CN116610864A

CN116610864A - PGC and UGC content layout thousand-person thousand-face-based content presentation algorithm and system

Info

Publication number: CN116610864A
Application number: CN202310659827.2A
Authority: CN
Inventors: 梁冉
Original assignee: Beijing Co Mall Internet Technology Co ltd
Current assignee: Beijing Co Mall Internet Technology Co ltd
Priority date: 2023-06-06
Filing date: 2023-06-06
Publication date: 2023-08-18

Abstract

The application provides a content presentation algorithm and a system based on thousands of people and thousands of faces of PGC and UGC content blocks, and relates to the technical field of data application. The method comprises the following steps: performing deep data analysis on the behavior of the user, associating brands with style labels, registering and selecting the preferred styles of the user as personal information of the user, analyzing the preference of the user styles, and recommending brand contents of similar styles; associating PGC and UGC contents with topics, wherein the topics belong to interest plates, confirming the consulting habit of a user, analyzing whether interest relation exists between the consulting contents, and recommending the contents of related interest plates; identifying and analyzing user preferences by analyzing user social media behaviors and behaviors of content blocks; and establishing a prediction model, taking brands and style labels, PGCs, UGC content and topics and social media behaviors of users as inputs of the prediction model, and predicting access, browsing and search trails of the users.

Description

PGC and UGC content layout thousand-person thousand-face-based content presentation algorithm and system

Technical Field

The application relates to the technical field of data application, in particular to a PGC and UGC content layout thousand-person and thousand-face based content presentation algorithm and system.

Background

In recent years, the information quantity is rapidly developed, the information explodes, the network brings more and more information to users, more and more choices are provided, but too various choices occupy too much time of the users, so that the users can hardly find interesting contents in a short time, lose patience and choose to leave. Therefore, the content interested by the user needs to be recommended out in a targeted manner based on the preference of each user, so that each user can find the content which the user wants to see in a short time, and the problems of incapacity of the user and low viscosity are solved.

The content recommendation of ' thousand people ' side ' brings content monopoly at the same time, the traffic is mainly concentrated on the first few screens, and many content creators cannot take the traffic, so that if most content creators lose activity, the content update of the platform is slow and not novel, and the user cannot be left. This requires that fairness in traffic distribution be resolved and traffic cannot be concentrated among several people.

Disclosure of Invention

The application aims to provide a content presentation algorithm based on thousands of people and thousands of faces of PGC and UGC content blocks, which can expose contents of more creators according to different users with different habits, interests and brand preferences, and can uniformly distribute more contents to traffic, so that traffic monopoly is avoided.

Another object of the present application is to provide a PGC, UGC content layout thousand-face based content presentation system capable of running a PGC, UGC content layout thousand-face based content presentation algorithm.

Embodiments of the present application are implemented as follows:

in a first aspect, an embodiment of the present application provides a content presentation algorithm based on PGC and UGC content formats and thousands of people and thousands of faces, which includes performing deep data analysis on a user's behavior, understanding a user's requirement through the data analysis algorithm, and then performing data processing to recommend appropriate content for the user; associating the brand with the style tag, registering and selecting the preferred style of the user as personal information of the user, analyzing the preference of the style of the user, and recommending brand content of similar style; associating PGC and UGC contents with topics, wherein the topics belong to interest plates, confirming the consulting habit of a user, analyzing whether interest relation exists between the consulting contents, and recommending the contents of related interest plates; identifying and analyzing user preferences by analyzing user social media behaviors and behaviors of content blocks; and establishing a prediction model, taking brands and style labels, PGCs, UGC content and topics and social media behaviors of users as inputs of the prediction model, and predicting access, browsing and search trails of the users.

In some embodiments of the application, the foregoing further comprises: collecting user behavior data, user attribute data and content data, wherein the user behavior data comprises click records, browse records and collection records, the user attribute data comprises gender, age and region information, and the content data comprises labels, topics and quality scoring information of PGC and UGC content.

