Information processing system, information processing method, and program

Abstract

In the field of content distribution, we provide appropriate support for feedback from content creators. [Solution] A program is disclosed that causes a computer to execute a process that manages activity information for each viewer regarding activities that can be performed by viewers who watch past content distributed by a broadcaster, and at least one of the past content and content currently being broadcast by the broadcaster; manages setting information regarding the broadcaster's reactions; obtains multiple actions from one or more viewers who are watching content currently being broadcast by the broadcaster; and automatically generates reaction candidates that the broadcaster can take for one or more specific actions among the multiple actions, based on the activity information and setting information.

Description

Technical Field

[0001] The present disclosure relates to an information processing system, an information processing method, and a program.

Background Art

[0002] In business negotiations carried out during flexible conversations, according to the classification that dynamically changes according to the progress of the negotiations and the intentions of the statements made by the sales staff and customers at that time, the sales staff makes proposals while dynamically changing the content to be spoken next, so as to appropriately navigate the sales talk according to the situation of the conversation at that time. Technologies are known.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the prior art as described above, in the field of content distribution, it is difficult to appropriately support the reaction from the distributor for a large number and diverse actions from a large number and diverse viewers.

[0005] Therefore, in one aspect, the present disclosure aims to appropriately support the reaction from the distributor in the field of content distribution.

Means for Solving the Problems

[0006] In one aspect, activity information regarding activities executable by viewers who view at least any one of the past content distributed by the distributor and the past content and the content currently being distributed by the distributor is managed for each viewer, setting information regarding the reaction of the distributor is managed, The aforementioned broadcaster obtains multiple actions from one or more viewers who are watching the content currently being streamed, A program is provided that causes a computer to perform a process that automatically generates reaction candidates that the distributor can take for one or more specific actions among the multiple actions, based on the activity information and the setting information. [Effects of the Invention]

[0007] In one respect, this disclosure makes it possible to appropriately support reactions from content distributors in the field of content distribution. [Brief explanation of the drawing]

[0008] [Figure 1] This is a block diagram of the information distribution system according to this embodiment. [Figure 2] This is a schematic diagram showing an example of the hardware configuration of a server device. [Figure 3] This is a schematic diagram showing an example of the hardware configuration of a terminal device. [Figure 4A] This is an explanatory diagram of a terminal image that can be viewed via a head-mounted display. [Figure 4B] This is an explanatory diagram for images viewable on a smartphone. [Figure 5] This is a schematic diagram illustrating the overall processing of this embodiment. [Figure 6] This figure shows an example of the content currently being streamed. [Figure 7] This is a schematic flowchart illustrating the process for deriving reaction candidates. [Figure 8] This is an explanatory diagram showing an example of the output screen for reaction candidates. [Figure 9] This is an example of a functional block diagram for Information Distribution System 1. [Figure 10] This is an explanatory diagram of information about the user (viewer). [Figure 11] This is an explanatory diagram of the activity information. [Figure 12] It is an explanatory diagram of setting information. [Figure 13] It is an explanatory diagram of the relationship between the charge amount per viewer and the number of comments. [Figure 14] It is an explanatory diagram of achievements. [Figure 15] It is an explanatory diagram of the analysis result of the emotions of viewers. [Figure 16] It is an explanatory diagram of the result of grouping viewers. [Figure 17] It is an explanatory diagram of an example of the relationship between comments and reaction candidates. [Figure 18] It is an explanatory diagram of another example of the relationship between comments and reaction candidates. [Figure 19] It is an explanatory diagram of yet another example of the relationship between comments and reaction candidates. [Figure 20] It is an explanatory diagram of a method for generating reaction candidates using a large language model. [Figure 21] It is a flowchart schematically showing an operation example of an information distribution system. [Figure 22] It is a flowchart related to the comment processing (step S2108) in FIG. 21. [Figure 23] It is a flowchart related to the reaction processing (step S2112) in FIG. 21.

Mode for Carrying Out the Invention

[0009] Hereinafter, each embodiment will be described in detail with reference to the accompanying drawings. In the accompanying drawings, for ease of viewing, in some cases, only some of the parts having the same attribute and existing in plurality are provided with reference numerals.

[0010] Hereinafter, the embodiments will be described with reference to the drawings.

[0011] Referring to FIG. 1, the outline of an information distribution system 1 according to an embodiment of the present invention will be described. FIG. 1 is a block diagram of the information distribution system 1 according to the present embodiment.

[0012] The information distribution system 1 comprises a server device 10 and one or more terminal devices 20. Although three terminal devices 20 are shown in Figure 1 for simplicity, the number of terminal devices 20 can be two or more.

[0013] The server device 10 is an information processing system, such as a server, managed by an operator providing an information distribution platform. The terminal device 20 is a device used by a user, such as a mobile phone, smartphone, tablet, PC (Personal Computer), head-mounted display, or game device. Typically, multiple terminal devices 20 can be connected to the server device 10 via the network 3 in different configurations for each user.

[0014] The terminal device 20 is capable of executing the information distribution application according to this embodiment. The information distribution application may be received by the terminal device 20 from the server device 10 or a predetermined application distribution server via the network 3, or it may be pre-stored in a storage device provided in the terminal device 20 or in a storage medium such as a memory card that the terminal device 20 can read. The server device 10 and the terminal device 20 are connected to each other via the network 3 so as to be able to communicate. For example, the server device 10 and the terminal device 20 cooperate to perform various processes related to information distribution.

[0015] In the information distribution system 1, users may be distinguished into distribution sides (content distribution sides) and viewing sides (content viewing sides). The organizer of a single distribution content may be a single user, or it may be multiple users (i.e., a collaboration by multiple distributionrs). When users are distinguished into distribution sides and viewing sides, the terminal device 20 includes a terminal device 20A for the distribution side (content distribution side) and a terminal device 20B for the viewing side (content viewing side). In the following explanation, the distribution side terminal device 20A and the viewing side terminal device 20B will be described as separate terminal devices, but it is possible that the distribution side terminal device 20A may be the viewing side terminal device 20B, or vice versa. In the following explanation, when terminal devices 20A and 20B are not specifically distinguished, they may simply be referred to as "terminal device 20".

[0016] Each terminal device 20 is connected to each other so as to be able to communicate via the server device 10. In the following, "one terminal device 20 transmits information to another terminal device 20" means "one terminal device 20 transmits information to another terminal device 20 via the server device 10." Similarly, "one terminal device 20 receives information from another terminal device 20" means "one terminal device 20 receives information from another terminal device 20 via the server device 10." However, in the modified example, each terminal device 20 may be connected so as to be able to communicate without going through the server device 10.

[0017] Network 3 may include wireless communication networks, the Internet, VPNs (Virtual Private Networks), WANs (Wide Area Networks), wired networks, or any combination thereof.

[0018] In the example shown in Figure 1, the information distribution system 1 includes studio units 30A and 30B. Studio units 30A and 30B are distribution-side devices, similar to the distribution-side terminal device 20A. Studio units 30A and 30B can be placed in a studio, room, hall, etc., for content production.

[0019] Each studio unit 30 may have the same functions as the distribution terminal device 20A and / or the server device 10. Hereinafter, when distinguishing between the distribution side and the viewing side, for the sake of simplicity, the explanation will mainly describe a configuration in which the distribution terminal device 20A distributes various content to each viewing terminal device 20B via the server device 10. However, alternatively or in addition to this, the studio units 30A and 30B facing the distributioner may have the same functions as the distribution terminal device 20A and distribute various content to each viewing terminal device 20B via the server device 10. Note that in the modified example, the information distribution system 1 may not include the studio units 30A and 30B.

