Information processing device, information processing method, program, and recording medium
The information processing device addresses the high cost of generating static dialogue scenarios by clustering presentation information and using related information to create efficient and cost-effective static dialogue scenarios.
Patent Information
- Application Number
- JP2025078962
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-05-09
- Publication Date
- 2026-11-19
- Estimated Expiration
- 2045-05-09
AI Technical Summary
The cost of generating static dialogue scenarios for scenario-based chatbots increases sharply with the amount of presented information, and existing technologies fail to efficiently reduce this cost.
An information processing device that acquires a group of presentation information, clusters it by referring to related information, and generates a static dialogue scenario using the clustering results, thereby reducing the effort required to create these scenarios.
The proposed solution reduces the cost and effort of generating static dialogue scenarios while ensuring an appropriate and efficient generation process.
Smart Images

Figure 2026182607000001_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to an information processing apparatus, an information processing method, a program, and a recording medium that generate a static interaction scenario.
Background Art
[0002] Automatic response systems (also referred to as chatbots) that automatically conduct interactions with users are known. Chatbots are broadly classified into a type that generates response content using a trained generation model, and a type that responds in accordance with a predefined static interaction scenario. While the former type is capable of flexible responses in accordance with inputs from users, a problem of presenting incorrect information (the so-called hallucination problem) has been pointed out. For this reason, in situations where it is necessary to convey accurate information to users, the latter type, that is, a chatbot that responds in accordance with a predefined static interaction scenario (also referred to as a scenario-type chatbot) is used.
[0003] As an example of a scenario-type chatbot, Patent Document 1 discloses a technique for generating a static interaction scenario including a partial interaction scenario specified by a label.
Prior Art Documents
Patent Documents
[0004]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0005] When using a scenario-based chatbot, it is necessary to acquire information that may be presented to the user (also called presented information) in advance and to generate static dialogue scenarios in advance using this presented information. As a result, there is a problem that the cost, including the effort required by developers to generate static dialogue scenarios, increases sharply as the amount of presented information increases, and this problem was difficult to resolve even with the technology described in Patent Document 1.
[0006] One aspect of this disclosure aims to provide a technology that can reduce the cost of generating static dialogue scenarios. [Means for solving the problem]
[0007] To solve the above problems, an information processing device according to one aspect of the present disclosure includes: an acquisition unit that acquires a group of information including a plurality of presentation information to be presented to a user; a clustering unit that performs clustering of the plurality of presentation information by referring to related information directly or indirectly associated with at least one of the plurality of presentation information; and a generation unit that generates a static dialogue scenario including the plurality of presentation information by referring to the results of clustering by the clustering unit.
[0008] To solve the above problems, an information processing method according to one aspect of the present disclosure includes an acquisition step in which one or more processors acquire a group of information including a plurality of presentation information to be presented to a user; a clustering step in which the plurality of presentation information is clustered by referring to related information directly or indirectly associated with at least one of the plurality of presentation information; and a generation step in which a static dialogue scenario including the plurality of presentation information is generated by referring to the results of the clustering in the clustering step.
[0009] Each aspect of the present invention may be implemented by a computer, in which case the program for the information processing device that enables the computer to implement the information processing device by operating the computer as each part (software element) of the information processing device, and the computer-readable recording medium on which the program is recorded, also fall within the scope of the present invention. [Effects of the Invention]
[0010] According to one aspect of this disclosure, the cost of generating static dialogue scenarios can be reduced. [Brief explanation of the drawing]
[0011] [Figure 1] This is a block diagram showing the configuration of the information processing system according to Embodiment 1. [Figure 2] This figure shows an example of information referenced by the information processing device according to Embodiment 1. [Figure 3] This is a flowchart showing an example of the processing flow executed by the information processing device according to Embodiment 1. [Figure 4] This is a diagram illustrating the processing performed by the information processing device according to Embodiment 1. [Figure 5] This is a diagram illustrating the processing performed by the information processing device according to Embodiment 1. [Figure 6] This is a diagram illustrating the processing performed by the information processing device according to Embodiment 1. [Figure 7] This is a diagram illustrating the processing performed by the information processing device according to Embodiment 1. [Figure 8] This is a diagram illustrating the processing performed by the information processing device according to Embodiment 1. [Figure 9] This is a diagram illustrating the processing performed by the information processing device according to Embodiment 1. [Figure 10] This figure shows an example of a static dialogue scenario generated by the information processing device according to Embodiment 1. [Figure 11] This is a flowchart showing a specific example of the processing flow executed by the information processing device according to Embodiment 1. [Figure 12] FIG. 3 is a flowchart illustrating a specific example of a flow of processing executed by the information processing apparatus according to Embodiment 1. [Figure 13] FIG. 4 is a diagram illustrating an example of display screen information generated by the information processing apparatus according to Embodiment 1. [Figure 14] FIG. 5 is a diagram for explaining an example implemented by the information processing apparatus according to Embodiment 1. [Figure 15] FIG. 6 is a diagram for explaining an example implemented by the information processing apparatus according to Embodiment 1. [Figure 16] FIG. 7 is a diagram for explaining an example implemented by the information processing apparatus according to Embodiment 1. [Figure 17] FIG. 8 is a diagram for explaining an example implemented by the information processing apparatus according to Embodiment 1. [Mode for Carrying Out the Invention]
[0012] [Embodiment 1] Hereinafter, one embodiment of the present disclosure will be described.
[0013] (Overview of Information Processing Apparatus 1)] First, an overview of the information processing apparatus 1 according to the present embodiment will be described. In overview, the information processing apparatus 1 according to the present embodiment is an apparatus that generates a dialogue scenario referred to by a dialogue execution unit (automatic response system, chatbot) that performs dialogue with a user. The relevant dialogue scenario may be simply referred to as "scenario" or "static dialogue scenario". Here, the "static dialogue scenario" refers to, as an example, a scenario that does not undergo changes in response to input from a user who interacts with the information processing apparatus 1. However, this does not limit the present embodiment. In addition, the information processing apparatus 1 may also be referred to as a "dialogue scenario generation apparatus" or a "scenario generation apparatus" and the like. The information processing apparatus 1 according to the present embodiment, · acquiring an information group including a plurality of pieces of presentation information to be presented to a user, - By referring to related information that is directly or indirectly associated with at least one of the aforementioned multiple pieces of presented information, clustering of the aforementioned multiple pieces of presented information is performed. • Refer to the results of the clustering described above to generate a static dialogue scenario containing the multiple pieces of presented information. This process is performed.
