CN112328888A - Information recommendation method and device, server and storage medium - Google Patents

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

Info

Publication number
CN112328888A
CN112328888A CN202011312572.5A CN202011312572A CN112328888A CN 112328888 A CN112328888 A CN 112328888A CN 202011312572 A CN202011312572 A CN 202011312572A CN 112328888 A CN112328888 A CN 112328888A
Authority
CN
China
Prior art keywords
image
user
determining
terminal
images
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011312572.5A
Other languages
Chinese (zh)
Inventor
薛致远
李亚乾
郭彦东
杨林
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.)
Guangdong Oppo Mobile Telecommunications Corp Ltd
Original Assignee
Guangdong Oppo Mobile Telecommunications Corp 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 Guangdong Oppo Mobile Telecommunications Corp Ltd filed Critical Guangdong Oppo Mobile Telecommunications Corp Ltd
Priority to CN202011312572.5A priority Critical patent/CN112328888A/en
Publication of CN112328888A publication Critical patent/CN112328888A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/5866—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, manually generated location and time information
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features

Abstract

The application relates to an information recommendation method, an information recommendation device, a server and a storage medium, and belongs to the technical field of internet. The method comprises the following steps: acquiring a plurality of images in an album application corresponding to a terminal; determining an image label for each image; determining a user portrait of a target user based on the image tags of the plurality of images, the target user being a user using the terminal; recommending target information matched with the user portrait to the terminal based on the user portrait, determining the user portrait through an image in an album application in the terminal, and further recommending that the target information matched with the user portrait is generally a favorite image of the user because the image in the album application in the terminal is generally the favorite image of the user; therefore, the interest of the user can be mined based on the image in the album application, so that the accuracy of the user portrait determined based on the image is improved, and the accuracy of the target information recommended based on the user portrait is improved.

