CN115379265A

CN115379265A - Live broadcast behavior control method and device of virtual anchor

Info

Publication number: CN115379265A
Application number: CN202110540971.5A
Authority: CN
Inventors: 孙艳; 丁建栋
Original assignee: Alibaba Singapore Holdings Pte Ltd
Current assignee: Alibaba Innovation Co
Priority date: 2021-05-18
Filing date: 2021-05-18
Publication date: 2022-11-22
Anticipated expiration: 2041-05-18
Also published as: CN115379265B

Abstract

The embodiment of the disclosure discloses a live action control method and a device of a virtual anchor, wherein the live action control method of the virtual anchor comprises the following steps: acquiring behavior parameters of a watching object of a target live broadcast room where a virtual anchor is located, wherein the behavior parameters are obtained based on fusion of real-time behavior parameters and historical behavior parameters; determining a main broadcasting interaction behavior instruction according to the behavior parameters of the watching object; and controlling the behavior of the virtual anchor according to the anchor interaction behavior instruction. This technical scheme can realize the virtual anchor to the real-time perception of live room environment to can realize carrying out real-time decision-making or adjustment to the virtual anchor action according to the live room environment, thereby promote the live effect greatly, improve live quality, strengthen user's live experience, promote the development of live platform.

Description

Technical Field

The disclosure relates to the technical field of data processing, in particular to a live action control method and device of a virtual anchor.

Background

With the development of internet technology, live broadcast platforms are in operation, and as live broadcast is more visual and interactive, more and more users are willing to know the characteristics of commodities and buy the commodities through the live broadcast platform, so that the live broadcast quality and the live broadcast cost are concerned by the live broadcast platform. To reduce the cost of live broadcasts, many live broadcast platforms enable virtual anchor, where, a virtual anchor refers to an anchor that uses an avatar to conduct live activities on a live platform. In the prior art, action and behavior data of a virtual anchor are generally preset, so that the virtual anchor executes anchor activities according to the preset action and behavior data, and further, in order to make the action and behavior of the virtual anchor closer to a real anchor, real-time data of a live broadcast room is collected, so that the tone, expression and body actions of the virtual anchor are adjusted according to the real-time data of the live broadcast room. However, limited by the technology, the virtual anchor in the prior art still cannot flexibly express words according to real-time data of a live broadcast room like a live anchor, and timely adjust action and behavior strategies, so that the live broadcast quality is greatly reduced, the live broadcast experience of a user is damaged, and the development of a live broadcast platform is hindered.

Disclosure of Invention

The embodiment of the disclosure provides a live action control method and device of a virtual anchor.

In a first aspect, a live action control method of a virtual anchor is provided in an embodiment of the present disclosure.

Specifically, the live action control method of the virtual anchor includes: the method comprises the steps of obtaining behavior parameters of a watching object of a target live broadcast room where a virtual anchor is located, wherein the behavior parameters are obtained based on fusion of real-time behavior parameters and historical behavior parameters; determining a main broadcasting interaction behavior instruction according to the behavior parameters of the watching object; and controlling the behavior of the virtual anchor according to the anchor interaction behavior instruction.

In an implementation manner of the present disclosure, the obtaining of the behavior parameter of the viewing object in the target live broadcast room where the virtual anchor is located includes: collecting real-time behavior data of a watching object, and determining real-time behavior parameters according to the real-time behavior data; obtaining historical behavior data of a watching object, and determining historical behavior parameters according to the historical behavior data; and fusing the real-time behavior parameters and the historical behavior parameters to obtain the behavior parameters of the watching object.

In an implementation manner of the present disclosure, the acquiring real-time behavior data of a viewing object, and determining real-time behavior parameters according to the real-time behavior data includes: acquiring real-time behaviors of a watching object; collecting real-time behavior data of the watching object in response to the fact that the real-time behavior meets a preset trigger condition; counting real-time behavior indexes with different time lengths and content dimensions according to the real-time behavior data, and counting real-time behavior parameters based on the real-time behavior indexes in different time lengths, wherein the real-time behavior indexes comprise: the occurrence time point of the behavior of the viewing object, the occurrence frequency of the behavior of the viewing object and/or the occurrence frequency of the behavior of the viewing object, wherein the real-time behavior parameters include: <xnotran> , / . </xnotran>

In an implementation manner of the present disclosure, the obtaining historical behavior data of a viewing object, and determining a historical behavior parameter according to the historical behavior data includes: acquiring historical behavior data of a watching object in a first preset historical time period; according to the historical behavior data, historical behavior indexes of different time lengths and content dimensions are counted, and historical behavior parameters are counted and obtained on the basis of the historical behavior indexes in different historical time lengths, wherein the historical behavior indexes comprise: the time point of occurrence of the behavior of the viewing object, the frequency of occurrence of the behavior of the viewing object, and the frequency of occurrence of the behavior of the viewing object, wherein the historical behavior parameters include: content of interest of the viewing object in different historical durations and behavior of interest of the viewing object in different historical durations.

In an implementation manner of the present disclosure, the determining an anchor interaction behavior instruction according to the behavior parameter of the viewing object includes: determining whether the behavior parameters of the watching object meet preset behavior triggering conditions; and when the behavior parameters of the watching objects meet a preset behavior triggering condition, generating an anchor interaction behavior instruction based on the preset behavior.

In one implementation manner of the present disclosure, the preset behavior triggering condition includes: the number of the viewing objects in the first preset duration reaches a first number threshold, the occurrence frequency of the behavior of the viewing objects in the second preset duration reaches a second number threshold and/or a preset viewing object behavior event occurs at a preset time point; when the behavior parameters of the watching object meet a preset behavior triggering condition, generating an anchor interaction behavior instruction based on the preset behavior, wherein the method comprises the following steps: when the number of the viewing objects in the first preset duration reaches a first number threshold, generating a first anchor interaction behavior instruction which enables the virtual anchor to execute a first preset behavior; when the occurrence frequency of the behavior of the watching object in a second preset duration reaches a second quantity threshold, generating a second anchor interaction behavior instruction which enables the virtual anchor to execute a second preset behavior; and/or generating a third anchor interaction behavior instruction for enabling the virtual anchor to execute a third preset behavior when the preset viewing object behavior event occurs at the preset time point.

In an implementation manner of the present disclosure, the determining an anchor interaction behavior instruction according to the behavior parameter of the viewing object further includes: when the number of the preset behaviors of which the trigger conditions are met is more than or equal to two, determining the preset behaviors of which the trigger conditions are met as candidate preset behaviors; determining a target preset behavior from the candidate preset behaviors according to the behavior parameters of the watching object; and generating a anchor interaction behavior instruction based on the target preset behavior.

In an implementation manner of the present disclosure, the determining a target preset behavior from the candidate preset behaviors according to the behavior parameter of the viewing object includes: and taking the behavior parameters of the watching object as input, and inputting the behavior parameters into a behavior determination model obtained by pre-training to obtain a target preset behavior.

In an implementation manner of the present disclosure, the determining a target preset behavior from the candidate preset behaviors according to the behavior parameter of the viewing object further includes: training the behavior determination model.

In one implementation of the present disclosure, the training the behavior determination model includes: determining an initial behavior determination model; acquiring training behavior parameters of a watching object; and taking the training behavior parameters as input, taking preset behaviors which are correspondingly marked based on the training behavior parameters as output, and training the initial behavior determination model to obtain a behavior determination model.

In one implementation manner of the present disclosure, the method further includes: <xnotran> , , ; </xnotran> The behavior according to the viewing object the parameters determine the anchor interaction behavior instructions, is implemented as: and determining an anchor interaction behavior instruction according to the behavior parameters of the viewing object and the historical behavior data of the virtual anchor of the target live broadcast room.

In an implementation manner of the present disclosure, the determining an anchor interaction behavior instruction according to the behavior parameter of the viewing object and the historical behavior data of the virtual anchor in the target live broadcast room includes: determining whether the behavior parameters of the watching object meet a preset behavior triggering condition; when the number of the preset behaviors of which the trigger conditions are met is greater than or equal to two, determining the preset behaviors of which the trigger conditions are met as candidate preset behaviors; determining candidate preset behaviors matched with the historical behavior data of the virtual anchor of the target live broadcast room as target preset behaviors; <xnotran> . </xnotran>

The present disclosure provides an embodiment of a live action control device of a virtual anchor. Specifically, the live action control device of the virtual anchor includes: the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is configured to acquire behavior parameters of a watching object of a target live broadcast room where a virtual anchor is located, and the behavior parameters are obtained based on fusion of real-time behavior parameters and historical behavior parameters; the determining module is configured to determine a anchor interaction behavior instruction according to the behavior parameters of the viewing object; and the control module is configured to control the behavior of the virtual anchor according to the anchor interaction behavior instruction.

In one implementation of the present disclosure, the first obtaining module is configured to: collecting real-time behavior data of a watching object, and determining real-time behavior parameters according to the real-time behavior data; obtaining historical behavior data of a watching object, and determining historical behavior parameters according to the historical behavior data; and fusing the real-time behavior parameters and the historical behavior parameters to obtain the behavior parameters of the watching object.

In one implementation of the present disclosure, the collecting real-time behavior data of the viewing object, and determining the real-time behavior parameters according to the real-time behavior data are configured to: acquiring real-time behaviors of a watching object; collecting real-time behavior data of the watching object in response to the fact that the real-time behavior meets a preset trigger condition; counting real-time behavior indexes with different time lengths and content dimensions according to the real-time behavior data, and counting real-time behavior parameters based on the real-time behavior indexes in different time lengths, wherein the real-time behavior indexes comprise: the time point of occurrence of the behavior of the viewing object, the frequency of occurrence of the behavior of the viewing object and/or the frequency of occurrence of the behavior of the viewing object, wherein the real-time behavior parameters include: the behavior of the viewing object occurs frequently in different time durations, the behavior of the viewing object occurs frequently in different time durations to reach a preset threshold value, and/or the behavior of the viewing object occurs at different time points.

In one implementation of the present disclosure, the determining module is configured to: determining whether the behavior parameters of the watching object meet preset behavior triggering conditions; and when the behavior parameters of the watching objects meet a preset behavior trigger condition, generating an anchor interaction behavior instruction based on the preset behavior.

In one implementation of the present disclosure, the determining module is further configured to: when the number of the preset behaviors of which the trigger conditions are met is more than or equal to two, determining the preset behaviors of which the trigger conditions are met as candidate preset behaviors; determining a target preset behavior from the candidate preset behaviors according to the behavior parameters of the watching object; and generating a main broadcasting interaction behavior instruction based on the target preset behavior.

In one implementation manner of the present disclosure, the method further includes: the second acquisition module is configured to acquire historical behavior data of a virtual anchor in a target live broadcast room in a second preset historical time period, wherein the historical behavior data comprises behaviors of which the virtual anchor historical evaluation score is higher than a preset score threshold value and behavior objects of which the virtual anchor historical evaluation score is higher than the preset score threshold value; the determination module is configured to: and determining an anchor interaction behavior instruction according to the behavior parameters of the viewing object and the historical behavior data of the virtual anchor of the target live broadcast room.

The technical scheme provided by the embodiment of the disclosure can have the following beneficial effects: the technical scheme obtains the behavior parameters of the watching objects based on the live broadcast room data, and makes decisions or adjusts the virtual anchor behavior based on the behavior parameters of the watching objects. This technical scheme can realize the real-time perception of virtual anchor to the live room environment to can realize carrying out real-time decision-making or adjustment to virtual anchor action according to the live room environment, thereby promote the live effect greatly, improve live broadcast quality, strengthen user's live experience, promote the development of live platform.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

Other features, objects, and advantages of the present disclosure will become more apparent from the following detailed description of non-limiting embodiments when taken in conjunction with the accompanying drawings. In the drawings:

fig. 1 is a flowchart of a live action control method of a virtual anchor according to an embodiment of the present disclosure;

fig. 2 is an overall flowchart of a live action control method of a virtual anchor according to an embodiment of the present disclosure;

fig. 3 is a block diagram illustrating a configuration of a live action control device of a virtual anchor according to an embodiment of the present disclosure;

FIG. 4 is a block diagram of an electronic device according to an embodiment of the present disclosure;

fig. 5 is a schematic structural diagram of a computer system suitable for implementing a live behavior control method of a virtual anchor according to an embodiment of the present disclosure.

Detailed Description

Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement them. Furthermore, parts that are not relevant to the description of the exemplary embodiments have been omitted from the drawings for the sake of clarity.

In the present disclosure, it is to be understood that terms such as "including" or "having," etc., are intended to indicate the presence of the disclosed features, numerals, steps, actions, components, parts, or combinations thereof in the specification, and are not intended to preclude the possibility that one or more other features, numerals, steps, actions, components, parts, or combinations thereof are present or added.

It should also be noted that the embodiments and features of the embodiments in the present disclosure may be combined with each other without conflict. The present disclosure will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.

The technical scheme provided by the embodiment of the disclosure obtains the behavior parameters of the watching object based on the live broadcast room data, and makes a decision or adjusts the virtual anchor behavior based on the behavior parameters of the watching object. This technical scheme can realize the virtual anchor to the real-time perception of live room environment to can realize carrying out real-time decision-making or adjustment to the virtual anchor action according to the live room environment, thereby promote the live effect greatly, improve live quality, strengthen user's live experience, promote the development of live platform.

Figure 1 shows a flow diagram of a live behavior control method of a virtual anchor according to an embodiment of the present disclosure, as shown in fig. 1, the live action control method of the virtual anchor includes the following steps S101 to S103.

In step S101, behavior parameters of a viewing object in a target live broadcast room where the virtual anchor is located are obtained, where the behavior parameters are obtained based on fusion of real-time behavior parameters and historical behavior parameters.

In step S102, a anchor interaction behavior instruction is determined according to the behavior parameter of the viewing object.

In step S103, the behavior of the virtual anchor is controlled according to the anchor interaction behavior instruction.

As mentioned above, with the development of internet technology, live broadcast platforms are in operation, and live broadcast is more intuitive and interactive, more and more users are willing to know the characteristics of commodities and buy commodities through the live broadcast platform, so live broadcast quality and live broadcast cost become the problems of the live broadcast platform. To reduce live broadcast costs, many live broadcast platforms enable a virtual anchor, which refers to an anchor that uses avatars to conduct live broadcast activities on the live broadcast platform. In the prior art, action and behavior data of a virtual anchor are generally preset, so that the virtual anchor executes anchor activities according to the preset action and behavior data, and further, in order to make the action and behavior of the virtual anchor closer to a real anchor, real-time data of a live broadcast room is collected, so that the tone, expression and body actions of the virtual anchor are adjusted according to the real-time data of the live broadcast room. However, limited by the technology, the virtual anchor in the prior art still cannot flexibly express words according to real-time data of a live broadcast room like a live anchor, and timely adjust action and behavior strategies, so that the live broadcast quality is greatly reduced, the live broadcast experience of a user is damaged, and the development of a live broadcast platform is hindered.

In view of the above drawbacks, in this embodiment, a live broadcast behavior control method for a virtual anchor is provided, where the method obtains a behavior parameter of a viewing object based on live broadcast room data, and makes a decision or adjusts a virtual anchor behavior based on the behavior parameter of the viewing object. This technical scheme can realize the real-time perception of virtual anchor to the live room environment to can realize carrying out real-time decision-making or adjustment to virtual anchor action according to the live room environment, thereby promote the live effect greatly, improve live broadcast quality, strengthen user's live experience, promote the development of live platform.

In an embodiment of the present disclosure, the live behavior control method of a virtual anchor may be applied to a behavior controller of a computer, a computing device, an electronic device, a server, a service cluster, or the like, which performs behavior control on the virtual anchor.

In an embodiment of the present disclosure, a target live broadcast room refers to a live broadcast room in which a virtual anchor is located and which wants to control actions and behaviors of the virtual anchor.

In one embodiment of the present disclosure, a viewing object refers to an object for viewing virtual anchor broadcast content in a live broadcast room, such as a live broadcast room user, a live broadcast room viewer, a live broadcast room buyer, and the like.

In an embodiment of the present disclosure, the behavior parameter of the viewing object refers to an index or a parameter capable of representing a behavior tendency of the viewing object in a certain direct broadcasting room, wherein, in order to comprehensively reflect behavior characteristics of the viewing object, the behavior parameter of the viewing object is obtained based on a real-time behavior parameter and a historical behavior parameter of the viewing object, that is, the behavior parameter of the viewing object is obtained based on a fusion of the real-time behavior parameter and the historical behavior parameter, and the following will describe the behavior parameter of the viewing object in detail.

In an embodiment of the present disclosure, the anchor interactive behavior instruction refers to an instruction for driving a virtual anchor in a target live broadcast room to execute one or more interactive behaviors, where the anchor interactive behavior may be, for example, a behavior of calling a viewing object, answering a question of the viewing object, replying a comment of the viewing object, exciting the viewing object, extracting an excitation, issuing a red envelope, and the like.

In an embodiment of the present disclosure, when controlling a behavior of a virtual anchor according to an anchor interaction behavior instruction, the anchor interaction behavior instruction may be sent to a control device of the virtual anchor, so that the control device controls a behavior of the virtual anchor in a target live broadcast room according to the anchor interaction behavior instruction, or when a behavior controller is the control device of the virtual anchor, the behavior of the virtual anchor in the target live broadcast room may be directly controlled according to the anchor interaction behavior instruction.

In the above embodiment, the behavior parameters of the viewing object in the target live broadcast room where the virtual anchor is located are first obtained, then the anchor interaction behavior instruction for driving the anchor to execute a certain behavior or behaviors is determined according to the behavior parameters of the viewing object, and finally the behavior of the virtual anchor is controlled according to the anchor interaction behavior instruction.

In an embodiment of the present disclosure, in step S101, the step of obtaining the behavior parameters of the viewing object in the target live broadcast room where the virtual anchor is located may include the following steps: and collecting real-time behavior data of the watching object, and determining real-time behavior parameters according to the real-time behavior data. And acquiring historical behavior data of the watching object, and determining historical behavior parameters according to the historical behavior data. And fusing the real-time behavior parameters and the historical behavior parameters to obtain the behavior parameters of the watching object.

In order to make the behavior parameters of the viewing object more comprehensive and embody richer contents, so as to improve the effectiveness and accuracy of the anchor interaction behavior instruction, in the embodiment, the behavior parameters of the viewing object are determined not only based on the real-time behavior data of the viewing object, but also by integrating the historical behavior data of the viewing object.

Specifically, the method comprises the following steps: firstly, real-time behavior data of a watching object is collected, wherein the real-time behavior data can comprise real-time behavior content and real-time behavior occurrence time information, and the real-time behavior data can comprise one or more of the following data: the watching object enters the live broadcasting room, leaves the live broadcasting room and the like to watch data; viewing object interactive data such as comments made by viewing objects, praise of viewing objects, sharing live broadcast rooms by viewing objects and the like; the viewing object transaction data may include the viewing object clicking a link to a certain product, the viewing object purchasing a certain product, and so on. The real-time behavior data can effectively reflect the live broadcast state of the current live broadcast room, such as the number of people entering the live broadcast room at the current time, the number of people leaving the live broadcast room at the current time, the number of people watching live broadcast at the current live broadcast room, the number of peak watching people, the number of people interacting with the main broadcast, the content with the most interaction, the number of praise obtained in the current live broadcast room, the content with the most comment, and the like; real-time behavior parameters of the viewing object are then determined from the real-time behavior data.

Then, historical behavior data of the viewing object is obtained, and then historical behavior parameters of the viewing object are determined according to the historical behavior data, wherein the historical behavior data can comprise one or more of the following data: the viewing object may be a commodity or a commodity category frequently browsed within a preset history time period, a commodity or a commodity category collected by the viewing object within the preset history time period, a commodity or a commodity category purchased by the viewing object within the preset history time period, a commodity or a commodity category shared by the viewing object within the preset history time period, and the like. Historical behavior data can effectively reflect the live broadcast state of a virtual anchor of a live broadcast room in a historical time period, the data is equally important for description of behavior characteristics of a watching object and determination of an anchor interaction behavior instruction, for example, the preference condition of the watching object to commodities or commodity categories can be determined through the historical behavior data, the commodities or commodity categories in which the watching object is most interested can be determined based on analysis of the preference of all the watching objects to the commodities or the commodity categories in the whole live broadcast room, and then a data basis can be provided for selection of broadcast content of the virtual anchor and control of broadcast strength of the virtual anchor to certain commodities or commodity categories; historical behavior parameters of the viewing object are then determined from the historical behavior data.

And finally, fusing the real-time behavior parameters and the historical behavior parameters together to obtain the behavior parameters of the watching object of the target live broadcasting room. The real-time behavior parameters and the historical behaviors can be merged together to realize the fusion of the real-time behavior parameters and the historical behavior parameters, for example, the real-time behavior parameters and the historical behaviors can be spliced together in a JSON form.

In an embodiment of the present disclosure, the step of collecting real-time behavior data of the viewing object and determining real-time behavior parameters according to the real-time behavior data may include the following steps: acquiring real-time behaviors of a watching object; collecting real-time behavior data of the watching object in response to the fact that the real-time behavior meets a preset trigger condition; counting real-time behavior indexes with different time lengths and content dimensions according to the real-time behavior data, and counting to obtain real-time behavior parameters based on the real-time behavior indexes in different time lengths, wherein the real-time behavior indexes comprise: the viewing object behavior occurrence time point, the viewing object behavior occurrence frequency and/or the viewing object behavior occurrence frequency, wherein the real-time behavior parameters comprise: the behavior of the viewing object occurs frequently in different durations, the behavior of the viewing object occurs frequently in different durations to reach a preset threshold, and/or the behavior of the viewing object occurs at different time points.

In order to reflect the live broadcast state of the current live broadcast room in multiple angles, in the embodiment, after the real-time behavior data of the watching object of the target live broadcast room is obtained, the real-time behavior indexes with different time lengths and content dimensions are counted according to the real-time behavior data, and the real-time behavior parameters are counted and obtained based on the real-time behavior indexes in different time lengths.

Specifically, first, real-time behaviors of a viewing object, such as entering or leaving a live broadcast room, commenting, praising and the like, are acquired, and when the real-time behaviors meet a preset trigger condition, real-time behavior data of the viewing object is acquired, for example, when a certain viewing object enters the live broadcast room, the number of viewing objects in the live broadcast room reaches a certain number threshold, the behavior that the viewing object enters the live broadcast room is considered to meet the trigger condition on the number of viewing objects in the live broadcast room, and behavior data of the behavior that the viewing object enters the live broadcast room, such as the time that the viewing object enters the live broadcast room, information of the viewing object and the like, can be acquired; then, according to the real-time behavior data, counting real-time behavior indexes of different time lengths and content dimensions, wherein the real-time behavior indexes of different time lengths and content dimensions may include one or more of the following indexes: the time point of occurrence of the behavior of the viewing object, the number of times of occurrence of the behavior of the viewing object, and the frequency of occurrence of the behavior of the viewing object, wherein the time point of occurrence of the behavior of the viewing object may be, for example, a time point when the viewing object enters a live broadcast room, a time point when a comment is made, and the like; the number of times of behavior occurrence of the viewing object may be, for example, the number of viewing objects entering the live broadcast room within a preset time length, the number of times that a comment is made by a certain viewing object or all viewing objects in the live broadcast room, the number of times that a comment is complied with by a certain viewing object or all viewing objects in the live broadcast room, or the like; the frequency of the viewing object behavior may be, for example, the number of viewing objects entering the live broadcast room per unit time, the number of comments made on a certain viewing object or all viewing objects in the live broadcast room, the number of praise made on a certain viewing object or all viewing objects in the live broadcast room, and the like. Then, in different time lengths, real-time behavior parameters are obtained based on the real-time behavior index statistics, wherein the real-time behavior parameters may include one or more of the following parameters: <xnotran> , , , 5 , 1 , 1 , 10 , , , 1 , 1 , 5 . </xnotran> The statistics of the real-time behavior parameters depends on different statistics durations and different statistics contents, and the statistics durations and the statistics contents can be determined according to different live broadcast scenes or different statistics purposes in the same live broadcast scene.

In an embodiment of the present disclosure, the current statistical parameters of the live broadcast room, the rolling time window statistical parameters of the live broadcast room, the sliding time window statistical parameters of the live broadcast room, the cumulative statistical parameters of the live broadcast room, and other different statistical durations and real-time behavior parameters of different contents may be obtained through setting time windows such as a rolling time window and a sliding time window.

In an embodiment of the present disclosure, the step of obtaining historical behavior data of the viewing object and determining the historical behavior parameters according to the historical behavior data may include the following steps: acquiring historical behavior data of a watching object in a first preset historical time period; according to historical behavior data, historical behavior indexes of different time lengths and content dimensions are counted, and historical behavior parameters are counted and obtained on the basis of the historical behavior indexes in different historical time lengths, wherein the historical behavior indexes comprise: the time point of occurrence of the behavior of the viewing object, the frequency of occurrence of the behavior of the viewing object, and the historical behavior parameters include: content of interest of the viewing object in different historical durations and behavior of interest of the viewing object in different historical durations.

In order to reflect the favorite conditions of the watching objects to the commodities or the commodity categories in multiple aspects, in the embodiment, after historical behavior data of the watching objects in a preset historical time period of the target live broadcast room are obtained, historical behavior indexes with different time lengths and content dimensions can be obtained through statistics according to the historical behavior data. The historical behavior indexes of different time lengths and content dimensions can comprise one or more of the following indexes: a viewing object behavior occurrence time point, a viewing object behavior occurrence frequency, and a viewing object behavior occurrence frequency, where the viewing object behavior occurrence time point may be, for example, a time when the viewing object browses a commodity or a commodity category, a time when the viewing object purchases a commodity or a commodity category, or the like; the number of times of viewing object behavior occurrence may be, for example, the number of times that the viewing object browses the goods or the goods category in different durations, the number of times that the viewing object purchases the goods or the goods category in different durations, or the like; the frequency of viewing object behaviors may be, for example, the number of times the viewing object browses a commodity or a commodity category per unit time, the number of times the viewing object purchases a commodity or a commodity category per unit time, or the like. Then, in different historical time lengths, historical behavior parameters are obtained through statistics based on historical behavior indexes, wherein the real-time behavior parameters comprise contents of interest of the watching objects in different historical time lengths and behaviors of interest of the watching objects in different historical time lengths, for example, commodities or commodity categories frequently browsed by the watching objects in the last 7 days, commodities or commodity categories bought by the watching objects in the last 1 month, commodities or commodity categories bought by the watching objects in the last 5 days and the like can be obtained through statistics based on the historical behavior indexes, and the contents of interest of the watching objects in different historical time lengths and the behaviors of interest of the watching objects in different historical time lengths can be determined based on the data. Similar to the statistics of the real-time behavior parameters, the statistics of the historical behavior parameters depend on different statistics durations and different statistics contents, and the statistics durations and the statistics contents can be determined according to different live broadcast scenes or different statistics purposes in the same live broadcast scene.

In an embodiment of the present disclosure, the step S102 of determining the anchor interaction behavior instruction according to the behavior parameter of the viewing object may include the following steps: determining whether the behavior parameters of the watching object meet preset behavior triggering conditions or not; and when the behavior parameters of the watching object meet the preset behavior triggering conditions, generating a main broadcasting interaction behavior instruction based on the preset behavior.

In this embodiment, in determining the anchor interaction behavior instruction: firstly, determining whether a behavior parameter of a viewing object meets a preset behavior triggering condition, where the preset behavior may be, for example: placing a call with the viewing object, answering a question of the viewing object, replying to a comment of the viewing object, exciting the viewing object, extracting an excitation, issuing a red envelope, and the like; the preset behavior triggering condition may be, for example: the number of the watching objects entering the live broadcast room in the last 5 seconds reaches a first number threshold value, and then the watching objects are stimulated; answering the question of the viewing object when the number of times a question has been asked within the last 1 minute reaches a second number threshold and the question having the largest number of times has been asked within the last 1 minute is determined; the number of times of posting a certain comment in the last 1 minute reaches a third quantity threshold value, and the comment with the largest posting number in the last 1 minute is determined, and the comment of the viewing object is replied; determining that the latest incoming viewing object information is in call with the viewing object within the last 5 minutes; extracting the incentive when the number of praise obtained in the live broadcast room in the last 10 minutes reaches a fourth quantity threshold value; and issuing red packages and the like when the number of praise obtained in the live broadcast room from the broadcast reaches a fifth number threshold and/or the accumulated number of praise obtained in the live broadcast room from the broadcast reaches a sixth number threshold.

If the behavior parameters of the viewing object are determined the preset action triggering condition is met, generating a corresponding anchor interaction behavior instruction based on the preset behavior so that the anchor executes the preset behavior. For example, if it is determined that the number of viewing objects entering the live broadcast room within the last 5 seconds reaches the first number threshold, it is considered that the behavior parameter of the viewing object satisfies the trigger condition for stimulating the preset behavior of the viewing object, and an anchor interaction behavior instruction may be generated based on the preset behavior, so that the virtual anchor executes the behavior for stimulating the viewing object; for example, if it is determined that the number of praise obtained in the live broadcast room in the last 10 minutes reaches the fourth number threshold, it is considered that the behavior parameter of the viewing object meets the trigger condition of the preset behavior of the extraction incentive, and an anchor interaction behavior instruction may be generated based on the preset behavior, so that the virtual anchor executes the behavior of the extraction incentive; for another example, if it is determined that the number of praise obtained in the live broadcast room since the current broadcast reaches the fifth number threshold, it is determined that the behavior parameter of the viewing object meets the trigger condition of the preset behavior of issuing the red packet, and the anchor interaction behavior instruction may be generated based on the preset behavior, so that the virtual anchor executes the behavior of issuing the red packet, and the like.

That is, in an embodiment of the present disclosure, the preset behavior triggering condition may include one or more of the following conditions: the number of the viewing objects in the first preset duration reaches a first number threshold, the occurrence frequency of the behavior of the viewing objects in the second preset duration reaches a second number threshold, and/or a preset viewing object behavior event occurs at a preset time point.

In this embodiment, when the behavior parameter of the viewing object satisfies the preset behavior trigger condition, the step of generating the anchor interaction behavior instruction based on the preset behavior may include the following steps: and when the number of the viewing objects in the first preset duration reaches a first number threshold, generating a first anchor interaction behavior instruction which enables the virtual anchor to execute a first preset behavior. And when the occurrence frequency of the behavior of the watching object in the second preset duration reaches a second quantity threshold, generating a second anchor interaction behavior instruction which enables the virtual anchor to execute a second preset behavior. And/or generating a third anchor interaction behavior instruction which enables the virtual anchor to execute a third preset behavior when a preset viewing object behavior event occurs at a preset time point.

In an embodiment of the present disclosure, step S102, namely, the step of determining the anchor interaction behavior instruction according to the behavior parameter of the viewing object, may further include the following steps: and when the number of the preset behaviors with the satisfied trigger conditions is more than or equal to two, determining the preset behaviors with the satisfied trigger conditions as candidate preset behaviors. And determining a target preset behavior from the candidate preset behaviors according to the behavior parameters of the viewing object. And generating a main broadcasting interaction behavior instruction based on the target preset behavior.

As mentioned above, the preset behaviors may be one or more, and the trigger conditions of the corresponding preset behaviors may also be one or more, so that in the process of determining the trigger conditions of the preset behaviors, a situation that the number of the preset behaviors whose trigger conditions are met is greater than or equal to two may occur, and at this time, two or more preset behaviors need to be selected, so as to effectively improve the live broadcast effect and live broadcast quality. In this embodiment, first, a preset behavior in which the trigger condition is satisfied is determined as a candidate preset behavior; then determining the most suitable target preset behavior from the candidate preset behaviors according to the behavior parameters of the watching object; and finally, generating a main broadcasting interaction behavior instruction based on the selected target preset behavior so as to control the virtual main broadcasting to execute the target preset behavior.

In an embodiment of the present disclosure, the step of determining the target preset behavior from the candidate preset behaviors according to the behavior parameter of the viewing object may include the steps of: and taking the behavior parameters of the watching object as input, and inputting the behavior parameters into a behavior determination model obtained by pre-training to obtain the target preset behavior.

In this embodiment, a target preset behavior is determined from the candidate preset behaviors by using a machine learning method, that is, a behavior parameter of the viewing object is used as an input and is input into the behavior determination model, so that the target preset behavior can be obtained. The behavior determining model is a pre-trained model, the input of the model is the behavior parameters of the viewing object, and the output of the model is the target preset behavior.

In an embodiment of the present disclosure, the step of determining the target preset behavior from the candidate preset behaviors according to the behavior parameter of the viewing object may further include the following steps: a behavior determination model is trained.

In this embodiment, in training the behavior determination model, the behavior determination model may be trained using the training behavior parameters of the viewing object as input, the labeled target preset behavior as output, the finally obtained behavior determination model with algorithm convergence is a model which can be used for judging the behavior parameters of a follow-up viewing object. The selection of the initial form of the behavior determination model and the training method of the behavior determination model are common technical means in the art, and the disclosure does not specifically limit the initial form and the training method of the behavior determination model.

In one embodiment of the present disclosure, the first and second electrodes are, the step of training the behaviour determination model may comprise the steps of: an initial behavior determination model is determined. And acquiring training behavior parameters of the watching object. And training the initial behavior determination model by taking the training behavior parameters as input and taking the preset behaviors which are correspondingly marked based on the training behavior parameters as output to obtain the behavior determination model.

When the behavior determination model is trained, firstly, an initial behavior determination model is determined, wherein the initial behavior determination model can be selected according to the requirements of practical application; then, acquiring a training behavior parameter of the viewing object, wherein the training behavior parameter can be a historical behavior parameter of the viewing object; and then training the initial behavior determination model by taking the training behavior parameters as input and the marked preset behaviors corresponding to the training behavior parameters as output, and obtaining the behavior determination model when the training result is converged.

In an embodiment of the present disclosure, the method may further include the steps of: and acquiring historical behavior data of the virtual anchor of the target live broadcasting room in a second preset historical time period, wherein the historical behavior data comprise behaviors of which the historical evaluation scores of the virtual anchor are higher than a preset score threshold value and behavior objects of which the historical evaluation scores of the virtual anchor are higher than the preset score threshold value.

The step of determining the anchor interaction behavior instruction according to the behavior parameter of the viewing object may be implemented as: and determining an anchor interaction behavior instruction according to the behavior parameters of the viewing object and the historical behavior data of the virtual anchor of the target live broadcast room.

In view of the fact that the broadcast characteristics of the virtual anchor also have a certain influence on the determination of the anchor interaction behavior commands, in this embodiment, the broadcast characteristics of the virtual anchor are also referred to when determining the anchor interaction behavior commands. That is, when acquiring real-time behavior data and historical behavior data of a viewing object of a target live broadcast room, historical behavior data of a virtual anchor of the target live broadcast room in a second preset historical time period is also acquired, where the historical behavior data includes a behavior of the virtual anchor with a historical evaluation score higher than a preset score threshold and a behavior object of the virtual anchor with a historical evaluation score higher than the preset score threshold, and the behavior of the virtual anchor with a historical evaluation score higher than the preset score threshold may be, for example: the method comprises the following steps of performing incentive behavior with the maximum use times of a virtual anchor or the most popular with a viewing object, calling mode with the maximum use times of the virtual anchor or the most popular with the viewing object, interaction behavior with the maximum use times of the virtual anchor or the most popular with the viewing object, and the like; the action objects with the virtual anchor historical evaluation score higher than the preset score threshold value may be, for example: and the virtual anchor broadcasts the commodities or commodity categories which have the most times or are most popular with the watching objects, and the like.