[0014] As described above, the information processing device 1 according to this embodiment performs clustering of the multiple presented pieces of information and generates a static dialogue scenario including the multiple presented pieces of information by referring to the clustering results, thereby reducing the cost, including the effort required by developers to generate the static dialogue scenario. Furthermore, as described above, the information processing device 1 performs clustering of the multiple presented pieces of information by referring to the related information, so an appropriate clustering result can be obtained. Therefore, according to the information processing device 1 according to this embodiment, an appropriate static dialogue scenario can be generated while reducing the cost of generating the static dialogue scenario.
[0015] (Information processing system 100) Next, the configuration of the information processing system 100, including the information processing device 1 according to this embodiment, will be described with reference to Figure 1. Figure 1 is a block diagram showing the configuration of the information processing system 100 according to this embodiment. As shown in Figure 1, the information processing system 100 includes the information processing device 1, a server device 70, and one or more terminal devices 81, 82.... Examples of the information processing device 1 include a PC (Personal Computer), a smartphone, and a tablet. The information processing device 1 may be a server that cooperates with the server device 70. Examples of terminal devices 81, 82... include a smartphone, a tablet, and a PC. However, these examples do not limit this embodiment.
[0016] The information processing device 1, the server device 70, and the terminal devices 81, 82, etc. are connected to each other via a network, as shown in Figure 1. The specific configuration of the network is not particularly limited, but as an example, a wireless LAN (Local Area Network), a wired LAN, a WAN (Wide Area Network), a public telephone network, a mobile data communication network, or a combination of these networks can be used.
[0017] (Server device 70) The server device 70 provides the information processing device 1 with various types of data that the information processing device 1 references. For example, the server device 70 provides the information processing device 1 with at least a portion of the information included in the reference information group RGI, which will be described later. However, this is not limited to this embodiment, and the information processing device 1 may be configured to acquire the reference information group RGI via its own input / output unit 40.
[0018] Furthermore, the server device 70 provides the information processing device 1 with at least a portion of the models included in the model group MD described later. However, this is not limited to this embodiment, and the information processing device 1 may also provide input information for the models to the server device 70, and the server device 70 may execute processing using the models. In this configuration, the server device 70 provides the execution results to the information processing device 1.
[0019] (Terminal devices 81, 82...) Each terminal device 81, 82... obtains user input from the user of the terminal device via its own user interface, such as a touch panel, and performs various processes based on the obtained user input. As an example, each terminal device may be configured to obtain a static dialogue scenario DS generated by the information processing device 1, and to present a response based on the static dialogue scenario DS and the user input to the user via the user interface.
[0020] Alternatively, each terminal device 81, 82... may provide the above-mentioned user input to the information processing device 1, and the information processing device 1 may be configured to generate display screen information DI including a static dialogue scenario DS and a response (presented information) based on the above-mentioned user input. In this configuration, the terminal device acquires the display screen information DI from the information processing device 1 and presents it to the user via the above-mentioned user interface.
[0021] (Information Processing Device 1) As shown in Figure 1, the information processing device 1 comprises a control unit 10, a storage unit 20, a communication unit 30, and an input / output unit 40.
[0022] (Communications Section 30) The communication unit 30 is an interface for sending and receiving data over a network. For example, the communication unit 30 acquires user input from one of the terminal devices 81, 82, etc. The communication unit 30 also provides the terminal device with display screen information DI, which includes a response (presentation information) generated based on the user input and a static dialogue scenario DS. However, this is not limited to this embodiment; the communication unit 30 may also provide the static dialogue scenario DS to each of the terminal devices 81, 82, etc. Examples of the communication unit 30 include, but are not limited to, communication chips in various communication standards such as Ethernet®, Wi-Fi®, and wireless communication standards for mobile data networks, as well as USB-compliant connectors.
[0023] (Input / output section 40) The input / output unit 40 is an interface to input devices that accept data input and output devices that output data. Examples of input devices include, but are not limited to, microphones, cameras, eye-tracking devices, keyboards, and touchpads. Examples of output devices include, but are not limited to, speakers and displays. When the information processing device 100 is used by a developer of a static dialogue scenario, the input / output unit 40 accepts input from that developer, who is different from the user of the terminal devices 81, 82, etc. This developer may also be referred to as the user of the information processing device 1.
[0024] (Storage unit 20) The memory unit 20 stores data referenced by the control unit 10, and data derived (generated) by the control unit 10, etc. As an example, the memory unit 20 stores: ·Reference information group RGI ·Generation information group GGI • Display screen information DI • Model group MD The following are stored. Examples of the storage unit 20 include, but are not limited to, flash memory, HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination thereof.
[0025] (Reference information group RGI) The reference information group RGI is data primarily referenced by the control unit 10, and as an example, • Multiple presentation information RP • Multiple related information RI • Number of options information NC This includes the following. Here, the presentation information RP included in the reference information group RGI is sometimes referred to as "referenced presentation information RP" in order to distinguish it from the presentation information GP included in the generated information group GGI, which will be described later.
[0026] Presentation Information RP is the information that is presented to the user during a conversation with the user according to a static dialogue scenario DS. The upper part of Figure 2 shows an example of presentation information RP. In the example shown in the upper part of Figure 2, the presentation information RP includes: • Title of the information presented: "How to sleep" • Description of the information presented: "It's tough when you can't sleep. Here are a few things you can try to help you sleep..." It includes.
[0027] Related information RI is information associated with presentation information RP. In this embodiment, one or more related information RIs may be associated with one presentation information RP. In the example shown in the upper part of Figure 2, three related information RIs (RI1, RI2, RI3) are associated with one presentation information RP. The lower part of Figure 2 shows an example of the configuration of related information RI1 and RI2. As shown in the lower part of Figure 2, each related information RI includes supplementary information CI and descriptive information EI. Here, supplementary information CI can be considered as the title of the related information RI. Furthermore, descriptive information EI can be considered as information that explains the supplementary information CI, or as descriptive information about the supplementary information CI. Note that the terms supplementary information and descriptive information are not limited to this embodiment.
[0028] In the example shown in the lower part of Figure 2, more specifically, related information RI1 contains: • Supplementary information CI1: "I can't sleep" • Description Information EI1: "I'm having trouble sleeping when I want to sleep, when I should sleep, when I should be sleeping." It includes. Also, related information RI2 includes: • Supplementary information CI2: "I want to sleep soundly." • Description Information EI2: "Even when I'm asleep, I feel like I'm still awake, like I'm not getting proper sleep, and it's bothering me." It includes.
[0029] The number of choices information NC specifies the number of choices (number of branches) for each branch included in the static dialogue scenario DS. The number of choices information NC is set by the generation unit 13, described later, in response to user input to the information processing device 1, which is input via the input / output unit 40, as an example.
[0030] (Generation information group GGI) The generated information group GGI is data primarily derived (generated) by the control unit 10, and as an example, • Clustering results CR ·Presentation information GP • Static dialogue scenario DS This includes the following. Here, the presentation information GP included in the generated information group GGI is sometimes referred to as "generated presentation information GP" in order to distinguish it from the presentation information RP included in the reference information group RGI mentioned above.
[0031] The clustering result CR is information indicating the result of the clustering process performed by the clustering unit 12, which will be described later. The clustering result CR includes, as an example, a hierarchical branching structure in which each of the multiple presentation information RPs mentioned above is included as a node. Specific examples of the clustering result CR will be described later.
[0032] The presented information GP is the presented information generated by the generation unit 13, which will be described later. Like the presented information RP described above, the presented information GP is the information that is presented to the user in a dialogue with the user according to the static dialogue scenario DS. Specific examples of the generated presented information GP will be described later.
[0033] The static dialogue scenario DS is a static dialogue scenario generated by the generation unit 13, which will be described later, and as an example, includes the presentation information RP and presentation information GP mentioned above. A specific example of the static dialogue scenario DS will be described later.
[0034] The display screen information DI is display information generated by the display control unit 15, which will be described later. Specific examples of the display screen information DI will be described later. The model group MD includes models for executing some of the processing of each part of the control unit 10. Such models include: • A model for feature vectorizing the information contained in the reference information set RGI. • Model for performing the summarization process described later. This includes, etc.
[0035] (Control Unit 10) As shown in Figure 1, the control unit 10 includes an acquisition unit 11, a clustering unit 12, a generation unit 13, an interaction execution unit 14, and a display control unit 15. Alternatively, the control unit 10 may be configured without the interaction execution unit 14 and the display control unit 15, with the terminal devices 81, 82, etc., each including the interaction execution unit 14 and the display control unit 15. Such a configuration is also included in this embodiment.
[0036] (Acquisition part 11) The acquisition unit 11 acquires a group of information including multiple presentation information RPs to be presented to the user. Here, the user refers, for example, to a user of any of the terminal devices 81, 82, etc., or to a user of the information processing device 1. The group of information acquired by the acquisition unit 11 refers, for example, to at least some of the information included in the reference information group RGI described above. A specific example of the acquisition process by the acquisition unit 11 will be described later.
[0037] (Clustering section 12) The clustering unit 12 performs clustering of the plurality of presentation information RPs by referring to related information RIs that are directly or indirectly associated with at least one of the plurality of presentation information RPs acquired by the acquisition unit 11.
[0038] Here, the supplementary information CI mentioned above can be cited as an example of related information RI directly associated with the presented information RP. Furthermore, the explanatory information EI associated with the supplementary information CI mentioned above can be cited as an example of related information RI indirectly associated with the presented information RP. However, how the above-mentioned information is related to each other is not limited to this embodiment, and the explanatory information EI can also be considered an example of related information RI directly associated with the presented information RP. Therefore, unless there is particular confusion, the expression "directly or indirectly associated" can also be written simply as "associated."
[0039] Furthermore, the clustering result CR, which is the result of the clustering process by the clustering unit 12, includes, for example, a hierarchical branching structure in which each of the multiple presented information RPs acquired by the acquisition unit 11 is included as a node. While the details of the clustering process by the clustering unit 12 are not limited to this embodiment, as an example, the clustering unit 12 performs a hierarchical clustering process that references the distance (similarity) between the information to be clustered. In such a hierarchical clustering process, the clustering unit 12 performs: Each presented information RP, associated with one or more related information RIs, is represented as a vector in the feature space (embedding into the feature space), • Sequentially create pairs of vectors that satisfy a predetermined rule (e.g., large / small distance between vectors (similarity)). • Generate a dendrogram containing each of the above pairs as a branch (each node) as the clustering result CR. This process is performed. Here, if the hierarchical clustering process is an algorithm that generates a tree diagram with a binary structure, the above set of vectors is a pair of vectors (a set of two vectors), and if the hierarchical clustering process is an algorithm that generates a tree diagram with a ternary structure, the above set of vectors is a set of three vectors. A specific example of the clustering process by the clustering unit 12 will be described later.
[0040] (Generation part 13) The generation unit 13 generates a static dialogue scenario DS that includes the multiple presentation information RPs by referring to the clustering result CR by the clustering unit 12. As an example, the generation unit 13 generates the static dialogue scenario DS such that the number of choices at each branch in the static dialogue scenario DS is greater than the number of branches at each branch (each node) included in the clustering result. The generation unit 13 stores the generated static dialogue scenario DS in the storage unit 20. The generated static dialogue scenario DS may be referenced by the dialogue execution unit 14, which will be described later, when performing dialogue processing with the user, or it may be provided to terminal devices 81, 82, etc., and referenced when performing dialogue processing with the user in those terminal devices. A specific example of the static dialogue scenario DS generation process by the generation unit 13 will be described later.
[0041] This invention also envisions support for disadvantages faced by minority groups, such as support for developmental disabilities. Therefore, in this invention, it is beneficial to deliberately avoid being influenced by the input of other users, so as not to suffer the disadvantage of the number of options, names (described in detail later), etc. being changed for the majority based on the access history and number of times other users, who represent the majority. In other words, the generation unit 13 can generate static dialogue scenarios without depending on the input of the user (other users).
[0042] (Dialogue execution unit 14) The dialogue execution unit 14 performs a dialogue with the user based on the static dialogue scenario DS generated by the generation unit 13. For example, the dialogue execution unit 14 obtains input from the user and determines the content of the response to the user based on the input and the static dialogue scenario DS. The content of the response includes at least one of the presentation information RP and presentation information GP included in the static dialogue scenario DS.
[0043] (Display control unit 15) The display control unit 15 generates a dialogue screen that includes the response content determined by the dialogue execution unit 14. In other words, the display control unit 15 controls the display of the dialogue screen that includes the dialogue content including the response content determined by the dialogue execution unit 14. The dialogue screen generated by the display control unit 15 may be presented, for example, via the display provided by the input / output unit 40, or it may be provided to each terminal device 81, 82, etc. via the communication unit 30.
[0044] (Main effects of the information processing device 1) Thus, the information processing device 1 according to this embodiment is • Obtain a set of information including multiple presentation information RPs to be presented to the user. - By referring to related information RI that is directly or indirectly associated with at least one of the multiple presentation information RPs, clustering of the multiple presentation information RPs is performed. • Referencing the clustering results CR, generate a static dialogue scenario DS containing the multiple presentation information RPs. This process is performed.
[0045] As described above, the information processing device 1 according to this embodiment performs clustering of the multiple presentation information RPs and generates a static dialogue scenario DS including the multiple presentation information RPs by referring to the clustering result CR. Therefore, it is possible to reduce the cost, including the effort required by developers to generate the static dialogue scenario DS. Furthermore, as described above, the information processing device 1 performs clustering of the multiple presentation information RPs by referring to the related information RI, so that an appropriate clustering result can be obtained. Accordingly, the information processing device 1 according to this embodiment can generate an appropriate static dialogue scenario DS while reducing the cost of generating the static dialogue scenario DS.
[0046] Furthermore, as mentioned above, in the information processing device 1, The clustering result CR includes a hierarchical branching structure in which each of the multiple presentation information RPs acquired by the acquisition unit 11 is included as a node. The generation unit 13 generates the static dialogue scenario DS such that the number of choices in each branch of the static dialogue scenario DS is greater than the number of branches in each branch included in the clustering result CR. This process is performed.
[0047] According to the above configuration, it is possible to generate a static dialogue scenario DS with fewer layers than the number of layers in the clustering result CR, thus preventing the number of layers in the static dialogue scenario DS from becoming excessively large. Therefore, according to the above configuration, it is possible to improve the convenience (usability) of users who engage in dialogue according to the static dialogue scenario DS.
[0048] (Processing flow by information processing device 1) Next, we will explain the processing flow by the information processing device 1, referring to Figure 3. Figure 3 is a flowchart showing the processing flow by the information processing device 1.
[0049] (Step S11) In step S11, the acquisition unit 11 acquires a group of information including multiple presentation information RPs to be presented to the user. The acquisition process by the acquisition unit 11 has been described above, so a redundant explanation will be omitted. A specific example of the acquisition process by the acquisition unit 11 will be described later.
[0050] (Step S12) In step S12, the clustering unit 12 performs clustering of the plurality of presentation information RPs by referring to related information RI that is directly or indirectly associated with at least one of the plurality of presentation information RPs acquired by the acquisition unit 11 in step S11. More specifically, the clustering unit 12 may perform clustering of the plurality of presentation information RPs by referring to one or more supplemental information CIs as related information RI associated with at least one of the plurality of presentation information RPs. Figure 4 is a diagram illustrating an example of clustering processing by the clustering unit 12. In the example shown in Figure 4, the information group (reference information group RGI) acquired by the acquisition unit 11 in step S11 includes presented information RP1, presented information RP2, ..., presented information RPn. Furthermore, presented information RP1 is associated with supplementary information CI_A and supplementary information CI_D, and presented information RP2 is associated with supplementary information CI_A, supplementary information CI_C, and supplementary information CI_K. Furthermore, presented information RPn is associated with supplementary information CI_D and supplementary information CI_F.
[0051] The clustering unit 12 features each of the presented information RP1 to RPn, referencing the related information RI (more specifically, supplementary information CI) associated with that presented information. As an example, the clustering unit 12: • Generate a feature vector V1 from the presented information RP1, supplementary information CI_A, and supplementary information CI_D. From the presented information RP2, supplementary information CI_A, supplementary information CI_C, and supplementary information CI_K, a feature vector V2 is generated, and... • Generate a feature vector Vn from the presented information RPn, supplementary information CI_D, and supplementary information CI_F. This process is performed. Here, feature vector V1 is sometimes referred to as the feature vector of the presented information RP1, feature vector V2 as the feature vector of the presented information RP2, and feature vector Vn as the feature vector of the presented information RPn.
[0052] The clustering unit 12 then performs hierarchical clustering by referring to each feature vector V1 to Vn. The lower part of Figure 4 shows the clustering unit 12. • Referencing each feature vector V1 to Vn, sequentially create pairs of vectors that satisfy a predetermined rule (e.g., large / small distance between vectors (similarity)). • Generate a tree diagram (a binary tree in the example in Figure 4) containing each of the above pairs as a branch, as the clustering result CR. This process is schematically illustrated.
[0053] (Step S13) Next, in step S13, the generation unit 13 generates a static dialogue scenario DS including the multiple presentation information RP by referring to the clustering result CR by the clustering unit 12 in step S12. The processing by the generation unit 13 in this step is, as an example, • Hierarchical search processing • Processing to generate presentation information • Static dialogue scenario generation process It includes.
[0054] (Hierarchical search process) The hierarchical search process is the process of searching for a hierarchy in the clustering result CR obtained by the clustering unit 12 where the number of choices matches the number of choices information NC. Figure 5 is a diagram for explaining the hierarchical search process in the clustering result CR obtained by the clustering unit 12 with reference to the presentation information RP (RP1 to RP22) obtained by the acquisition unit 11. Here, the clustering result CR shown in Figure 5 is a hierarchical branching structure (tree diagram) that includes each of the multiple presentation information (RP1 to RP22) obtained by the acquisition unit 11 as a node (for example, a terminal node). The horizontal axis in the tree diagram showing the clustering result CR in Figure 5 shows an index obtained by the cosine distance (similarity) and group mean between each presentation information RP.
[0055] The generation unit 13, in the tree diagram showing the clustering result CR, sequentially identifies the total number of branches (number of choices) starting from the top-level node (root node) in the tree diagram, as an example. In the example shown in Figure 5, the total number of branches (number of choices) is 2 in the hierarchy indicated by the leftmost vertical dotted line, but as the hierarchy moves to the right (towards lower nodes), the total number of branches (number of choices) changes to 3, 4, and 5. In the example shown in Figure 5, the number of choices indicated by the choice count information NC is 5, and the generation unit 13 identifies the hierarchy identified by the vertical dotted line indicating the number of choices 5 as the hierarchy of the search result. Note that each choice in the identified hierarchy may also be referred to as choices C11 to C15, as shown in Figure 5.
[0056] (Information generation process) The presentation information generation process is a process that generates presentation information GP to be attached to each option C11 to C15 in the hierarchy identified by the hierarchy search process described above. The generation unit 13 generates presentation information GP to be attached to a certain option included in a certain branch in the static dialogue scenario DS in the presentation information generation process. • In the clustering result CR, the presented information RP of nodes lower than the aforementioned option, and • Related information RI associated with the presentation information RP of the aforementioned lower node It is generated by referring to at least one of the following.
[0057] Figure 6 shows an example of the presentation information generation process executed by the generation unit 13 for the presentation information group RPG1, which includes presentation information RP1 to RP10 of nodes lower than option C11 in the clustering result CR shown in Figure 5.
[0058] In the example shown in Figure 6, for example of presentation information RP1 "Push yourself to the very last moment" included in the presentation information group RPG1, each related piece of information RI: Unable to meet deadlines I don't know how to make a schedule. • Unable to estimate the required time. The following are associated with it. Furthermore, for the information RP2 "Finish before the deadline" which is another example of the information RP included in the information group RPG1, each related information RI: Unable to meet deadlines • Forgetting the deadline • Unable to estimate the required time. I don't know how to make a schedule. These are associated. More specifically, each of these associated pieces of information RI is associated with each piece of supplemental information CI, as shown in Figure 6: Unable to meet deadlines I don't know how to make a schedule. • Forgetting the deadline And, for each supplementary information CI, the explanatory information EI: • Unable to meet deadlines for tasks such as work and homework. • Unable to meet deadlines for tasks such as work and homework. I forget deadlines for things I need to do, such as work or homework. It consists of the following. Furthermore, each of these supplementary information CIs and explanatory information EIs are also included in the presented information group RPG1.
[0059] The generation unit 13 generates presentation information GP to be attached to option C11, which is included in the hierarchy identified by the hierarchy search process described above, in a static dialogue scenario DS, by referring to the presentation information group RPG1, which includes presentation information RP1 to RP10 of nodes lower than option C11. As an example, the generation unit 13 generates presentation information GP to be attached to option C11 by referring to the explanatory information EI of each presentation information RP included in the presentation information group RPG1.
[0060] More specifically, as shown in Figure 6, the generation unit 13 generates the presentation information GP to be attached to option C11 by applying summarization processing to the explanatory information EI for each of the one or more supplementary information CIs associated with the presentation information RP1 to RP10 of the lower-level nodes. In the example shown in Figure 6, the explanatory information EI for the supplementary information CIs associated with the presentation information RP1 to RP10 of the lower-level nodes of option C11 is: • Unable to meet deadlines for tasks such as work and homework. • Unable to meet deadlines for tasks such as work and homework. I forget deadlines for things I need to do, such as work or homework. By performing a summarization process on the above, the following information GP11 is assigned to option C11: Work, homework, deadlines It is generating.
[0061] The specific examples of the summarization process performed by the generation unit 13 are not limited to this embodiment, but as an example, the summarization process can be performed by at least one of the topic model, key phrase extraction process, and generation model. Furthermore, one or more models (machine-learned models) included in the above-described model group MD can be used for the summarization process.
[0062] The generation unit 13 performs the same processing on at least one of the other options C12 to C15 in the hierarchy identified by the above-mentioned hierarchical search process. Figure 7 shows an example of the presentation information GP11 to GP15 assigned by the generation unit 13 to options C11 to C15 in the hierarchy identified by the above-mentioned hierarchical search process (denoted as hierarchy 1). As shown in Figure 7, for option C13, presentation information GP13: Missing something, understanding it, forgetting it The GP13 is generated by the generation unit 13 by referring to the explanatory information EI of each presentation information RP included in the presentation information group RPG3, which is composed of the presentation information RP12 to RP20 of the subordinate nodes of the choice C13.
[0063] On the other hand, in the example shown in Figure 7, the lower nodes of options C12, C14, and C15 each consist of one terminal node. In such a case, the generation unit 13 may use the presentation information RP11, RP21, and RP22 of the terminal node as they are, as shown in Figure 7, as the presentation information GP12, GP14, and GP15 of options C12, C14, and C15, respectively. In the example shown in Figure 7, for each of options C12, C14, and C15, the presentation information GP12, GP14, and GP15 are the presentation information RP of the lower node of each option: • List the tasks • Websites where you can search for lost and found items • Paid decluttering service It has been granted.
[0064] Figure 8 shows presentation information GP_N1, which includes presentation information GP11 to GP15 generated by the presentation information generation process, for each option included in the hierarchy (hierarchy 1) identified by the hierarchy search process described above. Presentation information GP_N1 is an example of a question included in the static dialogue scenario DS generated by the generation unit 13.
[0065] The generation unit 13 repeatedly applies the above-mentioned hierarchical search process and the above-mentioned presentation information generation process to the clustering result CR, • Identify the hierarchy that corresponds to the number of choices indicated by the number of choices information NC, • Generate presentation information GP for each option included in the identified hierarchy. The following process is performed. Figure 9 shows an example of another hierarchy (denoted as hierarchy 2) explored by the generation unit 13. The generation unit 13 also assigns the presentation information GP generated by the presentation information generation process described above to each of the choices included in these hierarchies.
[0066] (Dialogue scenario generation process) The generation unit 13 generates a static dialogue scenario DS using the results of the hierarchical search process and the presentation information generation process. Figure 10 shows an example of a static dialogue scenario DS generated by the generation unit 13. As shown in Figure 10, the generation unit 13 generates the static dialogue scenario DS such that the number of choices at each branch in the static dialogue scenario DS (e.g., 5 in the example in Figure 10) is greater than the number of branches at each branch included in the clustering result CR (e.g., 2 in the example above) by performing the hierarchical search process and the presentation information generation process. In the example shown in Figure 10, the presentation information RP1 to RP22 shown in Figure 5 etc. can be included in a scenario with 3 levels, which is a significant reduction compared to the number of levels in the clustering result CR. Generally, a larger number of levels in a scenario leads to a larger number of questions, reducing user convenience (ease of use). The above configuration improves the user convenience (ease of use) for users who engage in dialogue according to the static dialogue scenario DS.
[0067] Furthermore, the static dialogue scenario DS generated by the generation unit 13 may include a branch as part of it, which generates response content using a trained generative model (also called a dynamic scenario).
[0068] (Specific example of the processing flow by the information processing device 1) Next, with reference to Figures 11 and 12, a specific example of the processing flow by the information processing device 1 will be explained. Steps S111 to S115 shown in Figure 11 correspond to step S11 shown in Figure 3, and steps S131 to S133 shown in Figure 11 correspond to steps S12 and S13 shown in Figure 3. Also, steps S121 to S123 shown in Figure 12 correspond to step S12 shown in Figure 3.
[0069] (Step S111) First, in step S111 shown in Figure 11, the acquisition unit 11 acquires the presentation information RP and stores (records) the acquired presentation information RP in the storage unit 20 as part of the reference information group RGI. As an example, the acquisition unit 11 acquires the presentation information RP input from the user of the information processing device 1 via the input / output unit 40.
[0070] (Step S112) Next, in step S112, the acquisition unit 11, • Acceptance of selections regarding supplementary information CI and explanatory information EI to be associated with the presented information RP obtained in step S111, or • Acceptance of input regarding the above supplementary information CI and explanatory information EI The process is as follows: For example, the acquisition unit 11 presents the supplementary information CI and explanatory information EI included in the reference information group RGI at that time to the user of the information processing device 1 via the input / output unit 40. The user then selects the supplementary information CI and explanatory information EI to be associated with the presented information RP entered in step S111. If the supplementary information CI and explanatory information EI to be associated with the presented information RP are not included in the presentation via the input / output unit 40, the user inputs the supplementary information CI and explanatory information EI to be associated with the presented information RP by text input or the like.
[0071] In this step, if the user makes a "selection", proceed to step S114; if the user makes an "input", proceed to step S113.
[0072] (Step S113) In step S113, the acquisition unit 11 stores (records) the supplementary information CI and explanatory information EI entered by the user in step S112 in the storage unit 20 as part of the reference information group RGI.
[0073] (Step S114) In step S114, the acquisition unit 11 associates (links) the supplementary information CI and explanatory information EI selected or entered in step S112 with the presentation information RP acquired in step S111.
[0074] (Step S115) In step S115, the acquisition unit 11 determines whether there are other presentation information RPs that should be registered. If there are other presentation information RPs that should be registered, the unit returns to step S111; otherwise, it proceeds to step S131.
[0075] (Step S131) In step S131, the information processing device 1 obtains a static dialogue scenario generation command from the user. It also obtains the number of choices in the static dialogue scenario from the user. The obtained number of choices is stored in the storage unit 20 as choice number information NC.
[0076] (Step S132 (S12)) In step S132 (S12), the clustering unit 12 and the generation unit 13 generate a static dialogue scenario DS. The process for generating the static dialogue scenario DS has been described above, so a redundant explanation will be omitted here. Furthermore, a description of the process for generating the static dialogue scenario DS will be provided later with reference to Figure 12.
[0077] (Step S133) In step S133, the generation unit 133 outputs the static dialogue scenario DS generated in step S132 (S12). As an example, step S132 (S12) is stored in the storage unit 20 as part of the generated information group GGI.
[0078] (Details of step S132 (S12)) Figure 12 is a flowchart showing the detailed processing flow in step S132 (S12) described above.
[0079] (Step S121) In step S121, the acquisition unit 11 refers to the reference information group RGI and acquires the presentation information RP and supplementary information CI stored in steps S111 and S113.
[0080] (Step S122) In step S122, the clustering unit 12 creates feature vectors V1 to Vn from the acquired supplemental information CI, as explained with reference to Figure 4. Here, the set of feature vectors V1 to Vn is also called the feature matrix.
[0081] (Step S123) In step S123, the clustering unit 12 generates a tree diagram (clustering result CR) by referring to each feature vector V1 to Vn and performing the hierarchical clustering process described above. As the hierarchical clustering process, methods such as the cosine distance (similarity) and group average method described above are used.
[0082] (Step S1321) In step S1321, the acquisition unit 11 obtains the number of choices by referring to the number of choices information NC stored in step S131, and the generation unit 13 performs the above dialogue scenario generation process based on the number of choices and the tree diagram generated in step 123 to generate the paths of the static dialogue scenario DS. As an example, a static dialogue scenario DS including the five paths shown in Figure 10 is generated.
[0083] (Step S1322) In step S1322, the generation unit 13 generates the names (presentation information GP) of options whose paths follow a static dialogue scenario (for example, options with sub-nodes) by applying summarization processing to the explanatory information EI (or related information RI) (the above-mentioned presentation information generation processing). As explained with reference to Figure 6, these option names are automatically generated by the above-mentioned summarization processing (topic model, key phrase extraction processing, or generation model) from a list of explanatory information EI associated with presentation information that can be options, for nodes lower in the hierarchy of the specified option.
[0084] (Step S132 (Step S133)) In step S132 (step S133), the dialogue execution unit 14 records the dialogue scenario path and the names of the choices in the generated information group GGI, associating them. Here, the relationship between the dialogue scenario path, which choice name is displayed at which level and choice, is recorded in the generated information group GGI as a linked static dialogue scenario DS.
[0085] (Additional notes regarding step S1321) The "story scenario generation process based on the number of choices and the dendrogram resulting from clustering" in step S1321 may be performed as follows. • Referencing the tree diagram generated in step S123, determine the number of options so that the number of levels falls within a predetermined range. Based on the number of choices determined, a static dialogue scenario (DS) is generated.
[0086] By performing this processing, the number of options and the number of levels can be suitably determined according to the clustering results, allowing for flexible responses to changes in the presented information GP.
[0087] (Example of display screen information) Figure 13 shows an example of display screen information DI generated by the display control unit 15. The upper figure is an example of a title screen, which is an example of display screen information DI, and the lower figure is an example of an interaction screen, which is an example of display screen information DI.
[0088] The display control unit 15 of the information processing device 1 may visually present the display screen information DI to the users of each terminal device 81, 82, etc. via the input / output unit 40. As an example, the display control unit 15 may output the title DI1 "Dialogue App "Consultation Desk"" and the UI button DI2 "Start Dialogue" as the title screen of a static dialogue scenario, as shown in the upper part of Figure 13, to the upper center of the screen, and "Scenario Generation Date: 2025 / ** / ** Number of Options: 5" as clustering information DI3 to the lower right of the screen.
[0089] Furthermore, if the user selects (tap) "Start a conversation" DI2, as shown in the example of the conversation screen in the lower part of Figure 13, a speech bubble containing information GP_N1 such as "Please select what you want to know. 1. Work, homework, deadlines 2. List tasks 3. Missed, understood, forgotten 4. Websites where you can search for lost items 5. Paid tidying-up services" may be output from the left side of the screen.
[0090] Then, if the user selects (tap) "1. Work, homework, deadlines," a speech bubble with "1" may appear from the right side of the screen, followed by a speech bubble with suggested information GP_N11 appearing from the left side of the screen, such as "Please select what you want to know. 1. Time, schedule, estimate 2. Finish it all at once 3. Deadline, to-dos, forget 4. Priority, importance, deadline 5. Work on it a little bit each day."
[0091] Thus, the display control unit 15 displays a dialogue screen that includes the content of the dialogue, and the display control unit 15 may include the clustering information DI3 generated by the generation unit 13 in the dialogue screen or in a display screen other than the dialogue screen.
[0092] <Examples> Next, with reference to Figures 14 to 17, an example of the static dialogue scenario DS generation process by the information processing device 1 will be described. Figures 14 to 15 show the static dialogue scenario DS generated by the information processing device 1 in this example. Terminal nodes in the static dialogue scenario DS • Websites where you can search for lost and found items • List the tasks ... These correspond to the presented information RP (22 items in the example shown in Figures 14-15).
[0093] Furthermore, in this embodiment, the 22 pieces of information shown in Figures 14-15, and new information • Manage with a schedule • Use a planner Loss of items related to the incident The static dialogue scenario DS was updated by performing the clustering and dialogue scenario generation processes described above on the group of presented information, including the new presented information. Figures 16 and 17 show the static dialogue scenario DS obtained by this update process. As shown in Figures 16 and 17, it was confirmed that an appropriate static dialogue scenario DS including this new presented information was generated.
[0094] [Examples of implementation using software] The function of the information processing device 1 is a program that causes the device to function as a computer, and this can be realized by a program that causes each control block of the device (particularly each part included in the control unit 10) to function as a computer.
[0095] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.
[0096] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.
[0097] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.
[0098] Furthermore, each process described in the above embodiments may be performed by AI (Artificial Intelligence). In this case, the AI may operate on the control device described above, or it may operate on other devices (for example, an edge computer or a cloud server).
[0099] [summary] The information processing device according to embodiment 1 of this embodiment includes an acquisition unit that acquires an information group including a plurality of presentation information to be presented to a user, a clustering unit that performs clustering of the plurality of presentation information by referring to related information directly or indirectly associated with at least one of the plurality of presentation information, and a generation unit that generates a static dialogue scenario including the plurality of presentation information by referring to the results of clustering by the clustering unit.
[0100] The above configuration can reduce costs, including the effort required from developers to generate static dialogue scenarios.
[0101] Furthermore, the information processing device according to embodiment 2 of this embodiment is an information processing device according to embodiment 1, wherein the clustering result includes a hierarchical branching structure in which each of the plurality of presented information acquired by the acquisition unit is included as a node, and the generation unit generates the static dialogue scenario such that the number of choices in each branch of the static dialogue scenario is greater than the number of branches in each branch included in the clustering result.
[0102] The above configuration improves user convenience (ease of use) for users engaging in conversations according to static dialogue scenarios.
[0103] Furthermore, in the information processing apparatus according to embodiment 3 of this embodiment, the generation unit generates presentation information to be attached to a certain choice included in a certain branch in the static dialogue scenario by referring to at least one of the presentation information of a node lower than the certain choice and related information associated with the presentation information of the lower node in the clustering result.
[0104] With the above configuration, the names of the options in the scenario (presented information) can be generated by summarizing the information of the lower-level nodes.
[0105] Furthermore, the information processing device according to embodiment 4 of this embodiment is an information processing device according to embodiment 3, wherein the clustering unit performs clustering by referring to one or more supplementary information as related information associated with at least one of the plurality of presentation information, and the generation unit generates presentation information to be attached to a certain choice included in a certain branch in the static dialogue scenario by applying summarization processing to the explanatory information for each of the one or more supplementary information associated with the presentation information of the lower node.
[0106] Furthermore, in the information processing apparatus according to embodiment 5 of this embodiment, the generation unit performs the summarization process using at least one of the topic model, key phrase extraction process, and generation model, as described in embodiment 4.
[0107] With the above configuration, the names of the options in the scenario (presented information) can be generated by summarizing the information of the lower-level nodes.
[0108] Furthermore, in the information processing device according to embodiment 6 of this embodiment, the generation unit sets the number of choices at each branch in the static dialogue scenario according to user input, in the information processing device described in embodiment 5.
[0109] The above configuration allows you to set the number of options displayed to the user in a static dialogue scenario.
[0110] Furthermore, the information processing device according to embodiment 7 of this embodiment is an information processing device according to any one of embodiments 1 to 6, and includes a dialogue execution unit that performs dialogue with the user based on the static dialogue scenario generated by the generation unit.
[0111] With the above configuration, it is possible to perform user interactions based on static dialogue scenarios.
[0112] Furthermore, the information processing device according to embodiment 8 of this embodiment includes a display control unit that displays a dialogue screen including the content of the dialogue, in addition to the information processing device described in embodiment 7, and the display control unit includes the information related to clustering performed by the generation unit in the dialogue screen or a display screen other than the dialogue screen.
[0113] With the above configuration, the dialogue screen, including the content of the conversation, and information about clustering can be displayed to users using static dialogue scenarios.
[0114] Furthermore, the information processing method according to aspect 9 of this embodiment includes an acquisition step in which one or more processors acquire an information group including a plurality of presentation information to be presented to a user; a clustering step in which the plurality of presentation information is clustered by referring to related information directly or indirectly associated with at least one of the plurality of presentation information; and a generation step in which a static dialogue scenario including the plurality of presentation information is generated by referring to the results of the clustering by the clustering unit.
[0115] The above configuration provides the same effects as in Embodiment 1.
[0116] Furthermore, the program according to embodiment 10 of this embodiment is a program for causing a computer to function as an information processing device as described in embodiment 1, and causes the computer to function as the acquisition unit, the clustering unit, and the generation unit.
[0117] The above configuration provides the same effects as in Embodiment 1.
[0118] Furthermore, the computer-readable recording medium according to embodiment 11 of this embodiment records the program described in embodiment 10.
[0119] The above configuration provides the same effects as in Embodiment 1.
[0120] This disclosure is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. [Explanation of symbols]
[0121] 100 Information Processing Systems 1. Information Processing Device 10 Control Unit 11 Acquisition Department 12. Clustering section 13 Generation part 14 Dialogue Execution Unit 15 Display Control Unit 20 Memory section 30 Communications Department 40 Input / output section 70 Server Devices 81, 82… Terminal devices
Claims
1. An acquisition unit that acquires a group of information containing multiple pieces of information to be presented to the user, A clustering unit that performs clustering of the multiple presented pieces of information by referring to related information that is directly or indirectly associated with at least one of the multiple presented pieces of information, A generation unit generates a static dialogue scenario including the multiple pieces of presented information by referring to the clustering results by the clustering unit. An information processing device equipped with the following features.
2. The clustering result includes a hierarchical branching structure in which each of the multiple pieces of presented information acquired by the acquisition unit is included as a node. The generation unit generates the static dialogue scenario such that the number of choices in each branch of the static dialogue scenario is greater than the number of branches in each branch included in the clustering result. The information processing apparatus according to claim 1.
3. The generating unit is The presentation information to be attached to a certain choice included in a certain branch in the aforementioned static dialogue scenario, In the clustering results, the presented information of nodes lower than a certain option, and Related information associated with the presentation information of the lower node Generated by referring to at least one of the following The information processing apparatus according to claim 2.
4. The clustering unit, The clustering is performed by referring to one or more pieces of supplementary information as related information associated with at least one of the aforementioned multiple pieces of presented information. The generating unit is The presentation information to be assigned to a certain choice included in a certain branch in the static dialogue scenario is generated by applying summarization processing to the explanatory information for each of the one or more supplemental pieces of information associated with the presentation information of the lower node. The information processing apparatus according to claim 3.
5. The generation unit performs the summarization process using at least one of the topic model, key phrase extraction process, and generation model. The information processing apparatus according to claim 4.
6. The generation unit sets the number of choices at each branch in the static dialogue scenario according to user input. The information processing apparatus according to claim 5.
7. The system includes a dialogue execution unit that performs a dialogue with the user based on the static dialogue scenario generated by the generation unit. An information processing apparatus according to any one of claims 1 to 6.
8. The system includes a display control unit that displays a dialogue screen including the content of the dialogue, The display control unit includes the clustering information performed by the generation unit in the dialogue screen or other display screens. The information processing apparatus according to claim 7.
9. One or more processors, An acquisition step to obtain a group of information containing multiple pieces of information to be presented to the user, A clustering step of clustering the multiple presentation pieces by referring to related information that is directly or indirectly associated with at least one of the multiple presentation pieces, A generation step that generates a static dialogue scenario including the multiple pieces of presented information by referring to the clustering results in the clustering step, An information processing method that includes this.
10. A program for causing a computer to function as an information processing device according to claim 1, wherein the computer functions as the acquisition unit, the clustering unit, and the generation unit.
11. A computer-readable recording medium having the program described in claim 10 recorded on it.
Citation Information
Patent Citations
Dialog scenario production method, dialog scenario production device and program for producing dialog scenario
JP2004310628A