[0020] In the following, the information distribution system 1 implements an example of an information processing system. However, each element of a specific terminal device 20 (see Figure 3) may implement an example of an information processing system, or multiple terminal devices 20 may cooperate to implement an example of an information processing system. Furthermore, the server device 10 may implement an example of an information processing system by itself, or the server device 10 and one or more terminal devices 20 may cooperate to implement an example of an information processing system.

[0021] Figure 2 is a schematic diagram showing an example of the hardware configuration of server device 10. Figure 2 schematically illustrates peripheral devices 129 in relation to the hardware configuration of server device 10. Peripheral devices 129 are optional and may include displays, etc.

[0022] The server device 10 includes a CPU (Central Processing Unit) 111, RAM (Random Access Memory) 112, ROM (Read Only Memory) 113, auxiliary storage device 114, drive device 115, and communication interface 117 connected by a bus 119, as well as a wired transceiver 125 and a wireless transceiver 126 connected to the communication interface 117.

[0023] The auxiliary storage device 114 is, for example, an HDD (Hard Disk Drive) or an SSD (Solid State Drive), and is a storage device that stores data related to application software, etc.

[0024] The wired transceiver unit 125 includes a transceiver capable of communicating using a wired network. Peripheral devices 129 are connected to the wired transceiver unit 125. However, some or all of the peripheral devices 129 may be connected to the bus 119 or to the wireless transceiver unit 126.

[0025] The wireless transceiver 126 is a transceiver capable of communicating using a wireless network. The wireless network may include the network 3 described above.

[0026] The server device 10 may also be connectable to the recording medium 116. The recording medium 116 stores a predetermined program. The program stored on this recording medium 116 is installed on the auxiliary storage device 114 of the server device 10 via the drive device 115. The installed predetermined program becomes executable by the CPU 111 of the server device 10. For example, the recording medium 116 may be a recording medium that records information optically, electrically, or magnetically, such as a CD (Compact Disc)-ROM, flexible disk, or magneto-optical disk, or a semiconductor memory that records information electrically, such as a ROM or flash memory.

[0027] Figure 3 is a schematic diagram showing an example of the hardware configuration of terminal device 20. In Figure 3, peripheral devices 260 are schematically illustrated in relation to the hardware configuration of server device 10.

[0028] The terminal device 20 includes a CPU 211, RAM 212, ROM 213, auxiliary storage device 214, drive device 215, and communication interface 217 connected by a bus 219, as well as a wired transceiver 225 and a wireless transceiver 226 connected to the communication interface 217.

[0029] The auxiliary storage device 214 is, for example, an HDD or SSD, and is a storage device that stores data related to application software, etc.

[0030] The wired transceiver unit 225 includes a transceiver capable of communicating using a wired network. Peripheral devices 260 are connected to the wired transceiver unit 225. However, some or all of the peripheral devices 260 may be connected to the bus 219 or to the wireless transceiver unit 226.

[0031] The wireless transceiver 226 is a transceiver capable of communicating using a wireless network. The wireless network may include the network 3 described above.

[0032] The terminal device 20 may also be connectable to the recording medium 216. The recording medium 216 stores a predetermined program. The program stored on this recording medium 216 is installed on the auxiliary storage device 214 of the terminal device 20 via the drive device 215. The installed predetermined program becomes executable by the CPU 211 of the terminal device 20. For example, the recording medium 216 may be a recording medium that records information optically, electrically, or magnetically, such as a CD-ROM, flexible disk, or magneto-optical disk, or a semiconductor memory that records information electrically, such as a ROM or flash memory.

[0033] The peripheral device 260 is optional, but in this embodiment, it includes a display device 2623 and an input device 2624.

[0034] The display device 2623 includes, for example, a display device such as a liquid crystal display or an organic EL (Electro-Luminescence) display. The display device 2623 is capable of displaying a variety of images. The display device 2623 is configured, for example, as a touch panel and functions as an interface for detecting various user operations. The display device 2623 may also be built into a head-mounted display.

[0035] The input device 2624 may include physical keys, or it may further include any input interface, such as a pointing device like a mouse. The input device 2624 may also be capable of accepting non-contact user input, such as voice input, gesture input, or gaze input. For gesture input, sensors for detecting various user states (such as image sensors, acceleration sensors, distance sensors, etc.), dedicated motion capture systems integrating sensor technology and cameras, or controllers such as joysticks may be used. The camera for gaze detection may also be placed inside the head-mounted display. As mentioned above, various user states include, for example, the user's orientation, position, movement, or similar, and in this case, the user's orientation, position, and movement are not only the orientation, position, and movement of parts or all of the user's body, such as the face and hands, but also the orientation, position, and movement of the user's gaze or similar.

[0036] Viewers can view the distributed content via their terminal device 20B. The distributed content may be content in a virtual space in which an avatar can move around. For example, in the case of a head-mounted display, the terminal device 20B may generate stereoscopic images for the head-mounted display by generating images G200 and G201 that are seen by the left and right eyes, respectively, as shown in Figure 4A. Figure 4A schematically shows images G200 and G201 that are seen by the left and right eyes, respectively. In the following, unless otherwise specified, content in a virtual space refers to content represented by images G200 and G201. For example, the content may be content that involves various movements of an avatar in a virtual space. In this case, the distribution terminal device 20A may, for example, realize various movements of the avatar in the virtual space in response to various operations by the distributionr.

[0037] The virtual space described below is a concept that includes not only immersive spaces viewable using a head-mounted display or similar device, where users can move freely (as in reality) through an avatar, but also non-immersive spaces viewable using a smartphone or similar device, as shown in Figure 4B. Furthermore, non-immersive spaces viewable using a smartphone or similar device may be continuous three-dimensional spaces where users can move freely through an avatar, or they may be discontinuous two-dimensional spaces.

[0038] Figure 5 is a schematic diagram illustrating the overall processing of this embodiment. In Figure 5, the viewer, server device 10 (abbreviated as "server" in Figure 5), database, and distribution user are shown in the upper section, and the operation of each element during content distribution is shown below. Figure 6 is a diagram showing an example of content being distributed. Figure 7 is a schematic flowchart showing the overview of the reaction candidate derivation process. Figure 8 is an explanatory diagram showing an example of the reaction candidate output screen.

[0039] In this embodiment, while a piece of content is being distributed, viewers can freely generate comments via the terminal device 20B (S2100) and send them to the server device 10 (S2110). Although only one viewer is shown in Figure 5, there can actually be many. Therefore, the server device 10 can receive many comments at various timings while a piece of content is being distributed. Here, comments are described as actions from viewers, but the same applies to other actions from viewers to distributors, such as providing other payment methods to the distributor (the act of sending gifts to the distributor).

[0040] When the server device 10 receives (acquires) a comment (step S2200), it outputs the comment to the broadcaster (and all viewers) and updates the activity information of the viewer in the database (activity information DB2302) based on the comment (S2210). Activity information may be managed for each viewer. Note that the term "comment" is not limited to text messages, but includes the concept of voice messages, displays that express emotes, gestures by avatars, emojis, stamps, etc. In this case, even for comments of various forms, reaction candidates can be automatically generated by the reaction candidate creation process described later. The comment may be output within the content, as shown in Figure 6. In this case, the broadcaster and all viewers can see the comment. In the example shown in Figure 6, content including the broadcaster's avatar M2 is being broadcast, and the image G6 (an image of a certain frame) related to the content includes a comment display field G11. Comments from each viewer may be displayed (output) sequentially (for example, in the order they were received) in the comment display field G11. Since the number of comments that can be displayed in the comment display field G11 is limited, the comment display field G11 may be updated in such a manner that only the most recent predetermined number of comments are always displayed in the comment display field G11.

[0041] Furthermore, comments may be output in different locations depending on their attributes. For example, in the example shown in Figure 6, the image G6 related to the content (an image of a single frame) has comments G12 and G109 of other attributes superimposed in an area separate from the comment display field G11. Comment G109 indicates a request for collaborative streaming.

[0042] Furthermore, while a piece of content is being streamed, viewers, like with comments, can obtain (purchase, etc.) Super Chat, gifts, tips, gratuities, or similar payment methods via the terminal device 20B (S2100) and send them to the server device 10 to present to the streamer (S2110).

[0043] When the server device 10 receives (acquires) such billing medium (step S2200), it outputs the billing medium and updates the activity information pertaining to the viewer in the database (activity information DB2302) based on the billing medium (S2210). In the example shown in Figure 6, the gift G12A corresponding to the billing medium is superimposed on the image G6 (an image of a single frame) related to the content. Note that the method of outputting the billing medium is not limited to this and is arbitrary.

[0044] Activity information relating to a single viewer may include activity information related to content, and may also include activity information related to communication with the broadcaster, etc. Activity information relating to a single viewer may relate to at least one of the following: the content of the viewer's comments, the frequency of comments posted by the viewer, the history of comments posted, the history of payment methods (including total amount, average, etc.), and any one or more of the information that can be derived from these. Activity information may also be the value of one or more indicator parameters calculated based on some or all of these parameters (e.g., the number of comments, the total amount of payment methods, etc.). In this case, indicator value parameters that allow for a comprehensive assessment of the viewer's activity can be used.

[0045] The server device 10 performs comment output / storage processing in this manner while simultaneously performing reaction candidate derivation processing (step S2220). The reaction candidate derivation processing is a process that automatically generates reaction candidates (hereinafter referred to as "reaction candidates") that the broadcaster can take in response to one or more specific comments (actions) from one or more viewers.

[0046] The form of reaction output is arbitrary, but may be a visible output of comments or text, an audible output of comments (output of voice messages), a gesture output, or any combination thereof. Gesture output may be output via a display medium such as an avatar. Furthermore, gesture output may involve only movement of a part of the body (e.g., hands or face), or it may involve movement of the whole body or other tools. In the following, unless otherwise specified, the form of reaction output will be in the form of a comment as an example, and reaction candidates will also be in the form of a comment.

[0047] Specifically, as shown in Figure 7, the reaction candidate derivation process reads configuration information from the configuration information database DB2300 (step S700). If the configuration information in the configuration information database DB2300 is updated in real time, the reading of configuration information from the configuration information database DB2300 may also be performed at a high frequency.

[0048] The configuration information may be prepared individually for each broadcaster. In this case, configuration information tailored to the differing perspectives and preferences of each broadcaster will be available. The configuration information concerns the prioritization of various situations. Each situation may include at least one of the following: the content or situation of comments, and the situation of the viewers. The configuration information may be configurable (customizable) by the broadcaster, or it may be automatically updated based on artificial intelligence according to the characteristics of the broadcaster.

[0049] The content or circumstances of a comment may include at least one of the following: whether or not it is a question or inquiry from a viewer, and whether or not an answer should be given to the question or inquiry. Furthermore, the viewer's circumstances may include whether or not they are a first-time viewer, whether or not they are a regular viewer, whether or not they have been watching for a long period of time or longer, whether or not they are a viewer who has not watched in a long time, etc. Specific examples will be given later.

[0050] By the way, while it is possible for a content creator to react to every single comment received during a live stream, it is impractical. Therefore, it is more efficient to automatically generate reaction options only for one or more specific comments selected according to a certain priority order, rather than automatically generating reaction options for every single comment.

[0051] Therefore, in this embodiment, the reaction candidate derivation process selects (extracts) one or more specific comments according to the priority order related to the setting information provided by the broadcaster (step S702). Note that the extraction process for one or more specific comments may be performed collectively for a predetermined number of comments each time a predetermined number of comments are accumulated. Alternatively, it may be performed in a manner in which it is determined whether each comment is a specific comment (a comment targeted for generating reaction candidates) each time a comment is received.

[0052] In this case, the settings information, such as which of the various situations should take the highest priority and which situations should take priority over others, may be set automatically (for example, suggested based on artificial intelligence). This can reduce the burden on the broadcaster. Alternatively, the broadcaster may be allowed to freely set which of the various situations should take the highest priority and which situations should take priority over others. This allows for the generation of appropriate reaction candidates for each situation according to the priorities that may differ for each broadcaster.

[0053] Here, comments that are not extracted as one or more specific comments are comments for which no reaction candidates are generated. Therefore, no reaction based on reaction candidates will be given to such comments. However, in this case, it is also possible to interpret that the reaction to such comments will be "ignored" or "left alone".

[0054] In other embodiments, comments for which a reaction such as "ignore" or "leave alone" is appropriate may be selected and excluded before one or more specific comments are extracted. In this case, the population from which one or more specific comments are extracted can be reduced, allowing for efficient selection (extraction) of one or more specific comments and reducing the processing load.

[0055] In situations where disruptive remarks or excessively frequent comments from viewers are effective in maintaining order, deliberately ignoring or neglecting such comments can be beneficial. Therefore, in these situations, it is useful to prevent such comments from being extracted as one or more specific comments.

[0056] Furthermore, the settings information for each broadcaster may be updated (changed) at any time by the corresponding broadcaster. In addition, various conditions related to the settings information may be configurable for each broadcaster, and the granularity (level of detail) of these conditions may also be configurable for each broadcaster.

[0057] The reaction candidate derivation process automatically generates (derives) reaction candidates for one or more specific comments that have been extracted (step S704). In this process, the reaction candidate derivation process may automatically generate a common reaction candidate for two or more specific comments that have been extracted, or it may automatically generate separate reaction candidates for each of them. For example, if two or more specific comments that have been extracted are identical or similar in content, a common reaction candidate may be automatically generated.

[0058] The method for automatically generating reaction suggestions for a specific comment is optional. Details of the automatic reaction suggestion generation method will be described later.

[0059] Once the server device 10 has derived reaction candidates in this manner, it sends the reaction candidates to the corresponding distributor's terminal device 20A (step S2230). The transmission of reaction candidates to the distributor's terminal device 20A may be performed in real time each time a reaction candidate is derived, or it may be performed all at once after a certain number of reaction candidates have been derived.

[0060] When the broadcaster's terminal device 20A receives a reaction candidate, it presents (displays) the reaction candidate to the broadcaster by outputting it visually (or visually and audibly) (step S2400). Figure 8 shows an example of a suitable user interface for presenting reaction candidates to the broadcaster.

[0061] In the example shown in Figure 8, each User representing a viewer is associated with the number of comments, comment ranking, total amount spent, elapsed time (minutes), status, explanation, to-do, and reaction candidate. In this case, the number of comments, comment ranking, total amount spent, and elapsed time (minutes) are examples of activity information.

[0062] The comment count is the number of comments posted by the corresponding viewer, and may be the number of comments during the current content broadcast or the total number of comments across multiple past content broadcasts by the same broadcaster. The comment ranking is a ranking of the comment counts. The comment ranking may be a ranking during the current content broadcast or a ranking based on the total number of comments across multiple past content broadcasts by the same broadcaster. The total amount charged is the sum of the amounts charged to the corresponding viewer, and may be the total amount charged across all past content broadcasts by the same broadcaster. Note that the amount charged is an indicator of the economic value of the payment medium. The elapsed time (minutes) may be the elapsed time since the most recent comment was posted (the elapsed time at the present). Status is the status of the corresponding viewer. Status may preferably be represented by an emoji, as shown in Figure 8. In this case, each emoji may be associated with a predetermined status. The explanation is an explanation of the status (or the corresponding emoji), and may include a description of the status. ToDo indicates the reactions that the broadcaster should take, such as whether or not to react. ToDo may indicate potential reactions when a reaction is needed. ToDo may preferably be represented by an Emoji, as shown in Figure 8. In this case, each Emoji may be associated with a predetermined ToDo (task). The viewers to whom the ToDo Emoji is associated correspond to viewers who have made one or more specific comments as described above. Reaction candidate is the reaction candidate described above.