Description

Information recommendation method and device, server and storage medium
Technical Field
The embodiment of the application relates to the technical field of internet, in particular to an information recommendation method, an information recommendation device, a server and a storage medium.
Background
With the development of terminal technology, the functions of the terminal are more and more abundant. The terminal not only supports the search function, but also supports the recommendation function, namely the terminal actively recommends information which may be interested in for the user.
In the related technology, a terminal acquires browsing records and searching records of a user, performs interest analysis on the user according to the browsing records and the searching records to obtain interest characteristics of the user, determines target information interested by the user according to the interest characteristics of the user, and recommends the target information to the user.
Disclosure of Invention
The embodiment of the application provides an information recommendation method, an information recommendation device, a server and a storage medium, and can improve the information conversion rate. The technical scheme is as follows:
in one aspect, an information recommendation method is provided, and the method includes:
acquiring a plurality of images in an album application corresponding to a terminal;
determining an image label for each image;
determining a user portrait of a target user based on image tags of the plurality of images, the target user being a user using the terminal;
and recommending the target information matched with the user portrait to the terminal based on the user portrait.
In another aspect, an information recommendation apparatus is provided, the apparatus including:
the acquisition module is used for acquiring a plurality of images in the photo album application corresponding to the terminal;
a label determination module for determining an image label for each image;
a portrait determination module to determine a user portrait of a target user based on image tags of the plurality of images, the target user being a user using the terminal;
and the information recommending module is used for recommending the target information matched with the user portrait to the terminal based on the user portrait.
In another aspect, a server is provided, the server comprising a processor and a memory; the memory stores at least one program code for execution by the processor to implement the information recommendation method of the above aspect.
In another aspect, a computer-readable storage medium is provided, the storage medium storing at least one program code for execution by the processor to implement the information recommendation method of the above aspect.
In another aspect, a computer program product is provided, which stores at least one program code, and the at least one program code is loaded and executed by a processor to implement the information recommendation method of the above aspect.
In the embodiment of the application, the images in the album application in the terminal are generally favorite images of the user; therefore, the interest of the user can be mined based on the image in the album application, so that the accuracy of the user portrait determined based on the image is improved, and the accuracy of the target information recommended based on the user portrait is improved.
Drawings
FIG. 1 illustrates an implementation environment of an information recommendation method shown in an exemplary embodiment of the present application;
FIG. 2 illustrates a flow chart of a method of information recommendation shown in an exemplary embodiment of the present application;
FIG. 3 illustrates a flow chart of a method of information recommendation shown in an exemplary embodiment of the present application;
FIG. 4 illustrates a flow chart of a method of information recommendation shown in an exemplary embodiment of the present application;
FIG. 5 illustrates a flow chart of a method of information recommendation shown in an exemplary embodiment of the present application;
FIG. 6 illustrates a flow chart of a method of information recommendation shown in an exemplary embodiment of the present application;
fig. 7 is a block diagram illustrating a structure of an information recommendation apparatus according to an exemplary embodiment of the present application;
fig. 8 shows a block diagram of a server according to an exemplary embodiment of the present application.
Detailed Description
To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.
Reference herein to "a plurality" means two or more. "and/or" describes the association relationship of the associated objects, meaning that there may be three relationships, e.g., A and/or B, meaning: a exists alone, A and B exist simultaneously, and B exists alone. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship.
Referring to fig. 1, an implementation environment of an information recommendation method according to an exemplary embodiment of the present application is shown. The implementation environment includes: a terminal 101 and a server 102. Wherein, the terminal 101 and the server 102 perform data interaction through a wireless network or a data interface.
The terminal 101 is a terminal 101 having an image capturing function. The terminal 101 has an album application and other applications installed therein. The album application and other applications are applications installed by default in the system of the terminal 101, or the album application is an application installed by the terminal 101 through downloading of an application installer. In the embodiments of the present application, this is not particularly limited. The album application is used to store the images collected by the terminal 101. The other applications are applications with an information recommendation function and are used for recommending target information to the terminal. For example, other applications are shopping applications, information browsing applications, social applications, and the like.
The image is an image received by the terminal 101, an image captured by the terminal 101, or an image captured by the terminal 101. In the embodiment of the present application, the image type is not particularly limited. In some embodiments, the terminal 101 receives an image transmitted by another device, and stores the received image in the album application in response to the storing operation. In some embodiments, the terminal 101 is a terminal 101 with an image capture function, and the terminal 101 captures an image by capturing a current display content. In some embodiments, the terminal 101 is a camera-mounted device, and the terminal 101 captures images through the mounted camera. In some embodiments, the terminal 101 is connected to an image capturing device, and captures an image through the image capturing device, and sends the captured image to the terminal 101. The terminal 101 stores the captured image in a storage space of the album application.
The image is also a video frame image of a video file. The manner in which the terminal 101 acquires the video file is similar to the manner in which the terminal acquires the image, and is not described herein again.
In some embodiments, server 102 is a server 102 for user representation analysis, and server 102 also interacts with background servers of other applications for data. Correspondingly, the server 102 sends the user portrait obtained by analysis to the background servers of other applications, the background servers 102 of other applications determine target information matched with the user portrait based on the user portrait, and the target information is pushed to the terminal 101 used by the user. In some embodiments, the server 102 is a background server for other applications, and the target user logs in the server 102 through the terminal 101. Accordingly, the server 102 can analyze the user profile of the target user and recommend related information to the terminal 101 based on the user profile obtained by the analysis. For example, if the other application is a shopping application and the server 102 is a background server of the shopping application, the server 102 can determine a product that the user may like by analyzing the user image by the user, and push information related to the product to the terminal 101 used by the user.
In some embodiments, the terminal 101 is a mobile phone, a computer, a tablet, or a wearable device, etc. The server 102 is a single server, or the server 102 is a server cluster including a plurality of servers, or the server 102 is a cloud server or the like. In the embodiment of the present application, the terminal 101 and the server 102 are not particularly limited.
Referring to fig. 2, a flowchart of an information recommendation method according to an exemplary embodiment of the present application is shown. The execution subject in the embodiment of the present application is described as an example of a server. The method comprises the following steps:
step 201: the server acquires a plurality of images in the album application corresponding to the terminal.
The images are pictures in an album application or video frames of a video file in the album application. And the plurality of images are all images in the terminal album; or the images are newly added in a period of time before the current time in the terminal album. In the embodiment of the present application, the number of the plurality of images is not particularly limited.
