CN111143610A

CN111143610A - Content recommendation method and device, electronic equipment and storage medium

Info

Publication number: CN111143610A
Application number: CN201911397540.7A
Authority: CN
Inventors: 黄海兵
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2019-12-30
Filing date: 2019-12-30
Publication date: 2020-05-12
Anticipated expiration: 2039-12-30
Also published as: CN111143610B

Abstract

The application discloses a content recommendation method, a content recommendation device, electronic equipment and a storage medium; the method relates to natural language processing and video content understanding directions in the field of artificial intelligence, and can display a playing page of a target video; when the currently played video clip of the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in the playing area of the target video; displaying recommended content corresponding to the description information based on trigger operation aiming at the description information in the recommended content bullet screen; according to the method and the device, the recommended content bullet screen can be displayed in the playing area of the target video watched by the user, the recommended content corresponding to the recommended content bullet screen is associated with the video clip of the target video watched by the user, and therefore the recommended content is high in attraction to the user, the interest of the user can be increased, and the interaction degree of the user and the click rate of the recommended content can be improved.

Description

Content recommendation method and device, electronic equipment and storage medium

Technical Field

The present application relates to the field of computer technologies, and in particular, to a content recommendation method and apparatus, an electronic device, and a storage medium.

Background

With the development of computer technology, multimedia is more and more widely applied, and the number of videos is also sharply increased. In order to facilitate a user to quickly obtain a video to be watched from a large amount of videos, many video websites and video applications recommend videos which the user may be interested in on a playing page of the video watched by the user.

However, in the related art, the recommended content is generally displayed in the playing page of the video only in the non-full-screen playing mode. The recommended contents such as the recommended videos are generally located around the video playing area, and in the full-screen playing mode, the user needs to quit the full-screen playing mode of the watched videos to check the recommended videos, so that complicated operations are brought, user experience is reduced, the click rate of the recommended contents is low, and the interaction degree of the user is weak.

Disclosure of Invention

The embodiment of the application provides a content recommendation method, a content recommendation device, electronic equipment and a storage medium, which are beneficial to improving the click rate of a user on recommended content and increasing the interaction degree of the user.

The embodiment of the application provides a content recommendation method, which comprises the following steps:

displaying a playing page of a target video, wherein the playing page comprises a playing area of the target video;

when a video clip currently played by the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in a playing area of the target video, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip;

and displaying the recommended content based on the trigger operation aiming at the description information in the recommended content bullet screen.

Correspondingly, an embodiment of the present application provides a content recommendation device, including:

the display unit is used for displaying a playing page of a target video, and the playing page comprises a playing area of the target video;

the first display unit is used for displaying a recommended content bullet screen corresponding to a preset video clip in a playing area of the target video when the video clip currently played by the target video belongs to the preset video clip, wherein the recommended content bullet screen comprises description information of recommended content; the recommended content is associated with the content of the preset video clip;

and the second display unit is used for displaying the recommended content based on the trigger operation aiming at the description information in the bullet screen of the recommended content.

Optionally, in some embodiments of the present application, the first presentation unit may include a presentation subunit and an acquisition subunit.

Optionally, in some embodiments, the recommended content bullet screen includes a description information list of recommended content, where the description information list includes at least one piece of description information; the presentation subunit may be configured to present the description information list in a specific area within a playing area of the target video, where a correlation between video frames of the target video in the specific area is higher than a preset correlation.

Optionally, in some embodiments, the obtaining subunit may trigger to obtain a recommended content bullet screen corresponding to a preset video clip when a video clip currently played by the target video belongs to the preset video clip; and displaying the recommended content bullet screen in the playing area of the target video.

Optionally, in some embodiments, the obtaining subunit is specifically configured to send, when a video segment currently played by the target video belongs to a preset video segment, a recommended content obtaining request to a server, so as to trigger the server to determine, based on video content information of the preset video segment in multiple single modalities, recommended content associated with content of the preset video segment, and obtain a recommended content bullet screen including description information of the recommended content; and receiving the recommended content barrage sent by the server.

Optionally, in some embodiments of the present application, the obtaining subunit may further include a specific obtaining process of the recommended content, as follows:

the obtaining subunit may extract, when a video clip currently played by the target video belongs to a preset video clip, video content information of the preset video clip under a plurality of single modalities; acquiring content information of candidate content under at least one single mode; determining recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content; and obtaining the description information of the recommended content, and generating a recommended content bullet screen containing the description information.

Optionally, in some embodiments, the step of "extracting video content information of the preset video clip in multiple single modalities" may specifically include:

performing image extraction processing on the preset video clip to obtain an image sequence of the preset video clip, wherein the image sequence is video content information of the preset video clip in an image modality;

performing audio data extraction processing on the preset video clip to obtain a voice sequence of the preset video clip, wherein the voice sequence is video content information of the preset video clip in a voice mode;

and performing subtitle extraction processing on the preset video clip to obtain a text sequence of the preset video clip, wherein the text sequence is video content information of the preset video clip in a text mode.

Optionally, in some embodiments, the step "determining recommended content associated with the content of the preset video segment from the candidate content based on the video content information of the preset video segment and the content information of the candidate content" may include:

carrying out image recognition on the image sequence to obtain object description information of an image bearing object of the image sequence; performing voice recognition on the voice sequence, and translating the voice content in the voice sequence into corresponding text description information; obtaining identification information of the preset video clip based on the object description information, the character description information and the text sequence of the preset video clip; selecting candidate recommended content from the candidate content based on the identification information of the preset video segment, wherein the candidate recommended content is associated with the identification information of the preset video segment; extracting semantic description vectors from the identification information of the preset video segments through a content recommendation model; extracting semantic description vectors from the identification information of the candidate recommended content through a content recommendation model; calculating the similarity of semantic description vectors of the preset video clip and the candidate recommended content; and determining the candidate recommended content with the similarity higher than a preset similarity threshold as recommended content.

Optionally, in some embodiments of the present application, the first presentation unit of the content recommendation device may further include a training subunit, as follows:

the training subunit may be configured to train a content recommendation model. The method specifically comprises the following steps:

acquiring training data, wherein the training data comprises a sample video and recommended content corresponding to the sample video, the label information of the recommended content represents the expected similarity between the recommended content and the sample video, the recommended content clicked by a user in the recommended content is a positive sample, the label information is 1, the recommended content not clicked by the user is a negative sample, and the label information is 0; extracting semantic description vectors of the sample video and semantic description vectors of the recommended contents through a content recommendation model; calculating the actual similarity of the semantic description vector of the sample video and the semantic description vector of the recommended content; and adjusting parameters of the content recommendation model based on the actual similarity and the expected similarity corresponding to the positive sample in the recommended content and the actual similarity and the expected similarity corresponding to the negative sample in the recommended content.

Optionally, in some embodiments of the present application, the recommendation content includes a recommendation text; the description information of the recommended content comprises description information of a recommended text; the second presentation unit may be specifically configured to present the detailed text content of the recommended text based on a trigger operation for the description information in the recommended content bullet screen.

Optionally, in some embodiments of the present application, the recommended content is a recommended video; the second display unit may specifically play a target video clip of the recommended video based on a trigger operation for the description information in the recommended content bullet screen, where a degree of correlation between the target video clip and the preset video clip is higher than a preset degree of correlation.

Optionally, in some embodiments of the present application, the recommended content is a recommended video; the second presentation unit may include a display subunit and a play subunit, as follows:

and the display sub-unit is used for displaying a play sub-window of the recommended video on a play page of the target video based on the trigger operation aiming at the description information in the bullet screen of the recommended content.

And the playing subunit is configured to play the recommended video in the playing sub-window.

Optionally, in some embodiments of the present application, the second display unit may further include a closing subunit, as follows:

the closing subunit is configured to, when a closing operation of the playing sub-window for the recommended video is detected, display a playing page of the target video.

The electronic device provided by the embodiment of the application comprises a processor and a memory, wherein the memory stores a plurality of instructions, and the processor loads the instructions to execute the steps in the content recommendation method provided by the embodiment of the application.

In addition, a storage medium is further provided, on which a computer program is stored, where the computer program is executed by a processor to implement the steps in the content recommendation method provided in the embodiments of the present application.

