CN114896454B - Short video data recommendation method and system based on label analysis - Google Patents

Short video data recommendation method and system based on label analysis Download PDF

Info

Publication number
CN114896454B
CN114896454B CN202210812661.9A CN202210812661A CN114896454B CN 114896454 B CN114896454 B CN 114896454B CN 202210812661 A CN202210812661 A CN 202210812661A CN 114896454 B CN114896454 B CN 114896454B
Authority
CN
China
Prior art keywords
short video
interest
user
publishing
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202210812661.9A
Other languages
Chinese (zh)
Other versions
CN114896454A (en
Inventor
杨爽
谢匡华
朱福青
谢匡亮
何春
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.)
Changsha Meida Network Technology Co ltd
Original Assignee
Changsha Meida Network Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Changsha Meida Network Technology Co ltd filed Critical Changsha Meida Network Technology Co ltd
Priority to CN202210812661.9A priority Critical patent/CN114896454B/en
Publication of CN114896454A publication Critical patent/CN114896454A/en
Application granted granted Critical
Publication of CN114896454B publication Critical patent/CN114896454B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/73—Querying
    • G06F16/735—Filtering based on additional data, e.g. user or group profiles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Abstract

The application provides a short video data recommendation method and system based on label analysis, wherein each first user behavior activity data is loaded into a set knowledge entity network for regularization conversion, and corresponding second user behavior activity data is generated, so that the reliability of subsequent mining processing is improved, the corresponding second user behavior activity data is loaded into corresponding short video recommendation services by dividing different short video recommendation dimensions into a plurality of short video recommendation services, so that after all second user behavior activity data associated with each short video recommendation service are extracted and loaded into a corresponding interest thermodynamic trend output model to generate a corresponding user interest label thermodynamic trend, a key user interest label corresponding to a target user is further determined, short video data recommendation is performed on the target user according to the key user interest label, and further personalized recommendation is performed after user interest mining is performed on different short video recommendation dimensions, the accuracy of short video recommendation is improved.