In some embodiments of the present application, the foregoing deep data analysis on the behavior of the user, understanding the requirement of the user through the data analysis algorithm, and then performing the data processing to recommend the appropriate content for the user further includes: based on the user behavior data and the user attribute data, a user image is generated using a machine learning algorithm or a recommendation system algorithm.

In some embodiments of the present application, the associating the PGC and UGC content with topics, the topics being assigned to interest blocks, confirming a review habit of a user, analyzing whether interest links exist between review contents, and recommending the content of the relevant interest blocks includes: and carrying out label and feature extraction on the PGC and UGC contents, wherein the labels comprise topics, categories and emotion information, and the features comprise feature vectors of characters, pictures and videos.

In some embodiments of the application, the foregoing further comprises: and calculating the interest similarity of the user for each content according to the user portrait, the labels and the characteristics of the content.

In some embodiments of the application, the foregoing further comprises: and ordering the content according to the interest similarity of the user, and recommending the related content to the user, wherein the content recommendation is performed by using an ordering algorithm and a recommendation algorithm.

In some embodiments of the present application, the above includes a recommended algorithm flow: s1, setting an initial result cache, and acquiring a note id list which is given by a recommendation model and ordered according to the relevance according to user ids;

s2, setting a progressive pointer to point to a certain note in the initial result cache selected randomly. Invoking a recommendation model in advance, acquiring a similar note list of the note according to the similarity of the user behaviors, and storing the result into a note recommendation cache;

s3, maintaining an accessed note id list when filling the note recommending cache so as to ensure that the recommended notes are not recommended again;

s4, when a recommendation request is made, sequentially reading note ids from a note recommendation cache until the number of notes reaches the number of requested notes;

s5, returning the obtained note id list.

In a second aspect, an embodiment of the present application provides a content presentation system based on PGC and UGC content formats and thousand faces, which includes a user behavior analysis module, configured to perform deep data analysis on behaviors of a user, understand requirements of the user through a data analysis algorithm, and then perform data processing to recommend appropriate content for the user;

the style tag analysis module is used for associating brands with style tags, registering and selecting the preferred styles of users as personal information of the users, analyzing the preferred styles of the users, and recommending brand contents of similar styles;

the interest plate analysis module is used for associating PGC and UGC contents with topics, the topics belong to interest plates, the consulting habit of a user is confirmed, whether interest connection exists between the consulting contents is analyzed, and the contents of the relevant interest plates are recommended;

the user portrayal module is used for identifying and analyzing the user preference by analyzing the social media behaviors of the user and the behaviors of the content blocks;

the model training module is used for establishing a prediction model, taking brands and style labels, PGCs, UGC content and topics and social media behaviors of users as inputs of the prediction model, and predicting access, browsing and search trails of the users.

In a third aspect, embodiments of the present application provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method as any one of a PGC, UGC content layout thousand-sided based content rendering algorithm.

Compared with the prior art, the embodiment of the application has at least the following advantages or beneficial effects:

by means of accumulation of data and a back analysis model, crowd is subdivided, and different users are recommended to accurately touch different contents. Based on the behavior of the user, the algorithm recommends the content issued by the creator concerned or liked by the user to the user for browsing; based on the style labels of the users, the algorithm recommends brand contents preferred by the users to the users for browsing; based on interest preference of the user, recommending topics or some hotspots of interest to the user for browsing by the algorithm; accurate thousand people thousand faces have solved the user of different user figures, can both find oneself wanting the content of seeing in the short time on same platform, promote user's browsing duration, promoted user's viscidity. Aiming at users with different habits, interests and brand preferences, the contents of the first few screens of the users are different, so that more authors can expose the contents, more contents can be uniformly distributed to the traffic, and traffic monopoly is avoided.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic diagram of steps of a content presentation algorithm based on thousands of people and thousands of faces of PGC and UGC content blocks according to an embodiment of the present application;

FIG. 2 is a schematic diagram of a system architecture according to an embodiment of the present application;

FIG. 3 is a schematic diagram of a content presentation system module based on thousands of people and thousands of faces of PGC and UGC content blocks according to an embodiment of the present application;

fig. 4 is an electronic device provided in an embodiment of the present application.