[0063] As shown in Figure 8, broadcasters can intuitively understand the viewer's situation and To-Do items through emojis. In particular, unlike regular language, emojis do not require translation, are understandable to broadcasters with diverse native languages ​​(i.e., multilingual support is possible), and reduce the processing load of translation, etc. This allows broadcasters to present reaction candidates in a manner that suits their circumstances, such as not being able to spend a lot of time selecting reaction candidates.

[0064] The relationship between various emojis and their corresponding situations may be uniformly fixed, or it may be changeable by the broadcaster. Furthermore, some of the emojis may be independently generated by the broadcaster. In this case, the correspondence becomes easier for the broadcaster to understand, and the selection of reaction candidates becomes even easier.

[0065] Furthermore, each item shown in Figure 8 may be configurable by each broadcaster. For example, a broadcaster may configure the system to output only the following items from the list of comment count, comment ranking, total amount charged, elapsed time (minutes), status, explanation, to-do, and reaction candidate: comment ranking, elapsed time (minutes), status, and reaction candidate. The number of items may also vary depending on the attributes of the broadcaster's terminal device 20A. For example, the number of items may be fewer when the broadcaster's terminal device 20A has a relatively small screen (e.g., a smartphone) compared to when it has a relatively large screen (e.g., a desktop personal computer).

[0066] The broadcaster inputs a reaction instruction via the terminal device 20A (step S2410). The reaction instruction may be generated automatically or in response to a predetermined input from the broadcaster. The predetermined input is arbitrary, but may be input via a user interface such as the one shown in Figure 8. In this case, selecting an emoji may generate a reaction instruction that outputs a reaction candidate corresponding to the selected emoji as a reaction. In this case, a reaction instruction that outputs a corresponding reaction candidate as a reaction can be generated simply by the intuitive operation of selecting an emoji. If the broadcaster determines that a reaction is not necessary, they will not input a reaction instruction.

[0067] When the server device 10 receives a reaction instruction in this manner, it outputs (overlays) the reaction corresponding to the reaction instruction onto the content being distributed (step S2240). This makes the reaction viewable to viewers (step S2120). Figure 6 shows comment G13 as an example of a reaction. The reaction is output in a manner that can be seen by all viewers, but it may also be output in a manner that can be seen only by the target viewer.

[0068] In this way, according to this embodiment, by referring to the activity information of each viewer, one or more specific comments are extracted in order of priority according to the settings information set by the broadcaster. Then, reaction candidates are generated only for the extracted specific comments and presented to the broadcaster. The settings information can be set (customized, etc.) by the broadcaster, which increases the likelihood that the broadcaster can react appropriately and efficiently to desired comments (and consequently, desired viewers). As a result, it is possible to appropriately support reactions from broadcasters in the field of content distribution.

[0069] Referring to Figures 5 to 8, the various processes of the information distribution system 1 described above are executed in real time during content distribution, but some processes may be executed offline. For example, reaction candidate derivation processing is preferably executed in real time, but a certain effect can be expected even if it is executed offline. For example, the distributor may output reactions individually to specific viewers after distribution. The reaction candidate derivation processing described above can also be applied to such reactions.

[0070] Next, we will describe further details of the information distribution system 1 with reference to Figure 9 and subsequent figures.

[0071] Figure 9 is an example of a functional block diagram of the information distribution system 1. Figure 10 is an explanatory diagram of information about users (viewers). Figure 11 is an explanatory diagram of activity information. Figure 12 is an explanatory diagram of settings information. Figure 13 is an explanatory diagram of the relationship between the amount charged per viewer and the number of comments. Figure 14 is an explanatory diagram of achievements. Figure 15 is an explanatory diagram of the results of the analysis of viewer sentiment. Figure 16 is an explanatory diagram of the results of grouping the set of viewers. Figure 17 is an explanatory diagram of one example of the relationship between comments and reaction candidates. Figure 18 is an explanatory diagram of another example of the relationship between comments and reaction candidates. Figure 19 is an explanatory diagram of yet another example of the relationship between comments and reaction candidates. Figure 20 is an explanatory diagram of the method of generating reaction candidates using a large-scale language model. Note that in Figure 10 (and similar Figures 12 and below), "***" indicates a state in which some information is stored, and "···" indicates a state in which the storage of similar information is repeated. Figures 17 to 19 show a timeline (partial) of comments during content distribution at the top, and examples of corresponding reaction candidates at the bottom.

[0072] Note that each part of the information distribution system 1 described below may be implemented by the server device 10 alone. In this case, the terminal device 20A may receive the data output by the reaction candidate generation processing unit 172 and / or the reaction output unit 174 and output it to the display device 2623 (see Figure 3). Alternatively, each part of the information distribution system 1 described below may be implemented by the terminal device 20A alone (i.e., without going through the server device 10). In this case, some of the information in the activity information storage unit 150 described below (activity information regarding the content of other distributors) and some of the information in the setting information storage unit 152 (information regarding other distributors) do not need to be stored. Alternatively, each part of the information distribution system 1 described below may be implemented by a combination of the server device 10 and the terminal device 20A. In this case, for example, some or all of the functions of the reaction candidate generation processing unit 172 and / or the reaction output unit 174 may be implemented by the terminal device 20A alone.

[0073] The information distribution system 1 includes an activity information storage unit 150, a setting information storage unit 152, an activity information management unit 154, a setting information management unit 156, a distribution content processing unit 160, an action acquisition unit 162, a situation determination unit 164, an emotion analysis unit 166, an emoji mapping processing unit 168, a comment flow analysis unit 170, a reaction candidate generation processing unit 172, and a reaction output unit 174.

[0074] The activity information storage unit 150 and the setting information storage unit 152 can be implemented using, for example, the auxiliary storage device 114 shown in Figure 2 or the auxiliary storage device 214 shown in Figure 3.

[0075] Furthermore, the activity information management unit 154, the setting information management unit 156, and the distribution content processing unit 160 to the reaction output unit 174 can be realized by the CPU 111 shown in Figure 2 or the CPU 211 shown in Figure 3 executing one or more programs stored in the storage device (e.g., ROM 113, 213, etc.) shown in Figure 2 or Figure 3.

[0076] The activity information storage unit 150 stores activity information relating to viewers. The activity information is as described above. Activity information may be managed for each broadcaster. Figure 10 shows activity information relating to one broadcaster. In this case, the activity information is associated with the user ID along with the username (in Figure 10, "User"). Activity information relating to one broadcaster may be managed for each username (in Figure 10, "User") as shown in Figure 11. In the example shown in Figure 11, the activity information includes the number of comments, comment ranking, total amount charged, and elapsed time (minutes). Each of these parameters is as described above with reference to Figure 8.

[0077] The activity information storage unit 150 may store activity information relating to the broadcaster (for example, activity information relating to reactions) in addition to activity information relating to the viewer. In this case, the activity information relating to one broadcaster may be used to generate reaction candidates relating to one broadcaster.

[0078] The configuration information storage unit 152 stores configuration information. The configuration information is as described above. The configuration information may be managed for each broadcaster. Figure 12 shows the configuration information for one broadcaster. In this case, the configuration information associates reaction priority and priority with each of the various situations. The various situations are as described above. The various situations may be provided as default settings or may be customizable by the broadcaster. For example, situation A may include a situation related to the frequency of reactions from the broadcaster. For example, if a high priority is set for the frequency of reactions from the broadcaster, reaction candidates may be generated and presented to the broadcaster in a manner that allows for reactions for a relatively small number of comments.

[0079] Reaction priority represents absolute priority and can be indicated in three levels, such as "high," "medium," and "low." Priority level represents relative priority and can be indicated in rank, such as 1st place and 2nd place.

[0080] Incidentally, regarding setting information, the factors that broadcasters prioritize may vary from broadcaster to broadcaster, but it is possible to generate recommended setting information based on specific objectives, for example. For instance, recommended setting information could be generated based on artificial intelligence and presented to broadcasters for each objective, such as increasing the number of subscribers or participants, increasing total revenue, or increasing comment frequency (number of comments per unit of time). In this case, the results of surveys conducted among broadcasters may also be used. In this scenario, broadcasters can then build their own original setting information based on the recommended settings.

[0081] As shown in Figure 13, the relationship between the number of comments and the total amount spent may show certain trends. For example, in the example shown in Figure 13, viewers with a high number of comments are not necessarily also viewers with a high total amount spent. Therefore, simply prioritizing situations with a high number of comments over situations with a high total amount spent in the settings may not lead to an effective increase in total spending. Similarly, simply prioritizing situations with a high total amount spent over situations with a high number of comments in the settings may not lead to an effective increase in the number of comments. Therefore, the settings should be adapted according to the purpose and other factors. Furthermore, by adapting the settings, it becomes possible to create broadcasts that are not dominated by a few persistent or heavy spenders, and to prevent so-called spoilers.

[0082] Furthermore, the situations themselves that are included in the various situations related to the setting information may be defined by the broadcaster. For example, in Figure 12, the contents of Situation A and Situation B may be defined by the broadcaster. Also, situations where there is a question or inquiry from a viewer, or situations where a response to a question or inquiry is required, may be assigned a relatively high priority by default. In this case, the importance of the question or inquiry may also be determined. In addition, the situation of whether or not a comment contains a specific keyword may be determined. In this case, the specific keyword related to the situation that is assigned a high priority may be set by the broadcaster. Note that the specific keyword may include emojis, etc.

[0083] Furthermore, the various circumstances related to the setting information may include the viewer's emotional state. In this case, setting information may be configured such that situations where negative emotions such as anger are dominant take precedence over situations where positive emotions such as joy are dominant.

[0084] The activity information management unit 154 performs various management processes, such as storage processing to store activity information in the activity information storage unit 150, and update processing to update the activity information in the activity information storage unit 150. For example, when the action acquisition unit 162, described later, acquires various actions from viewers, the activity information management unit 154 may update the activity information in the activity information storage unit 150 accordingly. The activity information update may be performed in real time, or it may be performed periodically or irregularly. For example, the activity information update may be performed for each piece of content. In this case, the processing load can be reduced compared to when the update is performed in real time.

[0085] The configuration information management unit 156 performs various management processes, such as storage processing to store configuration information in the configuration information storage unit 152, and update processing to update activity information in the configuration information storage unit 152. For example, the configuration information management unit 156 performs management processing based on input from the distributor.

[0086] The content distribution processing unit 160 manages the distribution of content by each viewer. The format of content distribution is arbitrary and may be live streaming. Furthermore, the distribution may include distributions in which a human streamer appears directly, distributions using avatars related to a human streamer, distributions involving a robot (e.g., artificial intelligence) streamer, distributions involving collaborations between multiple streamers, viewer-participation distributions, or any combination of these. For example, the first half of a distribution may be a distribution involving a robot streamer, and the second half may be a distribution involving collaborations between multiple streamers. In a distribution, the speaking portion may be performed by a human streamer (or their corresponding avatar), while reactions (e.g., comments from the streamer to viewers in the comment section) may be performed by a robot (e.g., artificial intelligence) streamer. In addition, in very long distributions such as 24-hour distributions, a human streamer (or their corresponding avatar) may perform during specific time periods (e.g., from 9 PM to 11 PM), while a robot (e.g., artificial intelligence) streamer may perform during other time periods. In such diverse forms of distribution, various functions such as the action candidate generation processing unit 172 described above may be executed not only during the distribution period by a human broadcaster, but also during the distribution period by a robot (e.g., artificial intelligence) broadcaster.

[0087] The content distribution processing unit 160 may implement artificial intelligence to support the broadcaster, and / or artificial intelligence to control program production, cameras (e.g., virtual cameras for content generation), sound effects, etc. The content distribution processing unit 160 may suppress the volume and point the camera while the speaker is speaking. It may also control other sounds (e.g., interruptions such as reading out viewer comments) while the speaker is speaking to prevent conflicts between sounds and / or various effects.

[0088] The action acquisition unit 162 acquires various actions from viewers. The action acquisition unit 162 may function for each piece of content. These actions are actions that elicit reactions from the broadcaster, and may typically include posting comments or purchasing paid media. These actions are related to activity information.

[0089] The status determination unit 164 determines various statuses based on activity information. The status determination unit 164 may function for each piece of content. Preferably, the status determination unit 164 functions in real time during the distribution of the corresponding content and determines (detects) various statuses that may change dynamically. The various statuses to be determined may be the same for all content, but may only be those statuses described in the settings information of the distributor of the corresponding content. For example, if the settings information indicates that the reaction priority is "high" only for statuses A and B out of statuses A to X, then the various statuses to be determined (detected) may be limited to only statuses A and B out of statuses A to X. In this case, reducing the number of various statuses to be determined can reduce the amount of processing unmanageable.

[0090] The situation determination unit 164 may determine whether the content or situation of the comment is a question or inquiry from a viewer, whether or not an answer should be given to the question or inquiry, whether or not a specific keyword is included, etc. The content of the comment may be classified based on whether it is favorable or negative, violates public order and morals, or is novel, etc.

[0091] The situation determination unit 164 may determine the viewer's situation, such as whether they are a first-time viewer, whether it is their first time commenting, whether they have made a payment (such as a tip) for the first time, or whether it has been a long time since they last viewed the content. Such determinations regarding comments may be performed for each comment.

[0092] The situation determination unit 164 may determine the viewer's emotional state based on the emotional analysis results from the emotional analysis unit 166.

[0093] The circumstances of the viewer being judged may include, for example, at least one of the following: 1) A viewer who is watching content distributed by a distributor for the first time; 2) A viewer who pays above a predetermined threshold; 3) A viewer who is favorable to the distributor and does not fall under 1 or 2); and 4) A viewer who is undesirable to the distributor or hostile to the distributor. Paying above a predetermined threshold may be based on the aforementioned payment methods. Whether a viewer is favorable to a distributor may be determined based on the content of comments made by that viewer regarding the distribution of the distributor's content. Similarly, whether a viewer is undesirable to a distributor or hostile to that distributor may be determined based on the content of comments made by that viewer regarding the distribution of the distributor's content.

[0094] Regarding charges exceeding a predetermined standard, the standard is arbitrary and, as mentioned above, may be adjustable by the broadcaster. Incidentally, viewers who spend a relatively large amount on broadcasts are often favorable towards the broadcaster (i.e., they are often in situation 3), but they may also be in situation 4. Therefore, the situation of the viewer being judged may be further subdivided. For example, situation 4 may be further classified into more detailed situations depending on whether or not it also applies to situation 2. That is, situation 4 may include two sub-situations: a situation that also applies to situation 2, and a situation that does not apply to situation 2. This makes it possible to adapt the setting information to a variety of situations.

[0095] The status determination unit 164 may evaluate each viewer's achievement by applying an evaluation function to activity information when determining various statuses such as the viewer's status. Achievements may be evaluated based on whether or not certain requirements are met. Figure 14 shows a method for evaluating achievements using an API (Application Programming Interface) server and a database or model called the UserLTV model. For example, when determining whether or not a viewer has commented, a query to that effect (shown as "Query" in the figure) is given to the API server. The API server refers to the UserLTV model and outputs a response to the query (for example, a binary value of True / False). The UserLTV model may store information such as comments from each viewer, which is stored in the activity information storage unit 150 described above. In this case, the status determination unit 164 can be implemented by the API server. By using the API server, the processing load on the server device 10 can be reduced.

[0096] The emotion analysis unit 166 analyzes the emotions of each viewer at multiple points in time during content distribution. The emotion analysis unit 166 may function for each piece of content. Emotions may include anger, expectation, disgust, fear, joy, sadness, surprise, trust, etc. Figure 15 shows an example of emotion analysis results in chronological order. Such emotion analysis results may be used to generate reaction candidates, as will be described later. In other embodiments, each emotion such as anger may be further subdivided and analyzed, or its causes (factors) may be analyzed. Such analysis results can also be effectively used to generate reaction candidates.

[0097] The sentiment analysis unit 166 may analyze viewer comments and classify viewers into multiple groups. The functions of the sentiment analysis unit 166 may be implemented by artificial intelligence. The grouping method is arbitrary, but for example, the k-means clustering algorithm may be used. For example, in political content, viewers may be classified as right-wing, left-wing, etc. Figure 16 shows an example in which viewers are classified into two groups. Such classification results may be used to generate reaction candidates, as will be described later. Note that the number of groups to be classified is not limited to two, and there may be three or more. By using the sentiment analysis results to extract one or more specific comments or to make various situational judgments, processing that is in line with the viewer's emotions can be realized. For example, it is possible to prevent the generation of reaction candidates that would exacerbate anger in a situation where anger is the dominant emotion.

[0098] The Emoji mapping processing unit 168 outputs various situations with emojis in a manner that can be viewed by the distributor. The various situations are based on the determination results of the situation determination unit 164. The manner in which various situations are output with emojis and their effects may be as described above with reference to Figure 8.

[0099] The comment flow analysis unit 170 analyzes the flow of comments from viewers based on the time series of each comment during content distribution. The comment flow analysis unit 170 analyzes the flow of comments (the flow of conversations, etc., through comments) during content distribution. The flow of comments may include topics, etc., which may change dynamically. The flow analysis may be performed based on keyword extraction, etc. The results of such analysis may be used to generate reaction candidates, as described later.

[0100] The reaction candidate generation processing unit 172 generates reaction candidates. The reaction candidates are as described above. The reaction candidate generation processing unit 172 may generate reaction candidates based on the setting information and activity information, as described above.

[0101] For example, if the setting prioritizes situations where reactions occur relatively frequently, reaction candidates may be generated for each comment, as shown in Figure 17. Also, if the setting prioritizes the first situation over the second situation described above, reaction candidates may be generated with priority given to comments related to viewers in the first situation over comments related to viewers in the second situation, as shown in Figure 18.

[0102] In this embodiment, the reaction candidate generation processing unit 172 includes a first processing unit 1721 and a second processing unit 1722.

[0103] The first processing unit 1721 generates the same reaction candidate for two or more specific viewers who have a predetermined relationship. The predetermined relationship is arbitrary, but may be the same or common situation. For example, two or more viewers who are in the second situation described above may be two or more specific viewers who have the predetermined relationship. Alternatively, the predetermined relationship may be a relationship in which some or all of the activity information is the same or similar. For example, the predetermined relationship may be the same or common situation, and the elapsed time (in minutes) may be greater than or equal to a predetermined time.

[0104] The second processing unit 1722 generates different reaction candidates for each of two or more specific viewers who have a predetermined relationship. The second processing unit 1722 may function at a different time than the first processing unit 1721 described above. That is, there are multiple operating modes, and these multiple operating modes include a first mode in which the first processing unit 1721 functions and a second mode in which the second processing unit 1722 functions. In this case, the broadcaster can selectively implement a desired mode from among these multiple operating modes. For example, they can use multiple modes by selecting the second mode when they have time and the first mode when they are busy. Note that the mode selection may be automatically selected according to the broadcaster's situation.

[0105] Furthermore, in this embodiment, reaction candidates may be generated based on the emotion analysis results described above. In this case, reaction candidates that are in line with changes in the viewer's emotions can be generated. As a result, even in live streaming events where the viewer's emotions may change, such as as shown in Figure 15, it becomes possible to create variety, control the viewer's emotions, empathize with the viewer's emotions, and achieve diverse forms of communication. For example, during times when aversion is dominant, it may be possible to reduce the viewer's aversion by generating reaction candidates that eliminate the viewer's aversion.

[0106] Furthermore, in this embodiment, the reaction candidate generation processing unit 172 may generate reaction candidates based on the analysis results of the comment flow. In this case, reaction candidates that are in line with the comment flow can be generated. As a result, smoother communication can be promoted. For example, if the comment flow is "a request for another song (encore)", a reaction candidate "We'll play another song!" that is in line with the comment flow may be generated, as shown in Figure 19. This makes it possible to generate reaction candidates that are in line with the flow formed by the broadcast and viewer comments, and effectively enhance the sense of unity between the broadcaster and the viewers.

[0107] Here, the method for automatically generating reaction candidates is arbitrary, but it may be a method that utilizes artificial intelligence.

[0108] For example, the reaction candidate generation processing unit 172 may generate reaction candidates using generative artificial intelligence based on large language models (LLMs), etc. In this case, the automatic generation method for reaction candidates may include an automatic generation process for prompts to obtain reaction candidates. Large language models may be prepared (built) for each broadcaster, and in this case, various settings (e.g., persona settings) may be made by the broadcaster. Figure 20 shows three examples of prompts. In this case, a prompt or its base may be generated for each emoji associated with a situation. The form of the prompt is arbitrary, but it may be in a form that includes examples, as shown in Figure 20. Emojis represent various situations as described above. Therefore, broadcasters can intuitively check prompts according to various situations by relying on emojis. In the case of a prompt base, the broadcaster can complete the desired prompt by inputting or modifying a part of it.

[0109] In this case, the reaction candidate generation processing unit 172 may change the Prompt or its base depending on the situation. Alternatively, it may automatically generate reaction candidates according to the situation using a template. In this case, a variety of reaction candidates can be generated (presented) while reducing the burden on the broadcaster.

[0110] Furthermore, the reaction candidate generation processing unit 172 may generate two or more reaction candidates for a particular comment. In this case, the broadcaster can select one or more desired reaction candidates from among the two or more. In this case, the two or more reaction candidates may be comments that convey substantially different content, or they may convey substantially the same content but in different forms. In the latter case, for example, one reaction candidate may be a comment, while the other reaction candidates may be gestures that convey the same content as the comment. In this case, the likelihood of generating the broadcaster's desired reaction candidates increases, and an increase in processing load caused by rework (for example, regenerating reaction candidates based on instructions from the broadcaster) can be prevented.

[0111] Furthermore, when generating reaction candidates for an action, the reaction candidate generation processing unit 172 may consider not only the content of the action but also its form. For example, the reaction candidates for an action may be generated in different ways depending on whether the action is a comment or a gesture (a gesture that conveys substantially the same content as the comment). In this case, for example, if the action is a comment, a reaction candidate in the form of a comment may be generated, while if the action is a gesture, a reaction candidate in the form of a gesture may be generated. This allows for diversification of the reaction candidate forms according to the form of the action from the viewer, changes in the impact that reactions can have on the viewer, and as a result, can stimulate comments from viewers (and the resulting excitement of the broadcast).

[0112] Furthermore, while the reaction candidate generation processing unit 172 is generating reaction candidates, an image or sound effect indicating this may be output. This may also be the case during sentiment analysis or comment flow analysis.