Description

Short video data recommendation method and system based on label analysis
Technical Field
The application relates to the technical field of information recommendation, in particular to a short video data recommendation method and system based on label analysis.
Background
Short videos are short videos, which are a mode for transmitting internet content, and are videos transmitted on new internet media within a short time (e.g., within 5 minutes); with the popularization of mobile terminals and the increasing speed of networks, short and fast mass flow transmission contents are gradually favored by various large platforms, fans and capital. In the related art, in order to improve the user occupancy, personalized recommendation of short video data can be performed on each user by analyzing the user interest characteristics, however, in the current short video recommendation scheme, the accuracy of short video recommendation cannot meet the requirements of related users.
Disclosure of Invention
In order to overcome at least the above disadvantages in the prior art, the present application aims to provide a method and a system for recommending short video data based on tag analysis.
In a first aspect, the present application provides a short video data recommendation method based on tag analysis, which is applied to a short video data recommendation system based on tag analysis, and the method includes:
acquiring a plurality of first user behavior activity data based on a user behavior activity monitoring program configured on each short video service page of a target user;
loading each first user behavior activity data into a set knowledge entity network respectively for regularized conversion, and generating corresponding second user behavior activity data;
determining a plurality of short video recommendation services according to the short video subscription items of the user;
determining service association degree between each second user behavior activity data and the corresponding short video recommendation service according to each short video recommendation service, and if the service association degree is greater than a first target value, loading the corresponding second user behavior activity data into the corresponding short video recommendation service;
respectively configuring a plurality of interest thermodynamic trend output models according to a plurality of short video recommendation services, carrying out model tuning on each interest thermodynamic trend output model according to the collected user interest learning data, and outputting the interest thermodynamic trend output model after model tuning;
extracting all second user behavior activity data associated with each short video recommendation service, and loading the second user behavior activity data into the interest thermodynamic trend output model after the corresponding model is adjusted and optimized to generate a corresponding user interest label thermodynamic trend;
and summarizing the thermodynamic trend of each user interest label, determining a key user interest label corresponding to the target user based on a user interest decision model, and recommending the short video data to the target user based on the key user interest label corresponding to the target user.
In a possible implementation manner of the first aspect, the step of summarizing thermodynamic trends of user interest tags and determining key user interest tags corresponding to the target user based on a user interest decision model includes:
acquiring a short video distribution trend and hot search entry distribution of the target user in a current short video interaction scene;
according to the short video distribution trend and the hot search entry distribution, and according to an interest relation prediction model, predicting interest relation indexes of each short video recommendation service of the current short video interaction scene to a target user;
respectively fusing the thermodynamic trends of the interest labels of the users with the corresponding interest contact indexes, and determining and updating the thermodynamic trends of the interest labels of the users;
and determining key user interest labels corresponding to the target user according to the thermodynamic trend of each updated user interest label and a user interest decision model, wherein the key user interest labels represent user interest points and corresponding interest confidence degrees.
In a possible implementation manner of the first aspect, the step of performing short video data recommendation on the target user based on the key user interest tag corresponding to the target user includes:
determining interest keyword tag relation networks of short video recommendation services on the same dimension according to the thermodynamic trends of interest tags of users;
respectively fusing the interest keyword tag relationship networks of the short video recommendation services with the corresponding interest contact indexes to determine an updated interest keyword tag relationship network;
determining interest flow direction relations among interest keyword labels according to an updated interest keyword label relation network of each short video recommendation service;
determining short video content tag data for short video recommendation evaluation based on the key user interest tags;
dividing the short video content label data according to the interest flow direction relation among the short video recommendation services, and determining the arrangement distribution of the short video content labels corresponding to the short video recommendation services;
and respectively issuing the short video content labels to corresponding short video recommendation services in a distributed manner so as to perform short video recommendation.
In one possible implementation of the first aspect, after loading the corresponding second user behavioral activity data to the corresponding short video recommendation service, the method further comprises:
analyzing the attention intentions of the users according to all first user behavior activity data of the same short video recommendation service, and determining attention intention information of a plurality of users;
matching the user attention intention information of each second user behavior activity data of the same short video recommendation service with the user attention intention information of the rest second user behavior activity data respectively, and calculating corresponding matching deviation degrees;
after each second user behavior activity data of the same short video recommendation service is compared with the rest second user behavior activity data, determining a plurality of matching deviation degrees according to each second user behavior activity data;
analyzing whether each matching deviation degree of each second user behavior activity data is larger than a second target value or not, and if so, performing deviation label marking on the corresponding second user behavior activity data;
extracting the number of the deviated label marks of each second user behavior activity data;
and analyzing whether the number of the deviated label marks of each second user behavior activity data is greater than a third target value, and if so, removing the second user behavior activity data from the corresponding short video recommendation service.
In a possible implementation manner of the first aspect, after summarizing thermodynamic trends of respective user interest tags and determining a key user interest tag corresponding to the target user based on a user interest decision model, the method further includes:
obtaining past user interest mining data of different past short video interaction scenes of the target user, wherein each past user interest mining data at least comprises first past user behavior activity data corresponding to the past short video interaction scenes and a priori key user interest label;
respectively carrying out behavior characteristic coding according to first past user behavior activity data of each past short video interaction scene, and determining first behavior characteristic distribution of each past short video interaction scene;
performing behavior characteristic coding on first user behavior activity data of a current short video interaction scene, and determining second behavior characteristic distribution;
comparing the first behavior feature distribution with the second behavior feature distribution, and loading the past user interest mining data with the distinguishing parameter value smaller than the fourth target value into a target data set;
mining data of interest of each past user in the target data set, respectively loading corresponding first past user behavior activity data into a set knowledge entity network for regularization conversion, and generating corresponding second prior user behavior activity data;
determining the service association degree between each second prior user behavior activity data and the corresponding short video recommendation service according to each short video recommendation service, and if the service association degree is greater than a first target value, loading the corresponding second prior user behavior activity data into the corresponding short video recommendation service;
extracting all second prior user behavior activity data associated with each short video recommendation service, loading the second prior user behavior activity data into an interest thermodynamic trend output model, and generating a corresponding reference user interest label thermodynamic trend;
summarizing the thermodynamic trend of each reference user interest label, and determining a decision key user interest label corresponding to the target user based on a user interest decision model;
determining a plurality of interest distinguishing parameters based on decision key user interest tags of the prior user interest mining data in the target data set and corresponding prior key user interest tags;
carrying out averaging calculation on a plurality of interest distinguishing parameters according to the global interest point distribution of the prior user interest mining data of the target data set, and determining an interest output optimization parameter value;
optimizing the key user interest tags based on the interest output optimization parameter values, and determining the optimized key user interest tags.
In a possible implementation manner of the first aspect, after the step of performing short video data recommendation on the target user based on the key user interest tag corresponding to the target user, the method further includes:
determining the target user as a first target user, and acquiring user activity path data corresponding to the first target user when initiating a short video release event next time based on recommended short video data;