The embodiment of the application provides a content recommendation method, a content recommendation device, electronic equipment and a storage medium, which can display a playing page of a target video, wherein the playing page comprises a playing area of the target video; when a video clip currently played by the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in a playing area of the target video, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip; displaying the recommended content based on a trigger operation aiming at the description information in the recommended content bullet screen; according to the method and the device, the recommended content bullet screen can be displayed in the playing area of the target video watched by the user, the recommended content corresponding to the recommended content bullet screen is associated with the video clip of the target video watched by the user, and therefore the recommended content is high in attraction to the user, the interest of the user can be increased, and the interaction degree of the user and the click rate of the recommended content can be improved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a scene schematic diagram of a content recommendation method provided in an embodiment of the present application;

FIG. 2 is a flowchart of a content recommendation method provided in an embodiment of the present application;

FIG. 3 is a schematic page display diagram of a content recommendation method provided in an embodiment of the present application;

fig. 4 is a playing page of a video in the related art;

FIG. 5a is a schematic diagram illustrating a recommended content bullet screen in an embodiment of the present application;

FIG. 5b is another illustration of the recommended content bullet screen in the embodiment of the present application;

FIG. 5c is another illustration of the recommended content bullet screen in the embodiment of the present application;

FIG. 5d is another illustration of the recommended content bullet screen in the embodiment of the present application;

FIG. 6 is another flowchart of a content recommendation method provided in an embodiment of the present application;

fig. 7a is a schematic diagram of a video content information extraction process in an embodiment of the present application;

FIG. 7b is a schematic diagram of a process of obtaining recommended content in an embodiment of the present application;

FIG. 8 is a schematic diagram of an interface for playing a recommended video in the embodiment of the present application;

FIG. 9 is another flowchart of a content recommendation method provided in an embodiment of the present application;

fig. 10a is a schematic structural diagram of a content recommendation device according to an embodiment of the present application;

fig. 10b is a schematic structural diagram of a content recommendation device according to an embodiment of the present application;

fig. 10c is a schematic structural diagram of a content recommendation device according to an embodiment of the present application;

fig. 10d is a schematic structural diagram of a content recommendation device according to an embodiment of the present application;

fig. 10e is a schematic structural diagram of a content recommendation device provided in an embodiment of the present application;

fig. 11 is a schematic structural diagram of an electronic device provided in an embodiment of the present application;

fig. 12 is an alternative structural diagram of the distributed system 100 applied to the blockchain system according to the embodiment of the present application;

fig. 13 is an alternative schematic diagram of a block structure provided in the embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The embodiment of the application provides a content recommendation method and device, electronic equipment and a storage medium. The content recommendation device may be specifically integrated in an electronic device, and the electronic device may be a terminal or a server.

It is understood that the content recommendation method of the present embodiment may be executed on the terminal, may also be executed on the server, and may also be executed by both the terminal and the server. The above examples should not be construed as limiting the present application.

Take the case that the terminal and the server execute the content recommendation method together.

As shown in fig. 1, a terminal 10 may be configured to display a play page of a target video, where the play page includes a play area of the target video; when a video clip currently played by the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in a playing area of the target video, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip; and displaying the recommended content based on the trigger operation aiming at the description information in the recommended content bullet screen. The terminal 10 may include a mobile phone, a television, a tablet Computer, a notebook Computer, a Personal Computer (PC), and the like.

For example, when a video clip currently played by the target video belongs to a preset video clip, a recommended content obtaining request may be sent to the server 11 to trigger the server 11 to determine recommended content associated with content of the preset video clip based on video content information of the preset video clip in a plurality of single modalities, and obtain a recommended content bullet screen including description information of the recommended content; and when the terminal 10 receives the recommended content bullet screen fed back by the server 11, displaying the recommended content bullet screen in the playing area of the target video.

The server 11 may be configured to: receiving a recommended content acquisition request sent by a terminal 10, and acquiring a recommended content bullet screen corresponding to a preset video clip based on video content information of the preset video clip in a plurality of single modes; and then sends the acquired bullet screen of the recommended content to the terminal 10. Wherein the server 11 may comprise a single server or a cluster of servers, etc.

The process of the server 11 obtaining the recommended content may specifically include: extracting video content information of the preset video clip under a plurality of single modes; acquiring content information of candidate content under at least one single mode; determining recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content; and obtaining the description information of the recommended content, and generating a recommended content bullet screen containing the description information.

The steps executed by the server 11 may be executed by the terminal 10.

In this embodiment, the recommended content corresponding to the target video may be pre-calculated by a server or other devices, the corresponding relationship between the recommended content barrage and the target video may be pre-stored in the terminal 10 or the server 11, and the terminal 10 may obtain the recommended content barrage from a local or server based on the corresponding relationship when detecting that the video clip currently played by the target video belongs to the preset video clip. And then, displaying the recommended content bullet screen in the playing area of the target video.

For example, the recommended content corresponding to the target video may be obtained by real-time calculation by the terminal 10, and the terminal 10 may extract video content information of a preset video clip in a plurality of single modalities when detecting that a video clip currently played by the target video belongs to the preset video clip; acquiring content information of candidate content under at least one single mode; and determining recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content.

The content recommendation method provided in the embodiment of the present application relates to Natural Language Processing (NLP) and video content Understanding (VideoContent Understanding) in the field of Artificial Intelligence (AI). According to the method and the device, the video content of the video watched by the user can be analyzed, so that the recommended content related to the video content is obtained, the accuracy of the content recommendation result can be improved, and the interaction degree of the user is increased.

Among them, Artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result. In other words, artificial intelligence is a comprehensive technique of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a manner similar to human intelligence. Artificial intelligence is the research of the design principle and the realization method of various intelligent machines, so that the machines have the functions of perception, reasoning and decision making. The artificial intelligence technology is a comprehensive subject and relates to the field of extensive technology, namely the technology of a hardware level and the technology of a software level. The artificial intelligence software technology mainly comprises a computer vision technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions.

Among them, Natural Language Processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that enable efficient communication between humans and computers using natural language. Natural language processing is a science integrating linguistics, computer science and mathematics. Therefore, the research in this field will involve natural language, i.e. the language that people use everyday, so it is closely related to the research of linguistics. Natural language processing techniques typically include text processing, semantic understanding, machine translation, robotic question and answer, knowledge mapping, and the like.

The video content Understanding (VideoContent Understanding) is that a video is analyzed into structured and machine-readable intention and word slot information through a series of AI algorithms, and research of the method influences aspects such as face recognition, motion recognition, object detection, media production, video recommendation and the like. The following are detailed below. It should be noted that the following description of the embodiments is not intended to limit the preferred order of the embodiments.

The embodiment will be described from the perspective of a content recommendation apparatus, which may be specifically integrated in an electronic device, which may be a terminal or the like.

The content recommendation method can be applied to various types of video players, and recommended content barrages are displayed in playing areas of target videos watched by users, and recommended content corresponding to the recommended content barrages is obtained by analyzing the video content of the target videos, so that the recommended content has higher attraction to the users, the interaction degree of the users and the click rate of the recommended content can be improved, and the video players can be third-party application programs and the like. It can be understood that the content recommendation method can also be applied to the case that a webpage plays a video, and can also be applied to a live video and the like.

As shown in fig. 2, a specific flow of the content recommendation method may be as follows:

101. and displaying a playing page of the target video, wherein the playing page comprises a playing area of the target video.

In this embodiment, the target video may be a video watched by the user, and may be various types of videos, such as a news video, a history video, or an art video, and the like.

The playing page of the target video can be a video playing webpage or a video playing page in a video application, and under a mode of displaying video content in a non-full screen mode, the playing page of the target video can be divided into a plurality of areas, and besides the playing area of the target video, other related areas can be included, such as a comment area, an advertisement area, a collection area, a recommended video area and the like. Currently, the recommended video area is usually placed outside the playing area of the target video, and as shown in fig. 4, the recommended video is presented in the form of a video card below the playing area of the target video or in the form of a video list on the right side of the playing area of the target video. When a user watches a target video in a full-screen mode, if the user wants to watch a recommended video, the user needs to exit the full-screen mode, so that certain influence is caused to the user experience. In this embodiment, the recommended content may be directly displayed on the play area of the target video, where the recommended content includes the recommended video, and specific reference may be made to the specific description in step 102.

102. When the currently played video clip of the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in a playing area of the target video, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip.

The recommended content may be a single-mode content, such as a pure text content, or a multi-mode content, such as a text-text combined content, an audio and a video, and the like. The recommended content can specifically be from a video platform, a content interaction platform, a news information and shopping platform, and the like.

The recommended content bullet screen can be understood as recommended content information displayed by similar bullet screen information, and optionally, the recommended content bullet screen may include description information of the recommended content, where the description information may include a title, an abstract, and the like of a recommended text or a recommended video. Specifically, the description information may include a pure title of the recommended content, a title carrying the link information, a combination of the title and the link, and the like; in another example, the description information may be a pure abstract of the recommended content, or an abstract carrying link information, or a combination of an abstract and a link, etc. Wherein the link can be a presentation page address of the recommended content. For example, when the recommended content is news, the description information of the recommended content may be a news abstract, and when a trigger operation for the news abstract in the bullet screen of the recommended content is detected, the detailed news content is displayed; for another example, when the recommended content is a drama, the description information of the recommended content may be description information of a key drama, and when a trigger operation for the description information in the bullet screen of the recommended content is detected, the drama content of the drama is presented.