[0113] By the way, in collaborative broadcasts involving multiple streamers, or in viewer-participation broadcasts, it can sometimes be difficult to determine who a comment, such as a question from a single viewer, is directed to. For example, in a collaborative broadcast between streamer A and streamer B, if a viewer mentions A (@A), it is acceptable to generate reaction candidates as a comment (or request) directed to A. On the other hand, if there is no clear mention, it is unclear whether the comment is directed to streamer A or streamer B (which one should speak), and a situation may arise where both streamers speak at the same time.

[0114] Therefore, a method like the evaluation function described above may be used to check the response to broadcasters A and B (for example, activity information related to reactions). In this case, broadcaster A, who has not generated any particular reactions based on past history, etc., may be assumed not to respond, while other broadcasters B (broadcasters who can be expected to give some kind of reaction) may be assumed to respond, and reaction candidates may be generated accordingly. Alternatively, when reaction candidates are generated, the sentiment analysis unit 166 may perform sentiment analysis on the reaction candidates and activate negative comments (low joy values) as reactions first. Furthermore, if such determination is difficult, comments with shorter text lengths among the reaction candidates may be activated (output) as reactions first.

[0115] Furthermore, the function of the reaction candidate generation processing unit 172 may be switched on or off automatically in response to instructions from the broadcaster and / or by artificial intelligence, etc. For example, the function may be turned on when the number of comments from viewers per unit time exceeds a threshold, or when the broadcaster is taking a break, etc. Such on / off switching conditions may also be set for each broadcaster. This prevents a situation where reaction candidates are automatically generated uniformly, and the processing load can be reduced by turning off the function of the reaction candidate generation processing unit 172. When the function of the reaction candidate generation processing unit 172 is turned off, the functions for acquiring various parameters used to generate reaction candidates (such as the action acquisition unit 162, the situation determination unit 164, and the sentiment analysis unit 166) may also be stopped in conjunction with this. In this case as well, the processing load can be reduced.

[0116] The reaction output unit 174 automatically or in response to a predetermined input from the broadcaster outputs reaction candidates as reactions from the broadcaster. For example, the reaction output mode may have an automatic mode and a manual mode. In automatic mode, the reaction output unit 174 outputs the reaction candidates generated by the reaction candidate generation processing unit 172 as reactions as they are. In manual mode, when one or more reaction candidates are selected by the broadcaster from among the reaction candidates generated by the reaction candidate generation processing unit 172, the reaction output unit 174 may generate a reaction instruction to output the selected reaction candidates. Also in manual mode, after the reaction candidates generated by the reaction candidate generation processing unit 172 have been modified (edited, etc.) by the broadcaster, the reaction output unit 174 may generate a reaction instruction to output the modified reaction candidates.

[0117] In this embodiment, some or all of the functions of the reaction output unit 174 may be implemented by artificial intelligence. Furthermore, the function of the reaction output unit 174 that automatically outputs reaction candidates as reactions from the broadcaster may be switched on or off automatically in response to instructions from the broadcaster and / or by artificial intelligence, etc. For example, this function may be turned on when the number of comments from viewers per unit time exceeds a threshold or when the broadcaster is taking a break, etc. The conditions for such on / off switching may also be set for each broadcaster. This prevents situations in which reactions are automatically output when the broadcaster wants to control them, or when there are long periods in which no reactions are output at all when the broadcaster is unable to attend to them.

[0118] Next, an example of the operation of the information distribution system 1 will be described with reference to Figures 21 to 23.

[0119] Figures 21 to 23 are flowcharts illustrating an example of the operation of the information distribution system 1. Figure 21 may be executed at predetermined intervals during content distribution by one distributor.

[0120] In step S2100, the information distribution system 1 acquires comments from viewers. Note that comments from viewers may be stored in a predetermined storage area as pending comments, each time they are received, through an interrupt process separate from the process shown in Figure 21. In this case, in step S2100, the pending comments may be read from the predetermined storage area one by one or in batches.

[0121] In step S2102, the information distribution system 1 updates the content being distributed with activity information related to the corresponding viewer, based on the comments obtained in step S2100.

[0122] In step S2104, the information distribution system 1 outputs the comments acquired in step S2100 to the content being distributed (see, for example, the comment display field G11 in Figure 6).

[0123] In step S2106, the information distribution system 1 assigns a comment processing order to the comments acquired in step S2100. The comment processing order represents the order in which the comments will be processed, as described later. The comment processing order is assigned sequentially from 1, in chronological order of receipt.

[0124] In step S2108, the information distribution system 1 performs comment processing on the k-th comment in the comment processing order. The initial value of k is "1". A specific example of comment processing will be described later with reference to Figure 22.

[0125] In step S2110, the information distribution system 1 increments the variable k by "1".

[0126] In step S2112, the information distribution system 1 performs a reaction process. A specific example of the reaction process will be described later with reference to Figure 23.

[0127] In Figure 21, comments are processed one by one as an example, but multiple comments may be processed in parallel using multi-core processors, etc. Furthermore, depending on the processing load (e.g., the number of comments to be processed), a process such as thinning of the comments to be processed may be performed. This allows for comment processing (step S2108) to be performed while reducing the processing load, even when the number of comments per unit time is very large.

[0128] Figure 22 is a flowchart related to the comment processing (step S2108) in Figure 21.

[0129] In step S2200, the information distribution system 1 reads activity information relating to the poster of the k-th comment (hereinafter also simply referred to as the "comment poster"). The activity information may be read from the activity information storage unit 150 described above.

[0130] In step S2202, the information distribution system 1 determines the status of the comment poster based on the activity information read in step S2200. The determination of the comment poster's status may be a determination of which of the first status, second status, etc., described above it corresponds to.

[0131] In step S2204, the information distribution system 1 determines the content or situation of the k-th comment. The content or situation of the comment is as described above.

[0132] In step S2206, the information distribution system 1 performs sentiment analysis and comments flow analysis. These analyses are performed in relation to the sentiment analysis unit 166 and the comments flow analysis unit 170, as described above. The comments flow is analyzed including the k-th comment and the comments preceding it. Alternatively, it may be determined whether the k-th comment follows the comments flow analyzed based on the comments preceding it. Sentiment analysis and comments flow analysis do not need to be performed for each comment; they may be performed for multiple comments or periodically. In this case, sentiment analysis and comments flow analysis can be performed while reducing the processing load.

[0133] In step S2208, the information distribution system 1 compares the judgment and analysis results from steps S2202 to S2206 with the configuration information related to the distributor. Through this comparison, it is possible to determine whether the judgment and analysis results from steps S2202 to S2206 represent a high-priority situation based on the configuration information.

[0134] In step S2210, the information distribution system 1 determines whether it is necessary to generate a reaction candidate for the k-th comment. For example, if the results of the matching in step S2208 indicate that the judgment and analysis results from steps S2202 to S2206 are of high priority based on the configuration information, it is determined that a reaction candidate for the k-th comment should be generated. If the judgment result is "YES", the system proceeds to step S2212; otherwise, it terminates. In the latter case, no reaction candidate is generated for the k-th comment.

[0135] In step S2212, the information distribution system 1 generates reaction candidates for the k-th comment (reaction processing). The method for generating reaction candidates is as described above.

[0136] In step S2214, the information distribution system 1 outputs reaction candidates for the k-th comment. The method for outputting reaction candidates is as described above, with reference to Figure 8, etc.

[0137] Figure 23 is a flowchart relating to the reaction process (step S2112) in Figure 21.

[0138] In step S2300, the information distribution system 1 determines whether or not a reaction instruction has been generated from the distributor. If multiple reaction candidates are presented, the reaction instruction may include information indicating which reaction candidate the instruction is for. Details of the reaction instruction are as described above.