acquiring a pre-trained release intention mining model;
loading the user activity path data to the pre-trained publishing intention mining model so as to obtain publishing intention knowledge points output by the pre-trained publishing intention mining model, wherein one publishing intention knowledge point represents one online short video publishing element;
acquiring a plurality of online short video publishing element libraries, wherein one online short video publishing element library corresponds to one publishing intention knowledge point, and each online short video publishing element library comprises a plurality of online short video publishing elements with different parameters;
the following steps are performed for each publishing intention knowledge point:
acquiring one online short video publishing element in a corresponding online short video publishing element library based on the publishing intention knowledge point as a candidate short video publishing element, and constructing a content push network corresponding to the user activity path data based on each candidate short video publishing element so as to form a target content push network; the target content push network comprises candidate short video publishing elements which are connected through different video element logic relations;
and sharing data content to other second target users with content sharing association based on the target content push network.
In one possible implementation of the first aspect, the publishing intent knowledge point comprises a shared publishing intent knowledge point; when the publishing intention knowledge point is a shared publishing intention knowledge point, the acquiring each publishing intention knowledge point of the short video publishing event further comprises:
acquiring data in user activity path data corresponding to the shared publishing intention knowledge points, wherein the data is called shared data;
acquiring shared service node information of the shared data;
acquiring a preset service node database, wherein the service node database comprises a plurality of preset service node data and service node intention knowledge points corresponding to each service node data; analyzing whether shared service node information of the shared data and each preset service node data meet a first target requirement, wherein the first target requirement corresponds to a release intention knowledge point, and if so, acquiring the service node intention knowledge point corresponding to the preset service node data meeting the first target requirement as the release intention knowledge point of the shared release intention knowledge point.
In a possible implementation manner of the first aspect, the step of obtaining, based on the publishing intention knowledge point, one online short video publishing element in a corresponding online short video publishing element library as a candidate short video publishing element includes:
calculating matching parameter values of the release intention knowledge points and each online short video release element in the online short video release element library corresponding to the release intention knowledge points respectively;
and analyzing whether one matching parameter value exceeds a preset matching parameter value in the acquired matching parameter values, if so, acquiring an online short video publishing element corresponding to the matching parameter value exceeding the preset matching parameter value as a candidate short video publishing element of the publishing intention knowledge point.
In one possible implementation of the first aspect, the parameters of the online short video publishing elements include video tagging parameters, and each online short video publishing element has a unique video tagging parameter in each online short video publishing element library;
the obtaining of one online short video publishing element in the corresponding online short video publishing element library based on the publishing intention knowledge point as a candidate short video publishing element includes:
acquiring a pre-trained short video publishing element mining model;
loading the publishing intention knowledge points to the pre-trained short video publishing element mining model so as to obtain model mining information output by the pre-trained short video publishing element mining model, wherein the model mining information represents video label parameters of the publishing intention knowledge points;
acquiring online short video publishing elements corresponding to the unique video tag parameter which is the same as the video tag parameter of the publishing intention knowledge point as candidate short video publishing elements;
when there is no online short video publishing element corresponding to the unique video tag parameter which is the same as the video tag parameter of the publishing intention knowledge point, the publishing intention knowledge point of the online short video publishing element which is not the same as the video tag parameter of the publishing intention knowledge point is called a fuzzy publishing intention knowledge point, and the step of acquiring one online short video publishing element in the corresponding online short video publishing element library as a candidate short video publishing element based on the publishing intention knowledge point further comprises:
decomposing fuzzy release intention knowledge points to obtain each release intention knowledge component;
decomposing each online short video publishing element in an online short video publishing element library corresponding to the publishing intention knowledge point, thereby obtaining a preset publishing intention knowledge component of each online short video publishing element; performing the following steps for each of the publishing intent knowledge components:
calculating matching parameter values of the release intention knowledge components and the preset release intention knowledge components; analyzing whether one matching parameter value is larger than a preset matching parameter value or not, and if so, acquiring a preset release intention knowledge component of which the matching parameter value is larger than the preset matching parameter value;
and splicing the acquired preset publishing intention knowledge components to form candidate short video publishing elements of the fuzzy publishing intention knowledge points.
In a second aspect, an embodiment of the present application further provides a short video data recommendation system based on tag analysis, where the short video data recommendation system based on tag analysis includes a processor and a machine-readable storage medium, where a computer program is stored in the machine-readable storage medium, and the computer program is loaded and executed in conjunction with the processor to implement the short video data recommendation method based on tag analysis of the above first aspect.
By adopting the technical scheme of any one aspect, each first user behavior activity data is loaded into a set knowledge entity network respectively for regularization conversion, and corresponding second user behavior activity data is generated, so that the reliability of subsequent mining processing can be improved, on the basis, the corresponding second user behavior activity data is loaded into corresponding short video recommendation services by dividing different short video recommendation dimensions into a plurality of short video recommendation services, all second user behavior activity data associated with each short video recommendation service are extracted and loaded into corresponding interest thermodynamic trend output models to generate corresponding user interest thermodynamic trends, so that the key user interest tags corresponding to the target users are further determined, and the short video data recommendation is performed on the target users based on the key user interest tags corresponding to the target users, and then, personalized recommendation is performed after user interest mining is performed according to different short video recommendation dimensions, and the short video recommendation accuracy is improved.
Drawings
To more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be implemented in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and those skilled in the art can also obtain other related drawings by combining these drawings without inventive efforts.
Fig. 1 is a schematic flowchart of a short video data recommendation method based on tag analysis according to an embodiment of the present application.
Fig. 2 is a schematic block diagram of a structure of a short video data recommendation system based on tag analysis for implementing the short video data recommendation method based on tag analysis according to an embodiment of the present application.
Detailed Description
The following description is presented to enable one of ordinary skill in the art to make and use the invention and is provided in the context of a particular application and its requirements. It will be apparent to those of ordinary skill in the art that various changes can be made to the disclosed embodiments and that the general principles defined in this application can be applied to other embodiments and applications without departing from the principles and scope of the application. Thus, the present application is not limited to the described embodiments, but should be accorded the widest scope consistent with the claims.
The terminology used in the description presented herein is for the purpose of describing particular example embodiments only and is not intended to limit the scope of the present application. As used herein, the singular forms "a", "an" and "the" may include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, components, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, components, and/or groups thereof.
These and other features and characteristics of the present application, as well as the methods of operation and functions of the related elements of structure and the combination of parts and economies of manufacture, will become more apparent upon consideration of the following description of the drawings, all of which form a part of this specification. It is to be understood, however, that the drawings are designed solely for the purposes of illustration and description and are not intended as a definition of the limits of the application. It should be understood that the drawings are not to scale.