For example, taking the recommended content as the recommended video as an example, referring to fig. 3, 301, a target video is being played in a playing page, when a video clip currently played by the target video belongs to a preset video clip, in a playing area of the playing page shown in 301, a recommended content bullet screen 3021 (referring to the playing page 302 of the target video) corresponding to the preset video clip is displayed, and description information of the recommended content bullet screen may include a video cover thumbnail, a video title, and a video duration of the recommended video a 1. In one example, the recommended content bullet screen may be presented in the form of a small window (3021 as shown in the play page 302 of the target video). In other embodiments, the recommended content bullet screen of the recommended video may include, but is not limited to, a video title, a video cover thumbnail, a video summary, a video type, a video duration, a video playing amount, a total number of video comments, a number of video praise, and the like of the recommended video. If the recommended content is a pure text content, a content combined with graphics and text, an audio, or the like, the description information of the bullet screen of the recommended content may be a title, a content information summary, or the like of the recommended content.

The recommended content bullet screen may be displayed in a scrolling manner in a playing area of the target video together with a conventional bullet screen of the target video, where the displayed area may be an 1/2 area, a 1/4 area, and the like of the playing area, which is not limited in this embodiment. The recommended content barrage may also be displayed in a certain fixed area of the playing area of the target video, where the fixed area may be a background area where the video content of the target video is relatively static, and in the fixed area, the recommended content barrage of the recommended content corresponding to the video content at different times may be displayed. The preset moving direction may be set according to an actual situation, for example, the preset moving direction of the recommended content bullet screen of the recommended content may be set from right to left according to a left-to-right reading habit of the user, which is not limited in this embodiment.

In addition, when the recommended content bullet screen moves out of the playing area of the target video, a corresponding part of the recommended content bullet screen moving out of the playing area of the target video is not displayed.

Optionally, the step of displaying the recommended content bullet screen corresponding to the preset video segment in the playing area of the target video, where the recommended content bullet screen includes description information of recommended content, and the recommended content is associated with the content of the preset video segment, may specifically include:

and displaying a recommended content bullet screen corresponding to the preset video clip in a preset display mode in the playing area of the target video, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip. The preset display mode may be a mode similar to the display mode of the bullet screen to display the recommended content bullet screen, and refer to a page in fig. 5a, where the page is a schematic diagram of displaying the recommended content bullet screen on a playing page of the target video. For example, the content information of the recommended content may be summarized into a line of text, that is, a recommended content bullet screen of the video content, and when it is detected that the video segment currently played by the target video belongs to a preset video segment, the recommended content bullet screen corresponding to the preset video segment moves from right to left on the playing area of the target video along with other commenting bullet screens, where the recommended content bullet screen may be specially identified to be different from the other commenting bullet screens. For example, the bullet screen font size corresponding to the recommended content bullet screen may be set to be larger than the standard bullet screen font size, or other font colors may be used for distinguishing, or a special symbol may be added in front of the recommended content bullet screen for distinguishing, such as a special symbol such as a flame and a pentagram.

The preset display mode may also be a mode of displaying the recommended content bullet screen in a small window, taking the recommended content as a video as an example. Referring to page 302 in fig. 3, the page is a schematic diagram of a bullet screen of recommended content for displaying a recommended video on a playing page of a target video. The widget 3021 may display a recommended content bullet screen of the recommended video, and the description information of the recommended content bullet screen may be at least one of a video cover thumbnail and a video title. For example, in some embodiments, a small window corresponding to the recommended content bullet screen of the recommended video may be displayed in a moving manner from right to left on the upper half area of the playing area of the target video. In other embodiments, the widget corresponding to the recommended content bullet screen of the recommended video may be displayed in a preset area on the playing area of the target video, where the preset area may be a video content background area of the target video, and in the preset area, the recommended content bullet screen of the recommended video corresponding to the video content at different times may be displayed.

It should be understood that the preset display mode of the present embodiment is not limited to the above-listed mode.

In some embodiments, the recommended content bullet screen comprises a description information list of the recommended content, wherein the description information list contains at least one piece of description information; the step of displaying the recommended content barrage corresponding to the preset video clip in the playing area of the target video may specifically include:

and displaying the description information list in a specific area in the playing area of the target video, wherein the correlation between video frames of the target video in the specific area is higher than a preset correlation.

The description information list comprises description information of a plurality of recommended contents, the correlation degree between video frames of the target video in the specific area is higher than a preset correlation degree, the video content of the target video in the specific area is a relatively static background area, and the change degree of the video content in the specific area in a preset time period is not higher than a preset change degree. The preset correlation degree may be set according to an actual situation, for example, the preset correlation degree may be set according to a requirement for a preset variation degree. In this embodiment, the magnitude of the correlation between the video frames of the video content may be determined based on the magnitude of the degree of change of the video content. When the change degree of the video content is smaller, the correlation degree between the video frames is higher; conversely, when the degree of change of the video content is larger, the correlation between the video frames is smaller.

Referring to fig. 5b, a playing page is shown, where the page is a schematic diagram of a description information list of recommended content displayed on a playing page of a target video. The description information list of the recommended content is a bullet screen capable of displaying the recommended content based on the video content of the target video at different moments so as to update the recommended content in the description information list in real time. In addition, the description information list can also comprise a pull-up list control and a pull-down list control, and other recommended content barrages can be presented by triggering the pull-up list control and the pull-down list control. The description information of the recommended content bullet screen can carry a link of the corresponding recommended content, the link is an address of a display page of the recommended content, and the recommended content corresponding to the display link can be triggered through clicking operation or sliding operation on the link.

Optionally, in some embodiments, there are a plurality of recommended contents corresponding to the preset video segment; the step of displaying a recommended content bullet screen corresponding to the preset video segment in the playing area of the target video, where the recommended content bullet screen includes description information of recommended content, and the recommended content is associated with the content of the preset video segment, may include:

and in the playing area of the target video, sequentially moving and displaying a recommended content bullet screen corresponding to the preset video clip in a preset display mode according to a preset time interval, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip.

The preset display mode may refer to the description in the above embodiments. For example, the preset display manner may be to display a plurality of recommended content barrages in the form of a plurality of small windows (small windows with thick black frames), such as the play page in fig. 5c, where the page is a schematic diagram of displaying a plurality of recommended content barrages on the play page of the target video. The widget may include description information such as a content abstract or a title of the recommended content, and may be displayed in a moving manner in the playing area of the target video or in a preset area in the playing area of the target video, where the preset area may be a background area of the video content of the target video. For another example, the preset display mode may also be a mode similar to the display mode of the barrage to display a plurality of recommended content barrages, refer to the play page in fig. 5d, where the page is a schematic diagram of displaying a plurality of recommended content barrages on the play page of the target video. The content information of each recommended content can be summarized into a segment of text, namely description information of the recommended content, when it is detected that the current playing segment of the target video belongs to a preset video segment, each recommended content bullet screen corresponding to the preset video segment moves on the playing area of the target video along with other commenting bullet screens, wherein the recommended content bullet screens can be specially marked to be different from other commenting bullet screens. For example, the bullet screen font size corresponding to the recommended content bullet screen may be set to be larger than the standard bullet screen font size, or other font colors may be used for distinguishing, or a special symbol may be added in front of the recommended content bullet screen for distinguishing, such as a special symbol such as a flame and a pentagram.

The preset time interval is a time interval between every two adjacent recommended contents, and may be set according to actual needs, for example, the preset time interval may be set according to a requirement that only one recommended content is displayed in a playing area of the target video, which is not limited in this embodiment.

In this embodiment, the step of displaying the recommended content bullet screen corresponding to the preset video clip in the playing area of the target video when the video clip currently played by the target video belongs to the preset video clip may include:

when the video clip currently played by the target video belongs to a preset video clip, triggering to acquire a recommended content bullet screen corresponding to the preset video clip;

and displaying the recommended content bullet screen in the playing area of the target video.

Among them, there are many ways for the terminal to acquire the recommended content.

For example, the recommended content may be obtained from a local database, for example, the recommended content calculated in advance may be stored in the local database, and when a video clip currently played by the target video belongs to a preset video clip, the recommended content may be obtained from the database.

For example, the recommended content may be acquired by another device and provided to the content recommendation apparatus, that is, the content recommendation apparatus may specifically receive the recommended content transmitted by the other device.