The photo album application is a photo album application installed in the terminal, an electronic photo album in the social application platform, or a cloud photo album corresponding to the terminal, and accordingly, in one possible implementation manner, the server obtains a plurality of images of the photo album application installed in the terminal. And if the album application is the album application for storing the images in the terminal, the server calls the plurality of images stored in the storage area corresponding to the album application stored in the terminal. The server sends an acquisition request to the terminal, and the terminal sends a plurality of images to the server based on the acquisition request.
In one possible implementation manner, the server acquires a plurality of images of the album application in the application platform of the social application in the terminal. For example, a user uploads an image to a platform of a social application through a terminal, the uploaded image is automatically stored in a social application server of the social application, and the server calls an application in the album from the social application server through an account of the social application. The social application server and the server are the same server or different servers, which are not specifically limited in this embodiment of the application.
In one possible implementation manner, the server acquires a plurality of images of a cloud album in the server in which the terminal logs in. The cloud album server and the server are the same or different servers, and in the embodiment of the present application, this is not particularly limited. For example, the server obtains the plurality of images from a database corresponding to an album application of the terminal. Correspondingly, a plurality of images of the terminal are stored in the database, and correspondingly, the server calls the plurality of images in the album application of the terminal from the database. In the embodiment of the present application, the manner in which the server acquires the plurality of images is not particularly limited.
Step 202: the server determines an image tag for each image.
Wherein, the image label refers to identifying the elements existing in the image and describing the image in the form of a label. For example, the image tags are: sky, grass, trees, parks, lakes, people, etc.
With continued reference to FIG. 3, in this step, the server generates image tags for images in the album application. In some implementations, the server obtains the label of the image annotation directly. Correspondingly, when the terminal stores the image, the image is labeled, and the label is the name, the shooting time, the shooting place and the like of the image. In this step, the server acquires annotation information of the image, and determines an image tag of the image from the annotation information. In other embodiments, the server determines an image tag for the image based on the content of the image. Accordingly, the server extracts image features of the image and generates an image tag of the image based on the extracted image features. In some embodiments, the server performs a comprehensive analysis of all extracted features to obtain an image tag of the image. Accordingly, the last layer of the image tag determination model is a single-class activation layer, e.g., a softmax function layer. Thus, the image tag determination model can determine only one image tag of the image in the process of determining the image tag. In other embodiments, the server generates a plurality of image tags based on the extracted image features, respectively, from which image tags for the image are determined. Accordingly, the image tag determines the last layer of the model as a multi-class activation layer, e.g., a sigmoid function. Referring to fig. 4, the present step is implemented by the following steps (1) to (3) in response to the label determination model outputting a plurality of image labels, including:
(1) for each image, the server determines a plurality of image labels for the image from the label determination model.
The server acquires a label determination model, and the activation layer of the label determination model is a multi-classification activation layer. In some embodiments, the label determination model includes an input layer, a feature extraction layer, a label generation layer, and an activation layer. The label determination model is a model constructed using any type of Neural Network, for example, the label determination model is a model constructed using a Convolutional Neural Network (CNN). Wherein, any Residual Network type in a Residual Network (ResNet) is used as a feature extraction layer in the model. For example, ResNet101 is adopted as the CNN feature extraction layer.
The input layer is used for receiving an input image, convolving the image to obtain an image to be processed matched with the feature extraction layer, convolving the image to be processed through the feature extraction layer to obtain image features of the image to be processed, inputting the image features to the label generation layer, and performing feature analysis on the image features through the label generation layer to obtain a plurality of image labels of the image.
(2) The server determines a multi-classification activation layer of the model based on the label, performs probability prediction on the plurality of image labels of the image, and obtains prediction probabilities of the plurality of image labels of the image.
In this step, the server inputs the obtained image labels to the multi-classification activation layer, and performs probability prediction on the image labels through the multi-classification activation layer to obtain prediction probabilities of the image labels. In some embodiments, the server performs probability prediction on all the obtained image tags to obtain the probability of each image tag. For example, the server-derived image tags include: sky, grassland, tree, park, lake, people, in this step, the server predicts the probability of these labels of sky, grassland, tree, park, lake, people through the multi-classification active layer.
In some embodiments, the server classifies the obtained image tags to obtain multiple types of image tags, and performs probability prediction on each type of image tag respectively. For example, the server-derived image tags include: sky, grassland, tree, park, lake, people, then in this step, the server determines the category of these image labels first, then the label includes thing label: sky, tree, human; further comprising a scene tag: grassland, parks, lakes and the server carries out probability prediction on the affair class labels and the scene labels through a multi-classification activation layer.
(3) The server determines an image label with a prediction probability larger than a preset probability from the output image labels as the image label of the image.
The preset probability is set according to needs, and in the embodiment of the present application, the preset probability is not specifically limited.
It should be noted that the server can directly determine, from all image tags, an image tag with a prediction probability greater than a preset probability. The server can also respectively determine the image labels of which the prediction probability exceeds the preset probability corresponding to the category in each category based on the category of the image labels. The preset probabilities corresponding to each category are the same or different, and in the embodiment of the present application, this is not specifically limited.
The server may further determine an image tag having the highest prediction probability from among the plurality of image tags as the image tag of the image. Correspondingly, the steps are replaced by: the server determines an image label with the highest prediction probability from the output image labels as the image label of the image.
In the implementation mode, the label determination model is used for determining the plurality of image labels of the image, and the image label with the prediction probability meeting the condition is determined from the plurality of image labels as the image label of the image, so that the image label of the image is determined from the plurality of image labels, and the accuracy rate of determining the image label is improved.
Another point to be noted is that in some embodiments, the server can directly acquire the image in the terminal album application. In other embodiments, the server obtains the access right of the album application in the terminal, and the server can obtain the image from the album application after obtaining the access right. Correspondingly, before the step, the server sends an authority acquisition request to the terminal, correspondingly, the terminal displays an authority acquisition message, responds to the permission access operation, the terminal sends an access authority to the server, and the server executes the step 201 after receiving the access authority.
Step 203: the server determines a user representation of the target user based on the image tags from the plurality of images.
Wherein, the target user is a user using the terminal. And the server analyzes the user portrait of the user using the terminal through the image tags of the plurality of images in the photo album application to obtain the user portrait of the terminal. With continued reference to FIG. 3, in this step, the server models the user interests based on the image tags of the plurality of images to obtain a user representation. Wherein the server determines at least one valid image tag from the image tags of the plurality of images, from which the user representation of the terminal is determined. Referring to fig. 5, the process is realized by the following steps (1) - (2), including:
(1) the server determines at least one valid image tag from the plurality of image tags based on the image tags of the plurality of images.
Wherein the active image tag refers to an image tag that can be used to determine a user representation. The effective image tags are determined based on the image parameter requirements such as the number of the same image tags, the appearance frequency of the image tags, the generation time period of the image tags, and the shooting position information of the image.