Flow charts are used herein to illustrate operations performed by systems incorporating some embodiments of the present application. It should be understood that the operations in the flow diagrams may be performed out of order. Rather, various steps may be processed on a reverse order basis or simultaneously. Further, one or more other operations may be added to the flowchart. One or more operations may also be deleted from the flowchart.
The present application is described in detail below with reference to fig. 1 and 2 of the drawings accompanying the specification, and the particular methods of operation in the method embodiments may also be applied to apparatus embodiments or system embodiments. Fig. 1 is a schematic flowchart of a short video data recommendation method based on tag analysis according to an embodiment of the present application. The short video data recommendation method based on the label analysis comprises the following steps:
the STEP102 acquires a plurality of first user behavior activity data based on the user behavior activity monitoring programs configured on the short video service pages.
In this embodiment, the short video service page may exist in the form of an application program, or may exist in the form of a web page, which is not limited specifically. The user behavior activity monitoring program may monitor user behavior activity of the target user according to the set monitoring field dimension, and further obtain a plurality of first user behavior activity data.
STEP104, loading each first user behavior activity data into a set knowledge entity network respectively for regularized conversion, and generating corresponding second user behavior activity data.
In this embodiment, the setting of the knowledge entity network may be implemented based on a knowledge graph algorithm, and is used to perform regularized data conversion on each user behavior activity in the first user behavior activity data, so as to generate corresponding second user behavior activity data.
STEP106, determining a plurality of short video recommendation services according to the short video subscription items of the user. For example, if a user's short video subscription items include live launch items and video creation items, the corresponding short video recommendation services may include a main broadcast recommendation service as well as an originator recommendation service.
STEP108, determining a service association degree between each second user behavior activity data and the corresponding short video recommendation service according to each short video recommendation service, and if the service association degree is greater than a first target value, loading the corresponding second user behavior activity data into the corresponding short video recommendation service. For example, the target behavior feature in each second user behavior activity data may be analyzed, and then the matching feature matching the short video recommendation service in the target behavior feature may be analyzed, for example, when the short video recommendation service is an author recommendation service, the author-related feature in the target behavior feature may be analyzed, and then a ratio of the author-related feature to the target behavior feature may be calculated as the service association degree.
The STEP110 respectively configures a plurality of interest thermodynamic trend output models according to the plurality of short video recommendation services, performs model tuning on each interest thermodynamic trend output model according to the collected user interest learning data, and outputs the interest thermodynamic trend output model after model tuning.
For example, for different short video recommendation services, different user interest learning data samples can be collected in advance to perform model tuning on each interest thermodynamic trend output model, for example, the user interest learning data samples can include user behavior activity data samples and corresponding prior interest thermodynamic trends, then the user behavior activity data samples are input into the interest thermodynamic trend output model to determine corresponding target interest thermodynamic trends, then model tuning is performed on the interest thermodynamic trend output model based on characteristic differences between the target interest thermodynamic trends and the prior interest thermodynamic trends, and the interest thermodynamic trend output model after model tuning is output.
STEP112, extracting all second user behavior activity data associated with each short video recommendation service, and loading the second user behavior activity data into the interest thermodynamic trend output model after the corresponding model is tuned, so as to generate a corresponding user interest tag thermodynamic trend.
In this embodiment, the thermodynamic trend of the user interest tags may include thermodynamic value change information corresponding to each user interest tag arranged based on the development sequence of the time nodes.
STEP114, summarizing thermodynamic trends of the user interest tags, determining key user interest tags corresponding to the target user based on a user interest decision model, and performing short video data recommendation on the target user based on the key user interest tags corresponding to the target user.
For example, the user interest tag with the thermal value larger than the set thermal value and the thermal value larger than the set thermal value may be determined as a key user interest tag, so as to perform short video data recommendation on the target user based on the key user interest tag corresponding to the target user.
According to the embodiment of the application, the first user behavior activity data are obtained through the short video service page, and due to the fact that characteristics of the first user behavior activity data are greatly different and the data recording format of the single first user behavior activity data is irregular, the reliability of subsequent mining processing on the first user behavior activity data can be affected. Based on this, the embodiment of the application performs regularized conversion on each first user behavior activity data by setting a knowledge entity network (knowledge graph), so as to obtain second user behavior activity data. For a target user, the target user can be divided into a plurality of short video recommendation services from different short video recommendation dimensions (such as author recommendation dimension, anchor recommendation dimension, audience recommendation dimension, and the like), the first user behavior activity data of the short video service page is gathered according to the short video recommendation services, and then the analysis result corresponding to the short video recommendation service is analyzed through each interest thermodynamic trend output model. And finally, combining all user interest labels with thermodynamic trends, and accurately predicting the user interest of the target user based on a user interest decision model.
Therefore, in the embodiment of the application, each first user behavior activity data is loaded into a set knowledge entity network respectively for regularization conversion, and corresponding second user behavior activity data is generated, so that the reliability of subsequent mining processing can be improved, on the basis, the corresponding second user behavior activity data is loaded into corresponding short video recommendation services by dividing the short video recommendation dimensions into a plurality of short video recommendation services, so that all second user behavior activity data associated with each short video recommendation service are extracted and loaded into corresponding interest thermodynamic trend output models to generate corresponding user interest label thermodynamic trends, and further the key user interest labels corresponding to the target users are determined, so that the short video data recommendation is performed on the target users based on the key user interest labels corresponding to the target users, and then, personalized recommendation is performed after user interest mining is performed according to different short video recommendation dimensions, and the short video recommendation accuracy is improved.
Each short video recommendation service has a relevant definition or dimension, for example, some short video recommendation services are creator recommendation dimensions, and some short video recommendation services are anchor recommendation dimensions and audience recommendation dimensions; determining the service association degree between each second user behavior activity data and the corresponding short video recommendation service, for example, performing user attention intention analysis on the first user behavior activity data to obtain user attention intention information, and determining the service association degree between the user attention intention information and each short video recommendation service, and if the service association degree is greater than a first target value, determining that the first user behavior activity data falls into the short video recommendation service. The first user behavior activity data may include various information, that is, may correspond to a plurality of short video recommendation services.
In some exemplary design ideas, the step of summarizing thermodynamic trends of user interest tags and determining key user interest tags corresponding to the target user based on a user interest decision model includes:
acquiring a short video distribution trend and hot search entry distribution of the target user in a current short video interaction scene;
according to the short video distribution trend and the hot search entry distribution, and according to an interest relation prediction model, predicting interest relation indexes of each short video recommendation service of the current short video interaction scene to a target user;
respectively fusing the thermodynamic trends of the interest labels of the users with the corresponding interest contact indexes, and determining and updating the thermodynamic trends of the interest labels of the users;
and determining key user interest labels corresponding to the target users according to the thermodynamic trend of the updated user interest labels and the user interest decision model.
Wherein, the weight of each short video recommendation service to the target user is different in different periods. For example, in the live broadcast centralization stage, the interest contact index of the anchor recommendation dimension to the target user is relatively high, and therefore the corresponding interest contact index is relatively large. The interest relation indexes are updated in real time by combining the short video publishing trend of the current short video interaction scene and the interest relation indexes of hot search vocabulary entry distribution to each short video recommendation service, and the key user interest labels of the target users are dynamically predicted according to the updated interest relation indexes, so that the real-time reliability of user interest mining is improved.