In this embodiment, the step S102 of determining the anchor interaction behavior command according to the behavior parameter of the viewing object may be implemented as: behavior parameters according to viewing object and target live broadcast room virtualization the anchor's historical behavior data determines anchor interaction behavior instructions, the anchor interaction behavior instruction is comprehensively determined according to the behavior parameters of the viewing object and the historical behavior data of the virtual anchor of the target live broadcasting room.

In an embodiment of the present disclosure, the step of determining the anchor interaction behavior instruction according to the behavior parameter of the viewing object and the historical behavior data of the virtual anchor in the target live broadcast room may include the following steps: and determining whether the behavior parameters of the watching objects meet preset behavior triggering conditions. And when the number of the preset behaviors of which the trigger conditions are met is greater than or equal to two, determining the preset behaviors of which the trigger conditions are met as candidate preset behaviors. And determining candidate preset behaviors matched with the historical behavior data of the virtual anchor of the target live broadcast room as target preset behaviors. And generating a anchor interaction behavior instruction based on the target preset behavior.

As mentioned above, when determining the anchor interaction behavior command by considering only the behavior parameter of the viewing object, it is first determined whether the behavior parameter of the viewing object satisfies the preset behavior trigger condition, and if the number of the preset behaviors whose trigger conditions are satisfied is greater than or equal to two, and determining the preset behaviors with the trigger conditions met as candidate preset behaviors, then determining target preset behaviors from the candidate preset behaviors according to the behavior parameters of the viewing object, and generating a anchor interaction behavior instruction based on the target preset behaviors. In this embodiment, the anchor interaction behavior instruction is determined by integrating the behavior parameters of the viewing object and the historical behavior data of the virtual anchor in the target live broadcast room, specifically: similar to the case of determining the anchor interaction behavior command only by considering the behavior parameters of the viewing object, it is first determined whether the behavior parameters of the viewing object satisfy the preset behavior trigger condition, and if the number of the preset behaviors whose trigger conditions are satisfied is greater than or equal to two, all the preset behaviors whose trigger conditions are satisfied are determined as candidate preset behaviors. If the target preset behavior is determined to be the target preset behavior, the target preset behavior is determined to be the target preset behavior according to the historical behavior data of the virtual anchor in the target live broadcast room, for example, if the two candidate preset behaviors are: the incentives are extracted and the red packet is issued, and the preset behavior most popular with the watching objects can be determined as the target preset behavior according to the historical behavior data of the virtual anchor of the target live broadcast room; and finally, generating a main broadcasting interaction behavior instruction based on the target preset behavior.

Fig. 2 shows an overall flowchart of a live action control method of a virtual anchor according to an embodiment of the present disclosure, and as shown in fig. 2, first, real-time action data of a viewing object in a target live broadcast room where the virtual anchor is located is obtained, for example: the method comprises the following steps that a watching object enters a live broadcast room, the watching object leaves the live broadcast room and other watching object watching data, the watching object gives comments, the watching object likes, the watching object shares the live broadcast room and other watching object interaction data, the watching object clicks a link of a certain commodity, the watching object purchases a certain commodity and other watching object transaction data, and the like, and real-time behavior indexes of different durations and content dimensions are determined according to the real-time behavior data, such as: the method comprises the steps that current statistical indexes of a live broadcast room, rolling time window statistical indexes of the live broadcast room, sliding time window statistical indexes of the live broadcast room, accumulated statistical indexes of the live broadcast room and the like are obtained in different durations on the basis of real-time behavior index statistics; and then acquiring historical behavior data of the target live broadcast watching object in a first preset historical time period, such as: the method includes the steps that commodities or commodity categories which are frequently browsed by a watching object in a preset historical time period, commodities or commodity categories which are collected by the watching object in the preset historical time period, commodities or commodity categories which are purchased by the watching object in the preset historical time period and the like are determined, and historical behavior indexes of different time lengths and content dimensions are determined according to historical behavior data, such as: the preference of the watching object to the commodity or the commodity category and the like, and historical behavior parameters are obtained through statistics based on historical behavior indexes in different historical durations; the real-time behavior parameters and the historical behavior parameters are fused, so that the behavior parameters of the watching object of the target live broadcasting room can be obtained; determining a anchor interaction behavior instruction based on a preset behavior triggering condition and a pre-trained behavior determination model according to the behavior parameters of the viewing object; and finally, controlling the behavior of the virtual anchor according to the anchor interaction behavior instruction.

The live action control method of the virtual anchor can be used in various application scenes, such as a sales scene, an education scene, a performance scene, a display scene, a travel scene, a social scene and the like.

The following are embodiments of the disclosed apparatus that may be used to perform embodiments of the disclosed methods.

Fig. 3 shows a block diagram of a live action control apparatus of a virtual anchor according to an embodiment of the present disclosure, which may be implemented as part or all of an electronic device by software, hardware, or a combination of the two. As shown in fig. 3, the live action control device of the virtual anchor includes:

a first obtaining module 301 configured to obtain a behavior parameter of a viewing object of a target live broadcast room in which the virtual anchor is located, the behavior parameters are obtained by fusing real-time behavior parameters and historical behavior parameters. A determining module 302 configured to determine the anchor interaction behavior instruction according to the behavior parameter of the viewing object. And the control module 303 is configured to control the behavior of the virtual anchor according to the anchor interaction behavior instruction.

The above mentions, with the development of internet technology, live broadcast platform is coming, because the live broadcast is more directly perceived, rich in interactive, more and more users are willing to know the characteristics of commodity and buy commodity through live broadcast platform, consequently, live broadcast quality and live broadcast cost just become the problem that live broadcast platform is comparatively concerned about. To reduce the cost of live broadcasts, many live platforms enable virtual anchor, which refers to an anchor that uses avatars to conduct live activities on the live platform. In the prior art, the action and behavior data of the virtual anchor are usually preset, so that the virtual anchor executes anchor activities according to the preset action and behavior data, and further, in order to make the action and behavior of the virtual anchor closer to the real anchor, real-time data of a live broadcast room is collected, so as to adjust the tone, expression and body actions of the virtual anchor according to the real-time data of the live broadcast room. However, limited by the technology, the virtual anchor in the prior art still cannot flexibly express words according to real-time data of a live broadcast room like a live anchor, and timely adjust action and behavior strategies, so that the live broadcast quality is greatly reduced, the live broadcast experience of a user is damaged, and the development of a live broadcast platform is hindered.

In view of the above drawbacks, in this embodiment, a live action control device of a virtual anchor is provided, which obtains a behavior parameter of a viewing object based on live room data, and makes a decision or adjusts a virtual anchor action based on the behavior parameter of the viewing object. The technical scheme can realize the real-time perception of the virtual anchor on the environment of the live broadcasting room, can realize the real-time decision or adjustment on the behavior of the virtual anchor according to the environment of the live broadcasting room, therefore, the live broadcast effect is greatly improved, the live broadcast quality is improved, the live broadcast experience of a user is enhanced, and the development of a live broadcast platform is promoted.

In an embodiment of the present disclosure, the live behavior control device of the virtual anchor may be implemented as a behavior controller of a computer, a computing device, an electronic device, a server, a service cluster, or the like, which performs behavior control on the virtual anchor.

In an embodiment of the present disclosure, a viewing object refers to an object for viewing virtual anchor broadcast content in a certain live broadcast room, such as a live broadcast room user, a live broadcast room viewer, a live broadcast room buyer, and the like.

In an embodiment of the present disclosure, the behavior parameter of the viewing object refers to an index or a parameter capable of representing a behavior tendency of the viewing object in a certain live broadcasting room, wherein, in order to comprehensively reflect behavior characteristics of the viewing object, the behavior parameter of the viewing object is obtained based on a real-time behavior parameter and a historical behavior parameter of the viewing object, that is, the behavior parameter of the viewing object is obtained based on a fusion of the real-time behavior parameter and the historical behavior parameter, and the following will describe the behavior parameter of the viewing object in detail.

In an embodiment of the present disclosure, the anchor interactive behavior instruction refers to an instruction for driving a virtual anchor in a target live broadcast room to execute one or some interactive behaviors, where the anchor interactive behavior may be, for example, a behavior of calling a viewing object, answering a question of the viewing object, replying a comment of the viewing object, exciting the viewing object, extracting an excitation, and issuing a red envelope.

In an embodiment of the present disclosure, when controlling a behavior of a virtual anchor according to an anchor interaction behavior instruction, the anchor interaction behavior instruction may be sent to a control device of the virtual anchor, so that the control device controls a behavior of the virtual anchor in a target live broadcast room according to the anchor interaction behavior instruction, or when a behavior controller is a control device of the virtual anchor, the behavior of the virtual anchor in the target live broadcast room may be directly controlled according to the anchor interaction behavior instruction.

In the embodiment, firstly, the behavior parameters of the viewing object in the target live broadcasting room of the virtual anchor are obtained, then the anchor interaction behavior instruction for driving the anchor to execute one or more behaviors is determined according to the behavior parameters of the viewing object, and finally, the behavior of the virtual anchor is controlled according to the anchor interaction behavior instruction.

In an embodiment of the present disclosure, the first obtaining module 301 may be configured to: and collecting real-time behavior data of the watching object, and determining real-time behavior parameters according to the real-time behavior data. And acquiring historical behavior data of the watching object, and determining historical behavior parameters according to the historical behavior data. And fusing the real-time behavior parameters and the historical behavior parameters to obtain the behavior parameters of the watching object.

Specifically, the method comprises the following steps: firstly, real-time behavior data of a watching object is collected, wherein the real-time behavior data can comprise real-time behavior content and real-time behavior occurrence time information, and the real-time behavior data can comprise one or more of the following data: the watching object enters the live broadcasting room, leaves the live broadcasting room and the like to watch data; the watching objects give comments, like the watching objects, share the interactive data of the watching objects such as a live broadcast room and the like; the viewing object transaction data may include the viewing object clicking a link to a certain product, the viewing object purchasing a certain product, and so on. The real-time behavior data can effectively reflect the live broadcast state of the current live broadcast room, such as the number of people entering the live broadcast room at the current time, the number of people leaving the live broadcast room at the current time, the number of people watching the live broadcast in the current live broadcast room, the peak watching number of people, the number of people interacting with the main broadcast, the content with the most interaction, the number of praise obtained in the current live broadcast room, the content with the most comment number, and the like; real-time behavior parameters of the viewing object are then determined from the real-time behavior data.

And finally, fusing the real-time behavior parameters and the historical behavior parameters together to obtain the behavior parameters of the watching objects in the target live broadcast room. The real-time behavior parameters and the historical behaviors can be merged together to realize the fusion of the real-time behavior parameters and the historical behavior parameters, for example, the real-time behavior parameters and the historical behaviors can be spliced together in a JSON form.

In an embodiment of the present disclosure, the collecting of real-time behavior data of the viewing object, and the determining of the real-time behavior parameter according to the real-time behavior data may be configured to: real-time behavior of the viewing object is obtained. And collecting real-time behavior data of the watching object in response to the fact that the real-time behavior meets the preset trigger condition. Counting real-time behavior indexes with different durations and content dimensions according to the real-time behavior data, and counting real-time behavior parameters based on the real-time behavior indexes in different durations, wherein the real-time behavior indexes comprise: the viewing object behavior occurrence time point, the viewing object behavior occurrence frequency and/or the viewing object behavior occurrence frequency, wherein the real-time behavior parameters comprise: the behavior of the viewing object occurs frequently in different time durations, the behavior of the viewing object occurs frequently in different time durations to reach a preset threshold value, and/or the behavior of the viewing object occurs at different time points.

In order to reflect the live broadcast state of the current live broadcast room in multiple angles, in the embodiment, after the real-time behavior data of the watching object of the target live broadcast room is obtained, the real-time behavior indexes with different time lengths and content dimensions are counted according to the real-time behavior data, and the real-time behavior parameters are counted and obtained on the basis of the real-time behavior indexes in different time lengths.

Specifically, first, real-time behaviors of a viewing object, such as entering or leaving a live broadcast room, commenting, praising and the like, are acquired, and when the real-time behaviors meet a preset trigger condition, real-time behavior data of the viewing object is acquired, for example, when a certain viewing object enters the live broadcast room, the number of viewing objects in the live broadcast room reaches a certain number threshold, the behavior that the viewing object enters the live broadcast room is considered to meet the trigger condition on the number of viewing objects in the live broadcast room, and behavior data of the behavior that the viewing object enters the live broadcast room, such as the time that the viewing object enters the live broadcast room, information of the viewing object and the like, can be acquired; then, according to the real-time behavior data, counting real-time behavior indexes of different time lengths and content dimensions, wherein the real-time behavior indexes of different time lengths and content dimensions may include one or more of the following indexes: the time point of occurrence of the behavior of the viewing object, the frequency of occurrence of the behavior of the viewing object, and the frequency of occurrence of the behavior of the viewing object, wherein the time point of occurrence of the behavior of the viewing object may be, for example, a time point when the viewing object enters a live broadcast room, a time point when a comment is made, and the like; the number of times of behavior occurrence of the viewing object may be, for example, the number of viewing objects entering the live broadcast room within a preset time length, the number of times that a comment is made by a certain viewing object or all viewing objects in the live broadcast room, the number of times that a comment is complied with by a certain viewing object or all viewing objects in the live broadcast room, or the like; the frequency of occurrence of the viewing object behavior may be, for example, the number of viewing objects entering the live room per unit time, the number of times a comment is made on a certain viewing object or all viewing objects in the live room, and the like. Then, in different time lengths, real-time behavior parameters are obtained based on the real-time behavior index statistics, wherein the real-time behavior parameters may include one or more of the following parameters: the viewing object behaviors occurring in different durations, the viewing object behaviors occurring in different durations and occurring at different time points, where the frequency of the viewing object behaviors occurring in different durations reaches a preset threshold, may be, for example, the number of viewing objects entering a live broadcast room in the last 5 seconds, the number of times a question is asked in the last 1 minute, the number of times a comment is published in the last 1 minute, the number of votes received in a live broadcast room in the last 10 minutes, the number of votes received in a live broadcast room since the present announcement, the accumulated number of votes received in a live broadcast room since the present announcement, and the like, the viewing object behaviors occurring in different durations and reaching the preset threshold may be, for example, the question asked in the last 1 minute, the comment published in the most times in the last 1 minute, and the like, and the viewing object behaviors occurring at different time points may be, for example, the viewing object information entered latest in the last 5 minutes, and the like. The statistics of the real-time behavior parameters depends on different statistics durations and different statistics contents, and the statistics durations and the statistics contents can be determined according to different live broadcast scenes or different statistics purposes in the same live broadcast scene.

In an embodiment of the present disclosure, the obtaining of historical behavior data of the viewing object, the determining of the portion of the historical behavior parameter according to the historical behavior data, may be configured to: historical behavior data of the watching object in a first preset historical time period is obtained. According to historical behavior data, historical behavior indexes of different time lengths and content dimensions are counted, and historical behavior parameters are counted and obtained on the basis of the historical behavior indexes in different historical time lengths, wherein the historical behavior indexes comprise: a viewing object behavior occurrence time point, a viewing object behavior occurrence frequency, and a viewing object behavior occurrence frequency, the historical behavior parameters include: content of interest of the viewing object in different historical durations and behavior of interest of the viewing object in different historical durations.

In order to reflect the favorite conditions of the watching objects to the commodities or the commodity categories in multiple aspects, in the embodiment, after historical behavior data of the watching objects in a preset historical time period of the target live broadcast room are obtained, historical behavior indexes with different time lengths and content dimensions can be obtained through statistics according to the historical behavior data. Wherein, the first and the second end of the pipe are connected with each other, historical behavior indicators for different durations and content dimensions may include one or more of the following indicators: a viewing object behavior occurrence time point, a viewing object behavior occurrence number of times, and a viewing object behavior occurrence frequency, where the viewing object behavior occurrence time point may be, for example, a time at which the viewing object browses a commodity or a commodity category, a time at which the viewing object purchases a commodity or a commodity category, or the like. The number of times of occurrence of the viewing object behavior may be, for example, the number of times of viewing the product or the product category by the viewing object in different time periods, the number of times of purchasing the product or the product category by the viewing object in different time periods, or the like. The frequency of viewing object behavior may be, for example, the number of times the viewing object browses the goods or goods categories per unit time, the number of times the viewing object purchases the goods or goods categories per unit time, or the like. Then, in different historical time lengths, historical behavior parameters are obtained through statistics based on historical behavior indexes, wherein the real-time behavior parameters comprise the interesting content of the watching object in different historical time lengths and the interesting behavior of the watching object in different historical time lengths, for example, commodities or commodity categories frequently browsed by the watching object in the last 7 days, commodities or commodity categories purchased by the watching object in the last 1 month, commodities or commodity categories purchased by the watching object in the last 5 days and the like can be obtained through statistics based on the historical behavior indexes, and the interesting content of the watching object in different historical time lengths and the interesting behavior of the watching object in different historical time lengths can be determined based on the data. Similar to the statistics of the real-time behavior parameters, the statistics of the historical behavior parameters depend on different statistics durations and different statistics contents, and the statistics durations and the statistics contents can be determined according to different live broadcast scenes or different statistics purposes in the same live broadcast scene.

In an embodiment of the present disclosure, the determining module 302 may be configured to: <xnotran> . </xnotran> When the behavior parameter of the viewing object meets the preset behavior triggering condition, and generating a main broadcasting interaction behavior instruction based on the preset behavior.

In this embodiment, in determining the anchor interaction behavior instruction: firstly, determining whether a behavior parameter of a viewing object meets a preset behavior triggering condition, where the preset behavior may be, for example: placing a call with the viewing object, answering a question of the viewing object, replying to a comment of the viewing object, exciting the viewing object, extracting an excitation, issuing a red envelope, and the like; the preset behavior triggering condition may be, for example: the number of the watching objects entering the live broadcast room in the last 5 seconds reaches a first number threshold value, and then the watching objects are stimulated; answering the question of the viewing object when the number of times a question has been asked within the last 1 minute reaches a second number threshold and the question having the largest number of times has been asked within the last 1 minute is determined; when the number of times of posting a certain comment in the last 1 minute reaches a third number threshold and the comment with the largest posting number in the last 1 minute is determined, replying the comment of the viewing object; determining that the latest incoming viewing object information within the last 5 minutes is called with the viewing object; extracting the incentive when the number of praise obtained in the live broadcast room in the last 10 minutes reaches a fourth quantity threshold value; and when the accumulated number of praise obtained in the live broadcasting room from the broadcasting reaches a sixth number threshold, red packages are issued, and the like.

And if the behavior parameters of the watching objects meet the preset behavior triggering conditions, generating a corresponding anchor interaction behavior instruction based on the preset behaviors so that the anchor executes the preset behaviors. For example, if it is determined that the number of viewing objects entering the live broadcast room within the last 5 seconds reaches the first number threshold, it is considered that the behavior parameter of the viewing object satisfies the trigger condition for stimulating the preset behavior of the viewing object, and an anchor interaction behavior instruction may be generated based on the preset behavior, so that the virtual anchor executes the behavior for stimulating the viewing object; for example, if it is determined that the number of praise obtained in the live broadcast room in the last 10 minutes reaches the fourth number threshold, it is considered that the behavior parameter of the viewing object meets the trigger condition of the preset behavior of the extraction incentive, and an anchor interaction behavior instruction may be generated based on the preset behavior, so that the virtual anchor executes the behavior of the extraction incentive; for another example, if it is determined that the number of praise obtained in the live broadcast room since the current broadcast reaches the fifth number threshold, it is determined that the behavior parameter of the viewing object meets the trigger condition of the preset behavior of issuing the red packet, and the anchor interaction behavior instruction may be generated based on the preset behavior, so that the virtual anchor executes the behavior of issuing the red packet, and the like.

That is, in an embodiment of the present disclosure, the preset behavior triggering condition may include one or more of the following conditions: the number of the viewing objects in the first preset duration reaches a first number threshold, the occurrence frequency of the behavior of the viewing objects in the second preset duration reaches a second number threshold and/or a preset viewing object behavior event occurs at a preset time point.

In this embodiment, when the behavior parameter of the viewing object satisfies the preset behavior trigger condition, the portion for generating the anchor interaction behavior instruction based on the preset behavior may be configured to: and when the number of the viewing objects in the first preset duration reaches a first number threshold, generating a first anchor interaction behavior instruction which enables the virtual anchor to execute a first preset behavior. And when the occurrence frequency of the behavior of the watching object in the second preset duration reaches a second quantity threshold, generating a second anchor interaction behavior instruction for enabling the virtual anchor to execute a second preset behavior. And/or generating a third anchor interaction behavior instruction for enabling the virtual anchor to execute a third preset behavior when the preset viewing object behavior event occurs at the preset time point.

In an embodiment of the present disclosure, the determining module 302 may be further configured to: and when the number of the preset behaviors of which the trigger conditions are met is more than or equal to two, determining the preset behaviors of which the trigger conditions are met as candidate preset behaviors. And determining a target preset behavior from the candidate preset behaviors according to the behavior parameters of the viewing object. And generating a anchor interaction behavior instruction based on the target preset behavior.

As mentioned above, the preset behaviors may be one or more, and the trigger conditions of the corresponding preset behaviors may also be one or more, so that in the process of determining the trigger conditions of the preset behaviors, a situation that the number of the preset behaviors whose trigger conditions are met is greater than or equal to two may occur, and at this time, two or more preset behaviors need to be selected, so as to effectively improve the live broadcast effect and live broadcast quality. <xnotran> , ; </xnotran> Then determining the most suitable target preset behavior from the candidate preset behaviors according to the behavior parameters of the watching object; and finally, generating a main broadcasting interaction behavior instruction based on the selected target preset behavior so as to control the virtual main broadcasting to execute the target preset behavior.

In an embodiment of the present disclosure, the determining the target preset behavior from the candidate preset behaviors according to the behavior parameter of the viewing object may be configured to: and taking the behavior parameters of the watching object as input, and inputting the behavior parameters into a behavior determination model obtained by pre-training to obtain the target preset behavior.

In this embodiment, a machine learning method is used to determine a target preset behavior from candidate preset behaviors, that is, a behavior parameter of an object to be viewed is used as an input and is input into a behavior determination model, so that the target preset behavior can be obtained. The behavior determining model is a pre-trained model, the input of the behavior determining model is a behavior parameter of the viewing object, and the output of the behavior determining model is a target preset behavior.

In an embodiment of the present disclosure, the determining the target preset behavior from the candidate preset behaviors according to the behavior parameter of the viewing object may be further configured to: training behaviors a model is determined.

In this embodiment, when the behavior determination model is trained, the training behavior parameters of the viewing object may be used as input, the labeled target preset behavior may be used as output to train the behavior determination model, and the finally obtained behavior determination model with an algorithm convergence is a model that can be used to determine the behavior parameters of the subsequent viewing object. The initial form of the behavior determination model and the selection of the training method of the behavior determination model are common technical means in the art, and the present disclosure does not specifically limit the initial form and the training method.

In an embodiment of the present disclosure, the training of the portion of the behavior determination model may be configured to: determining initial behavior a model is determined. And acquiring the training behavior parameters of the watching object. And training the initial behavior determination model by taking the training behavior parameters as input and taking the preset behavior which is correspondingly marked based on the training behavior parameters as output to obtain the behavior determination model.

When the behavior determination model is trained, firstly, an initial behavior determination model is determined, wherein the initial behavior determination model can be selected according to the requirements of practical application; then acquiring a training behavior parameter of the viewing object, wherein the training behavior parameter can be a historical behavior parameter of the viewing object; and then training the initial behavior determination model by taking the training behavior parameters as input and the marked preset behaviors corresponding to the training behavior parameters as output, and obtaining the behavior determination model when the training result is converged.

In an embodiment of the present disclosure, the apparatus may further include: and the second acquisition module is configured to acquire historical behavior data of the virtual anchor of the target live broadcast room in a second preset historical time period, wherein the historical behavior data comprises behaviors of which the historical evaluation scores of the virtual anchor are higher than a preset score threshold value and behavior objects of which the historical evaluation scores of the virtual anchor are higher than the preset score threshold value.

Determining Module 302 may be configured to: and determining an anchor interaction behavior instruction according to the behavior parameters of the viewing object and the historical behavior data of the virtual anchor of the target live broadcast room.

In view of the fact that the broadcast characteristics of the virtual anchor also have a certain influence on the determination of the anchor interaction behavior command, in this embodiment, the broadcast characteristics of the virtual anchor are also referred to when determining the anchor interaction behavior command. That is, when acquiring real-time behavior data and historical behavior data of a viewing object of a target live broadcast room, historical behavior data of a virtual anchor of the target live broadcast room in a second preset historical time period is also acquired, where the historical behavior data includes a behavior in which a historical evaluation score of the virtual anchor is higher than a preset score threshold and a behavior object in which a historical evaluation score of the virtual anchor is higher than a preset score threshold, and the behavior in which the historical evaluation score of the virtual anchor is higher than the preset score threshold may be, for example: <xnotran> , , ; </xnotran> The action objects with the virtual anchor historical evaluation score higher than the preset score threshold value may be, for example: and the virtual anchor broadcasts the commodities or commodity categories which have the most times or are most popular with the watching objects, and the like. In this embodiment, the determination module 302 may be configured to: determining a anchor interaction behavior instruction according to the behavior parameters of the viewing object and the historical behavior data of the virtual anchor of the target live broadcast room, namely comprehensively determining the anchor interaction behavior instruction according to the behavior parameters of the viewing object and the historical behavior data of the virtual anchor of the target live broadcast room.

In an embodiment of the present disclosure, the determining, according to the behavior parameter of the viewing object and the historical behavior data of the target live broadcast virtual anchor, a part of the anchor interaction behavior instruction may be configured to: determining behavioral parameters of a viewing object whether a preset behavior triggering condition is met. When the number of preset actions for which the trigger condition is satisfied is greater than or equal to two, and determining the preset behaviors with the trigger conditions met as candidate preset behaviors. And determining candidate preset behaviors matched with the historical behavior data of the virtual anchor of the target live broadcast room as target preset behaviors. And generating a anchor interaction behavior instruction based on the target preset behavior.

As mentioned above, when the anchor interaction behavior command is determined by considering only the behavior parameters of the viewing object, it is first determined whether the behavior parameters of the viewing object satisfy the preset behavior trigger condition, if the number of the preset behaviors whose trigger conditions are satisfied is greater than or equal to two, the preset behaviors whose trigger conditions are satisfied are all determined as candidate preset behaviors, then a target preset behavior is determined from the candidate preset behaviors according to the behavior parameters of the viewing object, and the anchor interaction behavior command is generated based on the target preset behavior. In this embodiment, the anchor interaction behavior instruction is determined by integrating the behavior parameters of the viewing object and the historical behavior data of the virtual anchor in the target live broadcast room, specifically: similar to the case of determining the anchor interaction behavior instruction only by considering the behavior parameters of the viewing object, firstly, determining whether the behavior parameters of the viewing object meet the preset behavior triggering conditions, and if the number of the preset behaviors of which the triggering conditions are met is greater than or equal to two, determining all the preset behaviors of which the triggering conditions are met as candidate preset behaviors; if the target preset behavior is determined to be the target preset behavior, the target preset behavior is determined to be the target preset behavior according to the historical behavior data of the virtual anchor in the target live broadcast room, for example, if the two candidate preset behaviors are: the incentives are extracted and the red packet is issued, and then the preset behavior most popular with the watching object can be determined as the target preset behavior according to the historical behavior data of the virtual anchor of the target live broadcast room; finally, can be based on target preset behavior and generating a main broadcasting interaction behavior instruction.

The present disclosure also discloses an electronic device, fig. 4 shows a block diagram of the electronic device according to an embodiment of the present disclosure, as shown in fig. 4, the electronic device 400 includes a memory 401 and a processor 402; <xnotran> 401 , , 402 . </xnotran>

As shown in fig. 5, the computer system 500 includes a processing unit 501 that can execute various processes in the above-described embodiments according to a program stored in a Read Only Memory (ROM) 502 or a program loaded from a storage section 508 into a Random Access Memory (RAM) 503. In the RAM503, various programs and data necessary for the operation of the system 500 are also stored. The processing unit 501, the ROM502, and the RAM503 are connected to each other by a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.

The following components are connected to the I/O interface 505: an input portion 506 including a keyboard, a mouse, and the like; an output portion 507 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 508 including a hard disk and the like; and a communication section 509 including a network interface card such as a LAN card, a modem, or the like. The communication section 509 performs communication processing via a network such as the internet. The driver 510 is also connected to the I/O interface 505 as necessary. A removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 510 as necessary, so that a computer program read out therefrom is mounted into the storage section 508 as necessary. The processing unit 501 may be implemented as a CPU, a GPU, a TPU, an FPGA, an NPU, or other processing units.

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

The units or modules described in the embodiments of the present disclosure may be implemented by software or hardware. The units or modules described may also be provided in a processor, and the names of the units or modules do not in some cases constitute a limitation of the units or modules themselves.

<xnotran> , , ; </xnotran> Or it may be a separate computer readable storage medium not incorporated into the device. The computer readable storage medium stores one or more programs for use by one or more processors in performing the methods described in the present disclosure.

The foregoing description is only exemplary of the preferred embodiments of the disclosure and is illustrative of the principles of the technology employed. It will be appreciated by those skilled in the art that the scope of the invention in the present disclosure is not limited to the specific combination of the above-mentioned features, but also encompasses other embodiments in which any combination of the above-mentioned features or their equivalents is possible without departing from the inventive concept. For example, the above features and (but not limited to) the features disclosed in this disclosure having similar functions are replaced with each other to form the technical solution.

Claims

1. A live action control method of a virtual anchor comprises the following steps:

acquiring behavior parameters of a watching object of a target live broadcast room where a virtual anchor is located, wherein the behavior parameters are obtained based on fusion of real-time behavior parameters and historical behavior parameters;

determining a anchor interaction behavior instruction according to the behavior parameters of the viewing object;

and controlling the behavior of the virtual anchor according to the anchor interaction behavior instruction.

2. The method of claim 1, wherein the obtaining of the behavior parameters of the viewing object in the target live broadcast room of the virtual anchor comprises:

acquiring real-time behavior data of a watching object, and determining real-time behavior parameters according to the real-time behavior data;

acquiring historical behavior data of a watching object, and determining historical behavior parameters according to the historical behavior data;

and fusing the real-time behavior parameters and the historical behavior parameters to obtain the behavior parameters of the watching object.

3. The method of claim 2, the collecting real-time behavior data of the viewing object, determining real-time behavior parameters from the real-time behavior data, comprising:

acquiring real-time behaviors of a watching object;

collecting real-time behavior data of the watching object in response to the fact that the real-time behavior meets a preset trigger condition;

counting real-time behavior indexes with different time lengths and content dimensions according to the real-time behavior data, and counting real-time behavior parameters based on the real-time behavior indexes in different time lengths, wherein the real-time behavior indexes comprise: the time point of occurrence of the behavior of the viewing object, the frequency of occurrence of the behavior of the viewing object and/or the frequency of occurrence of the behavior of the viewing object, wherein the real-time behavior parameters include: <xnotran> , / . </xnotran>

4. The method of claim 2 or 3, wherein the obtaining historical behavior data of the viewing object and the determining historical behavior parameters according to the historical behavior data comprises:

acquiring historical behavior data of a watching object in a first preset historical time period;

according to the historical behavior data, historical behavior indexes of different time lengths and content dimensions are counted, and historical behavior parameters are counted and obtained on the basis of the historical behavior indexes in different historical time lengths, wherein the historical behavior indexes comprise: the viewing object behavior occurrence time point, the viewing object behavior occurrence frequency and the viewing object behavior occurrence frequency, wherein the historical behavior parameters comprise: content of interest of the viewing object in different historical durations and behavior of interest of the viewing object in different historical durations.

5. The method of any of claims 1-4, wherein determining the anchor interaction behavior instruction according to the behavior parameter of the viewing object comprises:

determining whether the behavior parameters of the watching object meet preset behavior triggering conditions;

and when the behavior parameters of the watching objects meet a preset behavior triggering condition, generating an anchor interaction behavior instruction based on the preset behavior.

6. The method of claim 5, the preset behavior triggering condition comprising: the number of the viewing objects in the first preset duration reaches a first number threshold, the occurrence frequency of the behavior of the viewing objects in the second preset duration reaches a second number threshold and/or a preset viewing object behavior event occurs at a preset time point;

when the behavior parameters of the watching object meet a preset behavior triggering condition, generating an anchor interaction behavior instruction based on the preset behavior, wherein the method comprises the following steps:

when the number of the viewing objects in the first preset duration reaches a first number threshold, generating a first anchor interaction behavior instruction which enables the virtual anchor to execute a first preset behavior;

when the occurrence frequency of the behavior of the watching object in a second preset duration reaches a second quantity threshold, generating a second anchor interaction behavior instruction which enables the virtual anchor to execute a second preset behavior; and/or the presence of a gas in the atmosphere,

and when a preset viewing object behavior event occurs at a preset time point, generating a third anchor interaction behavior instruction which enables the virtual anchor to execute a third preset behavior.

7. The method of claim 5 or 6, the determining an anchor interaction behavior instruction according to the behavior parameter of the viewing object, further comprising:

when the number of the preset behaviors of which the trigger conditions are met is more than or equal to two, determining the preset behaviors of which the trigger conditions are met as candidate preset behaviors;

determining a target preset behavior from the candidate preset behaviors according to the behavior parameters of the watching object;

and generating a anchor interaction behavior instruction based on the target preset behavior.

8. The method of claim 7, wherein determining a target preset behavior from the candidate preset behaviors according to the behavior parameter of the viewing object comprises:

and taking the behavior parameters of the watching object as input, and inputting the behavior parameters into a behavior determination model obtained by pre-training to obtain a target preset behavior.

9. The method of claim 8, wherein determining a target preset behavior from the candidate preset behaviors according to the behavior parameter of the viewing object, further comprises:

training the behavior determination model.

10. The method of claim 9, the training the behavior determination model, comprising:

determining an initial behavior determination model;

acquiring training behavior parameters of a watching object;

and taking the training behavior parameters as input, and taking preset behaviors which are correspondingly marked based on the training behavior parameters as output to train the initial behavior determination model to obtain a behavior determination model.

11. The method of any of claims 1-10, further comprising:

acquiring historical behavior data of a virtual anchor of a target live broadcast room in a second preset historical time period, wherein the historical behavior data comprises behaviors of the virtual anchor with historical evaluation scores higher than a preset score threshold value and behavior objects of the virtual anchor with historical evaluation scores higher than the preset score threshold value;

the determining of the anchor interaction behavior command according to the behavior parameter of the viewing object is implemented as:

according to the behavior parameters of the viewing object and the target live broadcast room historical behavior data of the virtual anchor determines anchor interaction behavior instructions.

12. The method of claim 11, wherein determining a anchor interaction behavior instruction according to the behavior parameters of the viewing object and historical behavior data of the target live broadcast room virtual anchor comprises:

when the number of the preset behaviors of which the trigger conditions are met is greater than or equal to two, determining the preset behaviors of which the trigger conditions are met as candidate preset behaviors;

determining candidate preset behaviors matched with the historical behavior data of the virtual anchor of the target live broadcast room as target preset behaviors;

and generating a main broadcasting interaction behavior instruction based on the target preset behavior.

13. A live action control device of a virtual anchor, comprising:

the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is configured to acquire behavior parameters of a watching object of a target live broadcast room where a virtual anchor is located, and the behavior parameters are obtained based on fusion of real-time behavior parameters and historical behavior parameters;

the determining module is configured to determine a anchor interaction behavior instruction according to the behavior parameters of the viewing object;

and the control module is configured to control the behavior of the virtual anchor according to the anchor interaction behavior instruction.

14. The apparatus of claim 13, the first acquisition module configured to:

collecting real-time behavior data of a watching object, and determining real-time behavior parameters according to the real-time behavior data;