be performed according to the above-mentioned first condition and second condition, so as to remove the multimedia resources that do not meet the condition, and recommend the remaining multimedia resources to the target user.
It should be noted that after determining the multimedia resource to be recommended according to the above conditions, the multimedia resource may be recommended to the target user, and the following processing may be performed during recommendation: and displaying the multimedia resources at the page positions corresponding to the sequencing positions according to the sequencing positions of the preference degrees of the target users to the multimedia resources in the determined multimedia resources to be recommended.
And sequencing the multimedia resources to be recommended according to the preference degree of the target user, and then displaying the multimedia resources to be recommended at the page positions corresponding to the sequencing sequence of the multimedia resources. For example, as shown in fig. 4, when a user opens a music client, 10 songs recommended on the recommendation page of the client can be seen, wherein, a singer picture of 3 songs can be displayed at the top of the page in a loop, and 7 songs are displayed in the "today's song recommendation" of the lower page. After the client determines the songs recommended to the user, the songs are ranked according to the preference degrees of the target user, the top 10 songs with the highest preference degree are selected, the top 3 songs with the highest preference degrees are circularly displayed at the top of the client page in sequence, and the remaining 7 songs are displayed in the 'song recommendation today' in sequence. The user may first see the 3 songs he is most likely to be interested in displayed in a loop, and may also see the songs he is likely to be interested in displayed in the "today's song recommendations" while browsing the page. For another example, when a user opens a song that the music client wants to search, a search page is opened, and when the user inputs a keyword that the user wants to search in a search box, recommended songs related to the keyword and recommended songs related to the user can be displayed below the search box. Therefore, the user can easily find the favorite multimedia resources according to the recommendation, the recommendation accuracy is improved, and the user experience is improved.
In this embodiment, taking the above processing manner as an example, there may be other processing manners, such as displaying the multimedia resources in the display batch corresponding to the sorting position at the determined sorting position in the multimedia resources to be recommended according to the preference degree of the target user for the multimedia resources, and the like, which are not described herein.
In this embodiment, taking the multimedia resource as an example, the network resource may also be other resources, such as news information, shopping information, such as a commodity link, and the like. Taking news information as an example, the resource recommendation model is trained based on the news information, the obtained resource recommendation model can be used for recommending news information to a target user, and other network resources can also be subjected to model training and recommendation in the same way, which is not limited by the invention.
According to the method provided by the embodiment of the invention, preference prediction and recommendation are carried out on the user to be recommended by utilizing the resource recommendation model obtained by training the network resource information, the user function use behavior information and the user historical use information, and as the source of sample data acquired during model training is more diversified, the resource recommendation model can describe the relationship between the user and the favorite network resources from different angles, so that the user preference degree on the network resources is accurately expressed, the accuracy rate of recommending the favorite network resources to the user is higher, and a strong basis is provided for commercial operation and the like.
Fig. 5 is a schematic structural diagram of a network resource recommendation device according to an embodiment of the present invention. Referring to fig. 5, the apparatus includes:
a first obtaining module 501, configured to obtain function usage behavior information of a target user to be recommended, where the function usage behavior information is used to indicate a usage behavior of the target user for at least one function of a network resource server;
a second obtaining module 502, configured to obtain a preference feature of the target user according to the function usage behavior information of the target user;
a determining module 503, configured to determine, according to the preference feature of the target user, the network resource to be recommended, and a resource recommendation model, a preference degree of the target user for the network resource;
a recommending module 504, configured to recommend the network resource to the target user when the preference degree of the target user for the network resource meets a preset recommending condition;
the preset recommendation model is obtained by training network resource information provided by the network resource server, function use behavior information of a plurality of sample users and historical use information of the plurality of sample users.
In any possible implementation, the function usage behavior information of the plurality of sample users includes at least one of the following information: user registration information, network list information, network publishing information, search click information, social network sharing information and interaction information of any network resource.
In any one of the possible implementations, the apparatus further includes:
the processing module is used for screening and extracting characteristics of sample original data to obtain characteristic information of a plurality of positive samples and characteristic information of a plurality of negative samples, wherein the sample original data comprises function use behavior information of sample users and historical use information of the plurality of sample users;
the resource recommendation model training module is used for carrying out model training based on the feature information of a plurality of positive samples, the feature information of a plurality of negative samples, the reference preference degrees of the positive and negative samples and the network resource feature information of each network resource to obtain the resource recommendation model;
the positive sample refers to a set of network resources and sample users with the stay time being longer than or equal to a preset time, and the negative sample refers to a set of network resources and sample users with the stay time being shorter than the preset time.
In any possible implementation manner, the resource recommendation model training module is configured to:
inputting the plurality of sample data into a current recommendation model, and outputting the prediction preference degrees of the plurality of sample data;
and adjusting the weight of each feature in the current recommendation model according to the prediction preference degree and the reference preference degree of the sample data to obtain the preset recommendation model.
In any possible implementation manner, the second obtaining module is configured to:
vector feature extraction is carried out on the historical use information of the target user and the plurality of network resource information included in the function use behavior information to obtain a user vector based on network resources, a user vector based on labels and feature information of the network resources;
extracting collaborative filtering characteristics of the historical use information of the target user to obtain collaborative filtering characteristics of the target user based on network resources and collaborative filtering characteristics based on users;
and extracting user characteristics according to the historical use information and the function use behavior information of the target user to obtain user portrait characteristics of the target user.
In any possible implementation, the recommendation module is to:
when the preference degree of the target user for the network resource meets a preset recommendation condition and at least one of a plurality of preset supplement conditions, recommending the network resource to the target user;
wherein the plurality of preset supplementary conditions include:
the historical use information of the target user does not contain the network resource; or, when the historical use information of the target user comprises the network resource, and the time interval between the stay time of the network resource and the current time is greater than or equal to the preset time length;
and/or the presence of a gas in the gas,
the score of the network resource is greater than a preset score;
and/or the presence of a gas in the gas,
and the network resource name is different from the determined network resource to be recommended.
In any possible implementation, the recommendation module is to:
and displaying the network resources at a page position corresponding to the sequencing position according to the sequencing position of the preference degree of the target user on the network resources in the determined network resources to be recommended.
According to the method provided by the embodiment of the invention, preference prediction and recommendation are carried out on the user to be recommended by utilizing the resource recommendation model obtained by training the network resource information, the user function use behavior information and the user historical use information, and as the source of sample data acquired during model training is more diversified, the resource recommendation model can describe the relationship between the user and the favorite network resources from different angles, so that the user preference degree on the network resources is accurately expressed, the accuracy rate of recommending the favorite network resources to the user is higher, and a strong basis is provided for commercial operation and the like.
Fig. 6 is a block diagram of an electronic device according to an embodiment of the present invention. The electronic device 600 may be a portable mobile electronic device such as: smart phones, tablet computers, MP3 players (Moving Picture Experts Group Audio Layer III, motion video Experts compression standard Audio Layer 3), MP4 players (Moving Picture Experts Group Audio Layer IV, motion video Experts compression standard Audio Layer 4). The electronic device 600 may also be referred to by other names such as user equipment, portable electronic devices, and the like.
In general, the electronic device 600 includes: a processor 601 and a memory 602.
The processor 601 may include one or more processing cores, such as a 4-core processor, an 8-core processor, and so on. The processor 601 may be implemented in at least one hardware form of a DSP (Digital Signal Processing), an FPGA (Field-Programmable Gate Array), and a PLA (Programmable Logic Array). The processor 601 may also include a main processor and a coprocessor, where the main processor is a processor for Processing data in an awake state, and is also called a Central Processing Unit (CPU); a coprocessor is a low power processor for processing data in a standby state. In some embodiments, the processor 601 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content required to be displayed on the display screen. In some embodiments, processor 601 may also include an AI (Artificial Intelligence) processor for processing computational operations related to machine learning.
Memory 602 may include one or more computer-readable storage media, which may be tangible and non-transitory. The memory 602 may also include high-speed random access memory, as well as non-volatile memory, such as one or more magnetic disk storage devices, flash memory storage devices. In some embodiments, a non-transitory computer readable storage medium in the memory 602 is used to store at least one instruction for execution by the processor 601 to implement the network resource recommendation method provided herein.
In some embodiments, the electronic device 600 may further optionally include: a peripheral interface 603 and at least one peripheral. Specifically, the peripheral device includes: at least one of a radio frequency circuit 604, a touch screen display 605, a camera 606, an audio circuit 607, and a power supply 609.
The peripheral interface 603 may be used to connect at least one peripheral related to I/O (Input/Output) to the processor 601 and the memory 602. In some embodiments, the processor 601, memory 602, and peripheral interface 603 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 601, the memory 602, and the peripheral interface 603 may be implemented on a separate chip or circuit board, which is not limited in this embodiment.
The Radio Frequency circuit 604 is used for receiving and transmitting RF (Radio Frequency) signals, also called electromagnetic signals. The radio frequency circuitry 604 communicates with communication networks and other communication devices via electromagnetic signals. The rf circuit 604 converts an electrical signal into an electromagnetic signal to transmit, or converts a received electromagnetic signal into an electrical signal. Optionally, the radio frequency circuit 604 comprises: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a subscriber identity module card, and so forth. The radio frequency circuitry 604 may communicate with other electronic devices via at least one wireless communication protocol. The wireless communication protocols include, but are not limited to: the world wide web, metropolitan area networks, intranets, generations of mobile communication networks (2G, 3G, 4G, and 5G), Wireless local area networks, and/or WiFi (Wireless Fidelity) networks. In some embodiments, the rf circuit 604 may further include NFC (Near Field Communication) related circuits, which are not limited in this application.
The touch display 605 is used to display a UI (User Interface). The UI may include graphics, text, icons, video, and any combination thereof. The touch display screen 605 also has the ability to acquire touch signals on or over the surface of the touch display screen 605. The touch signal may be input to the processor 601 as a control signal for processing. The touch display 605 is used to provide virtual buttons and/or a virtual keyboard, also referred to as soft buttons and/or a soft keyboard. In some embodiments, the touch display 605 may be one, providing the front panel of the electronic device 600; in other embodiments, the touch screen display 605 may be at least two, respectively disposed on different surfaces of the electronic device 600 or in a folded design; in still other embodiments, the touch display 605 may be a flexible display disposed on a curved surface or on a folded surface of the electronic device 600. Even more, the touch screen display 605 may be arranged in a non-rectangular irregular pattern, i.e., a shaped screen. The touch screen 605 may be made of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), and the like.
The camera assembly 606 is used to capture images or video. Optionally, camera assembly 606 includes a front camera and a rear camera. Generally, a front camera is used for realizing video call or self-shooting, and a rear camera is used for realizing shooting of pictures or videos. In some embodiments, the number of the rear cameras is at least two, and each of the rear cameras is any one of a main camera, a depth-of-field camera and a wide-angle camera, so that the main camera and the depth-of-field camera are fused to realize a background blurring function, and the main camera and the wide-angle camera are fused to realize a panoramic shooting function and a VR (Virtual Reality) shooting function. In some embodiments, camera assembly 606 may also include a flash. The flash lamp can be a monochrome temperature flash lamp or a bicolor temperature flash lamp. The double-color-temperature flash lamp is a combination of a warm-light flash lamp and a cold-light flash lamp, and can be used for light compensation at different color temperatures.
The audio circuit 607 is used to provide an audio interface between a user and the electronic device 600. Audio circuitry 607 may include a microphone and a speaker. The microphone is used for collecting sound waves of a user and the environment, converting the sound waves into electric signals, and inputting the electric signals to the processor 601 for processing or inputting the electric signals to the radio frequency circuit 604 to realize voice communication. For stereo capture or noise reduction purposes, the microphones may be multiple and disposed at different locations of the electronic device 600. The microphone may also be an array microphone or an omni-directional pick-up microphone. The speaker is used to convert electrical signals from the processor 601 or the radio frequency circuit 604 into sound waves. The loudspeaker can be a traditional film loudspeaker or a piezoelectric ceramic loudspeaker. When the speaker is a piezoelectric ceramic speaker, the speaker can be used for purposes such as converting an electric signal into a sound wave audible to a human being, or converting an electric signal into a sound wave inaudible to a human being to measure a distance. In some embodiments, audio circuitry 607 may also include a headphone jack.