The interest contact prediction model is configured, model tuning and optimization are carried out according to a large amount of training sample data, so that the interest contact prediction model after model tuning is obtained, the interest contact prediction model after model tuning is adopted to carry out interest contact index prediction, and accuracy of interest contact index prediction is improved.
In some exemplary design approaches, after determining, by the user interest decision model, the key user interest tag corresponding to the target user, the method further includes:
determining interest keyword tag relation networks of short video recommendation services on the same dimension according to the thermodynamic trends of interest tags of users;
respectively fusing the interest keyword tag relationship networks of the short video recommendation services with the corresponding interest contact indexes to determine an updated interest keyword tag relationship network;
determining interest flow direction relations among interest keyword labels according to an updated interest keyword label relation network of each short video recommendation service;
determining short video content tag data for short video recommendation evaluation based on the key user interest tags;
dividing the short video content label data according to the interest flow direction relation among the short video recommendation services, and determining the arrangement distribution of the short video content labels corresponding to the short video recommendation services;
and respectively issuing the short video content labels to corresponding short video recommendation services in a distributed manner so as to perform short video recommendation.
The short video recommendation services may have corresponding interest keyword label relationship networks in the same dimension, the interest flow direction relationship may be determined according to the interest keyword label relationship networks of the short video recommendation services, then the short video content label data according to the interest flow direction relationship is determined based on the interest labels of the key users, and then the short video content label data is distributed to the short video recommendation services according to the interest flow direction relationship.
In some exemplary design considerations, after loading the corresponding second user behavioral activity data to the corresponding short video recommendation service, the method further comprises:
analyzing the attention intentions of the users according to all first user behavior activity data of the same short video recommendation service, and determining attention intention information of a plurality of users;
matching the user attention intention information of each second user behavior activity data of the same short video recommendation service with the user attention intention information of the rest second user behavior activity data respectively, and calculating corresponding matching deviation degrees;
after each second user behavior activity data of the same short video recommendation service is compared with the rest second user behavior activity data, determining a plurality of matching deviation degrees according to each second user behavior activity data;
analyzing whether each matching deviation degree of each second user behavior activity data is larger than a second target value or not, and if yes, performing deviation label marking on the corresponding second user behavior activity data;
extracting the number of the deviated label marks of each second user behavior activity data;
and analyzing whether the quantity of the deviation label marks of each second user behavior activity data is larger than a third target value, and if so, removing the second user behavior activity data from the corresponding short video recommendation service.
In some exemplary design ideas, after summarizing thermodynamic trends of user interest tags and determining a key user interest tag corresponding to the target user based on a user interest decision model, the method further includes:
acquiring prior user interest mining data of different prior short video interaction scenes, wherein each prior user interest mining data at least comprises first prior user behavior activity data corresponding to the prior short video interaction scenes and a prior key user interest tag;
respectively carrying out behavior feature coding according to first prior user behavior activity data of each prior short video interaction scene, and determining first behavior feature distribution of each prior short video interaction scene;
performing behavior feature coding on first user behavior activity data of a current short video interaction scene, and determining second behavior feature distribution;
comparing the first behavior feature distribution with the second behavior feature distribution, and loading the prior user interest mining data with the distinguishing parameter value smaller than the fourth target value into a target data set;
mining data of interest of each prior user in the target data set, respectively loading corresponding first prior user behavior activity data into a set knowledge entity network for regularization conversion, and generating corresponding second prior user behavior activity data;
determining the service association degree of each second priori user behavior activity data before the corresponding short video recommendation service according to each short video recommendation service, and if the service association degree is greater than a first target value, loading the corresponding second priori user behavior activity data into the corresponding short video recommendation service;
extracting all second prior user behavior activity data associated with each short video recommendation service, loading the second prior user behavior activity data into an interest thermodynamic trend output model, and generating a corresponding reference user interest label thermodynamic trend;
summarizing thermodynamic trends of the interest labels of the reference users, and determining the interest labels of decision key users corresponding to the target users based on a user interest decision model;
determining a plurality of interest distinguishing parameters based on decision key user interest tags of the prior user interest mining data in the target data set and corresponding prior key user interest tags;
carrying out averaging calculation on a plurality of interest distinguishing parameters according to the global interest point distribution of prior user interest mining data of a target data set, and determining an interest output optimization parameter value;
optimizing the key user interest tags based on the interest output optimization parameter values, and determining the optimized key user interest tags.
On the basis of the above embodiment, in the embodiment of the present application, the target user may further be determined as a first target user, user activity path data corresponding to the first target user when initiating a short video distribution event next time based on recommended short video data is obtained, a pre-trained distribution intention mining model is obtained, and the user activity path data is loaded to the pre-trained distribution intention mining model, so as to obtain distribution intention knowledge points output by the pre-trained distribution intention mining model, where one distribution intention knowledge point represents an online short video distribution element. The method comprises the steps of obtaining a plurality of online short video publishing element libraries, wherein one online short video publishing element library corresponds to a publishing intention knowledge point, and each online short video publishing element library comprises a plurality of online short video publishing elements with different parameters. The following steps are performed for each publishing intention knowledge point:
acquiring one online short video publishing element in a corresponding online short video publishing element library based on the publishing intention knowledge point as a candidate short video publishing element, and constructing a content push network corresponding to the user activity path data based on each candidate short video publishing element so as to form a target content push network; the target content push network comprises candidate short video publishing elements which are connected through different video element logic relations; and sharing data content to other second target users with content sharing association based on the target content push network.
Based on the steps, the online short video publishing elements basically matched with the publishing intention knowledge points corresponding to the user activity path data are obtained through the online short video publishing element library to serve as candidate short video publishing elements, and the influence of factors such as noise and the like caused by directly pushing the content by using the user activity path data is overcome.
In some exemplary design concepts, the publishing intent knowledge points comprise shared publishing intent knowledge points; when the publishing intention knowledge point is a shared publishing intention knowledge point, the acquiring each publishing intention knowledge point of the short video publishing event further comprises: acquiring data in user activity path data corresponding to the shared publishing intention knowledge points, wherein the data is called shared data; acquiring shared service node information of the shared data; acquiring a preset service node database, wherein the service node database comprises a plurality of preset service node data and service node intention knowledge points corresponding to each service node data; analyzing whether shared service node information of the shared data and each preset service node data meet a first target requirement, wherein the first target requirement corresponds to a release intention knowledge point, and if so, acquiring the service node intention knowledge point corresponding to the preset service node data meeting the first target requirement as the release intention knowledge point of the shared release intention knowledge point.