For a scene in which the device is a server, the step "triggering to acquire a recommended content bullet screen corresponding to a preset video clip when a video clip currently played by the target video belongs to the preset video clip" may include:

when the video clip currently played by the target video belongs to a preset video clip, sending a recommended content acquisition request to a server to trigger the server to determine recommended content associated with the content of the preset video clip based on video content information of the preset video clip under a plurality of single modes, and acquiring a recommended content bullet screen containing description information of the recommended content;

and receiving the recommended content barrage sent by the server.

For example, in some embodiments, the recommended content may be calculated in advance based on the video content of the target video, and then stored in the shared ledger of the blockchain, and when the recommended content needs to be acquired, the recommended content is extracted from the shared ledger of the blockchain and then provided to the terminal.

Optionally, in other embodiments, as shown in fig. 6, the recommended content corresponding to the preset video segment of the target video may also be determined through real-time calculation and analysis, and it is understood that the process of real-time calculation and analysis may be executed on a server or a terminal.

Optionally, reference may be made to the following description for a specific process of obtaining recommended content corresponding to a preset video segment of a target video.

In this embodiment, the step of triggering to acquire a recommended content bullet screen corresponding to a preset video clip when a video clip currently played by the target video belongs to the preset video clip may include:

when the video clip currently played by the target video belongs to a preset video clip, extracting video content information of the preset video clip under a plurality of single modes;

acquiring content information of candidate content under at least one single mode;

determining recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content;

and obtaining the description information of the recommended content, and generating a recommended content bullet screen containing the description information.

Before the step "the video segment currently played by the target video belongs to a preset video segment", the method may further include: the target video is divided into at least one video segment.

For example, the target video may be divided into several video segments at certain time intervals. For example, the target video may be divided by treating each sentence in the target video as a time interval. For another example, the target video may be divided into a certain number of video segments, and when the duration of the target video is longer, the length of the time segment of the divided video segment is also longer.

The candidate content may be content in a preset database. The content information of the candidate content may be stored in a database after being calculated in advance, or may be acquired through real-time calculation and analysis. If the content information of the candidate content is pre-calculated, the content information of the candidate content can be extracted from the database and provided to the content recommendation device when the content information of the candidate content needs to be acquired.

The video content information is content information contained in the video, and the video contains rich visual, auditory and subtitle information, so that the video has information of multiple modes, wherein the modes can be an image mode, a voice mode, a text mode and the like, and the video content information of the multiple modes of the video can be extracted based on analysis processing of the information in each mode.

Optionally, in some implementations, the step of "extracting video content information of the preset video clip in a plurality of single modalities" may specifically include:

Among them, the subtitles can be divided into soft subtitles and hard subtitles. Soft captions are separately stored caption files that can be extracted directly from the video data stream; the hard captions are captions embedded in the video, and the caption files and the video stream are pressed together, so that the caption files cannot be separated.

Optionally, in some embodiments, as shown in fig. 7a, the video data stream of the preset video segment includes a video stream, an audio stream, and a subtitle stream of the preset video segment. For the preset video segment, the corresponding subtitle is a soft subtitle, that is, a subtitle stream can be extracted from the video data stream. Before analyzing and extracting information from the video data stream, the video data stream needs to be decompressed, and the uncompressed video original data, audio original data and subtitle original data can be obtained by respectively decoding the video stream, the audio stream and the subtitle stream of the preset video segment. By decoding, the compression-encoded video stream is converted into uncompressed color video data, and the compression-encoded audio stream is converted into uncompressed audio sample data.

Optionally, in other embodiments, if the preset video segment does not compress the video data, the decoding process is not required.

Optionally, in some embodiments, if the subtitles corresponding to the preset video segment are hard subtitles, extracting the subtitles in the image sequence by performing character recognition on the image sequence obtained after processing the preset video segment; alternatively, the speech may be translated into the subtitles by performing speech recognition on a speech sequence obtained by processing a preset video segment.

Optionally, in some embodiments, for a preset video segment with subtitles, audio data of the preset video segment may not be processed, that is, the subtitle information is preferentially used; for a preset video segment without subtitles, voice recognition can be performed on audio data to obtain corresponding subtitle information.

In this embodiment, the candidate content may be a single-mode content, such as a pure text content, or a multi-mode content, such as a text-text combined content, an audio and a video, and the like. The candidate content may specifically originate from a video platform, a content interaction platform, a news information and shopping platform, etc.

In the step of "obtaining content information of the candidate content in at least one single mode", when the candidate content is a single-mode content, only the content information of the candidate content in one single mode can be obtained, for example, if the candidate content is a pure text content, only the content information of the pure text content in a text mode can be obtained. When the candidate content is a multi-modal content, content information of the candidate content in a single mode can be acquired, and content information of the candidate content in a plurality of single modes can also be acquired, for example, when the candidate content is a text-text combined content, content information of the text-text combined content in an image mode or a text mode can be acquired, and content information of the text-text combined content in the image mode and the text mode can also be acquired; for another example, when the candidate content is a candidate video, video content information of the candidate video in the text modality may be obtained, or even only a video title of the candidate video may be obtained, video content information of the candidate video in the image modality, the voice modality, and the text modality may also be obtained, and a process of extracting video content information of a plurality of single modalities of the candidate video may refer to the above-mentioned video content information extraction process of the preset video segment, which is not described herein again. It is to be understood that the above list is not to be construed as limiting the present application.

Optionally, in some embodiments, if the candidate content is a content of a single modality, the recommended content may be determined by using video content information of a preset video clip in the same modality. For example, if the candidate content is a pure text content, only the text mode is available, and only the video content information in the text mode of the preset video clip may be extracted to determine the recommended content.

Alternatively, in some embodiments, as shown in FIG. 7b, the recommended content may be determined using a candidate recommended content recall and candidate recommended content ranking method.

The step of determining recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content may include:

carrying out image recognition on the image sequence to obtain object description information of an image bearing object of the image sequence;

performing voice recognition on the voice sequence, and translating the voice content in the voice sequence into corresponding text description information;

obtaining identification information of the preset video clip based on the object description information, the character description information and the text sequence of the preset video clip;

selecting candidate recommended content from the candidate content based on the identification information of the preset video segment, wherein the candidate recommended content is associated with the identification information of the preset video segment;

extracting semantic description vectors from the identification information of the preset video segments through a content recommendation model;

extracting semantic description vectors from the identification information of the candidate recommended content through a content recommendation model;

calculating the similarity of semantic description vectors of the preset video clip and the candidate recommended content;

and determining the candidate recommended content with the similarity higher than a preset similarity threshold as recommended content.

In some embodiments, the step of "obtaining object description information of an image bearing object of the image sequence by performing image recognition on the image sequence" may specifically include: the image classification algorithm is used for classifying the images, then the image recognition is carried out, the characteristics of the bearing object in the image sequence are extracted, and the bearing object in the images can be recognized by carrying out characteristic matching on the images based on the preset data set. The image bearing object can comprise characters, places and the like, the preset data set can comprise famous person portraits in various fields and famous scenery spot ancient trails all over the world, and can also comprise some classical images (including cartoon character images) of movie and television works and the like, and the names and the place names of the characters in the image sequence can be identified based on the preset data set and the image identification process. For example, based on images of the west lake pre-stored in the preset data set, scenes of the west lake included in the image sequence are identified, and the geographic position where the image sequence is located, that is, the object description information of the image bearing object is "the west lake".

Optionally, in some embodiments, a keyword may be extracted from text information obtained by converting subtitles or audio data, where the keyword is a word having an identification function and is identification information of a preset video segment. For example, the caption "jia-a-really severe" is extracted as a keyword.

Among them, the candidate recommended content can be selected from the candidate contents using an open source search framework (ElasticSearch). The open source search framework is a search server which provides a distributed multi-user full-text search engine and can conveniently enable a large amount of data to have the capabilities of searching, analyzing and exploring. In this embodiment, an inverted index may be established through an open source search framework, and content in the candidate content is scored by using a Term Frequency-Inverse text Frequency (TF-IDF) function, so as to obtain candidate recommended content. For TF-IDF, the word frequency is the frequency of occurrence of each word in the text, namely the identification information; the inverse text frequency is used to reflect the importance of a word and thus modify the word vector represented by the word frequency only. The inverse text frequency may reflect the frequency of occurrence of a word in all texts, if a word occurs in many texts, its inverse text frequency has a lower value; conversely, if a word appears only twice in all text, its inverse text frequency has a higher value.

Specifically, the video content information (the identification information of the preset video clip) of the preset video clip in the multiple single modes can be used as input data of an open source search framework, then the open source search framework performs word segmentation on sentences in the video content information through a word segmentation controller, a TF-IDF function is used to obtain weights of the word segmentation, further ranking or scoring is performed on the content in the candidate content, the content with the ranking or score meeting the preset requirement is regarded as being associated with the video content information of the preset video clip in the multiple single modes, and the content is selected as the candidate recommended content.