Icon: 101-memory; 102-a processor; 103-communication interface.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments of the present application. The components of the embodiments of the present application generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.

Thus, the following detailed description of the embodiments of the application, as presented in the figures, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.

It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures.

It should be noted that the term "comprises," "comprising," or any other variation thereof is intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

Some embodiments of the present application are described in detail below with reference to the accompanying drawings. The various embodiments and features of the embodiments described below may be combined with one another without conflict.

Example 1

Referring to fig. 1, fig. 1 is a schematic diagram of steps of a content presentation algorithm based on PGC and UGC content blocks and thousands of faces, which is provided in an embodiment of the present application, and is as follows:

1. recommendation model: the recommender model employs CCO (Correlated Cross Occurrence) algorithms to correlate with the master events (views) through various behavioral events of the user. Notes transformed by users with similar behavior are taken as recommended notes, and essentially belong to collaborative filtering algorithms.

The number of valid associations refers to the maximum number of choices that the model training events consider to be valid to represent the user's preferences. The number of valid associations should generally not be greater than or equal to the cardinality of event enumeration. The effective correlation number can be obtained by exploratory analysis of note business data.

2. Recommendation algorithm

2.1 design goals: by pre-filling the buffer, the result is obtained from the buffer for a majority of the time of the recommended request, thereby achieving high throughput and low latency. The recommendation effect of thousands of people and the same user is relatively random each time is realized through randomization. Typically, the user recommends have a relatively sharp decrease in relevance, and the model may appear as a under-fit. The problem of under fitting is effectively solved by integrating the user recommendation result and the note similarity data.

2.2 algorithm flow:

s1, setting initial result cache, and acquiring a note id list which is given by a recommendation model and ordered according to the relevance according to user ids, wherein the default size is 96.

S2, setting a progressive pointer to point to a certain note in the initial result cache selected randomly. And calling a recommendation model in advance, acquiring a similar note list of the note according to the user behavior similarity, and storing the result into a note recommendation cache, wherein the default size is 32.

S3, maintaining an accessed note id list when filling the note recommendation cache so as to ensure that the recommended notes are not recommended again.

S4, when a recommendation request is made, sequentially reading note ids from a note recommendation cache until the number of notes reaches the number of requested notes.

S5, returning the obtained note id list.

3. Recommending query timing: the state is recommended to be operated with a state, and the state is saved by using a Session object. Which stores the working context of the current recommendation algorithm (query object, initial result cache, note recommendation cache, pointer location, query re-list). After receiving the recommendation request, the Session manager first tries to acquire the Session of the user. If the Session is already owned by another thread, then a Session is recreated and the previous Session is removed. The current Session is marked as occupied. If it is a newly created Session, the relevant work context is initialized and configured.

Executing a randomization algorithm to obtain a query result. If the note recommendation buffer has been consumed, an asynchronous HTTP query to the recommender system service will be triggered to refill the buffer.

The Session is marked as unoccupied to release ownership of the Session. I.e., to ensure that all instructions of the previous thread have been executed and that the execution results are visible to the new thread.

The current access timestamp is marked using the Session and the removal queue is maintained using the heap structure. Reaching the maximum Session capacity (default 8192) removes the earliest time stamped Session.

If a Session does not have any action within 10 minutes, it is removed.

4. As shown in fig. 2, the system architecture includes: the system architecture of the recommendation system comprises a data source, an event processor, a sample database, a model database, a recommendation engine service and a recommendation system service.

The component description includes: data source: notes, events, and related metadata tables in the business system. An event handler: grammar verification, normalization, de-duplication and merging are carried out on the data source change event to generate standard sample data. Sample database: sample data is stored, including note document data and user behavior event data. Model database: and storing the trained recommendation system model.