In some exemplary design ideas, the step of obtaining, based on the publishing intention knowledge point, one online short video publishing element in a corresponding online short video publishing element library as a candidate short video publishing element includes: calculating matching parameter values of the release intention knowledge points and each online short video release element in the online short video release element library corresponding to the release intention knowledge points respectively;
and analyzing whether one matching parameter value exceeds a preset matching parameter value in the acquired matching parameter values, if so, acquiring an online short video publishing element corresponding to the matching parameter value exceeding the preset matching parameter value as a candidate short video publishing element of the publishing intention knowledge point.
In some exemplary design considerations, the parameters of the online short video publishing elements include video tagging parameters, each online short video publishing element having a unique video tagging parameter in each online short video publishing element library; the obtaining of one online short video publishing element in the corresponding online short video publishing element library based on the publishing intention knowledge point as a candidate short video publishing element includes: acquiring a pre-trained short video publishing element mining model; loading the publishing intention knowledge points to the pre-trained short video publishing element mining model so as to obtain model mining information output by the pre-trained short video publishing element mining model, wherein the model mining information represents video label parameters of the publishing intention knowledge points; and acquiring online short video publishing elements corresponding to the unique video tag parameters which are the same as the video tag parameters of the publishing intention knowledge points as candidate short video publishing elements.
When there is no online short video publishing element corresponding to the unique video tag parameter that is the same as the video tag parameter of the publishing intention knowledge point, the publishing intention knowledge point of the online short video publishing element corresponding to the unique video tag parameter that is the same as the video tag parameter of the publishing intention knowledge point is called a fuzzy publishing intention knowledge point, and the step of obtaining one online short video publishing element in the corresponding online short video publishing element library as a candidate short video publishing element based on the publishing intention knowledge point further comprises: decomposing fuzzy release intention knowledge points to obtain each release intention knowledge component; decomposing each online short video publishing element in an online short video publishing element library corresponding to the publishing intention knowledge point, thereby obtaining a preset publishing intention knowledge component of each online short video publishing element; performing the following steps for each of the publishing intent knowledge components: calculating matching parameter values of the release intention knowledge components and the preset release intention knowledge components; analyzing whether one matching parameter value is larger than a preset matching parameter value or not, and if so, acquiring a preset release intention knowledge component of which the matching parameter value is larger than the preset matching parameter value; and splicing the acquired preset publishing intention knowledge components to form candidate short video publishing elements of the fuzzy publishing intention knowledge points.
Fig. 2 schematically illustrates a short video data recommendation system 100 based on tag analysis that may be used to implement various embodiments described in the present application.
For one embodiment, fig. 2 shows a tag analysis based short video data recommendation system 100, the tag analysis based short video data recommendation system 100 having one or more processors 102, a control module (chipset) 104 coupled to at least one of the processor(s) 102, a memory 106 coupled to the control module 104, a non-volatile memory (NVM)/storage 108 coupled to the control module 104, one or more input/output devices 110 coupled to the control module 104, and a network interface 112 coupled to the control module 106.
The processor 102 may include one or more single-core or multi-core processors, and the processor 102 may include any combination of general-purpose or special-purpose processors (e.g., graphics processors, application processors, baseband processors, etc.). In some embodiments, the tag analysis-based short video data recommendation system 100 can be used as a server device such as a gateway in the embodiments of the present application.
In some embodiments, the tag analysis-based short video data recommendation system 100 may include one or more computer-readable media (e.g., memory 106 or NVM/storage 108) having instructions 114 and one or more processors 102 in combination with the one or more computer-readable media and configured to execute the instructions 114 to implement modules to perform the actions described in the present disclosure.
For one embodiment, control module 104 may include any suitable interface controller to provide any suitable interface to at least one of the processor(s) 102 and/or any suitable device or component in communication with control module 104.
Control module 104 may include a memory controller module to provide an interface to memory 106. The memory controller module may be a hardware module, a software module, and/or a firmware module.
The memory 106 may be used, for example, to load and store data and/or instructions 114 for the tag analysis based short video data recommendation system 100. For one embodiment, memory 106 may comprise any suitable volatile memory, such as suitable DRAM. In some embodiments, the memory 106 may comprise a double data rate type four synchronous dynamic random access memory (DDR4 SDRAM).
For one embodiment, control module 104 may include one or more input/output controllers to provide an interface to NVM/storage 108 and input/output device(s) 110.
For example, NVM/storage 108 may be used to store data and/or instructions 114. NVM/storage 108 may include any suitable non-volatile memory (e.g., flash memory) and/or may include any suitable non-volatile storage device(s) (e.g., one or more Hard Disk Drives (HDDs), one or more Compact Disc (CD) drives, and/or one or more Digital Versatile Disc (DVD) drives).
The NVM/storage 108 may include storage resources that are physically part of the device on which the tag analysis based short video data recommendation system 100 is installed, or it may be accessible by the device and may not be necessary as part of the device. For example, NVM/storage 108 may be accessible via input/output device(s) 110 over a network.
The input/output device(s) 110 may provide an interface for the tag analysis based short video data recommendation system 100 to communicate with any other suitable device, the input/output device 110 may include a communication component, a pinyin component, a sensor component, and so forth. The network interface 112 may provide an interface for the tag analysis based short video data recommendation system 100 to communicate wirelessly according to one or more networks, and the tag analysis based short video data recommendation system 100 may communicate wirelessly according to any of one or more wireless network standards and/or protocols with one or more components of a wireless network, such as accessing a communication standard based wireless network, such as WiFi, 2G, 3G, 4G, 5G, etc., or a combination thereof.
For one embodiment, at least one of the processor(s) 102 may be packaged together with logic for one or more controllers (e.g., memory controller modules) of the control module 104. For one embodiment, at least one of the processor(s) 102 may be packaged together with logic for one or more controller(s) of the control module 104 to form a System In Package (SiP). For one embodiment, at least one of the processor(s) 102 may be integrated on the same die with logic for one or more controller(s) of the control module 104. For one embodiment, at least one of the processor(s) 102 may be integrated on the same die with logic of one or more controllers of the control module 104 to form a system on a chip (SoC).
In various embodiments, the tag analysis based short video data recommendation system 100 may be, but is not limited to being: a server, a desktop computing device, or a mobile computing device (e.g., a laptop computing device, a handheld computing device, a tablet, a netbook, etc.) among other terminal devices. In various embodiments, the tag analysis based short video data recommendation system 100 may have more or fewer components and/or different architectures. For example, in some embodiments, the tag analysis based short video data recommendation system 100 includes one or more cameras, a keyboard, a Liquid Crystal Display (LCD) screen (including touch screen displays), a non-volatile memory port, multiple antennas, a graphics chip, an Application Specific Integrated Circuit (ASIC), and speakers.
An embodiment of the present application provides an electronic device, including: one or more processors; and one or more machine readable media having instructions stored thereon that, when executed by the one or more processors, cause the electronic device to perform a data processing method as described in one or more of the present applications.
For the device embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, refer to the partial description of the method embodiment.
The embodiments in the present specification are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and the basis of a flow and/or block of the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present application have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including the preferred embodiment and all alterations and modifications as fall within the scope of the embodiments of the application.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal 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 terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, article, or terminal equipment comprising the element.
The data processing method and apparatus provided by the present application are introduced in detail, and specific examples are applied in this document to explain the principles and embodiments of the present application, and the descriptions of the above embodiments are only used to help understand the method and core ideas of the present application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.

