Information presentation device, information presentation method, and information presentation program

Abstract

An information presentation device according to the present invention acquires non-verbal information pertaining to a first user or the surroundings of the first user, extracts characteristic objects from the acquired non-verbal information, and creates first information indicating a relationship between the objects in the form of a graph or vectors. The information presentation device extracts characteristic matters from the profile of a second user who is a conversation partner of the first user, and creates second information indicating a relationship between the extracted matters in the form of a graph or vectors. Then, the information presentation device extracts a word representing an object the distance of which from a matter included in the second information is not more than a prescribed threshold value, from among the objects included in the first information, and outputs the word.

Description

Information presentation device, information presentation method, and information presentation program

[0001] The present invention relates to an information presentation device, an information presentation method, and an information presentation program used to generate utterances for a digital human agent.

[0002] Traditionally, digital human agents have been developed that can autonomously converse based on user voice or text inputs. Most of these agents generate responses to user utterances using pre-trained individual models or large language models (LLMs).

[0003] Gemini: A Family of Highly Capable Multimodal Models, Gemini Team, Google, [online], [Retrieved April 9, 2024], Internet, <URL: https: / / arxiv.org / pdf / 2312.11805.pdf> YOLO-World, Tianheng Cheng et al., [online], [Retrieved April 9, 2024], Internet, <URL: https: / / www.yoloworld.cc / > SIMMC 2.0: A Task-oriented Dialog Dataset for Immersive Multimodal Conversations, Satwik Kottur et al., [online], [Retrieved April 9, 2024], Internet, <https: / / arxiv.org / pdf / 2104.08667.pdf> MediaGnosis, [online], [Retrieved April 9, 2024], Internet, <https: / / www.rd.ntt / mediagnosis / >

[0004] However, when interacting with a user who is shy at first meeting or who has difficulty in self-disclosure, sufficient information about the user cannot be obtained from the user's linguistic information alone. As a result, the interaction with the user remains routine, resulting in a lack of high quality and intimacy. Therefore, the present invention aims to solve the above-mentioned problem and provide high-quality interaction even for users who have difficulty in self-disclosure.

[0005] In order to solve the above-mentioned problems, the present invention is characterized by comprising a first information creation unit that acquires non-verbal information about a first user or the first user's surroundings, extracts characteristic objects from the acquired non-verbal information, and creates first information that shows the relationships between the objects in a graph or vector form; a second information creation unit that extracts characteristic items from the profile of a second user who speaks to the first user, and creates second information that shows the relationships between the items in a graph or vector form; and a filtering unit that outputs words that represent objects from among the objects included in the first information whose distance from the items included in the second information is less than a predetermined threshold.

[0006] According to the present invention, it is possible to provide a high-quality dialogue even to a user who is not good at self-disclosure.

[0007] FIG. 1 is a diagram illustrating an example of a dialogue between agents in a virtual space. FIG. 2 is a diagram illustrating an example of a dialogue between agents in a virtual space and an example of a dialogue between a user and an agent in a physical space. FIG. 3 is a diagram illustrating an example of a dialogue system including an information presentation device. FIG. 4 is a diagram illustrating an outline of the operation of the information presentation device. FIG. 5 is a diagram illustrating an example of converting non-verbal information into text and a graph. FIG. 6 is a diagram illustrating an example of converting profile information into text and a graph. FIG. 7 is a diagram illustrating an example in which a graph of non-verbal information and a graph of agent profile information are arranged on a common sense graph. FIG. 8A is a diagram illustrating an example of a configuration of an information processing device according to each embodiment. FIG. 8B is a diagram illustrating an example of non-verbal information. FIG. 9 is a flowchart illustrating an example of a processing procedure executed by the information presentation device according to the first embodiment. FIG. 10 is a diagram illustrating an example of converting non-verbal information into text and a vector. FIG. 11 is a diagram illustrating an example of converting profile information into text and a vector. FIG. 12 is a diagram illustrating an example of calculating the distance between a word in non-verbal information and a word in an agent's profile information. FIG. 13 is a flowchart illustrating an example of a processing procedure executed by the information presentation device according to the second embodiment. FIG. 14 is a diagram illustrating an example of a computer that executes an information presentation program.

[0008] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a description will be given of an embodiment of the present invention with reference to the drawings, but the present invention is not limited to the embodiment.

[0009] [Overview] First, an overview of the information presentation device of this embodiment will be described with reference to Fig. 1. The information presentation device outputs, for example, a topic (e.g., words used in the topic) when agent A of user A converses with agent B of user B in the virtual space shown in Fig. 1.

[0010] For example, the information presentation device acquires non-verbal information of agent B (e.g., information about agent B or the environment around agent B). Then, the information presentation device refers to profile information of agent A (information indicating agent A's personality, thoughts, etc.) and extracts non-verbal information that is important to agent A from the non-verbal information of agent B. Then, the information presentation device outputs words that represent the extracted non-verbal information as words to be used as topics of conversation between agent A and agent B. This allows agent A to have an intimate conversation with agent B.

[0011] For example, in the example shown in Fig. 2, if the information presentation device extracts "hair" as non-verbal information important to the agent, the agent can say to the conversation partner, "Your hairstyle is cute today!", etc. Also, if the information presentation device extracts "XX's mark" as non-verbal information important to the agent, the agent can say to the conversation partner, "You're a fan of XX, aren't you?", etc.

[0012] The information output by the information presentation device may be used for dialogue between agents in a virtual space, as described above, or may be used for dialogue between a user and an agent in a physical space.

[0013] The information presentation device 10 is applied to, for example, the dialogue system shown in Fig. 3. The dialogue system includes the information presentation device 10 and a dialogue device. The dialogue device includes a dialogue simulation unit that simulates dialogue between agents, a human DTDB (human digital twin database) that stores information indicating the personality and thoughts of users (agents), and a visualization unit that creates a video of the dialogue between the agents.

[0014] In the above-described dialogue system, the information presentation device 10 acquires information (profile information) indicating the personality and thoughts of the user (agent) engaging in dialogue from the dialogue device's human DTDB. The information presentation device 10 also acquires non-verbal information of the agent's dialogue partner from a visualization unit or an external device. The information presentation device 10 then extracts non-verbal information important to the agent based on the acquired information. The information presentation device 10 then outputs words highly associated with text representing the extracted non-verbal information to the dialogue simulation unit as words to be used as topics of dialogue.

[0015] By the information presentation device 10 outputting words to be used in the topic of conversation in the above-described manner, the conversation device can provide a high-quality conversation even with a conversation partner who is not good at self-disclosure.

[0016] Next, an overview of the operation of the information presentation device 10 will be described with reference to FIG. 4. For example, when the information presentation device 10 acquires non-verbal information of the conversation partner of agent A, it converts objects included in the non-verbal information into text and graphs or vectorizes the objects in the non-verbal information. For example, as shown in FIG. 5, the information presentation device 10 converts an image of a scene viewed by the conversation partner into text using a caption system. Then, the information presentation device 10 applies LLM to the text to graph the objects included in the scene.

[0017] Returning to the explanation of Fig. 4, the information presentation device 10 also refers to the common sense DB and graphs or vectorizes the items (text) included in the profile information of agent A. For example, the information presentation device 10 graphs the items included in the profile by applying LLM to the profile information of agent A, as shown in Fig. 6.

[0018] Returning to the explanation of Figure 4, the information presentation device 10 then refers to a common sense database (a database showing the semantic similarity between words and the association between words; details will be given later) and compares the graph or vector of the objects included in the non-verbal information with the graph or vector of the items included in agent A's profile. Then, using the results of this comparison, the information presentation device 10 selects from the objects included in the non-verbal information objects whose distance to items included in agent A's profile (for example, name) is equal to or less than a predetermined threshold and outputs them to agent A as important non-verbal information (performs a filtering process).

[0019] For example, as shown in Fig. 7, the information presentation device 10 arranges a graph of objects included in the non-verbal information and a graph of items included in agent A's profile information on a graph (common sense graph) showing the associations between words in the common sense DB. Then, the information presentation device 10 extracts, from the nodes of the non-verbal information, nodes whose distance from a node of agent A's profile information (e.g., a node for the name "Audrey") is equal to or less than a predetermined threshold (e.g., 2) as non-verbal information important to agent A. For example, the information presentation device 10 extracts "Shibuya Station" and "Billboard" shown in Fig. 7.

[0020] The information display device 10 may output the above-mentioned "Shibuya Station" and "Billboard" themselves as words used in the conversation topic, or may output words related to "Shibuya Station" and "Billboard."

[0021] For example, the information presentation device 10 may output nodes (e.g., "Tokyo" and "design") that are located in the direction of the group of nodes in agent A's profile information, among the nodes whose distance from "Shibuya Station" and "Billboard" is equal to or less than a predetermined threshold (e.g., 1) on the graph shown in Fig. 7. This allows the information presentation device 10 to output new topic words related to important non-verbal information to agent A. As a result, the dialogue device can more easily develop the topic of dialogue, thereby improving the quality of the dialogue.

[0022] In the following embodiments, the information display device 10 outputs the new topic words as words to be used in the conversation topic.

[0023] First Embodiment Next, a description will be given of an information presentation device 10 according to a first embodiment. The information presentation device 10 according to the first embodiment performs filtering processing by graphing non-verbal information and agent profile information.

[0024] 8A, a configuration example of the information presentation device 10 will be described. The information presentation device 10 includes, for example, an input / output unit 11, a storage unit 12, and a control unit 13.

[0025] The input / output unit 11 is an interface that controls the input and output of various data, and receives inputs such as profile information of an agent and non-verbal information of a conversation partner of the agent.

[0026] As described above, the agent profile information is information indicating the agent's profile, thoughts of the agent, and the like.

[0027] The non-verbal information is information that indicates the user and the environment around the user, such as the user's physical characteristics, information about objects around the user, etc. The non-verbal information is input to the information presentation device 10, for example, in the form of a screen capture (image) of the environment around the agent's conversation partner, text showing a list of objects around the conversation partner, etc.

[0028] For example, the non-verbal information input to the information presentation device 10 may be a screen capture of the gaze of the conversation partner in the virtual space (see reference numeral 801 in Figure 8B), a list of objects around the conversation partner and information indicating the positions of the objects (see reference numeral 802 in Figure 8B), etc.

[0029] Returning to the explanation of Fig. 8A, the storage unit 12 stores data, programs, etc. that are referenced when the control unit 13 executes various processes. The storage unit 12 is realized by a semiconductor memory element such as a random access memory (RAM) or a flash memory, or a storage device such as a hard disk or an optical disk.

[0030] For example, the storage unit 12 includes a common sense DB. The common sense DB is realized by a common sense graph, such as ConceptNet, which represents the associations between words in a graph. The storage unit 12 also stores non-verbal information received by the input / output unit 11, agent profile information, and the like.

[0031] The control unit 13 is responsible for overall control of the information presentation device 10. The functions of the control unit 13 are realized, for example, by a central processing unit (CPU) executing a program stored in the storage unit 12.

[0032] The control unit 13 includes, for example, a first information creation unit 131 , a second information creation unit 134 , and a filtering unit 135 .

[0033] [First Information Creation Unit] The first information creation unit 131 acquires non-verbal information of a conversation partner, extracts characteristic objects from the acquired non-verbal information, and creates a first graph (a graph of non-verbal information) that graphs the relationships between the objects in the non-verbal information. The first information creation unit 131 includes a text conversion unit 132 and an information creation unit 133.

[0034] [Text Conversion Unit] The text conversion unit 132 converts input non-verbal information into text. For example, if the input non-verbal information is a photograph or screen capture of a physical space or a virtual space, the text conversion unit 132 converts the scenery into text using an automatic caption generation system or the like. Also, for example, if the input non-verbal information is the name of a room in a virtual space, the time, and the names and location information of surrounding objects, the text conversion unit 132 converts the scenery around the conversation partner into text based on this information. The text conversion unit 132 then outputs the converted information.

[0035] [Information Creation Unit] The information creation unit 133 extracts characteristic objects from the non-language information that has been converted into text, and creates a graph of the non-language information that graphs the relationships between the objects in the non-language information.

[0036] For example, the information creation unit 133 uses a learning model for understanding text (for example, ChatGPT) to extract topic words from the non-verbal information converted into text, express them as nodes, and create information (a graph of non-verbal information) that graphs the relationships between the nodes. For example, the information creation unit 133 creates the graph shown in FIG. 5.

[0037] [Second Information Creation Unit] Returning to the explanation of Fig. 8A, the second information creation unit 134 extracts characteristic items from the profile information of the agent, and creates a second graph (a graph of profile information) that graphically represents the relationships between the items in the profile information.

[0038] For example, the second information creation unit 134 uses a learning model for understanding text to extract characteristic items from the profile information, express them as nodes, and create information (a profile information graph) that graphs the relationships between the nodes, similar to the information creation unit 133. For example, the second information creation unit 134 creates the graph shown in FIG.

[0039] [Filtering Unit] Returning to the explanation of Figure 8A, the filtering unit 135 refers to the common sense graph and extracts, from among the objects included in the non-verbal information graph, words of objects whose distance from items included in the profile information graph is equal to or less than a predetermined threshold, as important words. The filtering unit 135 then extracts and outputs nodes whose distance from the important words is within the predetermined threshold and that are located in the direction of the profile information nodes, as words used in the conversation topic.

[0040] For example, the filtering unit 135 receives the common sense graph, the non-verbal information graph, the profile information graph, and the threshold value as input, and executes the following processing.

[0041] In addition, ・Common sense graph: G cs =(Vc s ;Ecs ) ・Graph of non-verbal information: G env =(V env ;E env ) Profile information graph: G ag =(V ag ;E ag )

[0042] The filtering unit 135 plots the relationships (edges) of the nodes of the graph of non-verbal information and the graph of profile information on a common sense graph, and calculates G' cs Create ({G env, G ag}⊆G' cs}).

[0043] Then, the filtering unit 135 calculates the above G' cs The filtering unit 135 then flags pairs of nodes (words) whose inter-node distance is equal to or less than the threshold value as being "important."

[0044] For example, the filtering unit 135 flags nodes in the graph of non-verbal information that are less than a threshold distance from a pre-specified node in the graph of profile information (for example, a node indicating a name) as "important."

[0045] The distance between nodes is: - Infinite (∞) if there is no edge between nodes - 0 if a node in the non-verbal information graph matches a node in the profile information graph - [0, ..., n, ∞], n ≤ |V(G'cs)|.

[0046] Furthermore, when there are multiple types of relationships between nodes, the filtering unit 135 may reflect the types of relationships between the nodes in setting the weights of the edges connecting the nodes. Furthermore, the filtering unit 135 may calculate the number of edges with the shortest distance between the nodes as the distance between the nodes.

[0047] The filtering unit 135 extracts and outputs, as words to be used as topics of conversation, the words of nodes whose distance from a node flagged as "important" (a node of an important word) is equal to or less than a predetermined threshold (for example, 1) and which are located in the direction of the group of nodes in the graph of profile information.

[0048] 9, an example of a processing procedure executed by the information presentation device 10 will be described. First, the first information creation unit 131 acquires non-verbal information of the conversation partner (S1) and converts it into text (S2). Then, the first information creation unit 131 creates a graph of the non-verbal information from the non-verbal information converted into text in S2 (S3).

[0049] The second information creation unit 134 also acquires profile information of the agent (S4), and creates a graph of the profile information from the profile information acquired in S4 (S5).

[0050] Then, the filtering unit 135 refers to the common sense graph and extracts, from among the objects included in the graph of non-verbal information, objects whose distance from the items included in the graph of profile information is equal to or less than a predetermined threshold, as important words (S6).

[0051] After S6, the filtering unit 135 extracts and outputs, as words to be used as topics of conversation, the words of nodes that are located in the direction of the group of nodes in the profile information graph, among the nodes whose distance from the node of the important word extracted in S6 is equal to or less than a predetermined threshold (for example, 1) (S7).

[0052] By executing the above-described process, the information presentation device 10 can provide a high-quality conversation even with a conversation partner who is not comfortable disclosing himself / herself.

[0053] Second Embodiment Next, an information presentation device 10 according to a second embodiment will be described. The information presentation device 10 according to the second embodiment vectorizes non-verbal information and agent profile information and performs filtering processing.

[0054] For the above vectorization, the information presentation device 10 creates a common sense vector that represents the semantic similarity between words in the common sense DB as a vector. The common sense vector is created using, for example, Gensim word2vec, doc2vec, etc. The created common sense vector is stored in the storage unit 12. The same components as those in the first embodiment are designated by the same reference numerals, and their description will be omitted.

[0055] [Information Creation Unit] When the information creation unit 133 converts non-language information into text, it refers to common sense vectors and creates information by vectorizing words included in the converted non-language information into text.

[0056] For example, the information creation unit 133 extracts topic words from the non-verbal information converted into text, and creates information (non-verbal information vectors) by vectorizing the extracted words using common sense vectors. Note that Fig. 10 shows an example of a two-dimensional representation of the non-verbal information vectors.

[0057] [Vectorization] The vectorization of non-verbal information will be described in detail below. Upon receiving input of the text of non-verbal information and the common sense vector, the information creation unit 133 executes the following process.

[0058] In addition, let V be the common sense vector. cs Let each word vector in the common sense vector be w cs Expressed as V cs ={w cs_1 ,w cs_2 ,…,w cs_n}.

[0059] First, the information creation unit 133 checks whether the words included in the text of the non-verbal information are already included in the common sense vector. Then, the information creation unit 133 adds the words included in the text of the non-verbal information that are not included in the common sense vector to V. cs For example, the information creating unit 133 converts the words that need to be added into an embedding representation using a technique such as a one-hot vector, and adds them to V cs Add to.

[0060] The common sense vector with the added word is V' cs Let V' be the common sense vector. csLet w' be the vector of each word in cs Expressed as V' cs ={w' cs_1 ,w' cs_2 ,…,w' cs_(n+m)}, where m is the number of added words. cs Both V′ and V′ are stored in the storage unit 12.

[0061] Thereafter, the information creating unit 133 creates V' cs With reference to the vector of non-verbal information V env ={w env_1 ,w env_2 ,…,w env_nenv Then, the information creating unit 133 creates a vector V of non-language information. env The information creating unit 133 outputs the vector V' of non-language information. env When creating the vector, a technique such as one-hot vector may be used.

[0062] [Second Information Creation Unit] The second information creation unit 134 also refers to the common sense vector and creates information in which words of items included in the profile information of the agent are vectorized.

[0063] For example, the second information creation unit 134 extracts words representing characteristic items from the agent's profile information, and creates information (profile information vectors) by vectorizing the extracted words using common sense vectors. Note that Fig. 11 shows an example of a two-dimensional representation of the profile information vectors.

[0064] For example, the second information creation unit 134 executes the same process as the information creation unit 133 described above, and obtains V' cs Refer to the profile information vector V ag ={w ag_1 ,w ag_2 ,…,w ag_nag Then, the second information creation unit 134 creates a vector V ag The second information creation unit 134 outputs the profile information vector V ag When creating the vector, a technique such as one-hot vector may be used.

[0065] [Filtering Unit] The filtering unit 135 filters the non-verbal information vector V env , a vector of profile information V ag ,The filtering process is performed using a distance threshold.

[0066] For example, the filtering unit 135 may filter the non-language information vector V env element of {w env_1 ,w env_2 ,…,w env_nenv} and a vector of profile information V ag element of {w ag_1 ,w ag_2 ,…,w ag_nag} are plotted in the same vector space, and the distance between the elements of the vector of the non-verbal information and the elements of the vector of the profile information is measured. For example, the Euclidean distance is used to measure the distance.

[0067] The filtering unit 135 then flags elements (words) of the vector of non-language information that are at a distance equal to or less than a threshold value from the vector element of the profile information as "important."

[0068] For example, the filtering unit 135 calculates the distance between the elements of the vector of the non-language information and the elements of the vector of the profile information, as shown in Fig. 12. Then, the filtering unit 135 extracts elements (e.g., "Shibuya" and "Billboard") from the elements of the vector of the non-language information whose distance from the elements of the vector of the profile information is equal to or less than a threshold value (e.g., 0.2), and marks them with a flag indicating "important."

[0069] Then, the filtering unit 135 extracts and outputs words (e.g., "Shibuya station" and "Billboard design") whose distance from the above "Shibuya" and "Billboard" is less than a predetermined threshold (e.g., 0.1) from among the elements of the vector of non-verbal information and the elements of the vector of profile information as words to be used as topics of conversation.

[0070] [Example of Processing Procedure] An example of processing procedure executed by the information presentation device 10 will be described using Fig. 13. The processes of S11 and S12 in Fig. 13 are the same as S1 and S2 in Fig. 9, so the description will begin with S13 in Fig. 13. The first information creation unit 131 creates a vector of non-verbal information from the non-verbal information converted into text in S12, with reference to the common sense vector (S13).

[0071] The second information creation unit 134 also acquires profile information of the agent (S14), and then references the common sense vector to create a vector of the profile information acquired in S14 (S15).

[0072] Thereafter, the filtering unit 135 calculates the distance between the elements of the vector of non-verbal information created in S13 and the elements of the vector of profile information created in S15 (S16).

[0073] Then, the filtering unit 135 extracts, from among the elements contained in the vector of non-verbal information, words of elements whose distance from the elements of the vector of profile information is less than a predetermined threshold, as important words of the agent (S17: Extraction of important words).

[0074] After S17, the filtering unit 135 extracts and outputs, from among the elements of the vector of non-verbal information and the elements of the vector of profile information, words whose distance from the important words extracted in S17 is less than a predetermined threshold, as words to be used in the topic of conversation (S18).

[0075] By executing the above-described process, the information presentation device 10 can provide a high-quality conversation even with a conversation partner who is not comfortable disclosing himself / herself.

[0076] In other words, the information presentation device 10 can extract information highly relevant to the user's personality and preferences from non-verbal information about the user and the user's surroundings and use that information to generate a dialogue. This makes it possible to have a high-quality dialogue tailored to the user, even for a user who is not good at self-disclosure, based on the user's non-verbal information.

[0077] [Other Embodiments] Note that the filtering unit 135 in each embodiment may use a score representing the distance between words (for example, the shorter the distance, the lower the score) when extracting important words. In this case, the filtering unit 135 may extract all words with scores equal to or less than a predetermined threshold, or may extract a predetermined number of words in ascending order of scores. Furthermore, the filtering unit 135 may also use a score representing the distance between words in the same manner as above when extracting words used in conversation topics using the distance from important words.

[0078] Furthermore, the information presentation device 10 of each embodiment may further include an utterance generation unit (e.g., a dialogue simulation unit in FIG. 3 ) that generates an utterance from the agent to the dialogue partner based on the words output from the filtering unit 135. For example, the utterance generation unit generates and outputs an utterance from the agent using the words output from the filtering unit 135 as a topic.

[0079] [System Configuration, etc.] The components of each unit shown in the figure are conceptual functional units and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown, and all or part of them can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc. Furthermore, all or any part of the processing functions performed by each device can be realized by a CPU and a program executed by the CPU, or can be realized as hardware using wired logic.

[0080] Furthermore, among the processes described in the above embodiments, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using a known method.In addition, the information including the processing procedures, control procedures, specific names, various data and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified.

[0081] [Program] The information presentation device 10 can be implemented by installing a program (information presentation program) as package software or online software on a desired computer. For example, by executing the program on an information processing device, the information processing device can function as the information presentation device 10. The information processing device referred to here includes mobile communication terminals such as smartphones, mobile phones, and PHS (Personal Handyphone Systems), as well as terminals such as PDAs (Personal Digital Assistants).

[0082] 14 is a diagram showing an example of a computer that executes an information presentation program. The computer 1000 includes, for example, a memory 1010 and a CPU 1020. The computer 1000 also includes a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

[0083] The memory 1010 includes a read-only memory (ROM) 1011 and a random access memory (RAM) 1012. The ROM 1011 stores a boot program such as a basic input / output system (BIOS). The hard disk drive interface 1030 is connected to a hard disk drive 1031. The disk drive interface 1040 is connected to a disk drive 1041. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1041. The serial port interface 1050 is connected to a mouse 1110 and a keyboard 1120, for example. The video adapter 1060 is connected to a display 1130, for example.

[0084] The hard disk drive 1031 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. That is, the programs that define the processes executed by the information presentation device 10 are implemented as program modules 1093 in which computer-executable code is written. The program modules 1093 are stored, for example, in the hard disk drive 1031. For example, the program modules 1093 for executing processes similar to those of the functional configuration of the information presentation device 10 are stored in the hard disk drive 1031. The hard disk drive 1031 may be replaced with an SSD (Solid State Drive).

[0085] Data used in the processing of the above-described embodiment is stored as program data 1094 in, for example, the memory 1010 or the hard disk drive 1031. The CPU 1020 then reads the program module 1093 or the program data 1094 stored in the memory 1010 or the hard disk drive 1031 into the RAM 1012 as necessary and executes them.

[0086] The program module 1093 and program data 1094 may not necessarily be stored in the hard disk drive 1031, but may also be stored in, for example, a removable storage medium and read by the CPU 1020 via the disk drive 1041 or the like. Alternatively, the program module 1093 and program data 1094 may be stored in another computer connected via a network (such as a local area network (LAN) or a wide area network (WAN)). The program module 1093 and program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

[0087] REFERENCE SIGNS LIST 10 Information presentation device 11 Input / output unit 12 Storage unit 13 Control unit 131 First information creation unit 132 Text conversion unit 133 Information creation unit 134 Second information creation unit 135 Filtering unit

Claims

1. An information presentation device comprising: a first information creation unit that acquires non-verbal information about a first user or the first user's surroundings, extracts characteristic objects from the acquired non-verbal information, and creates first information that shows the relationships between the objects in a graph or vector form; a second information creation unit that extracts characteristic items from the profile of a second user who will be speaking to the first user, and creates second information that shows the relationships between the items in a graph or vector form; and a filtering unit that extracts objects from the objects included in the first information whose distance to the items included in the second information is less than a predetermined threshold, and outputs words that represent the objects.

2. The information presentation device according to claim 1, characterized in that the non-verbal information is information about the physical characteristics of the first user or information about objects present around the first user.

3. The information presentation device described in claim 1, characterized in that the first information creation unit converts the acquired non-verbal information into text, extracts words that represent characteristic objects from the converted non-verbal information, and creates the first information by graphing the relationships between the words or vectorizing the semantic closeness between the words; the second information creation unit extracts words that represent characteristic matters from the profile of the second user, and creates the second information by graphing the relationships between the words or vectorizing the semantic closeness between the words; and the filtering unit refers to a database that indicates the semantic closeness or relevance between words, and calculates the distance between the words included in the first information and the words included in the second information, thereby calculating the distance between the object included in the first information and the matter included in the second information.

4. The information presentation device according to claim 3, wherein the filtering unit extracts and outputs words from among the words contained in the first information and the words contained in the second information whose distance from the word representing the object is equal to or less than a predetermined threshold.

5. The information presentation device according to claim 1, further comprising an utterance generation unit that generates an utterance from the second user to the first user based on the words output from the filtering unit.

6. An information presentation method executed by an information presentation device, comprising the steps of: acquiring non-verbal information about a first user or the first user's surroundings, extracting characteristic objects from the acquired non-verbal information, and creating first information that shows the relationships between the objects in a graph or vector form; extracting characteristic items from the profile of a second user who will be speaking to the first user, and creating second information that shows the relationships between the items in a graph or vector form; and extracting and outputting words that represent objects whose distance from the items included in the second information is less than a predetermined threshold from the objects included in the first information.

7. An information presentation program for causing a computer to execute the steps of: acquiring non-verbal information about a first user or the first user's surroundings, extracting characteristic objects from the acquired non-verbal information, and creating first information that shows the relationships between the objects in a graph or vector form; extracting characteristic items from the profile of a second user who will be speaking to the first user, and creating second information that shows the relationships between the items in a graph or vector form; and extracting and outputting words that represent objects whose distance from the items included in the second information is equal to or less than a predetermined threshold from the objects included in the first information.

Images