The power supply 609 is used to supply power to various components in the electronic device 600. The power supply 609 may be ac, dc, disposable or rechargeable. When the power supply 609 includes a rechargeable battery, the rechargeable battery may be a wired rechargeable battery or a wireless rechargeable battery. The wired rechargeable battery is a battery charged through a wired line, and the wireless rechargeable battery is a battery charged through a wireless coil. The rechargeable battery may also be used to support fast charge technology.
In some embodiments, the electronic device 600 also includes one or more sensors 610. The one or more sensors 610 include, but are not limited to: acceleration sensor 611, gyro sensor 612, pressure sensor 613, optical sensor 615, and proximity sensor 616.
The acceleration sensor 611 may detect the magnitude of acceleration in three coordinate axes of a coordinate system established with the electronic device 600. For example, the acceleration sensor 611 may be used to detect components of the gravitational acceleration in three coordinate axes. The processor 601 may control the touch screen display 605 to display the user interface in a landscape view or a portrait view according to the gravitational acceleration signal collected by the acceleration sensor 611. The acceleration sensor 611 may also be used for acquisition of motion data of a game or a user.
The gyro sensor 612 may detect a body direction and a rotation angle of the electronic device 600, and the gyro sensor 612 and the acceleration sensor 611 may cooperate to acquire a 3D motion of the user on the electronic device 600. The processor 601 may implement the following functions according to the data collected by the gyro sensor 612: motion sensing (such as changing the UI according to a user's tilting operation), image stabilization at the time of photographing, game control, and inertial navigation.
The pressure sensor 613 may be disposed on a side bezel of the electronic device 600 and/or on an underlying layer of the touch display screen 605. When the pressure sensor 613 is disposed on the side frame of the electronic device 600, a user's grip signal on the electronic device 600 can be detected, and left-right hand recognition or shortcut operation can be performed based on the grip signal. When the pressure sensor 613 is disposed at the lower layer of the touch display screen 605, it is possible to control an operability control on the UI interface according to a pressure operation of the user on the touch display screen 605. The operability control comprises at least one of a button control, a scroll bar control, an icon control and a menu control.
The optical sensor 615 is used to collect the ambient light intensity. In one embodiment, processor 601 may control the display brightness of touch display 605 based on the ambient light intensity collected by optical sensor 615. Specifically, when the ambient light intensity is high, the display brightness of the touch display screen 605 is increased; when the ambient light intensity is low, the display brightness of the touch display screen 605 is turned down. In another embodiment, the processor 601 may also dynamically adjust the shooting parameters of the camera assembly 606 according to the ambient light intensity collected by the optical sensor 615.
Proximity sensor 616, also referred to as a distance sensor, is typically disposed on the front side of electronic device 600. The proximity sensor 616 is used to capture the distance between the user and the front of the electronic device 600. In one embodiment, when the proximity sensor 616 detects that the distance between the user and the front face of the electronic device 600 gradually decreases, the processor 601 controls the touch display screen 605 to switch from the bright screen state to the dark screen state; when the proximity sensor 616 detects that the distance between the user and the front surface of the electronic device 600 gradually becomes larger, the processor 601 controls the touch display screen 605 to switch from the breath screen state to the bright screen state.
Those skilled in the art will appreciate that the configuration shown in fig. 6 does not constitute a limitation of the electronic device 600, and may include more or fewer components than those shown, or combine certain components, or employ a different arrangement of components.
Fig. 7 is a block diagram of a server according to an embodiment of the present invention. Referring to fig. 7, server 700 includes a processing component 722 that further includes one or more processors and memory resources, represented by memory 732, for storing instructions, such as applications, that are executable by processing component 722. The application programs stored in memory 732 may include one or more modules that each correspond to a set of instructions. Further, the processing component 722 is configured to execute the instructions to perform the network resource recommendation method provided by the embodiments shown in fig. 2 and 3.
The server 700 may also include a power component 726 configured to perform power management of the server 700, a wired or wireless network interface 750 configured to connect the server 700 to a network, and an input output (I/O) interface 758. The Server 700 may operate based on an operating system, such as Windows Server, stored in a memory 732TM，Mac OS XTM，UnixTM,LinuxTM，FreeBSDTMOr the like.
In an exemplary embodiment, a computer readable storage medium is also provided, in which at least one instruction, at least one program, a set of codes, or a set of instructions is stored, and the at least one instruction, the at least one program, the set of codes, or the set of instructions is loaded and executed by a processor to implement the network resource recommendation method in the above-described embodiment. For example, the computer readable storage medium may be a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (11)

1. A method for recommending network resources, the method comprising:
acquiring function use behavior information of a target user, wherein the function use behavior information is used for expressing the use behavior of the target user on at least one function of a network resource server;
acquiring preference characteristics of the target user according to the function use behavior information of the target user;
determining the preference degree of the target user to the network resources according to the preference characteristics of the target user, the network resources to be recommended and a resource recommendation model;
when the preference degree of the target user for the network resource meets a preset recommendation condition, judging whether the network resource meets a preset supplement condition, if so, recommending the network resource to the target user;
wherein the preset supplementary condition comprises at least one of the following conditions:
the historical use information of the target user does not contain the network resource;
when the historical use information of the target user comprises the network resource, and the time interval between the stay time of the network resource and the current time is greater than or equal to a preset time length;
the score of the network resource is greater than a preset score;
the network resource name is different from the determined network resource to be recommended;
the resource recommendation model is obtained by training network resource information provided by the network resource server, function use behavior information of a plurality of sample users and historical use information of the plurality of sample users, and the sample users are determined according to the function use behavior information of the users;
wherein, according to the historical usage information and the function usage behavior information of the target user, acquiring the preference characteristics of the target user comprises:
vector feature extraction is carried out on the historical use information of the target user and the plurality of network resource information included in the function use behavior information to obtain a user vector based on network resources, a user vector based on labels and feature information of the network resources;
extracting collaborative filtering characteristics of the historical use information of the target user to obtain collaborative filtering characteristics of the target user based on network resources and collaborative filtering characteristics based on users;
and extracting user characteristics from the historical use information and the function use behavior information of the target user to obtain user portrait characteristics of the target user.
2. The method of claim 1, wherein the function usage behavior information of the plurality of sample users comprises at least one of:
user registration information, network resource list information, network resource publishing information, search click information, social network sharing information and interaction information for any network resource.
3. The method of claim 1, wherein the resource recommendation model is trained by the following process:
screening and extracting characteristics of sample original data to obtain characteristic information of a plurality of positive samples and characteristic information of a plurality of negative samples, wherein the sample original data comprises function use behavior information of sample users and historical use information of the plurality of sample users;
performing model training based on the feature information of a plurality of positive samples, the feature information of a plurality of negative samples, the reference preference degrees of the positive and negative samples and the network resource feature information of each network resource to obtain the resource recommendation model;
the positive sample refers to a set of network resources and sample users with the stay time being longer than or equal to a preset time, and the negative sample refers to a set of network resources and sample users with the stay time being shorter than the preset time.
4. The method according to claim 3, wherein the performing model training based on the feature information of the positive samples, the feature information of the negative samples, the reference preference degrees of the positive and negative samples, and the network resource feature information of each network resource to obtain the resource recommendation model comprises:
inputting a plurality of sample data into a current recommendation model, and outputting the prediction preference degrees of the sample users to the plurality of sample data;
and adjusting the weight of each feature in the current recommendation model according to the prediction preference degree and the reference preference degree of the sample data to obtain the resource recommendation model.
5. The method of claim 1, wherein recommending the network resource to the target user comprises:
and displaying the network resources at a page position corresponding to the sequencing position according to the sequencing position of the preference degree of the target user on the network resources in the determined network resources to be recommended.
6. A network resource recommendation apparatus, the apparatus comprising:
the system comprises a first acquisition module, a second acquisition module and a recommendation module, wherein the first acquisition module is used for acquiring function use behavior information of a target user to be recommended, and the function use behavior information is used for expressing the use behavior of the target user on at least one function of a network resource server;
the second acquisition module is used for acquiring the preference characteristics of the target user according to the function use behavior information of the target user;
the determining module is used for determining the preference degree of the target user on the network resources according to the preference characteristics of the target user, the network resources to be recommended and a resource recommendation model;
the recommendation module is used for judging whether the network resource meets a preset supplement condition or not when the preference degree of the target user for the network resource meets a preset recommendation condition, and recommending the network resource to the target user if the preference degree of the target user for the network resource meets the preset supplement condition;
wherein the preset supplementary condition comprises at least one of the following conditions:
the historical use information of the target user does not contain the network resource;
when the historical use information of the target user comprises the network resource, and the time interval between the stay time of the network resource and the current time is greater than or equal to a preset time length;
the score of the network resource is greater than a preset score;
the network resource name is different from the determined network resource to be recommended;
the resource recommendation model is obtained by training network resource information provided by the network resource server, function use behavior information of a plurality of sample users and historical use information of the plurality of sample users, and the sample users are determined according to the function use behavior information of the users;
the second obtaining module is configured to:
vector feature extraction is carried out on the historical use information of the target user and the plurality of network resource information included in the function use behavior information to obtain a user vector based on network resources, a user vector based on labels and feature information of the network resources;
extracting collaborative filtering characteristics of the historical use information of the target user to obtain collaborative filtering characteristics of the target user based on network resources and collaborative filtering characteristics based on users;
and extracting user characteristics according to the historical use information and the function use behavior information of the target user to obtain user portrait characteristics of the target user.
7. The apparatus of claim 6, further comprising:
the processing module is used for screening and extracting characteristics of sample original data to obtain characteristic information of a plurality of positive samples and characteristic information of a plurality of negative samples, wherein the sample original data comprises function use behavior information of sample users and historical use information of the plurality of sample users;
the resource recommendation model training module is used for carrying out model training based on the feature information of a plurality of positive samples, the feature information of a plurality of negative samples, the reference preference degrees of the positive and negative samples and the network resource feature information of each network resource to obtain a resource recommendation model;
the positive sample refers to a set of network resources and sample users with the stay time being longer than or equal to a preset time, and the negative sample refers to a set of network resources and sample users with the stay time being shorter than the preset time.
8. The apparatus of claim 7, wherein the resource recommendation model training module is configured to:
inputting a plurality of sample data into a current recommendation model, and outputting the prediction preference degrees of the sample users to the plurality of sample data;
and adjusting the weight of each feature in the current recommendation model according to the prediction preference degree and the reference preference degree of the sample data to obtain the resource recommendation model.
9. An electronic device, comprising a processor and a memory, wherein at least one instruction, at least one program, a set of codes, or a set of instructions is stored in the memory, and wherein the at least one instruction, the at least one program, the set of codes, or the set of instructions is loaded and executed by the processor to implement the network resource recommendation method according to any one of claims 1 to 5.
10. A server, comprising a processor and a memory, wherein the memory has stored therein at least one instruction, at least one program, a set of codes, or a set of instructions, which is loaded and executed by the processor to implement the network resource recommendation method of any of claims 1 to 5.
11. A computer readable storage medium having stored therein at least one instruction, at least one program, a set of codes, or a set of instructions, which is loaded and executed by a processor to implement the network resource recommendation method of any of claims 1 to 5.
CN201711120052.2A 2017-11-14 2017-11-14 Network resource recommendation method and device, electronic equipment, server and storage medium Active CN108304441B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711120052.2A CN108304441B (en) 2017-11-14 2017-11-14 Network resource recommendation method and device, electronic equipment, server and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711120052.2A CN108304441B (en) 2017-11-14 2017-11-14 Network resource recommendation method and device, electronic equipment, server and storage medium

Publications (2)

Publication Number Publication Date
CN108304441A CN108304441A (en) 2018-07-20
CN108304441B true CN108304441B (en) 2022-04-26

Family

ID=62869590

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711120052.2A Active CN108304441B (en) 2017-11-14 2017-11-14 Network resource recommendation method and device, electronic equipment, server and storage medium

Country Status (1)

Country Link
CN (1) CN108304441B (en)

Families Citing this family (59)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109271550B (en) * 2018-07-27 2022-05-24 华南理工大学 Music personalized recommendation method based on deep learning
CN110796505B (en) * 2018-08-03 2023-07-04 淘宝（中国）软件有限公司 Business object recommendation method and device
CN109344314B (en) * 2018-08-20 2021-11-16 腾讯科技（深圳）有限公司 Data processing method and device and server
CN110874639A (en) * 2018-08-31 2020-03-10 珠海格力电器股份有限公司 Method and device for acquiring operation information
CN110147882B (en) * 2018-09-03 2023-02-10 腾讯科技（深圳）有限公司 Neural network model training method, crowd diffusion method, device and equipment
CN109447622B (en) * 2018-09-30 2022-02-08 中国银行股份有限公司 Transaction type recommendation method and system and intelligent transaction terminal
CN109408714B (en) * 2018-10-15 2021-02-05 上海趣蕴网络科技有限公司 Multi-model fusion recommendation system and method
CN109408724B (en) * 2018-11-06 2020-10-02 北京达佳互联信息技术有限公司 Method and device for determining pre-estimated click rate of multimedia resource and server
CN109299314B (en) * 2018-11-13 2019-12-27 百度在线网络技术（北京）有限公司 Music retrieval and recommendation method, device, storage medium and terminal equipment
CN111274472A (en) * 2018-12-04 2020-06-12 北京嘀嘀无限科技发展有限公司 Information recommendation method and device, server and readable storage medium
CN109284445B (en) * 2018-12-11 2020-12-29 北京达佳互联信息技术有限公司 Network resource recommendation method and device, server and storage medium
CN109547814B (en) * 2018-12-13 2021-07-16 北京达佳互联信息技术有限公司 Video recommendation method and device, server and storage medium
CN109508426A (en) * 2018-12-21 2019-03-22 深圳市智搜信息技术有限公司 A kind of intelligent recommendation method and its system and storage medium based on physical environment
CN109697258A (en) * 2018-12-27 2019-04-30 丹翰智能科技(上海)有限公司 It is a kind of for determining the method and apparatus of the customization financial information of target user
CN109740056A (en) * 2018-12-28 2019-05-10 丹翰智能科技(上海)有限公司 It is a kind of for provide a user customization financial information method and apparatus
CN109885719A (en) * 2019-01-24 2019-06-14 广州小鹏汽车科技有限公司 A kind of song recommendations method, system, terminal and storage medium
CN109994000B (en) * 2019-03-28 2021-10-19 掌阅科技股份有限公司 Reading accompanying method, electronic equipment and computer storage medium
CN111753124A (en) * 2019-03-29 2020-10-09 Tcl集团股份有限公司 Music recommendation method and server
CN110134820B (en) * 2019-04-26 2020-12-11 湖南大学 Feature increasing based hybrid personalized music recommendation method
CN111858970B (en) * 2019-04-30 2024-01-02 北京达佳互联信息技术有限公司 Multimedia content recommendation method and device, electronic equipment and readable storage medium
CN112000819B (en) * 2019-05-27 2023-07-11 北京达佳互联信息技术有限公司 Multimedia resource recommendation method and device, electronic equipment and storage medium
CN110413837B (en) * 2019-05-30 2023-07-25 腾讯科技（深圳）有限公司 Video recommendation method and device
CN110297939A (en) * 2019-06-21 2019-10-01 山东科技大学 A kind of music personalization system of fusion user behavior and cultural metadata
CN110287418B (en) * 2019-06-26 2021-10-19 拉扎斯网络科技（上海）有限公司 Data processing and tendency characteristic value identification method and device, and electronic equipment
CN112148964B (en) * 2019-06-29 2022-11-18 阿里巴巴集团控股有限公司 Information processing and recommending method, system and equipment
CN112182460B (en) * 2019-07-04 2024-03-15 深圳市雅阅科技有限公司 Resource pushing method and device, storage medium and electronic device
CN110297848B (en) * 2019-07-09 2024-02-23 深圳前海微众银行股份有限公司 Recommendation model training method, terminal and storage medium based on federal learning
CN112241327A (en) * 2019-07-17 2021-01-19 腾讯科技（深圳）有限公司 Shared information processing method and device, storage medium and electronic equipment
CN110704739B (en) * 2019-09-30 2022-07-15 汉海信息技术(上海)有限公司 Resource recommendation method and device and computer storage medium
CN110851706B (en) * 2019-10-10 2022-11-01 百度在线网络技术（北京）有限公司 Training method and device for user click model, electronic equipment and storage medium
CN110866180B (en) * 2019-10-12 2022-07-29 平安国际智慧城市科技股份有限公司 Resource recommendation method, server and storage medium
CN110830812B (en) * 2019-10-31 2021-11-30 广州市网星信息技术有限公司 Similar anchor classification model training method, anchor recommendation method and related device
CN110910209A (en) * 2019-11-12 2020-03-24 腾讯科技（深圳）有限公司 Data processing method and device and computer readable storage medium
CN112804566A (en) * 2019-11-14 2021-05-14 中兴通讯股份有限公司 Program recommendation method, device and computer readable storage medium
CN110874767A (en) * 2019-11-19 2020-03-10 山东爱城市网信息技术有限公司 Business hall accurate marketing method based on collaborative filtering
CN110889020B (en) * 2019-11-22 2022-08-23 百度在线网络技术（北京）有限公司 Site resource mining method and device and electronic equipment
CN111159435B (en) * 2019-12-27 2023-09-05 新方正控股发展有限责任公司 Multimedia resource processing method, system, terminal and computer readable storage medium
CN113365115B (en) * 2020-03-03 2022-11-04 北京达佳互联信息技术有限公司 Characteristic code determining method, device, server and storage medium
CN113495985B (en) * 2020-03-18 2024-03-19 北京达佳互联信息技术有限公司 Article recommendation method and device, short video recommendation method and device and server
CN111556366A (en) * 2020-04-02 2020-08-18 北京达佳互联信息技术有限公司 Multimedia resource display method, device, terminal, server and system
CN113535991B (en) * 2020-04-14 2024-01-26 北京达佳互联信息技术有限公司 Multimedia resource recommendation method and device, electronic equipment and storage medium
CN111625713A (en) * 2020-04-30 2020-09-04 平安国际智慧城市科技股份有限公司 Resource recommendation method and device based on big data, electronic equipment and medium
CN111737573A (en) * 2020-06-17 2020-10-02 北京三快在线科技有限公司 Resource recommendation method, device, equipment and storage medium
CN113836390B (en) * 2020-06-24 2023-10-27 北京达佳互联信息技术有限公司 Resource recommendation method, device, computer equipment and storage medium
CN111814051B (en) * 2020-07-08 2023-07-28 北京奇艺世纪科技有限公司 Resource type determining method and device
CN114071119B (en) * 2020-07-31 2024-03-19 北京达佳互联信息技术有限公司 Resource testing method and device, electronic equipment and storage medium
CN113760521A (en) * 2020-09-22 2021-12-07 北京沃东天骏信息技术有限公司 Virtual resource allocation method and device
CN112199022A (en) * 2020-10-23 2021-01-08 Oppo广东移动通信有限公司 Content display method and device, electronic equipment and storage medium
CN112200639A (en) * 2020-10-30 2021-01-08 杭州时趣信息技术有限公司 Information flow model construction method, device and medium
CN112507163B (en) * 2020-12-02 2023-07-21 北京奇艺世纪科技有限公司 Duration prediction model training method, recommendation method, device, equipment and medium
CN112765470B (en) * 2021-01-25 2022-02-22 腾讯科技（深圳）有限公司 Training method of content recommendation model, content recommendation method, device and equipment
CN115079908A (en) * 2021-03-12 2022-09-20 北京字节跳动网络技术有限公司 Media content collection method and device, electronic equipment and computer readable storage medium
CN113486373A (en) * 2021-07-13 2021-10-08 苏州医沃智控科技有限公司 eCTD universal technology document submission management method and system
CN113344647B (en) * 2021-07-14 2023-05-23 杭州网易云音乐科技有限公司 Information recommendation method and device
CN113539457A (en) * 2021-07-16 2021-10-22 挂号网（杭州）科技有限公司 Medical resource recommendation method and device, electronic equipment and storage medium
CN113382289B (en) * 2021-08-11 2021-11-02 北京达佳互联信息技术有限公司 Live broadcast room delivery method and device, electronic equipment and storage medium
CN113722588B (en) * 2021-08-12 2023-09-05 北京达佳互联信息技术有限公司 Resource recommendation method and device and electronic equipment
TR2021020670A2 (en) * 2021-12-22 2022-01-21 Turkcell Technology Research And Development Co A SYSTEM THAT PROVIDES CREATING PERSONAL SONG
CN114779999A (en) * 2022-04-27 2022-07-22 北京达佳互联信息技术有限公司 Information display method and device, electronic equipment and storage medium

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102402625A (en) * 2011-12-28 2012-04-04 深圳市五巨科技有限公司 Method and system for recommending music
CN102523511A (en) * 2011-11-09 2012-06-27 中国传媒大学 Network program aggregation and recommendation system and network program aggregation and recommendation method
CN104021233A (en) * 2014-06-30 2014-09-03 电子科技大学 Social network friend recommendation method based on community discovery
CN105868254A (en) * 2015-12-25 2016-08-17 乐视网信息技术（北京）股份有限公司 Information recommendation method and apparatus
CN106874522A (en) * 2017-03-29 2017-06-20 珠海习悦信息技术有限公司 Information recommendation method, device, storage medium and processor

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102523511A (en) * 2011-11-09 2012-06-27 中国传媒大学 Network program aggregation and recommendation system and network program aggregation and recommendation method
CN102402625A (en) * 2011-12-28 2012-04-04 深圳市五巨科技有限公司 Method and system for recommending music
CN104021233A (en) * 2014-06-30 2014-09-03 电子科技大学 Social network friend recommendation method based on community discovery
CN105868254A (en) * 2015-12-25 2016-08-17 乐视网信息技术（北京）股份有限公司 Information recommendation method and apparatus
CN106874522A (en) * 2017-03-29 2017-06-20 珠海习悦信息技术有限公司 Information recommendation method, device, storage medium and processor

Also Published As

Publication number Publication date
CN108304441A (en) 2018-07-20

Similar Documents

Publication Publication Date Title
CN108304441B (en) Network resource recommendation method and device, electronic equipment, server and storage medium
CN109740068B (en) Media data recommendation method, device and storage medium
US11540019B2 (en) Video recommendation method and device, computer device and storage medium
WO2019233219A1 (en) Dialogue state determining method and device, dialogue system, computer device, and storage medium
CN109918669B (en) Entity determining method, device and storage medium
CN109657236B (en) Guidance information acquisition method, apparatus, electronic apparatus, and storage medium
CN110163066B (en) Multimedia data recommendation method, device and storage medium
CN111506758B (en) Method, device, computer equipment and storage medium for determining article name
WO2021164652A1 (en) Method for displaying and method for providing multimedia resource
CN111897996B (en) Topic label recommendation method, device, equipment and storage medium
CN110572716B (en) Multimedia data playing method, device and storage medium
CN108270794B (en) Content distribution method, device and readable medium
CN111291200B (en) Multimedia resource display method and device, computer equipment and storage medium
WO2022057435A1 (en) Search-based question answering method, and storage medium
CN111611490A (en) Resource searching method, device, equipment and storage medium
CN111858971A (en) Multimedia resource recommendation method, device, terminal and server
CN111368127B (en) Image processing method, image processing device, computer equipment and storage medium
CN112235635B (en) Animation display method, animation display device, electronic equipment and storage medium
CN111031391A (en) Video dubbing method, device, server, terminal and storage medium
CN113918767A (en) Video clip positioning method, device, equipment and storage medium
CN113032587B (en) Multimedia information recommendation method, system, device, terminal and server
CN112069350A (en) Song recommendation method, device, equipment and computer storage medium
CN113987326B (en) Resource recommendation method and device, computer equipment and medium
CN113377976B (en) Resource searching method and device, computer equipment and storage medium
CN110929137A (en) Article recommendation method, article recommendation device, article recommendation 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
GR01 Patent grant
GR01 Patent grant