In some embodiments, the step of extracting a semantic description vector from the identification information of the preset video segment through a content recommendation model may specifically include:

mapping identification information of a preset video segment into a semantic vector space through a content recommendation model to obtain a high-dimensional sparse vector of the identification information of the preset video segment; and then, performing dimensionality reduction on the high-dimensional sparse vector to obtain a low-dimensional semantic vector, namely a semantic description vector of the identification information of the preset video segment.

In some embodiments, the step of "extracting a semantic description vector for the identification information of the candidate recommended content through a content recommendation model" may include:

mapping identification information of the candidate recommended content into a semantic vector space through a content recommendation model to obtain a high-dimensional sparse vector of the identification information of the candidate recommended content; and then, performing dimensionality reduction on the high-dimensional sparse vector to obtain a low-dimensional semantic vector, namely a semantic description vector of the identification information of the candidate recommended content.

The identification information of the candidate recommended content may be content information in a single modality, or may be content information in multiple single modalities, which is not limited in this scheme.

In this embodiment, the step of "calculating the similarity between the semantic description vectors of the preset video segment and the candidate recommended content" may include:

and calculating the vector distance of the semantic description vectors of the preset video segment and the candidate recommended content, wherein the vector distance represents the similarity of the two semantic description vectors.

Determining recommended content by measuring a vector distance between semantic description vectors, wherein the vector distance can be a cosine distance, namely cosine similarity, and the cosine similarity is obtained by evaluating the similarity of two semantic description vectors by calculating a cosine value of an included angle between the two semantic description vectors; in information retrieval, the cosine similarity ranges from 0 to 1, and when the cosine similarity is closer to 1, the higher the similarity of two semantic description vectors is; when the cosine similarity value is closer to 0, the lower the similarity of the two semantic description vectors is, the two semantic description vectors are independent. It should be noted that the vector distance is not limited to the cosine distance.

The preset similarity threshold in this embodiment may be set according to actual situations, and this embodiment does not limit this. For example, when the similarity is measured by cosine similarity, the preset similarity threshold may be specifically set to 0.7, and when the cosine similarity between the semantic description vector of the preset video segment and the semantic description vector of the candidate recommended content is greater than 0.7, the candidate recommended content corresponding to the semantic description vector of the candidate recommended content may be used as the recommended content.

It should be noted that the content recommendation model may be specifically provided to the content recommendation apparatus after being trained by another device, or may be trained by the content recommendation apparatus itself.

If the content recommendation device performs training by itself, before the step "extracting semantic description vectors from the identification information of the preset video segment through a content recommendation model", the video recommendation method may further include:

acquiring training data, wherein the training data comprises a sample video and recommended content corresponding to the sample video, the label information of the recommended content represents the expected similarity between the recommended content and the sample video, the recommended content clicked by a user in the recommended content is a positive sample, the label information is 1, the recommended content not clicked by the user is a negative sample, and the label information is 0;

extracting semantic description vectors of the sample video and semantic description vectors of the recommended contents through a content recommendation model;

calculating the actual similarity of the semantic description vector of the sample video and the semantic description vector of the recommended content;

and adjusting parameters of the content recommendation model based on the actual similarity and the expected similarity corresponding to the positive sample in the recommended content and the actual similarity and the expected similarity corresponding to the negative sample in the recommended content.

The content recommendation Model may specifically adopt a Semantic Model (DSSM) based on a deep web, where the DSSM is a correlation calculation Model of a Query (Query) and a document (document), and may map a search keyword and document content in a web page into a Semantic space Model through a deep learning frame to calculate similarity between the search keyword and the document content in the web page, so as to rank the similarity of the document content in the web page and provide a search result. The data input by the DSSM input layer is a ternary data group which is a retrieval keyword, the clicked document content and the un-clicked document content are input after Query is input, and the model is trained through ternary data, so that the similarity between the clicked document content and the retrieval keyword output by the model approaches to 1, and the similarity between the un-clicked document content and the retrieval keyword approaches to 0. Query is a message sent by a search engine or a database to find a specific file, website, record or a series of records in the database.

In the present embodiment, the triple data input by the model is the video content information of the sample video, the content information of the recommended content clicked by the user (positive sample), and the content information of the recommended content not clicked by the user (negative sample). The model is trained on the triplet data. The training process comprises the steps of firstly calculating the actual similarity between semantic description vectors of a sample video and a positive sample and calculating the actual similarity between the semantic description vectors of the sample video and a negative sample, then adjusting parameters of a content recommendation model by using a back propagation algorithm, and optimizing the parameters of the content recommendation model based on the actual similarity and the expected similarity corresponding to the positive sample in the recommendation content and the actual similarity and the expected similarity corresponding to the negative sample in the recommendation content to make the actual similarity of the positive sample approach to the expected similarity of the positive sample; the actual similarity of the negative samples approaches to the expected similarity of the negative samples, and a trained content recommendation model is obtained. Specifically, the calculated actual similarity of the positive sample may be higher than the expected similarity of the positive sample, and the actual similarity of the negative sample may be lower than the expected similarity of the negative sample, where the values of the expected similarity of the positive sample and the expected similarity of the negative sample may be set according to actual conditions.

103. And displaying the recommended content based on the trigger operation aiming at the description information in the recommended content bullet screen.

The recommended content corresponding to the description information may be a text, a picture, an audio, a video, or the like. In this embodiment, when a trigger operation such as clicking or sliding on the description information in the recommended content bullet screen is detected, the recommended content corresponding to the description information may be displayed. And if the recommended content corresponding to the description information is the pure text content, displaying the pure text content when the trigger operation aiming at the description information is detected.

Optionally, in some embodiments, when a trigger operation for the description information in the recommended content bullet screen is detected, a jump selection window is displayed, where the jump selection window is an application program for a user to select whether to jump to a content provider of the recommended content corresponding to the description information.

Optionally, in some embodiments, the recommended content includes recommended text; the description information of the recommended content comprises description information of a recommended text; the step "displaying the recommended content based on the trigger operation for the description information in the recommended content bullet screen" may include:

and displaying the detailed text content of the recommended text based on the trigger operation aiming at the description information in the recommended content bullet screen.

The description information of the recommended text can be an abstract or a title of the recommended text, the description information in the bullet screen of the recommended content can carry a link, and when a trigger operation on the description information is detected, such as a click operation or a sliding operation, the detailed text content of the recommended text corresponding to the link is displayed.

The description is specifically given by taking the recommended content corresponding to the description information as a recommended video.

For example, referring to fig. 3, a recommended content bullet screen with a recommended video is displayed in a playing area of a target video shown in 302, when a trigger operation of a user on description information in the recommended content bullet screen is detected, a playing sub-window of the recommended video is displayed on a playing page shown in 302, the recommended video is played in the playing sub-window, and the playing page with the playing sub-window is displayed as shown in 303.

In some embodiments, the recommended content is a recommended video; the step "displaying the recommended content based on the trigger operation for the description information in the recommended content bullet screen" may include: and when the trigger confirmation operation aiming at the description information in the recommended content bullet screen is detected, playing the recommended video corresponding to the description information. When a trigger operation for the description information in the recommended content bullet screen is detected, the recommended video corresponding to the description information can be played from the starting time point of the video, as shown in the page 303 of fig. 3; the recommended video corresponding to the description information may also be played directly from a certain time point of the video, where the time point is a starting time point of a video segment whose degree of correlation with the preset video segment is higher than the preset degree of correlation, as shown in fig. 8, the recommended video is played from time point a. For a video with a longer video duration, the user can directly start watching from a video segment with a higher correlation degree with the preset video segment.

Specifically, in some embodiments, the recommended content is a recommended video; the step "displaying the recommended content based on the trigger operation for the description information in the recommended content bullet screen" may include:

and playing a target video clip of the recommended video based on the trigger operation aiming at the description information in the bullet screen of the recommended content, wherein the correlation degree of the target video clip and the preset video clip is higher than the preset correlation degree.

The step of playing the target video clip of the recommended video based on the trigger operation for the description information in the recommended content bullet screen may include:

and when the trigger confirmation operation aiming at the description information in the recommended content bullet screen is detected, playing the target video clip of the recommended video.

The preset correlation degree can be set according to actual conditions, and the method is not limited in the application. For example, the recommended video is divided into a plurality of video segments, and the setting is performed according to the degree of correlation between each video segment and a preset video segment. Or, the video clip with the highest degree of correlation with the preset video clip can be directly used as the target video clip; and when the trigger operation aiming at the description information in the recommended content bullet screen is detected, directly starting playing from the target video clip of the recommended video corresponding to the description information.