In some embodiments, the server determines, based on the image tags of the plurality of images, a number of images to which each image tag corresponds; from the plurality of image tags, at least one valid image tag is determined, the number of which exceeds a preset number.
The server determines image tags existing in the plurality of image tags, counts images corresponding to different image tags to obtain the number of images corresponding to each image tag, and then determines at least one effective image tag of which the number exceeds a preset number from the plurality of image tags.
It should be noted that the preset number is set as needed, and in the embodiment of the present application, the preset number is not particularly limited. For example, the predetermined number is 20, 30, 50, etc. And, the number of the at least one effective image tag is set as required. In some embodiments, the server takes as the at least one valid image tag all tags for which the number of images exceeds a preset number. In some embodiments, the server determines the number of the at least one valid image tag in advance, and in response to the number of the image tags meeting the preset number exceeding the predetermined number of the at least one valid image tag, selects a higher number of image tags from the image tags whose number of images exceeds the preset number as the at least one valid image tag.
The server may be further configured to determine, based on the number of the plurality of image tags, an image tag having a largest number from among the plurality of image tags as an image tag satisfying the portrait requirement.
In the implementation mode, the image label corresponding to the image with the largest number in the terminal is determined, and the image label is determined to be the effective image label, so that the favorite features of the user using the terminal are known based on the number of the image labels, and then the related information is recommended to the user, and the conversion rate of the recommended information is improved.
In some embodiments, the server determines a frequency of occurrence of each image tag based on the image tags of the plurality of images; at least one valid image tag, whose frequency of occurrence exceeds a preset frequency, is determined from the plurality of image tags.
Wherein, the appearance frequency refers to the proportion of each image label in the image labels in the total image. In this implementation, the server determines, as an effective image tag, an image tag having a high frequency of appearance from among the plurality of kinds of image tags.
In this implementation manner, the server determines at least one valid image tag based on the occurrence frequency of each image tag, thereby determining the favorite features of the user using the terminal based on the at least one valid image tag, and further recommending related information to the user, thereby improving the conversion rate of the recommended information.
And, the server can also determine at least one valid image tag in combination with the number of images and the frequency of occurrence of the image tags. At least one valid image tag is determined by determining the frequency of occurrence of each image tag when the number of images in the album application is low. Therefore, the problem that the effective image label cannot be determined due to the fact that the number of the images does not exceed the preset number when the number of the images in the album application is low is solved.
In some embodiments, the server determines the generation time of each image, and determines at least one effective image tag from a plurality of image tags based on the generation time of each image, wherein the interval between the generation time of the image corresponding to the at least one effective image tag and the current time is less than the preset time length.
The preset duration is set according to needs, and in the embodiment of the present application, the preset duration is not specifically limited. For example, the preset time period is 5 hours, 24 hours, one week, or the like. In the implementation mode, the server determines the image tag of the image as the effective image tag by determining the image tag of the image appearing within the preset time length, so that the information which is recently interested by the user using the terminal is known, the information is pushed to the user based on the information, and the conversion rate of the pushed information is improved.
In some embodiments, the server determines at least one valid image tag for the target time period for the generation time of the image based on the image tags of the plurality of images. In this implementation, the server determines a target time period, determines image tags for images generated within the target time period based on the time period, and determines the image tags as at least one valid image tag. For example, the server determines that the target time period is eleven o 'clock to one point at noon, and determines the image label of the image generated between eleven o' clock and one point as the valid image label. Or, in response to the time period in which the current time is a festival period, the server determines the image tag of the image generated during the festival period to be a valid image tag.
In an implementation mode, the image tags of the images generated in the target time period are determined, so that the server pushes the information related to the target time period to the terminal, and the conversion rate of the pushed information is improved.
In some embodiments, the server determines at least one valid image tag for which the generation location of the image matches the target location based on the image tags of the plurality of images. In the implementation manner, at least one effective image tag shot at the target position is determined based on the position information, so that the server can determine the user preference of the user at the target position, so that the information related to the target position is pushed to the terminal, and the conversion rate of the pushed information is improved.
It should be noted that all the above technical solutions for determining at least one valid image tag can be combined arbitrarily to form an optional embodiment of the present application, and are not described herein again.
(2) The server determines a user representation of the target user based on the tag type of the at least one active image tag.
Wherein the server constructs the user representation based on the different image tags. The process is realized by the following steps (2-1) - (2-4), and comprises the following steps:
(2-1) in response to the image tag being a subject, the server determines subject preference characteristics of the target user based on the subject.
In some embodiments, the server directly determines the subject as a subject preference feature of the target user. For example, the subject of the at least one valid image tag is a car, book, food, etc., which can establish that the subject of the target user prefers to purchase or understand the product of the car, book, food, etc. In other embodiments, the server determines the subject-related content as a subject preference characteristic of the target user based on the subject of the at least one active image tag. For example, if the image tag is an infant, the subject preference of the target user is constructed to be a tendency to buy mother and infant products.
(2-2) in response to the at least one valid image tag being a scene, the server determines a travel preference feature of the target user based on the scene.
For example, if the image tag is a restaurant, bar, etc., then the user representation is constructed to be the target user's preference for dining at the restaurant, bar, etc. For another example, if the image tag is a scene of a sea, a park, a scenic spot and an ancient site, the target user portrait is constructed as a favorite tour.
And (2-3) in response to the image tag being a behavior, the server determines a behavior preference characteristic of the terminal according to the behavior.
For example, the image tag is running, sports, dinner, etc., and the user representation of the target user is constructed as a favorite sports product, a favorite organization building, etc.
(2-4) the server constructing a user representation of the target user based on at least one of the subject preference feature, the travel preference feature and the behavior preference feature.
In this step, the server can determine the user representation for each preference feature, and determine these user representations as the user representation of the target user, or combine these user representations into a user representation of the target user. The process is as follows: the server determines a preference main body of the target user based on the main body preference characteristic, and determines the preference main body as a user portrait of the target user; the server determines a preference scene of the target user based on the travel preference characteristics, and determines the preference scene as a user portrait of the target user; and determining the preferred behavior of the target user based on the behavior preference characteristics, and determining the main body and the scene generated by the preferred behavior as the user portrait of the target user.