Recommendation engine service:

1) The documents and behavioral event data in the sample database are trained in full volume, periodically once a day, generating a recommendation model, and persisting new model parameters.

2) In response to a call to a recommendation system service.

Recommendation system service: the service of note recommendation logic is encapsulated.

5. Results of offline experiments

The off-line experiment is an accuracy analysis of the algorithm made off-line. It uses the sample data itself to perform the accuracy experiments, not based on user feedback, and therefore only represents the mathematical validity of the algorithm, not the operational effect. The operation effect is further realized by the data warehouse through collecting and analyzing the user behavior data.

1) Algorithm group:

random: randomly recommending, namely randomly selecting one from all notes as a recommending result.

Engine_range: an algorithm that randomly samples UR recommendations and blends in similar entries (i.e., the recommendation algorithm described in the second chapter).

Engine_range_bandwidth: and carrying out equalization processing on the sample data.

Training/validation set settings:

the users with the viewing event amounts of [50,200] are taken as verification sets, and the rest users are taken as training sets.

All events of the training set are input, each behavior of the verification set inputs 5 events at most, and the rest view events are used as verification. I.e., all view events in the verification set are not present in the training set, whereby the verification model recommends the accuracy of the user's other view notes from the small number of events of the user of the verification set.

And taking the users in the verification set, taking the first 600 recommendations of each user, and calculating the accuracy rate.

The median of the accuracy rate of the random selection is 0%, while the median of the accuracy rate of the recommendation algorithm is obviously not 0%, and the significance of the recommendation algorithm cannot be explained by using random factors. So that the algorithm is mathematically efficient.

Example 2

Referring to fig. 3, fig. 3 is a schematic diagram of a content presentation system based on PGC and UGC content blocks and thousands of people and thousands of faces, which is provided in an embodiment of the present application, and is as follows:

the user behavior analysis module is used for carrying out deep data analysis on the behaviors of the user, understanding the requirements of the user through a data analysis algorithm, and then carrying out data processing to recommend proper contents for the user;

As shown in fig. 4, an embodiment of the present application provides an electronic device including a memory 101 for storing one or more programs; a processor 102. The method of any of the first aspects described above is implemented when one or more programs are executed by the processor 102.

And a communication interface 103, where the memory 101, the processor 102 and the communication interface 103 are electrically connected directly or indirectly to each other to realize data transmission or interaction. For example, the components may be electrically connected to each other via one or more communication buses or signal lines. The memory 101 may be used to store software programs and modules that are stored within the memory 101 for execution by the processor 102 to perform various functional applications and data processing. The communication interface 103 may be used for communication of signaling or data with other node devices.

The Memory 101 may be, but is not limited to, a random access Memory (Random Access Memory, RAM), a Read Only Memory (ROM), a programmable Read Only Memory (Programmable Read-Only Memory, PROM), an erasable Read Only Memory (Erasable Programmable Read-Only Memory, EPROM), an electrically erasable Read Only Memory (Electric Erasable Programmable Read-Only Memory, EEPROM), etc.

The processor 102 may be an integrated circuit chip with signal processing capabilities. The processor 102 may be a general purpose processor including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; but also digital signal processors (Digital Signal Processing, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

In the embodiments provided in the present application, it should be understood that the disclosed method and system may be implemented in other manners. The above-described method and system embodiments are merely illustrative, for example, flow charts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of methods and systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

In addition, functional modules in the embodiments of the present application may be integrated together to form a single part, or each module may exist alone, or two or more modules may be integrated to form a single part.