For example, the recommended video may be divided into a plurality of video segments, a degree of correlation between each video segment and a preset video segment is calculated, and when the degree of correlation between the video segment and the preset video segment is higher than the preset degree of correlation, the video segment is regarded as the target video segment. And when the trigger operation aiming at the description information in the recommended content bullet screen is detected, playing a target video clip of the recommended video corresponding to the description information. In some embodiments, after the last target video segment is played, the next target video segment may be skipped to play.

In this embodiment, the recommended content is a recommended video; the step "displaying the recommended content based on the trigger operation for the description information in the recommended content bullet screen" may specifically include:

displaying a play sub-window of the recommended video on a play page of the target video based on a trigger operation for the description information in the recommended content bullet screen;

and playing the recommended video in the playing sub-window.

The step of displaying a play sub-window of the recommended video on a play page of the target video based on the trigger operation for the description information in the recommended content bullet screen may include:

and when the triggering confirmation operation of the user for the description information in the recommended content bullet screen is detected, displaying a play sub-window of the recommended video on the play page of the target video.

The playing operation may specifically be a click operation or a sliding operation of a user on description information in a recommended content bullet screen, and when a trigger operation for the description information in the recommended content bullet screen is detected, video content of a recommended video corresponding to the description information is obtained, where the description information may carry a link of the recommended video.

In some embodiments, as shown in fig. 8, the play sub-window of the recommended video may be provided with a full-screen play control, and by triggering the full-screen play control, the recommended video may be displayed in a full screen.

In this embodiment, after the step "based on the trigger operation for the description information in the recommended content bullet screen, displaying the play sub-window of the recommended video on the play page of the target video", the method may further include:

and when the closing operation of the playing sub-window of the recommended video is detected, displaying the playing page of the target video.

And the playing page of the target video is the playing page of the target video before the playing sub-window is displayed.

In some embodiments, the recommended content may be a recommended text, where when a trigger operation for the description information in the recommended content bullet screen is detected, the recommended text may be displayed from the beginning of the text, or may be displayed directly from a certain segment of the text, where a degree of correlation between the segment and a preset video segment is higher than a first preset degree of correlation. For longer text, the user may view the text directly from the segment that is more relevant to the predetermined video segment.

Specifically, in some embodiments, the recommended content is a recommended text; the step "displaying the recommended content based on the trigger operation for the description information in the recommended content bullet screen" may include:

and displaying a target segment of the recommended text based on trigger operation aiming at the description information in the recommended content bullet screen, wherein the correlation degree of the target segment and the preset video segment is higher than a first preset correlation degree.

The first preset correlation degree may be set according to an actual situation, which is not limited in the present application. For example, the recommended text is divided into a plurality of segments, and the setting is performed according to the degree of correlation between each segment and a preset video segment. Or, the segment with the highest degree of correlation with the preset video segment can be directly used as the target segment; and when the trigger operation aiming at the description information in the recommended content bullet screen is detected, directly starting to display from the target segment of the recommended text.

For example, the recommended text may be divided into a plurality of segments, a degree of correlation between each segment and a preset video segment may be calculated, and when the degree of correlation between the segment and the preset video segment is higher than a first preset degree of correlation, the segment may be regarded as the target segment. And when the trigger operation aiming at the description information in the recommended content bullet screen is detected, displaying the target segment of the recommended text. In some embodiments, after the last target segment is displayed, the next target segment may be skipped to display.

In this embodiment, the recommended content is a recommended text; the step "displaying the recommended content based on the trigger operation for the description information in the recommended content bullet screen" may specifically include:

displaying a display sub-window of the recommended text on a playing page of the target video based on a trigger operation aiming at the description information in the recommended content bullet screen;

and displaying the recommended text in the display sub-window.

In this embodiment, after the step "based on the trigger operation for the description information in the recommended content bullet screen, displaying the display sub-window of the recommended text on the play page of the target video", the method may further include:

and when the closing operation of the display sub-window aiming at the recommended text is detected, displaying the playing page of the target video.

It can be understood that, when the recommended content corresponding to the description information is in other forms, such as audio or graphics, reference may be made to the embodiment where the recommended content corresponding to the description information is a recommended video or a recommended text, which is not described herein again.

As can be seen from the above, the present embodiment may display a play page of a target video, where the play page includes a play area of the target video; when a video clip currently played by the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in a playing area of the target video, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip; displaying the recommended content based on a trigger operation aiming at the description information in the recommended content bullet screen; according to the method and the device, the recommended content bullet screen can be displayed in the playing area of the target video watched by the user, the recommended content corresponding to the recommended content bullet screen is associated with the video clip of the target video watched by the user, and therefore the recommended content is high in attraction to the user, the interest of the user can be increased, and the interaction degree of the user and the click rate of the recommended content can be improved.

The method described in the foregoing embodiment will be described in further detail below by way of example with the content recommendation device being specifically integrated in a server.

An embodiment of the present application provides a content recommendation method, and as shown in fig. 9, a specific flow of the content recommendation method may be as follows:

901. the method comprises the steps that a server obtains a video to be processed, the video to be processed is divided into at least one video clip, and each video clip is used as a preset video clip.

The video to be processed is specifically a video whose recommended content is to be calculated.

The video to be processed is divided in various ways.

For example, the video to be processed may be divided into several video segments at certain time intervals. For example, each sentence in the video to be processed may be regarded as a time interval to divide the video to be processed.

For another example, the video to be processed may be divided into a certain number of video segments, and when the duration of the video to be processed is longer, the length of the time segment of the divided video segment is also longer.

902. And the server extracts the video content information of the preset video clip under a plurality of single modes.

The specific process of extracting the video content information of the preset video segment in multiple single modes may refer to the description of the embodiment in the foregoing 102, and details are not repeated here.

903. The server acquires content information of the candidate content in at least one single modality.

The candidate content may be content in a preset database. The content information of the candidate content may be pre-calculated and stored in a database.

It is to be understood that, the above embodiments may be referred to in the process of obtaining content information of candidate content in at least one single modality, and details are not described herein again.

904. The server determines recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content.

The step "the server determines recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content", may include:

Among them, the candidate recommended content can be selected from the candidate contents using an open source search framework (ElasticSearch). In this embodiment, an inverted index may be established through an open source search framework, and content in the candidate content is scored by using a term Frequency-Inverse text Frequency (TF-IDF) function, so as to obtain candidate recommended content. For TF-IDF, the word frequency is the frequency of occurrence of each word in the text, namely the identification information; the inverse text frequency is used to reflect the importance of a word and thus modify the word vector represented by the word frequency only. The inverse text frequency may reflect the frequency of occurrence of a word in all texts, if a word occurs in many texts, its inverse text frequency has a lower value; conversely, if a word appears only twice in all text, its inverse text frequency has a higher value.

Specifically, the video content information (the identification information of the preset video clip) of the preset video clip in the multiple single modes can be used as input data of an open source search framework, then the open source search framework performs word segmentation on sentences in the video content information through a word segmentation controller, obtains words and weights by using a TF-IDF function, further ranks or scores the content in the candidate content, regards the content with the ranking or score meeting the preset requirement as being associated with the video content information of the preset video clip in the multiple single modes, and selects the content as the candidate recommended content.

The recommended content is determined by measuring the vector distance between semantic description vectors, wherein the vector distance can be cosine distance, namely cosine similarity. It should be noted that the vector distance is not limited to the cosine distance.

In the present embodiment, the triple data input by the model is the video content information of the sample video, the content information of the clicked recommended content (positive sample), and the content information of the non-clicked recommended content (negative sample). The model is trained on the triplet data. The training process comprises the steps of firstly calculating the actual similarity between semantic description vectors of a sample video and a positive sample and calculating the actual similarity between the semantic description vectors of the sample video and a negative sample, then adjusting parameters of a content recommendation model by using a back propagation algorithm, and optimizing the parameters of the content recommendation model based on the actual similarity and the expected similarity corresponding to the positive sample in the recommendation content and the actual similarity and the expected similarity corresponding to the negative sample in the recommendation content to make the actual similarity of the positive sample approach to the expected similarity of the positive sample; the actual similarity of the negative samples approaches to the expected similarity of the negative samples, and a trained content recommendation model is obtained.

905. And the server stores the corresponding relation between the preset video clip and the recommended content.

In this embodiment, the preset video clip and the identification information of the recommended content corresponding to the preset video clip may be stored, where the identification information of the preset video clip and the identification information of the recommended content have a corresponding relationship, and the identification information may be a label or the like.

For example, a preset video segment and recommended content corresponding to the preset video segment may be labeled, for example, the preset video segment is labeled as a, and the recommended content corresponding to the preset video segment is labeled as a1, a2, A3, and the like. Through the correspondence of the labels, recommended content associated with the content of the preset video clip can be quickly determined.

Optionally, in this embodiment, the preset video clip and the description information of the recommended content corresponding to the preset video clip may also be correspondingly stored in the relationship table.

906. When the server receives a recommended content acquisition request sent by the terminal, the video clip currently played by the target video is determined based on the recommended content acquisition request.

Wherein the target video is a video being watched by the user.

907. The server determines recommended content corresponding to a video clip currently played by the target video.

For example, the recommended content associated with the content of the video clip currently played by the target video may be determined based on the correspondence between the video clip currently played by the target video and the identification information of the recommended content corresponding thereto.

Optionally, in some embodiments, the description information of the recommended content corresponding to the currently played video clip may be searched from a relationship table between a preset video clip and the recommended content corresponding to the preset video clip.

908. And the server sends the description information of the recommended content to the terminal so that the terminal generates a bullet screen of the recommended content according to the description information of the recommended content.

In an embodiment, the server may further generate a recommended content barrage based on the description information of the recommended content, and send the recommended content barrage to the terminal, so that the terminal displays the recommended content barrage corresponding to the preset video segment in the playing area of the target video.

As can be seen from the above, in this embodiment, a to-be-processed video may be obtained by a server, and the to-be-processed video is divided into at least one video segment, where each video segment is taken as a preset video segment; extracting video content information of the preset video clip under a plurality of single modes; acquiring content information of candidate content under at least one single mode; determining recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content; storing the corresponding relation between the preset video clip and the recommended content; when a server receives a recommended content acquisition request sent by a terminal, determining a video clip currently played by a target video based on the recommended content acquisition request; determining recommended content corresponding to a video clip currently played by a target video; and sending the description information of the recommended content to the terminal so that the terminal generates a recommended content bullet screen according to the description information of the recommended content. According to the method and the device, the recommended content bullet screen can be displayed in the playing area of the target video watched by the user, the recommended content corresponding to the recommended content bullet screen is associated with the video clip of the target video watched by the user, and therefore the recommended content is high in attraction to the user, the interest of the user can be increased, and the interaction degree of the user and the click rate of the recommended content can be improved.

In order to better implement the above method, an embodiment of the present application further provides a content recommendation device, as shown in fig. 10 a. The content recommendation apparatus may include a display unit 1001, a first presentation unit 1002, and a second presentation unit 1003 as follows:

(1) a display unit 1001;

a display unit 1001, configured to display a play page of a target video, where the play page includes a play area of the target video.

(2) A first display unit 1002;

the first display unit 1002 is configured to, when a video segment currently played by the target video belongs to a preset video segment, display a recommended content bullet screen corresponding to the preset video segment in a playing area of the target video, where the recommended content bullet screen includes description information of recommended content, and the recommended content is associated with content of the preset video segment.

In this embodiment, the first display unit 1002 can include a display subunit 10021 and an obtaining subunit 10022, see fig. 10 b.

Optionally, in some embodiments, the recommended content bullet screen includes a description information list of recommended content, where the description information list includes at least one piece of description information; the presentation subunit 10021 may be configured to present the description information list in a specific area within a playing area of the target video, where a correlation between video frames of the target video in the specific area is higher than a preset correlation.

Optionally, in some embodiments, the obtaining subunit 10022 may trigger obtaining of a recommended content bullet screen corresponding to a preset video clip when the video clip currently played by the target video belongs to the preset video clip; and displaying the recommended content bullet screen in the playing area of the target video.

Optionally, in some embodiments, the obtaining subunit 10022 is specifically configured to, when a video clip currently played by the target video belongs to a preset video clip, send a recommended content obtaining request to a server, so as to trigger the server to determine, based on video content information of the preset video clip in a plurality of single modalities, recommended content associated with content of the preset video clip, and obtain a recommended content bullet screen including description information of the recommended content; and receiving the recommended content barrage sent by the server.

Optionally, in some embodiments of the present application, the acquiring subunit 10022 may further include a specific acquiring process of the recommended content, which is as follows:

the obtaining subunit 10022 may extract video content information of a preset video segment in multiple single modalities when a video segment currently played by the target video belongs to the preset video segment; acquiring content information of candidate content under at least one single mode; determining recommended content associated with the content of the preset video clip from the candidate content based on the video content information of the preset video clip and the content information of the candidate content; and obtaining the description information of the recommended content, and generating a recommended content bullet screen containing the description information.

Optionally, in some embodiments, the step "determining recommended content associated with the content of the preset video segment from the candidate content based on the video content information of the preset video segment and the content information of the candidate content" may specifically include:

Optionally, in some embodiments of the present application, referring to fig. 10c, the first presentation unit 1002 of the content recommendation apparatus may further include a training subunit 10023, as follows:

the training subunit 10023 may be configured to train a content recommendation model. The method specifically comprises the following steps:

(3) A second presentation unit 1003;

the second presentation unit 1003 is configured to present, based on a trigger operation for the description information in the recommended content bullet screen, the recommended content corresponding to the description information.

Optionally, in some embodiments, the recommended content includes recommended text; the description information of the recommended content comprises description information of a recommended text; the second presenting unit 1003 may be specifically configured to present the detailed text content of the recommended text based on a trigger operation for the description information in the recommended content bullet screen.

Optionally, in some embodiments, the recommended content is a recommended video; the second displaying unit 1003 may be specifically configured to play a target video clip of the recommended video based on a trigger operation for the description information in the bullet screen of the recommended content, where a degree of correlation between the target video clip and the preset video clip is higher than a preset degree of correlation.

In the embodiment of the application, the recommended content is a recommended video; referring to fig. 10d, the second presentation unit 1003 may include a display subunit 10031 and a play subunit 10032, as follows:

the display subunit 10031 is configured to display a play sub-window of the recommended video on the play page of the target video based on the trigger operation for the description information in the recommended content bullet screen.

The playing subunit 10032 is configured to play the recommended video in the playing sub-window.

Optionally, in some embodiments of the present application, referring to fig. 10e, the second display unit 1003 may further include a closing subunit 10033, as follows:

the closing subunit 10033 is configured to, when a closing operation of the play sub-window for the recommended video is detected, display a play page of the target video.

As can be seen from the above, in this embodiment, the display unit 1001 may display a playing page of a target video, where the playing page includes a playing area of the target video; when a video clip currently played by the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in a playing area of the target video through a first display unit 1002, where the recommended content bullet screen includes description information of recommended content, and the recommended content is associated with content of the preset video clip; and displaying the recommended content through a second display unit 1003 based on a trigger operation for the description information in the recommended content bullet screen. According to the method and the device, the recommended content bullet screen can be displayed in the playing area of the target video watched by the user, the recommended content corresponding to the recommended content bullet screen is associated with the video clip of the target video watched by the user, and therefore the recommended content is high in attraction to the user, the interest of the user can be increased, and the interaction degree of the user and the click rate of the recommended content can be improved.

An electronic device according to an embodiment of the present application is further provided, as shown in fig. 11, which is a schematic structural diagram of the electronic device according to the embodiment of the present application, where the electronic device may be a terminal or a server, and the following description is given with reference to the electronic device as a terminal, as follows:

the terminal may include components such as a processor 1101 of one or more processing cores, memory 1102 of one or more computer-readable storage media, a power supply 1103, and an input unit 1104. Those skilled in the art will appreciate that the terminal structure shown in fig. 11 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.

Wherein:

the processor 1101 is a control center of the terminal, connects various parts of the entire terminal using various interfaces and lines, and performs various functions of the terminal and processes data by operating or executing software programs and/or modules stored in the memory 1102 and calling data stored in the memory 1102, thereby performing overall monitoring of the terminal. Optionally, processor 1101 may include one or more processing cores; preferably, the processor 1101 may integrate an application processor, which mainly handles operating systems, user interfaces, application programs, etc., and a modem processor, which mainly handles wireless communications. It will be appreciated that the modem processor described above may not be integrated into the processor 1101.

The memory 1102 may be used to store software programs and modules, and the processor 1101 executes various functional applications and data processing by operating the software programs and modules stored in the memory 1102. The memory 1102 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory 1102 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, the memory 1102 may also include a memory controller to provide the processor 1101 with access to the memory 1102.

The terminal further includes a power supply 1103 for supplying power to the various components, and preferably, the power supply 1103 is logically connected to the processor 1101 via a power management system, so that functions of managing charging, discharging, and power consumption are implemented via the power management system. The power supply 1103 may also include any component, such as one or more dc or ac power sources, recharging systems, power failure detection circuitry, power converters or inverters, power status indicators, and the like.