Claims (9)

1. A short video data recommendation method based on label analysis is characterized by being applied to a short video data recommendation system based on label analysis, and the method comprises the following steps:
acquiring a plurality of first user behavior activity data based on a user behavior activity monitoring program configured on each short video service page of a target user;
loading each first user behavior activity data into a set knowledge entity network respectively for regularized conversion, and generating corresponding second user behavior activity data;
determining a plurality of short video recommendation services according to the short video subscription items of the user;
determining service association degree between each second user behavior activity data and the corresponding short video recommendation service according to each short video recommendation service, and if the service association degree is greater than a first target value, loading the corresponding second user behavior activity data into the corresponding short video recommendation service;
respectively configuring a plurality of interest thermodynamic trend output models according to a plurality of short video recommendation services, carrying out model tuning on each interest thermodynamic trend output model according to the collected user interest learning data, and outputting the interest thermodynamic trend output model after model tuning;
extracting all second user behavior activity data associated with each short video recommendation service, and loading the second user behavior activity data into the interest thermodynamic trend output model after the corresponding model is adjusted and optimized to generate a corresponding user interest label thermodynamic trend;
summarizing the thermodynamic trend of each user interest label, determining a key user interest label corresponding to the target user based on a user interest decision model, and recommending short video data to the target user based on the key user interest label corresponding to the target user;
summarizing the thermodynamic trend of each user interest label, and determining a key user interest label corresponding to the target user based on a user interest decision model, wherein the steps comprise:
acquiring a short video distribution trend and hot search entry distribution of the target user in a current short video interaction scene;
according to the short video distribution trend and the hot search entry distribution, and according to an interest relation prediction model, predicting interest relation indexes of each short video recommendation service of the current short video interaction scene to a target user; the method comprises the steps of configuring an interest contact prediction model, carrying out model tuning and optimization on the interest contact prediction model according to training sample data to obtain an interest contact prediction model after model tuning, and carrying out interest contact index prediction by adopting the interest contact prediction model after model tuning;
fusing the thermodynamic trends of the user interest labels with corresponding interest contact indexes respectively, and determining and updating the thermodynamic trends of the user interest labels, wherein the thermodynamic trends of the user interest labels comprise thermodynamic value change information corresponding to the user interest labels arranged based on the development sequence of time nodes;
determining key user interest labels corresponding to the target user according to the thermodynamic trend of each updated user interest label and a user interest decision model, wherein the user interest labels of which the thermodynamic values are greater than the set thermodynamic values and of which the thermodynamic values are greater than the set thermodynamic values are determined as the key user interest labels, and the key user interest labels represent user interest points and corresponding interest confidence coefficients;
for different short video recommendation services, different user interest learning data samples are collected in advance to perform model tuning on each interest thermodynamic trend output model, the user interest learning data samples comprise user behavior activity data samples and corresponding priori interest thermodynamic trends, the user behavior activity data samples are input into the interest thermodynamic trend output model to determine corresponding target interest thermodynamic trends, then model tuning is performed on the interest thermodynamic trend output model based on characteristic differences between the target interest thermodynamic trends and the priori interest thermodynamic trends, and the interest thermodynamic trend output model after model tuning is output.
2. The method of claim 1, wherein the step of performing short video data recommendation on the target user based on the key user interest tag corresponding to the target user comprises:
determining interest keyword tag relation networks of short video recommendation services on the same dimension according to the thermodynamic trends of interest tags of users;
respectively fusing the interest keyword tag relation networks of the short video recommendation services with the corresponding interest contact indexes to determine and update the interest keyword tag relation networks;
determining interest flow direction relations among interest keyword labels according to an updated interest keyword label relation network of each short video recommendation service;
determining short video content tag data for short video recommendation evaluation based on the key user interest tags;
dividing the short video content label data according to the interest flow direction relation among the short video recommendation services, and determining the arrangement distribution of the short video content labels corresponding to the short video recommendation services;
and respectively issuing the short video content labels to corresponding short video recommendation services in a distributed manner so as to perform short video recommendation.
3. The tag analysis-based short video data recommendation method of claim 1, wherein after loading the corresponding second user behavioral activity data to the corresponding short video recommendation service, the method further comprises:
analyzing the attention intentions of the users according to all first user behavior activity data of the same short video recommendation service, and determining attention intention information of a plurality of users;
matching the user attention intention information of each second user behavior activity data of the same short video recommendation service with the user attention intention information of the rest second user behavior activity data respectively, and calculating corresponding matching deviation degrees;
after each second user behavior activity data of the same short video recommendation service is compared with the rest second user behavior activity data, determining a plurality of matching deviation degrees according to each second user behavior activity data;
analyzing whether each matching deviation degree of each second user behavior activity data is larger than a second target value or not, and if so, performing deviation label marking on the corresponding second user behavior activity data;
extracting the number of the deviated label marks of each second user behavior activity data;
and analyzing whether the quantity of the deviation label marks of each second user behavior activity data is larger than a third target value, and if so, removing the second user behavior activity data from the corresponding short video recommendation service.
4. The method for recommending short video data based on tag analysis according to claim 1, wherein after summarizing thermodynamic trends of individual user interest tags and determining key user interest tags corresponding to the target user based on a user interest decision model, the method further comprises:
obtaining past user interest mining data of different past short video interaction scenes of the target user, wherein each past user interest mining data at least comprises first past user behavior activity data corresponding to the past short video interaction scenes and a priori key user interest label;
respectively carrying out behavior feature coding according to first past user behavior activity data of each past short video interaction scene, and determining first behavior feature distribution of each past short video interaction scene;
performing behavior feature coding on first user behavior activity data of a current short video interaction scene, and determining second behavior feature distribution;
comparing the first behavior feature distribution with the second behavior feature distribution, and loading the past user interest mining data with the distinguishing parameter value smaller than the fourth target value into a target data set;
mining data of interest of each past user in the target data set, respectively loading corresponding first past user behavior activity data into a set knowledge entity network for regularization conversion, and generating corresponding second prior user behavior activity data;
determining service association degree between each second priori user behavior activity data and the corresponding short video recommendation service according to each short video recommendation service, and if the service association degree is greater than a first target value, loading the corresponding second priori user behavior activity data into the corresponding short video recommendation service;
extracting all second prior user behavior activity data associated with each short video recommendation service, loading the second prior user behavior activity data into an interest thermodynamic trend output model, and generating a corresponding reference user interest label thermodynamic trend;
summarizing the thermodynamic trend of each reference user interest label, and determining a decision key user interest label corresponding to the target user based on a user interest decision model;
determining a plurality of interest distinguishing parameters based on decision key user interest tags of the prior user interest mining data in the target data set and corresponding prior key user interest tags;
carrying out averaging calculation on a plurality of interest distinguishing parameters according to the global interest point distribution of prior user interest mining data of a target data set, and determining an interest output optimization parameter value;
optimizing the key user interest tags based on the interest output optimization parameter values, and determining the optimized key user interest tags.
5. The method for recommending short video data based on tag analysis according to any of claims 1-4, wherein after the step of recommending short video data to the target user based on the key user interest tag corresponding to the target user, the method further comprises:
determining the target user as a first target user, and acquiring user activity path data corresponding to the first target user when initiating a short video release event next time based on recommended short video data;
acquiring a pre-trained release intention mining model;
loading the user activity path data to the pre-trained publishing intention mining model so as to obtain publishing intention knowledge points output by the pre-trained publishing intention mining model, wherein one publishing intention knowledge point represents one online short video publishing element;
acquiring a plurality of online short video publishing element libraries, wherein one online short video publishing element library corresponds to one publishing intention knowledge point, and each online short video publishing element library comprises a plurality of online short video publishing elements with different parameters;
the following steps are performed for each publishing intention knowledge point:
acquiring one online short video publishing element in a corresponding online short video publishing element library based on the publishing intention knowledge point as a candidate short video publishing element, and constructing a content push network corresponding to the user activity path data based on each candidate short video publishing element so as to form a target content push network; the target content push network comprises candidate short video publishing elements which are connected through different video element logic relations;
and sharing data content to other second target users with content sharing association based on the target content push network.
6. The tag analysis-based short video data recommendation method according to claim 5, wherein said distribution-intention knowledge points comprise shared distribution-intention knowledge points; when the publishing intent knowledge point is a shared publishing intent knowledge point, the method further comprises:
acquiring data in user activity path data corresponding to the shared publishing intention knowledge points, wherein the data is called shared data;
acquiring shared service node information of the shared data;
acquiring a preset service node database, wherein the service node database comprises a plurality of preset service node data and service node intention knowledge points corresponding to each service node data; analyzing whether shared service node information of the shared data and each preset service node data meet a first target requirement, wherein the first target requirement corresponds to a release intention knowledge point, and if so, acquiring the service node intention knowledge point corresponding to the preset service node data meeting the first target requirement as the release intention knowledge point of the shared release intention knowledge point.
7. The tag analysis-based short video data recommendation method according to claim 6, wherein said step of obtaining one online short video publishing element in the corresponding online short video publishing element library as a candidate short video publishing element based on the publishing intention knowledge point comprises:
calculating matching parameter values of the release intention knowledge points and each online short video release element in the online short video release element library corresponding to the release intention knowledge points respectively;
and analyzing whether one matching parameter value exceeds a preset matching parameter value in the acquired matching parameter values, if so, acquiring an online short video publishing element corresponding to the matching parameter value exceeding the preset matching parameter value as a candidate short video publishing element of the publishing intention knowledge point.
8. The tag analysis-based short video data recommendation method according to claim 7, wherein the parameters of the online short video publishing elements comprise video tag parameters, each online short video publishing element having a unique video tag parameter in each online short video publishing element library;
the obtaining of one online short video publishing element in the corresponding online short video publishing element library based on the publishing intention knowledge point as a candidate short video publishing element includes:
acquiring a pre-trained short video publishing element mining model;
loading the publishing intention knowledge points to the pre-trained short video publishing element mining model so as to obtain model mining information output by the pre-trained short video publishing element mining model, wherein the model mining information represents video label parameters of the publishing intention knowledge points;
acquiring online short video publishing elements corresponding to the unique video tag parameter which is the same as the video tag parameter of the publishing intention knowledge point as candidate short video publishing elements;
when there is no online short video publishing element corresponding to the unique video tag parameter which is the same as the video tag parameter of the publishing intention knowledge point, the publishing intention knowledge point of the online short video publishing element which is not the same as the video tag parameter of the publishing intention knowledge point is called a fuzzy publishing intention knowledge point, and the step of acquiring one online short video publishing element in the corresponding online short video publishing element library as a candidate short video publishing element based on the publishing intention knowledge point further comprises:
decomposing fuzzy release intention knowledge points to obtain each release intention knowledge component;
decomposing each online short video publishing element in an online short video publishing element library corresponding to the publishing intention knowledge point, thereby obtaining a preset publishing intention knowledge component of each online short video publishing element; performing the following steps for each release intention knowledge component:
calculating matching parameter values of the release intention knowledge components and the preset release intention knowledge components; analyzing whether one matching parameter value is larger than a preset matching parameter value or not, and if so, acquiring a preset release intention knowledge component of which the matching parameter value is larger than the preset matching parameter value;
and splicing the acquired preset publishing intention knowledge components to form candidate short video publishing elements of the fuzzy publishing intention knowledge points.
9. A tag analysis based short video data recommendation system, comprising a processor and a machine-readable storage medium, wherein the machine-readable storage medium has stored therein machine-executable instructions, which are loaded and executed by the processor, to implement the tag analysis based short video data recommendation method of any one of claims 1 to 8.
CN202210812661.9A 2022-07-12 2022-07-12 Short video data recommendation method and system based on label analysis Active CN114896454B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210812661.9A CN114896454B (en) 2022-07-12 2022-07-12 Short video data recommendation method and system based on label analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210812661.9A CN114896454B (en) 2022-07-12 2022-07-12 Short video data recommendation method and system based on label analysis