For example, if the image tag is an animal, it is determined that the subject preference characteristic of the target user is an animal, that is, if the subject of preference is an animal, the favorite animal is taken as the user portrait of the target user. Or, if the image label is the scenic spot, determining the target to enable the scene preference characteristic of the user to be the scenic spot, namely, if the preferred scene is the scenic spot, taking the scenic spot which is liked to visit as the user portrait of the target user. Or if the image tag is running, determining that the behavior preference characteristic of the target user is running, namely the preference behavior is running, determining that the main body and the scene corresponding to the preference behavior are respectively sports shoes and a gymnasium according to the preference behavior, and determining that the user of the target user looks like the sports shoes and the gymnasium.
The server directly determines any preference characteristic as the user portrait of the target user, or the server performs combined analysis on the multiple preference characteristics to obtain the user portrait of the target user. For example, the server determines that the at least one active image tag is a sea, park, attraction scene, and can then determine that the user prefers to go to the sea, park, or attraction, respectively. The server may also be capable of combining these preference characteristics to determine a user representation of the target user. For example, if the subject preference of the target user is dog and the behavior preference of the target user is walking, the user portrait of the target user is determined to be a dog walking preference.
Additionally, the at least one active image tag is an image tag determined based on different portrait parameter requirements. Accordingly, in this step, the server may request certain image tags in conjunction with different portrait parameters. For example, the at least one valid image tag is an image tag in which the number of images is greater than a preset number, and the generation time of the image is within a target time period; the image labels with the image quantity larger than the preset quantity are restaurant A, the target time period is twelve o 'clock to one o' clock at noon, and the corresponding labels are gourmet. The server determines that the user portrait of the target user likes to go to restaurant a at twelve o 'clock to one o' clock at noon.
In the implementation mode, the image label is determined by combining different portrait parameter requirements, so that the determination dimensionality of the portrait of the user is enriched, and the accuracy of the portrait of the user is improved.
Step 204: the server recommends the target information matched with the user portrait to the terminal based on the user portrait.
The target information is push messages of other applications or first screen display contents in the other applications. The display form of the target information is not particularly limited in the embodiments of the present application. The target information includes: at least one of advertisement information, download information of an application program, and recommendation information of a subject. For example, the target information is commodity information recommended by a shopping application program; or, an application download link recommended for the application installation platform; or theme information for the terminal to use, etc.
In some embodiments, the server determines target information of the user portrait matching from an information base, and recommends the target information to the terminal. Referring to fig. 6, the process is realized by the following steps (1) - (2), including:
(1) the server determines, based on the user representation, target information from an information repository for which the user representation matches.
The information base is corresponding to other application programs, and relevant information of the other applications is stored in the information base.
In this step, the server determines a need of a user using the terminal from the user representation based on the user representation, and determines the target information from the information base of the other application based on the need. For example, if the user portrait is a favorite car, and the other application is a shopping application, a purchase link of the car is recommended to the terminal; or, the other application is a message browsing application, and then the news information related to the automobile is pushed to the terminal.
The server may recommend the target information directly to the terminal, and the server may transmit the user image to another server corresponding to the other application, and the other server may determine the target information from the information base. The process of determining the target information by other servers is similar to the process of determining the target information by the server, and is not described herein again.
(2) The server recommends the target information to the terminal.
Wherein the plurality of images carry a terminal identification of the terminal. In some embodiments, the server sends the target information to the terminal directly based on the terminal identification of the terminal. In some embodiments, the server determines a target account number logged in by other applications in the terminal based on the terminal identifier, and sends the target information to the terminal based on the target account number.
In the implementation mode, the target information is determined through the user portrait, and the target information is recommended to the terminal, so that the recommended target information is matched with the user portrait, and the conversion rate of the recommended target information is improved.
Additionally, the server may be further capable of determining the target information in conjunction with a portrait parameter request for generating an image tag of the user portrait. In some embodiments, the server recommends, to the terminal, the target information corresponding to at least one valid image tag whose number of images exceeds a preset number. For example, if the server determines that at least one valid image tag having the number of images exceeding the preset number is a book, it determines that the target user likes reading a book, and recommends the bookstore or the book to the terminal.
In some embodiments, the server recommends, to the terminal, target information corresponding to at least one valid image tag whose frequency of occurrence of images exceeds a preset frequency. For example, if the server determines that at least one valid image tag with the image appearance frequency exceeding the preset frequency is an automobile, it determines that the target user likes the automobile, and recommends automobile exhibition information or automobile information to the terminal.
In some embodiments, the server recommends, to the terminal, target information corresponding to at least one valid image tag for which an interval between the generation time of the image and the current time is less than a preset duration. For example, the server determines that an image of a baby does not appear in the user's album before, and an image of a baby appears within a preset time period, and recommends a mother-baby product to the terminal.
In some embodiments, the server determines a current position, and recommends, to the terminal, target information corresponding to at least one valid image tag of the target position in response to the current position matching the target position. For example, the server acquires the current position information of the terminal as city a, determines the possible things that the target user may do in city a based on the image tags of the images with the generation position of the image in the album application as city a, and recommends the time-related information to the terminal. Or the server acquires the current position information of the terminal as the city B and directly recommends the scenic spot information of the city B to the terminal. Or the server acquires the current position information of the terminal as C city, determines that the user likes nice matters by combining other user figures of the target user, and recommends the gourmet information of the C city to the terminal.
In some embodiments, the server determines a current time, and recommends target information corresponding to at least one valid image tag in the target time period to the terminal in response to the current time being within the target time period. For example, if the server determines that the current time is in the lunch time period, the server recommends a restaurant for lunch to the terminal; the server can also recommend information to the terminal in combination with other user images. For example, if the user is portrayed as liking a food and the current time is during a holiday, the target information is determined to be a food related to the holiday.
In the implementation mode, the server combines the user portrait with other information to determine the target information to be recommended, so that the pertinence of the target information to the terminal is further improved, and the information conversion rate is further improved.
The above embodiment can also be performed by a terminal, and accordingly, the terminal determines a user portrait of the terminal based on an image in a terminal album application, and when target information needs to be recommended to the user, the user portrait is sent to a backend server of another application program, and the backend server of the other application program recommends the target information to the terminal based on the user portrait. The process of determining the user profile by the terminal is similar to the process of determining the user profile by the server in step 201 and 203, and is not described herein again.
In the embodiment of the application, the image in the album application corresponding to the terminal is generally a favorite image of the user; therefore, the interest of the user can be mined based on the image in the album application, so that the accuracy of the user portrait determined based on the image is improved, and the accuracy of the target information recommended based on the user portrait is improved.