The terminal may further include an input unit 1104, and the input unit 1104 may be used to receive input numeric or character information and generate a keyboard, mouse, joystick, optical or trackball signal input in relation to user setting and function control.

Although not shown, the terminal may further include a display unit and the like, which will not be described in detail herein. Specifically, in this embodiment, the processor 1101 in the terminal loads the executable file corresponding to the process of one or more application programs into the memory 1102 according to the following instructions, and the processor 1101 runs the application programs stored in the memory 1102, thereby implementing various functions as follows:

displaying a playing page of a target video, wherein the playing page comprises a playing area of the target video; when a video clip currently played by the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in a playing area of the target video, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip; and displaying the recommended content based on the trigger operation aiming at the description information in the recommended content bullet screen.

The above operations can be implemented in the foregoing embodiments, and are not described in detail herein.

As can be seen from the above, the present embodiment may display a play page of a target video, where the play page includes a play area of the target video; when a video clip currently played by the target video belongs to a preset video clip, displaying a recommended content bullet screen corresponding to the preset video clip in a playing area of the target video, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip; and displaying the recommended content based on the trigger operation aiming at the description information in the recommended content bullet screen. According to the method and the device, the recommended content bullet screen can be displayed in the playing area of the target video watched by the user, the recommended content corresponding to the recommended content bullet screen is associated with the video clip of the target video watched by the user, and therefore the recommended content is high in attraction to the user, the interest of the user can be increased, and the interaction degree of the user and the click rate of the recommended content can be improved.

It will be understood by those skilled in the art that all or part of the steps of the methods of the above embodiments may be performed by instructions or by associated hardware controlled by the instructions, which may be stored in a computer readable storage medium and loaded and executed by a processor.

To this end, the present application provides a storage medium, in which a plurality of instructions are stored, where the instructions can be loaded by a processor to execute the steps in any one of the content recommendation methods provided in the present application. For example, the instructions may perform the steps of:

Wherein the storage medium may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like.

Since the instructions stored in the storage medium may execute the steps in any content recommendation method provided in the embodiments of the present application, beneficial effects that can be achieved by any content recommendation method provided in the embodiments of the present application may be achieved, which are detailed in the foregoing embodiments and will not be described herein again.

The system related to the embodiment of the application can be a distributed system formed by connecting a client, a plurality of nodes (any form of electronic equipment in an access network, such as a server and a terminal) through a network communication mode.

Taking a distributed system as an example of a blockchain system, referring To fig. 12, fig. 12 is an optional structural schematic diagram of the distributed system 100 applied To the blockchain system provided in this embodiment of the present application, and is formed by a plurality of nodes 200 (computing devices in any form in an access network, such as servers and user terminals) and a client 300, a Peer-To-Peer (P2P, Peer To Peer) network is formed between the nodes, and the P2P Protocol is an application layer Protocol operating on a Transmission Control Protocol (TCP). In a distributed system, any machine, such as a server or a terminal, can join to become a node, and the node comprises a hardware layer, a middle layer, an operating system layer and an application layer. In this embodiment, information such as video content information of the target video and recommended content corresponding to the preset video segment of the target video may be stored in a shared ledger of the area chain system through the node, and the electronic device (e.g., a terminal or a server) may obtain the video content information of the target video and the recommended content corresponding to the preset video segment of the target video based on record data stored in the shared ledger.

Referring to the functions of each node in the blockchain system shown in fig. 12, the functions involved include:

1) routing, a basic function that a node has, is used to support communication between nodes.

Besides the routing function, the node may also have the following functions:

2) the application is used for being deployed in a block chain, realizing specific services according to actual service requirements, recording data related to the realization functions to form recording data, carrying a digital signature in the recording data to represent a source of task data, and sending the recording data to other nodes in the block chain system, so that the other nodes add the recording data to a temporary block when the source and integrity of the recording data are verified successfully.

For example, the services implemented by the application include:

2.1) wallet, for providing the function of transaction of electronic money, including initiating transaction (i.e. sending the transaction record of current transaction to other nodes in the blockchain system, after the other nodes are successfully verified, storing the record data of transaction in the temporary blocks of the blockchain as the response of confirming the transaction is valid; of course, the wallet also supports the querying of the remaining electronic money in the electronic money address;

and 2.2) sharing the account book, wherein the shared account book is used for providing functions of operations such as storage, query and modification of account data, record data of the operations on the account data are sent to other nodes in the block chain system, and after the other nodes verify the validity, the record data are stored in a temporary block as a response for acknowledging that the account data are valid, and confirmation can be sent to the node initiating the operations.

2.3) Intelligent contracts, computerized agreements, which can enforce the terms of a contract, implemented by codes deployed on a shared ledger for execution when certain conditions are met, for completing automated transactions according to actual business requirement codes, such as querying the logistics status of goods purchased by a buyer, transferring the buyer's electronic money to the merchant's address after the buyer signs for the goods; of course, smart contracts are not limited to executing contracts for trading, but may also execute contracts that process received information.

3) And the Block chain comprises a series of blocks (blocks) which are mutually connected according to the generated chronological order, new blocks cannot be removed once being added into the Block chain, and recorded data submitted by nodes in the Block chain system are recorded in the blocks.

Referring to fig. 13, fig. 13 is an optional schematic diagram of a Block Structure (Block Structure) provided in this embodiment, where each Block includes a hash value of a transaction record stored in the Block (hash value of the Block) and a hash value of a previous Block, and the blocks are connected by the hash value to form a Block chain. The block may include information such as a time stamp at the time of block generation. A block chain (Blockchain), which is essentially a decentralized database, is a string of data blocks associated by using cryptography, and each data block contains related information for verifying the validity (anti-counterfeiting) of the information and generating a next block.

The content recommendation method, device, electronic device and storage medium provided by the embodiments of the present application are described in detail above, and a specific example is applied in the present application to explain the principle and the implementation of the present application, and the description of the above embodiments is only used to help understand the method and the core idea of the present application; meanwhile, for those 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

1. A content recommendation method, comprising:

2. The method according to claim 1, wherein the recommended content bullet screen comprises a description information list of recommended content, the description information list containing at least one of the description information; the displaying of the recommended content bullet screen corresponding to the preset video clip in the playing area of the target video includes:

3. The method of claim 1, wherein the recommended content comprises recommended text; the description information of the recommended content comprises description information of a recommended text;

the displaying the recommended content based on the trigger operation aiming at the description information in the recommended content bullet screen comprises the following steps:

4. The method of claim 1, wherein the recommended content is a recommended video; the displaying the recommended content based on the trigger operation aiming at the description information in the recommended content bullet screen comprises the following steps:

5. The method of claim 1, wherein the recommended content is a recommended video; the displaying the recommended content based on the trigger operation aiming at the description information in the recommended content bullet screen comprises the following steps:

and playing the recommended video in the playing sub-window.

6. The method according to claim 5, wherein the triggering operation based on the description information in the bullet screen of the recommended content further comprises, after displaying a play sub-window of the recommended video on the play page of the target video:

7. The method according to claim 1, wherein when the currently played video segment of the target video belongs to a preset video segment, displaying a recommended content bullet screen corresponding to the preset video segment in a playing area of the target video comprises:

8. The method according to claim 7, wherein when the currently played video segment of the target video belongs to a preset video segment, triggering to obtain a recommended content bullet screen corresponding to the preset video segment comprises:

and receiving the recommended content barrage sent by the server.

9. The method according to claim 7, wherein when the currently played video segment of the target video belongs to a preset video segment, triggering to obtain a recommended content bullet screen corresponding to the preset video segment comprises:

10. The method according to claim 9, wherein the extracting video content information of the preset video segment in a plurality of single modalities includes:

11. The method according to claim 10, wherein the determining recommended content associated with the content of the preset video segment from the candidate content based on the video content information of the preset video segment and the content information of the candidate content comprises:

12. The method according to claim 11, wherein before extracting the semantic description vector from the identification information of the preset video segment through the content recommendation model, the method further comprises:

13. A content recommendation apparatus characterized by comprising:

the first display unit is used for displaying a recommended content bullet screen corresponding to a preset video clip in a playing area of the target video when the video clip currently played by the target video belongs to the preset video clip, wherein the recommended content bullet screen comprises description information of recommended content, and the recommended content is associated with the content of the preset video clip;

14. An electronic device comprising a memory and a processor; the memory stores an application program, and the processor is configured to execute the application program in the memory to perform the steps of the content recommendation method according to any one of claims 1 to 12.

15. A storage medium storing instructions adapted to be loaded by a processor to perform the steps of the content recommendation method according to any one of claims 1 to 12.