[0139] In step S2302, the information distribution system 1 outputs a corresponding reaction in response to the reaction instruction in step S2300. The method for outputting the reaction is as described above (see, for example, comment G13 in Figure 6).

[0140] In step S2304, the information distribution system 1 updates the reaction waiting queue by deleting the reaction output in step S2302 from the reaction waiting queue. The reaction waiting queue is a queue in which reactions that can be output, as described later, are waiting. Reactions in the reaction waiting queue may be deleted even if they are not output if predetermined conditions are met. The predetermined conditions are arbitrary, but may be met, for example, when a predetermined upper limit time has elapsed or when the number of reactions in the reaction waiting queue exceeds a predetermined upper limit. In this case, the predetermined condition time and predetermined upper limit may be set for each distributor.

[0141] Although each embodiment has been described in detail above, the invention is not limited to any particular embodiment, and various modifications and changes are possible within the scope described in the claims. Furthermore, it is possible to combine all or more of the components of the embodiments described above.

[0142] For example, in the embodiments described above, emojis that can be used in various ways as described above may be given different meanings based on differences in display characteristics such as color and size. For example, emojis representing various situations may indicate priority according to their color and size. For example, a red color may indicate high priority. Similarly, a larger size may indicate high priority. Alternatively, flashing or other effects may indicate high priority. This allows the broadcaster to visually grasp the level of priority, etc., and efficiently take the necessary actions. [Explanation of Symbols]

[0143] 1. Information distribution system 3 Network 10 Server devices 20 Terminal devices 20A Terminal device 20B Terminal device 30 Studio Units 30A Studio Unit 30B Studio Unit 150 Activity information storage section 152 Configuration Information Storage Unit 154 Activity Information Management Department (1st Management Department) 156 Setting information management department (second management department) 160 Distribution Content Processing Section 162 Action Acquisition Unit (Acquisition Unit) 164 Situation Determination Unit 166 Emotion Analysis Department 168 Emoji mapping processing unit 170 Comment Flow Analysis Department 172 Reaction Candidate Generation Processing Unit (Reaction Candidate Generation Unit) 1721 First Processing Unit 1722 Second Processing Unit 174 Reaction Output Section

Claims

1. Activity information regarding activities that viewers can perform when viewing past content distributed by a broadcaster, and at least one of the past content and content currently being broadcast by the broadcaster, is managed for each viewer. Information relating to the broadcaster's reactions, where the priority of the viewer's actions or the viewer's situation is managed by setting information established by the broadcaster. The aforementioned broadcaster obtains multiple actions from one or more viewers who are watching the content currently being broadcast and manages them as activity information. Based on the activity information and the setting information, one or more of the multiple actions are identified according to the priority order. A program that causes a computer to perform a process that automatically generates candidate reactions that the distributor can take in response to the identified action.

2. The program according to claim 1, wherein the viewer's action includes at least one of the content of the action and the circumstances of the action relating to the content.

3. The program according to claim 2, wherein the content of the action relating to the content includes at least one of the following: whether or not it is a question or inquiry from the viewer, whether or not an answer should be given to the question or inquiry, and whether or not a specific keyword is included.

4. The viewer's situation includes at least one of the following: 1) A viewer who is watching content distributed by the broadcaster for the first time; 2) A viewer who provides visible or financial support exceeding a predetermined standard through Super Chat, gifts, tips, or similar means; 3) A viewer who is favorable to the broadcaster and does not fall under 1) or 2) The aforementioned computer, If the aforementioned viewer situation includes the third situation, Whether a viewer is favorably disposed towards a particular streamer is determined based on the content of the comments that viewer makes regarding the streamer's content. If the aforementioned viewer situation includes the fourth situation, The program according to claim 1, which determines whether a viewer is an undesirable viewer or a viewer hostile to a broadcaster based on the content of comments made by that viewer in response to the broadcaster's content.

5. The program according to claim 4, wherein the fourth situation is further classified into more detailed situations depending on whether or not it also applies to the second situation.

6. The program according to claim 4, further comprising a computer performing a process to determine the viewer's status based on the activity information.

7. The program according to claim 6, wherein the activity information relates to at least one of the following: the frequency of posting comments, the history of posted comments, the content of posted comments, the history of visible support or financial support, and information that can be derived based on one or more of these.

8. The program according to claim 7, wherein the process for determining the viewer's actions or the viewer's status includes applying an evaluation function to the activity information to evaluate the achievement of each viewer.

9. The program according to claim 7, further comprising a computer performing a process to output the actions of the viewer or the status of the viewer in an emoji format, in a manner that can be viewed by the broadcaster.

10. The process for generating the reaction candidates includes a first process for generating the same reaction candidates for two or more specific viewers who have a predetermined relationship, The program according to claim 7, wherein the computer determines whether the two or more specific viewers have the predetermined relationship based on whether the circumstances of the viewers are the same or common.

11. The program according to claim 10, wherein the process for generating reaction candidates selectively includes the first process and the second process for generating different reaction candidates for each of two or more specific viewers having the predetermined relationship.

12. The program according to claim 1, wherein the setting information relates to the frequency of reactions from the broadcaster.

13. The process involves analyzing viewer comments posted to the content to analyze viewer sentiment at multiple points in time during the distribution of the content, The program according to claim 1, further comprising causing a computer to perform a process of generating candidate reactions from the broadcaster based on the results of the emotion analysis.

14. The program according to claim 1, further comprising a computer performing a process to analyze the flow of comments from viewers based on the time series of each comment during the distribution of the aforementioned content.

15. The program according to claim 14, further comprising causing a computer to perform a process of generating reaction candidates from the broadcaster based on the results of the analysis of the comment flow.

16. The program according to claim 1, wherein the process for generating reaction candidates includes generating an input to a generative artificial intelligence based on the activity information, and obtaining the reaction candidates or the base of the reaction candidates by providing the input to the generative artificial intelligence.

17. The program according to claim 1, further comprising a computer performing a process of outputting the generated reaction candidates as reactions from the distributor, either automatically or in response to a predetermined input from the distributor.

18. The program according to claim 17, wherein the reaction from the broadcaster includes at least one of the following: a visible output of comments or text, an audible output of comments, and a gesture output.

19. The program according to claim 1, wherein the broadcast by the broadcaster includes at least one of the following: broadcast using an avatar relating to a human broadcaster, broadcast by a robot broadcaster, broadcast through collaboration by multiple broadcasters, and broadcast in which viewers participate.

20. The program according to claim 1, wherein the action includes at least one of a visible output of a comment and an acoustic output of a comment.

21. Activity information regarding activities that viewers can perform when viewing past content distributed by a broadcaster, and at least one of the past content and content currently being broadcast by the broadcaster, is managed for each viewer. Information relating to the broadcaster's reactions, where the priority of the viewer's actions or the viewer's situation is managed by setting information established by the broadcaster. The aforementioned broadcaster obtains multiple actions from one or more viewers who are watching the content currently being broadcast and manages them as activity information. Based on the activity information and the setting information, one or more of the multiple actions are identified according to the priority order. A computer-based information processing method, comprising automatically generating candidate reactions that the distributor can take in response to the identified action.

22. A first management unit manages activity information for each viewer regarding past content distributed by the broadcaster, and activities that viewers can perform when viewing at least one of the past content and content currently being broadcast by the broadcaster. Information relating to the broadcaster's reactions, the priority of the viewer's actions or the viewer's situation, is managed by a second management unit which manages setting information set by the broadcaster. The aforementioned distribution unit acquires multiple actions from one or more viewers who are watching the content currently being distributed by the distributioner, An information processing system including a reaction candidate generation processing unit that, based on the activity information and setting information managed by the acquisition unit, identifies one or more of the multiple actions according to the priority order, and automatically generates reaction candidates that the distributor can take for the identified actions.

Images