Referring to fig. 7, a block diagram of an information recommendation apparatus according to an embodiment of the present application is shown. The device includes:
an obtaining module 701, configured to obtain multiple images in an album application corresponding to a terminal;
a label determination module 702 for determining an image label for each image;
a portrait determination module 703 for determining a user portrait of a target user based on the image tags of the plurality of images, the target user being a user using the terminal;
and an information recommending module 704, configured to recommend target information matching the user portrait to the terminal based on the user portrait.
In some embodiments, the representation determination module 703 includes:
a tag determination unit configured to determine at least one valid image tag from the plurality of image tags based on image tags of the plurality of images, the valid image tag referring to an image tag that can be used to determine a user portrait;
a portrait determination unit to determine a user portrait of the target user based on the tag type of the at least one active image tag.
In some embodiments, the tag determining unit is configured to determine, based on image tags of the plurality of images, a number of images corresponding to each image tag; determining at least one effective image tag of which the number exceeds a preset number from a plurality of image tags;
the label determining unit is used for determining the appearance frequency of each image label based on the image labels of the plurality of images; determining at least one effective image tag with the occurrence frequency exceeding a preset frequency from a plurality of image tags;
the label determining unit is used for determining the generation time of each image, and determining at least one effective image label from a plurality of image labels based on the generation time of each image, wherein the interval between the generation time of the image corresponding to the at least one effective image label and the current time is less than the preset time length;
the label determining unit is used for determining at least one effective image label of the generation time of the image in the target time period based on the image labels of the plurality of images;
the label determining unit is used for determining at least one effective image label of which the generation position of the image is matched with the target position based on the image labels of the plurality of images.
In some embodiments, the portrait determination unit is to determine a subject preference feature of the target user based on the subject in response to the image tag being a subject; responding to the image label as a scene, and determining the travel preference characteristics of the target user based on the scene; in response to the image tag being a behavior, determining behavior preference characteristics of the terminal based on the behavior; and constructing the user portrait of the target user based on at least one of the subject preference feature, the travel preference feature and the behavior preference feature.
In some embodiments, the portrait determination unit is configured to determine a preferred subject of the target user based on the subject preference feature, the preferred subject being determined to be a user portrait of the target user;
the portrait determining unit is used for determining a preference scene of the target user based on the travel preference characteristics, and determining the preference scene as a user portrait of the target user;
the portrait determining unit is used for determining the preference behavior of the target user based on the behavior preference characteristics, and determining the main body and the scene generated by the preference behavior as the user portrait of the target user.
In some embodiments, the tag determination module 702 includes:
the label determining unit is used for determining a plurality of image labels of each image through a label determining model, wherein the activation layer of the label determining model is a multi-classification activation layer;
the probability prediction unit is used for performing probability prediction on a plurality of image labels of the image based on the multi-classification activation layer of the label determination model to obtain the prediction probability of the plurality of image labels of the image;
and the label selection unit is used for determining an image label with the prediction probability larger than the preset probability from the output plurality of image labels as the image label of the image.
In some embodiments, the information recommendation module 704 includes:
an information determination unit to determine target information from an information repository for the user representation based on the user representation;
and the information recommending unit is used for recommending the target information to the terminal.
In some embodiments, the obtaining module 701 is configured to obtain a plurality of images of an album application installed in a terminal;
the obtaining module 701 is configured to obtain a plurality of images of an album application in an application platform of a social application in a terminal;
the obtaining module 701 is configured to obtain a plurality of images of a cloud album in a server in which a terminal logs in.
In some embodiments, the target information includes: at least one of advertisement information, download information of application program, recommendation information of theme
In the embodiment of the application, the images in the album application in the terminal are generally favorite images of the user; therefore, the interest of the user can be mined based on the image in the album application, so that the accuracy of the user portrait determined based on the image is improved, and the accuracy of the target information recommended based on the user portrait is improved.
Referring to fig. 8, a block diagram of a server according to an embodiment of the present application is shown. The server in the present application comprises one or more of the following components: a processor 810, a memory 820, and an information transmitter 830.
In some embodiments, processor 810 includes one or more processing cores. The processor 810 connects various parts within the entire server 101 using various interfaces and lines, performs various functions of the server and processes data by operating or executing program codes stored in the memory 820 and calling data stored in the memory 820. In some embodiments, the processor 810 is implemented in hardware using at least one of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor 810 may integrate one or more of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a Neural-Network Processing Unit (NPU), a modem, and the like. Wherein, the CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for rendering and drawing contents required to be displayed by the touch display screen; the NPU is used for realizing an Artificial Intelligence (AI) function; the modem is used to handle wireless communications. It is to be appreciated that in some embodiments, the modem is not integrated into the processor 810 and is implemented solely on a single chip.
In some embodiments, Memory 820 comprises a Random Access Memory (RAM), and in other embodiments, Memory 820 comprises a Read-Only Memory (ROM). In some embodiments, the memory 820 includes a non-transitory computer-readable medium. Memory 820 may be used to store program code. The memory 820 may include a stored program area and a stored data area, wherein the stored program area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing various method embodiments described below, and the like; the storage data area may store data (such as audio data, a phonebook) created by use of the server, and the like.
The information transmitter 830 is a component for transmitting recommendation information, and in some embodiments, the information transmitter 830 is an information transmitting chip.
In addition, those skilled in the art will appreciate that the configurations of the servers illustrated in the above-described figures do not constitute limitations on the servers, which can include more or fewer components than illustrated, or some components in combination, or a different arrangement of components. For example, the server further includes a radio frequency circuit, an input unit, a sensor, an audio circuit, a Wireless Fidelity (WiFi) module, a power supply, a bluetooth module, and other components, which are not described herein again.
The embodiment of the present application also provides a computer-readable medium, in which at least one program code is stored, and the at least one program code is loaded and executed by the processor to implement the information recommendation method shown in the above embodiments.
The embodiment of the present application further provides a computer program product, where at least one program code is stored, and the at least one program code is loaded and executed by the processor to implement the information recommendation method shown in the above embodiments.
Those skilled in the art will recognize that, in one or more of the examples described above, the functions described in the embodiments of the present application can be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, the functions can be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media is any available media that can be accessed by a general purpose or special purpose computer.
The above description is only exemplary of the present application and should not be taken as limiting, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims (12)

1. An information recommendation method, characterized in that the method comprises:
acquiring a plurality of images in an album application corresponding to a terminal;
determining an image label for each image;
determining a user portrait of a target user based on image tags of the plurality of images, the target user being a user using the terminal;
and recommending the target information matched with the user portrait to the terminal based on the user portrait.
2. The method of claim 1, wherein determining a user representation of a target user based on image tags of the plurality of images comprises:
determining at least one active image tag from the plurality of image tags based on image tags of the plurality of images, the active image tag referring to an image tag that can be used to determine a user representation;
determining a user representation of the target user based on the tag type of the at least one active image tag.
3. The method of claim 2, wherein determining at least one valid image tag from the plurality of image tags based on the image tags of the plurality of images comprises at least one of:
determining the number of images corresponding to each image label based on the image labels of the plurality of images; determining at least one effective image tag of which the number exceeds a preset number from a plurality of image tags;
determining a frequency of occurrence of each image tag based on the image tags of the plurality of images; determining at least one effective image tag with the occurrence frequency exceeding a preset frequency from a plurality of image tags;
determining the generation time of each image, and determining at least one effective image label from a plurality of image labels based on the generation time of each image, wherein the interval between the generation time of the image corresponding to the at least one effective image label and the current time is less than the preset time length;
determining at least one valid image tag of the generation time of the image within a target time period based on the image tags of the plurality of images;
based on the image tags of the plurality of images, at least one valid image tag is determined for which the generation location of the image matches the target location.
4. The method of claim 2, wherein determining the user representation of the target user based on the tag type of the at least one active image tag comprises:
in response to the image tag being a subject, determining subject preference characteristics of the target user based on the subject;
in response to the image tag being a scene, determining travel preference characteristics of the target user based on the scene;
in response to the image tag being a behavior, determining behavior preference characteristics of the terminal based on the behavior;
constructing a user representation of the target user based on at least one of the subject preference feature, the travel preference feature, and the behavior preference feature.
5. The method of claim 4, wherein constructing the user representation of the target user based on at least one of the subject preference feature, the travel preference feature, and the behavior preference feature comprises at least one of:
determining a preferred subject of the target user based on the subject preference feature, the preferred subject being determined to be a user representation of the target user;
determining a preference scene of the target user based on the travel preference characteristics, and determining the preference scene as a user portrait of the target user;
determining a preferred behavior of the target user based on the behavior preference characteristics, and determining a subject and a scene generated by the preferred behavior as a user representation of the target user.
6. The method of claim 1, wherein determining the image label for each image comprises:
for each image, determining a plurality of image labels of the image through a label determination model, wherein an activation layer of the label determination model is a multi-classification activation layer;
based on a multi-classification activation layer of the label determination model, performing probability prediction on a plurality of image labels of the image to obtain prediction probabilities of the plurality of image labels of the image;
and determining an image label with a prediction probability greater than a preset probability as an image label of the image from the output plurality of image labels.
7. The method of claim 1, wherein recommending, to the terminal, target information for the user representation matching based on the user representation comprises:
determining target information for the user representation match from an information repository based on the user representation;
and recommending the target information to the terminal.
8. The method according to claim 1, wherein the obtaining of the plurality of images in the album application corresponding to the terminal includes at least one of the following implementation manners:
acquiring a plurality of images of an album application installed in a terminal;
acquiring a plurality of images of album application in an application platform of a social application in a terminal;
the method includes the steps that a plurality of images of a cloud photo album in a server logged in by a terminal are obtained.
9. The method of any of claims 1-8, wherein the target information comprises: at least one of advertisement information, download information of an application program, and recommendation information of a subject.
10. An information recommendation apparatus, characterized in that the apparatus comprises:
the acquisition module is used for acquiring a plurality of images in the photo album application corresponding to the terminal;
a label determination module for determining an image label for each image;
a portrait determination module to determine a user portrait of a target user based on image tags of the plurality of images, the target user being a user using the terminal;
and the information recommending module is used for recommending the target information matched with the user portrait to the terminal based on the user portrait.
11. A server, comprising a processor and a memory; the memory stores at least one program code for execution by the processor to implement the information recommendation method of any of claims 1 to 9.
12. A computer-readable storage medium, characterized in that the storage medium stores at least one program code for execution by a processor to implement the information recommendation method according to any one of claims 1 to 9.
CN202011312572.5A 2020-11-20 2020-11-20 Information recommendation method and device, server and storage medium Pending CN112328888A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011312572.5A CN112328888A (en) 2020-11-20 2020-11-20 Information recommendation method and device, server and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011312572.5A CN112328888A (en) 2020-11-20 2020-11-20 Information recommendation method and device, server and storage medium

Publications (1)

Publication Number Publication Date
CN112328888A true CN112328888A (en) 2021-02-05

Family

ID=74321896

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011312572.5A Pending CN112328888A (en) 2020-11-20 2020-11-20 Information recommendation method and device, server and storage medium

Country Status (1)

Country Link
CN (1) CN112328888A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112883269A (en) * 2021-02-26 2021-06-01 上海连尚网络科技有限公司 Method and equipment for adjusting label data information
CN113094586A (en) * 2021-04-21 2021-07-09 腾讯音乐娱乐科技（深圳）有限公司 UGC (user generated content) pushing method, device, equipment and storage medium
CN113378082A (en) * 2021-07-05 2021-09-10 北京百度网讯科技有限公司 Information recommendation method and device, electronic equipment and storage medium
CN113515696A (en) * 2021-05-17 2021-10-19 上海众源网络有限公司 Recommendation method and device, electronic equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108875820A (en) * 2018-06-08 2018-11-23 Oppo广东移动通信有限公司 Information processing method and device, electronic equipment, computer readable storage medium
CN109376313A (en) * 2018-09-28 2019-02-22 Oppo广东移动通信有限公司 Information recommendation method, device, terminal and storage medium
CN110866168A (en) * 2018-08-27 2020-03-06 阿里巴巴集团控股有限公司 Information recommendation method and device, terminal and server

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108875820A (en) * 2018-06-08 2018-11-23 Oppo广东移动通信有限公司 Information processing method and device, electronic equipment, computer readable storage medium
CN110866168A (en) * 2018-08-27 2020-03-06 阿里巴巴集团控股有限公司 Information recommendation method and device, terminal and server
CN109376313A (en) * 2018-09-28 2019-02-22 Oppo广东移动通信有限公司 Information recommendation method, device, terminal and storage medium

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112883269A (en) * 2021-02-26 2021-06-01 上海连尚网络科技有限公司 Method and equipment for adjusting label data information
CN113094586A (en) * 2021-04-21 2021-07-09 腾讯音乐娱乐科技（深圳）有限公司 UGC (user generated content) pushing method, device, equipment and storage medium
CN113515696A (en) * 2021-05-17 2021-10-19 上海众源网络有限公司 Recommendation method and device, electronic equipment and storage medium
CN113378082A (en) * 2021-07-05 2021-09-10 北京百度网讯科技有限公司 Information recommendation method and device, electronic equipment and storage medium

Similar Documents

Publication Publication Date Title
US11290775B2 (en) Computerized system and method for automatically detecting and rendering highlights from streaming videos
US10447645B2 (en) Computerized system and method for automatically creating and communicating media streams of digital content
CN112328888A (en) Information recommendation method and device, server and storage medium
KR101959368B1 (en) Determining an active persona of a user device
CN110139162B (en) Media content sharing method and device, storage medium and electronic device
CN106021586B (en) Information processing method and server
US20150193472A1 (en) Generating user insights from user images and other data
CN109840793B (en) Promotion information pushing method and device, storage medium and electronic device
CN109074358A (en) Geographical location related with user interest is provided
CN107977928B (en) Expression generation method and device, terminal and storage medium
CN112765373B (en) Resource recommendation method and device, electronic equipment and storage medium
WO2020044099A1 (en) Service processing method and apparatus based on object recognition
CN110225398B (en) Multimedia object playing method, device and equipment and computer storage medium
CN110782289B (en) Service recommendation method and system based on user portrait
CN107547922B (en) Information processing method, device, system and computer readable storage medium
CN112328823A (en) Training method and device for multi-label classification model, electronic equipment and storage medium
CN113382301A (en) Video processing method, storage medium and processor
WO2022111071A1 (en) User profile generation method, apparatus, server, and storage medium
CN110427499B (en) Method and device for processing multimedia resources, storage medium and electronic device
CN114298767A (en) Live broadcast platform information pushing method and device, equipment, medium and product thereof
CN110661693A (en) Methods, computing device-readable storage media, and computing devices facilitating media-based content sharing performed in a computing device
CN111698563A (en) Content sending method and device based on AI virtual anchor and storage medium
JP7266356B1 (en) Program, information processing device, information processing system, and information processing method
CN112764553B (en) Chat expression collection method and device and electronic equipment
CN117252206A (en) Intention recognition method, apparatus, device, medium, and program product

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