In another aspect, an embodiment of the application provides a computer readable storage medium having stored thereon a computer program which, when executed by the processor 102, implements a method as in any of the first aspects described above. The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

In summary, according to the content presentation algorithm and system based on thousands of people and thousands of faces of PGC and UGC content formats provided by the embodiments of the present application, people are subdivided through accumulation of data and an analysis model behind the data, and different users are recommended to accurately reach different contents. Based on the behavior of the user, the algorithm recommends the content issued by the creator concerned or liked by the user to the user for browsing; based on the style labels of the users, the algorithm recommends brand contents preferred by the users to the users for browsing; based on interest preference of the user, recommending topics or some hotspots of interest to the user for browsing by the algorithm; accurate thousand people thousand faces have solved the user of different user figures, can both find oneself wanting the content of seeing in the short time on same platform, promote user's browsing duration, promoted user's viscidity. Aiming at users with different habits, interests and brand preferences, the contents of the first few screens of the users are different, so that more authors can expose the contents, more contents can be uniformly distributed to the traffic, and traffic monopoly is avoided.

The above is only a preferred embodiment of the present application, and is not intended to limit the present application, but various modifications and variations can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

It will be evident to those skilled in the art that the application is not limited to the details of the foregoing illustrative embodiments, and that the present application may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.

Claims

1. A content presentation algorithm based on thousands of people and thousands of faces of PGC and UGC content blocks, comprising:

deep data analysis is carried out on the behaviors of the user, the requirements of the user are understood through a data analysis algorithm, and then data processing is carried out to recommend proper contents for the user;

associating the brand with the style tag, registering and selecting the preferred style of the user as personal information of the user, analyzing the preference of the style of the user, and recommending brand content of similar style;

associating PGC and UGC contents with topics, wherein the topics belong to interest plates, confirming the consulting habit of a user, analyzing whether interest relation exists between the consulting contents, and recommending the contents of related interest plates;

identifying and analyzing user preferences by analyzing user social media behaviors and behaviors of content blocks;

and establishing a prediction model, taking brands and style labels, PGCs, UGC content and topics and social media behaviors of users as inputs of the prediction model, and predicting access, browsing and search trails of the users.

2. The content rendering algorithm of claim 1, further comprising:

collecting user behavior data, user attribute data and content data, wherein the user behavior data comprises click records, browse records and collection records, the user attribute data comprises gender, age and region information, and the content data comprises labels, topics and quality scoring information of PGC and UGC content.

3. The content presentation algorithm based on PGC and UGC content layout thousand-face as set forth in claim 1, wherein the performing deep data analysis on the behavior of the user, understanding the user's requirement through the data analysis algorithm, and then performing data processing to recommend suitable content for the user further includes:

based on the user behavior data and the user attribute data, a user image is generated using a machine learning algorithm or a recommendation system algorithm.

4. The content presentation algorithm based on thousands of PGC and UGC content blocks as claimed in claim 1, wherein associating PGC and UGC content with topics, the topics belonging to interest blocks, confirming a user's review habit, analyzing whether interest links exist between review content, and recommending content of related interest blocks includes:

and carrying out label and feature extraction on the PGC and UGC contents, wherein the labels comprise topics, categories and emotion information, and the features comprise feature vectors of characters, pictures and videos.

5. The content rendering algorithm based on PGC, UGC content pieces thousand-sided as set forth in claim 4, further comprising:

and calculating the interest similarity of the user for each content according to the user portrait, the labels and the characteristics of the content.

6. The content rendering algorithm based on PGC, UGC content pieces thousand-people thousand-plane as set forth in claim 5, further comprising:

and ordering the content according to the interest similarity of the user, and recommending the related content to the user, wherein the content recommendation is performed by using an ordering algorithm and a recommendation algorithm.

7. The content presentation algorithm based on thousands of PGC, UGC content pieces as set forth in claim 1, further comprising a recommendation algorithm flow:

s1, setting an initial result cache, and acquiring a note id list which is given by a recommendation model and ordered according to the relevance according to user ids;

s2, setting a progressive pointer to point to a certain note in the initial result cache selected randomly, calling a recommendation model in advance, acquiring a similar note list of the note according to the similarity of user behaviors, and storing the result in the note recommendation cache;

s5, returning the obtained note id list.

8. A PGC, UGC content layout thousand-person thousand-face based content presentation system, comprising:

9. A computer readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, implements the algorithm of any one of claims 1-7.