Publications (2)

Publication Number Publication Date
CN114896454A CN114896454A (en) 2022-08-12
CN114896454B true CN114896454B (en) 2022-09-27

Family

ID=82729533

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210812661.9A Active CN114896454B (en) 2022-07-12 2022-07-12 Short video data recommendation method and system based on label analysis

Country Status (1)

Country Link
CN (1) CN114896454B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115270003B (en) * 2022-09-27 2023-01-06 创域智能(常熟)网联科技有限公司 Information recommendation method and system based on Internet of things platform behavior data mining
CN116567294A (en) * 2023-05-19 2023-08-08 上海国威互娱文化科技有限公司 Panoramic video segmentation processing method and system
CN117150075B (en) * 2023-10-30 2024-02-13 轻岚(厦门)网络科技有限公司 Short video intelligent recommendation system based on data analysis

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113360763A (en) * 2021-06-19 2021-09-07 阚忠建 Service attention tendency prediction method based on artificial intelligence and artificial intelligence cloud system
CN114048387A (en) * 2021-11-23 2022-02-15 赵运柱 Content recommendation method based on big data and AI prediction and artificial intelligence cloud system
CN114419501A (en) * 2022-01-11 2022-04-29 平安普惠企业管理有限公司 Video recommendation method and device, computer equipment and storage medium

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10579632B2 (en) * 2017-12-18 2020-03-03 Microsoft Technology Licensing, Llc Personalized content authoring driven by recommendations
US11636394B2 (en) * 2020-06-25 2023-04-25 Microsoft Technology Licensing, Llc Differentiable user-item co-clustering

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113360763A (en) * 2021-06-19 2021-09-07 阚忠建 Service attention tendency prediction method based on artificial intelligence and artificial intelligence cloud system
CN114048387A (en) * 2021-11-23 2022-02-15 赵运柱 Content recommendation method based on big data and AI prediction and artificial intelligence cloud system
CN114419501A (en) * 2022-01-11 2022-04-29 平安普惠企业管理有限公司 Video recommendation method and device, computer equipment and storage medium

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Approach for Video Classification with Multi-label on YouTube-8M Dataset;Shin, Kwangsoo等;《Springer》;20181231;全文 *
基于兴趣推荐算法的短视频传播模式研究;任立国等;《江西电力》;20220430;全文 *
抖音seo短视频领域用户的阅读模式趋于标签化;抖盈;《https://baijiahao.baidu.com/s?id=1735512998514542658&wfr=spider&for=pc》;20220613;全文 *

Also Published As

Publication number Publication date
CN114896454A (en) 2022-08-12

Similar Documents

Publication Publication Date Title
US20210027160A1 (en) End-to-end deep collaborative filtering
CN114896454B (en) Short video data recommendation method and system based on label analysis
CN110321422B (en) Method for training model on line, pushing method, device and equipment
US11036811B2 (en) Categorical data transformation and clustering for machine learning using data repository systems
US10430661B2 (en) Generating a compact video feature representation in a digital medium environment
WO2022041982A1 (en) Data recommendation method and apparatus, computer device, and storage medium
CN111783810B (en) Method and device for determining attribute information of user
US11907817B2 (en) System and methods for machine learning training data selection
CN112989146B (en) Method, apparatus, device, medium and program product for recommending resources to target user
CN113688310A (en) Content recommendation method, device, equipment and storage medium
CN113515690A (en) Training method of content recall model, content recall method, device and equipment
CN112182281B (en) Audio recommendation method, device and storage medium
US20190164083A1 (en) Categorical Data Transformation and Clustering for Machine Learning using Natural Language Processing
CN111858969B (en) Multimedia data recommendation method, device, computer equipment and storage medium
CN114817692A (en) Method, device and equipment for determining recommended object and computer storage medium
CN112308644A (en) Method and device for processing description information
CN116070027A (en) Online pushing topic analysis method based on artificial intelligence and big data system
CN116030375A (en) Video feature extraction and model training method, device, equipment and storage medium
CN110555131B (en) Content recommendation method, content recommendation device and electronic equipment
CN110555135B (en) Content recommendation method, content recommendation device and electronic equipment
CN113888216A (en) Advertisement information pushing method and device, electronic equipment and storage medium
CN112685516A (en) Multi-channel recall recommendation method and device, electronic equipment and medium
CN114866845B (en) Information detection method and system based on short video release
CN112905885B (en) Method, apparatus, device, medium and program product for recommending resources to user
CN110555157B (en) Content recommendation method, content recommendation device and electronic equipment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant