JP7429926B2 - Communication support devices, methods and programs - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Act 1. December 1, 2019 - December 5, 2019 (Publication date: December 3, 2019) Published at OzCHI2019-31st Australian Conference on Human-Computer-Interaction

Embodiments of the present invention relate to a device, method, and program for supporting communication between users using an agent such as a chatbot.

In recent years, chat has been widely used as a communication tool between users in remote locations. However, in chat, communication between users does not occur unless one of the users actively starts a conversation.

Therefore, in order to promote communication between users, a technology has been proposed that combines chat with an agent called a chatbot or the like. In this technology, for example, an agent monitors the behavior of one user, generates a message that broadcasts the behavior, and reflects the message on the chat (for example, see Non-Patent Document 1).

In addition, as a related technology, research is being conducted to deeply understand the intention of human behavior while taking into account cultural background and individual routines in order to support human behavior (for example, see Non-Patent Document 2). reference).

Kosuke Sato, Naoki Oshima, Ryosuke Aoki, Naoki Takegawa, "Multiple chatbots that connect parents and children who live apart by chatting with each other based on sensed cooking behavior - Two-week user test using an automated system", HAI (Human-Agent Interaction) ) Symposium 2018, G-11 Genevieve Bell and Joseph Jofish Kaye, “Designing Technology for Domestic Spaces: A Kitchen Manifesto”, Gastronomica: The Journal of Critical Food Studies, 46-52, Vol.2 No.2, Spring 2002.

However, Non-Patent Document 1 discloses that an agent determines a user's behavior, for example, cooking status, based on detection data from multiple types of sensors, generates a corresponding message, and reflects it on a chat between users. However, it does not disclose how a message is specifically generated based on sensor detection data. Furthermore, what kind of messages are effective in promoting communication has not yet been considered.

On the other hand, Non-Patent Document 2 provides a direction for deeply understanding the intentions of people's actions while taking into consideration cultural backgrounds and individual routines in order to support people's cooking actions. , there is no mention of what kind of messages are effective in promoting communication between users, or how to generate messages.

The present invention has been made in view of the above-mentioned circumstances, and aims to provide a technique for generating messages that are effective in promoting communication between users.

In order to solve the above problems, one aspect of the device or method according to the present invention is that when supporting communication between a first user and a second user via a network, Detection data related to the behavior of the first user is acquired, and the information value of the context information related to the behavior of the first user included in the detected data satisfies a preset condition. The data is classified into first detection data and second detection data that does not satisfy the above conditions. Next, from the classified first detection data, a plurality of pieces of first context information related to the behavior of the first user are decomposed and extracted, and based on the second detection data, the detection data is Second context information related to the first user's behavior is inferred from the context information included in the first user's behavior. Then, the information value of the plurality of first context information or the second context information is evaluated, and based on the evaluation result, first weighted context information using an assertive expression and ambiguous expression selectively generates second weighted context information using the second user, and outputs the generated first or second weighted context information in a state where it can be viewed by the second user. It is something.

According to one aspect of the present invention, for example, the information value of context information related to user behavior that is decomposed and extracted or inferred from detected data is evaluated, and if the evaluation value is low, based on the context information, Weighted context information using an assertive expression is generated, while if the evaluation value is high, weighted context information using an ambiguous expression is generated based on the context information. will be presented. For this reason, the context information representing the user's behavior that is extracted or inferred from the detected data can be presented using either assertive and accurate expressions or ambiguous and interesting expressions, depending on the information value. This makes it possible to effectively provide the user with an opportunity to communicate while ensuring the reliability of the presented information.

That is, according to one aspect of the present invention, it is possible to provide a technique for generating messages that are effective in promoting communication between users.

FIG. 1 is an overall configuration diagram of a system including a server device functioning as a communication support device according to an embodiment of the present invention. FIG. 2 is a block diagram showing the hardware configuration of the server device shown in FIG. 1. FIG. 3 is a block diagram showing the software configuration of the server device shown in FIG. 1. FIG. 4 is a flowchart showing an example of the overall processing procedure and processing contents of the communication support processing executed by the server device shown in FIGS. 2 and 3. FIG. 5 is a flowchart showing an example of the processing procedure and processing contents of the context classification process among the entire processing procedure shown in FIG. 4. FIG. 6 is a flowchart showing an example of the processing procedure and processing contents of the context decomposition processing among the processing steps shown in FIG. 4. FIG. 7 is a flowchart showing an example of the processing procedure and processing contents of the context analogy processing among the processing steps shown in FIG. 4. FIG. 8 is a flowchart illustrating an example of the processing procedure and processing contents of the weighted context generation process among the processing procedures shown in FIG. 4. FIG. 9 is a diagram illustrating an example of the relationship between a context and its information value. FIG. 10 is a diagram illustrating an example of a weighted context expressed in an assertive expression. FIG. 11 is a diagram illustrating an example of a weighted context expressed in an ambiguous expression.

Embodiments of the present invention will be described below with reference to the drawings.

[One embodiment]
(Configuration example)
(1) System FIG. 1 is an overall configuration diagram of a system including a server device functioning as a communication support device according to an embodiment of the present invention.

In FIG. 1, UT1 and UT2 indicate user terminals owned by users US1 and US2, respectively. Each of the user terminals UT1 and UT2 is a mobile information terminal such as a smartphone or a tablet terminal. A communication tool is installed in the user terminals UT1 and UT2, and by using the communication tool, the users US1 and US2 can communicate with each other via the network NW, for example, by chatting. Note that it is also possible to use notebook personal computers or wearable terminals as the user terminals UT1 and UT2.

By the way, a plurality of sensor devices for detecting the behavior of the user US1 are installed in the residence of the user US1. In this example, sensors SS1 and SS2 are installed on the cooking appliance ST and the lighting appliance LG to detect their operating states, respectively, and a camera CM is installed on the ceiling. Furthermore, a transmitting/receiving device TR is provided in the residence. The transmitting/receiving device TR is composed of, for example, a wireless router, and transfers detection data output from the sensors SS1, SS2 and the camera CM to the server device SV via the network NW.

The sensor SS1 may be, for example, an acceleration sensor that is attached to a knob of the cooking utensil ST, a knife (not shown), a handle of a frying pan, etc. to detect its movement, a temperature sensor that detects the temperature of the cooking utensil ST, or a sensor that detects the temperature of the cooking utensil ST. A detection microphone is used, and an illuminance sensor is used as the sensor SS2.

(2) Server device SV
2 and 3 are block diagrams showing an example of the hardware configuration and software configuration of the server device SV, respectively.

The server device SV is composed of, for example, a server computer or a cloud computer, and includes a control unit 1 having a hardware processor such as a central processing unit (CPU). A storage unit having a program storage section 2 and a data storage section 3 and a communication interface (communication I/F) 4 are connected to the control section 1 via a bus 5 .

The communication I/F 4 sends and receives various information to and from the user terminals UT1 and UT2 under the control of the control unit 1 using a communication protocol defined by the network NW.

The program storage unit 2 uses a nonvolatile memory such as a hard disk drive (HDD) or a solid state drive (SSD) that can be written to and read out at any time, and a nonvolatile memory such as a read only memory (ROM) as a storage medium. In addition to middleware such as an OS (Operating System), the computer stores programs necessary for executing various control processes according to an embodiment of the present invention.

The data storage unit 3 is a combination of a nonvolatile memory such as an HDD or SSD that can be written to and read from at any time as a storage medium, and a volatile memory such as a RAM (Random Access Memory). A detected data storage section 31, a first context storage section 32, a second context storage section 33, and a weighted context storage section 34 are provided.

The detection data storage unit 31 is used to store detection data transmitted from the sensors SS1, SS2 and the camera CM in time series. The detection data is given, for example, the identification information (user ID) of the user US1, the identification information (sensor ID) of the transmitting sensor, and the detection time.

The first context storage unit 32 is used by the control unit 1, which will be described later, to store first context information extracted and decomposed from the detected data after classification. A user ID is given to this first context information.

The second context storage unit 33 is used by the control unit 1, which will be described later, to store second context information inferred from the context included in the detected data after classification. This second context information is also given a user ID.

The weighted context storage unit 34 is used to store weighted context information generated by the control unit 1, which will be described later, in association with a user ID.

The control unit 1 has a function of, for example, a chatbot or an agent, which provides communication support to the users US1 and US2 who communicate by chatting or the like. Processing units for realizing this function include a detection data acquisition processing unit 11, a context classification processing unit 12, a context decomposition processing unit 13, a context analogy processing unit 14, and a weighted context generation processing unit 15. , and a message distribution processing section 16. Each of these processing units 11 to 16 is realized, for example, by causing a hardware processor of the control unit 1 to execute a program stored in the program storage unit 2.

For example, when the user US1 starts cooking, the detected data acquisition processing unit 11 receives the detected data of the sensors SS1, SS2 and the camera CM transmitted from the transmitter/receiver TR in the residence of the user US1 via the communication I/F4. A process is performed to receive the detected data and store it in the detected data storage unit 31 in chronological order.

The context classification processing unit 12 reads the detected data for a certain period of time from the detected data storage unit 31, and divides the detected data into information values according to the information value of the context information related to the behavior of the user US1 included in the detected data. A process of classifying the first detection data into high-value first detection data and the second detection data with low information value is performed.

Here, the information value of the context indicates the degree to which the viewer can infer the actions and intentions of the monitored user when a person views the detected data, and the higher the information value, the more The viewer can grasp a lot of information related to the user's behavior. In this embodiment, the context classification processing unit 12 determines the information value of the context based on the type of sensor, and classifies the detected data into first detected data with high value and second detected data with low value. An example of the processing will be described in the operation example.

The context decomposition processing unit 13 takes in the high-value first detection data classified by the context classification processing unit 12, and performs a process of decomposing and extracting a plurality of pieces of low context information with low information value from the first detection data. conduct. The context decomposition processing unit 13 also performs a process of generating high context information with higher information value, which is inferred by combining the plurality of extracted low context information. Then, the context decomposition processing unit 13 stores the plurality of low context information and high context information that have been decomposed and extracted and inferred as the first context information in the first context storage unit 32 with the user ID assigned thereto. Note that an example of the above-mentioned context decomposition and extraction processing will be described in an operation example.

The context analogy processing unit 14 takes in the second detection data of low value classified by the context classification processing unit 12, and based on the low context information included in the second detection data, the context analogy processing unit 14 extracts the second detection data of higher information value based on the low context information included in the second detection data. Performs processing to infer high context information. Then, the context analogy processing unit 14 stores the inferred high context information in the second context storage unit 33 as second context information, with the user ID attached thereto. Note that an example of the analogy processing of the high context information will be described in the operation example.

The weighted context generation processing section 15 reads new context information every time new context information is stored in the first context storage section 32 or the second context storage section 33. Then, the information value of the read context information is evaluated. At this time, new context information is read from both the first context storage unit 32 and the second context storage unit 33, and if the contents of these contexts are related to the same action of the user US1, The information value of the context information is evaluated, and the context information with the lower evaluation value is selected.

Based on the information value evaluation results, the weighted context generation processing unit 15 generates first weighted context information in which context information with a low evaluation value is weighted using an assertive expression, while first weighted context information with a high evaluation value is weighted with an assertive expression. Regarding the context, second weighted context information weighted using ambiguous expressions is generated. Then, a process is performed in which each piece of generated weighted context information is stored in the weighted context storage unit 34 with a user ID assigned thereto.

When new weighted context information is stored in the weighted context storage unit 34, the message delivery processing unit 16 reads out the new weighted context information. Then, based on the user ID, processing is performed to post the weighted context information on a communication tool between the user terminal UT1 used by the corresponding user US1 and the user terminal UT2 with which the corresponding user US1 communicates.

(Operation example)
Next, a series of operations related to communication support performed in the server device SV configured as above will be described.

Here, for example, when a university student living alone and his mother communicate through chat, the server device SV uses its chatbot or agent function to carry out processing to support communication between the university student and his mother. An example of execution will be explained below.

FIG. 4 is a flowchart showing the overall processing procedure and processing contents of the communication support processing executed by the control unit 1 of the server device SV.

(1) Acquisition of detection data When the user US1 turns on the lighting fixture LG in the kitchen to cook, the lighting of the lighting fixture LG is detected by the sensor SS2, which is, for example, an illuminance sensor, and the detection data is transmitted to the transmitting/receiving device TR. is transferred to the server device SV.

The control unit 1 of the server device SV monitors the start of cooking in step S1. In this state, when detection data representing lighting of the lighting fixture LG is received, it is determined that the user US1 has started cooking, and under the control of the detection data acquisition processing unit 11, in step S2, the following processing is performed. Executes detection data acquisition processing.

That is, when the user US1 ignites, for example, the cooking utensil ST and operates a knob to set the firepower, the operation of this knob is detected by the sensor SS1, which is composed of, for example, an acceleration sensor, and the detected data is transmitted from the transmitting/receiving device TR to the server device. Sent to SV. Further, the cooking state of the user US1 is imaged by the camera CM, and the video data is transmitted from the transmitting/receiving device TR to the server device SV as detection data representing the cooking state. Note that even when the user US1 cuts a material with a knife or shakes a frying pan, the movement or sound is detected by an acceleration sensor or a microphone, and the detected data is sent from the transmitting/receiving device TR to the server device SV. be done.

On the other hand, in step S2, the detection data acquisition processing unit 11 of the server device SV receives each detection data transmitted from the transmitting/receiving device TR in the home of the user US1 via the communication I/F 4, and receives each of the received detection data. The data is stored in the detected data storage section 31. Note that at this time, each of the above detection data is given a user ID, a sensor ID, and a detection time, respectively. Thereafter, the detection data acquisition processing unit 11 continues the detection data acquisition process until the end of cooking is detected in step S7.

Note that if there are other service target users, the detection data acquisition process is similarly executed for these other users as well.

(2) Context classification process When new detection data is stored in the detection data storage unit 31, the control unit 1 of the server device SV, under the control of the context classification processing unit 12, performs context classification as follows in step S3. Perform classification processing.

FIG. 5 is a flowchart showing an example of the processing procedure and processing contents of the context classification process by the context classification processing section 12.

That is, when the detection data for a certain period of time (for example, 2 minutes) is stored in the detection data storage section 31, the context classification processing section 12 stores the detection data for the certain period of time from the detection data storage section 31 in step S31. Load. Next, in step S32, the context classification processing unit 12 determines the sensor type from the sensor ID assigned to the read detection data, and in step S33, classifies the detection data according to the determination result of the sensor type. .

For example, since the detection data representing the operation of the cooking utensil ST is data representing only the operation of the knob obtained by the acceleration sensor, the detection data representing the operation of the cooking utensil ST is "detection data with low context value." are categorized. On the other hand, the video data output from the camera commercial is detection data that captures the cooking state of user US1 in the kitchen and includes multiple pieces of information. data.

Note that "detection data with low context value" also includes detection data from illuminance sensors, temperature sensors, and vibration sensors, and "detection data with high context value" includes sounds from various appliances and user information. Audio data including US1's voice, recipe search data obtained from websites, etc. are classified.

Next, the context classification processing unit 12 determines the context value of the classified detection data in step S34, and according to the determination result, sends the detection data with high context value to the context decomposition processing unit 13, and sends the detection data with low context value to the context decomposition processing unit 13. The data is sent to the context analogy processing section 14.

(3) Context decomposition process When the detection data with high context value is sent, the context decomposition processing unit 13 executes the context decomposition process as follows in step S4.

FIG. 6 is a flowchart showing an example of the processing procedure and processing contents of the context decomposition process by the context decomposition processing unit 13.

That is, the context decomposition processing unit 13 first takes in the detection data to be processed in step S41, and then decomposes and extracts a plurality of low contexts with low information value from this detection data. For example, if the detected data is video data showing the state of cooking, for example, "cooking utensil ST is being used," "a knife is being used," "food is on the cutting board," " Objects with low information value, such as "Food is being held in the user's hand" and "A frying pan is being shaken," are extracted and decomposed as low context information.

Note that the object decomposition and extraction process can be carried out by, for example, storing basic image patterns of each cooking-related appliance in advance, and performing pattern matching processing on the acquired video data and the basic image pattern. be.

Continuously, in step S43, the context decomposition processing unit 13 determines whether high context information with higher information value can be inferred from each of the plurality of decomposed and extracted low context information alone or in combination. . If the analogy is possible, the high context information is inferred in step S44, and the inferred high context information is stored in the first context storage unit 32 in association with the user ID.

For example, in this example, from the combination of "A knife is being used" and "Food is on a cutting board," the high context information "Food is being cut on a cutting board" is inferred. Further, from the combination of "a knife is being used" and "food is being held in the user's hand", high context information such as "food is being peeled" can be inferred. An example of analogy processing for high context information will also be explained in the subsequent context analogy processing.

Note that if the high context information cannot be inferred, the plural pieces of low context information may be stored in the first context storage unit 32 as they are or may be selected.

The context decomposition and extraction processing in steps S41 to S44 described above is repeatedly executed until it is determined in step S45 that processing for all detection data to be processed has been completed.

(4) Context analogy processing On the other hand, when the detection data with low context value is sent, the context analogy processing unit 14 executes context analogy processing as follows in step S5.

FIG. 7 is a flowchart showing an example of the processing procedure and processing contents of the context analogy processing by the context analogy processing section 14.

That is, the context analogy processing unit 14 first takes in the detection data to be processed in step S51, and then extracts low context information with low information value from the detection data in step S52. Through this process, for example, when the detected data is acceleration data representing the operation of the knob of the cooking utensil ST, information representing "the amount of operation (operation angle) of the knob of the cooking utensil ST" is extracted as low context information. .

Next, in step S53, the context analogy processing unit 14 infers high context information having higher information value based on the extracted low context information. If it is possible to infer the high context information with even higher information value, the process returns from step S54 to step S53, and the analogy process for the high context information is repeatedly executed.

For example, when the information representing the above-mentioned "amount of operation (operation angle) of the knob of cooking utensil ST" is extracted as the low context information, first from "operation of the knob of cooking utensil ST" included in the information, " "A cooking utensil (stove) ST is being used" is inferred, and then "strength of heat" is inferred from "knob operating angle". Furthermore, the "purpose of use (for example, boiling, stir-frying, steaming)" can be inferred from the "transition state of heat intensity."

The analogy process described above is performed by storing transition information on heat level in advance in association with the type of cooking such as boiling, frying, steaming, etc., and comparing the context information with the transition information on heat level by pattern matching. This is achieved by identifying the transition pattern with the highest degree of similarity based on the degree of similarity.

Next, in step S55, the context analogy processing unit 14 stores the high context information finally obtained through the iterative analogy process in the second context storage unit 33 in association with the user ID. Further, in step S56, the context analogy processing unit 14 causes the second context storage unit 33 to store the number of repetitions (the number of analogy steps) when the repeated analogy processing is performed in association with the high context information. . The number of analogy steps is used as an evaluation index of the information value of the context information.

The context analogy processing in steps S51 to S56 described above is repeatedly executed until it is determined in step S57 that processing for all detection data to be processed has been completed.

FIG. 9 shows an example of the relationship between the context information generated by the analogy process described above and its value. As shown in the figure, "a cooking utensil (stove) is being used" and "a frying pan is being used" are close to the states indicated by the sensor detection data and are low-context information with low value. In other words, this low context information is context information that represents a clear state with a low degree of ambiguity, and is preferably presented in a definitive manner.

On the other hand, "the stove is kept on high heat" and "the frying pan is being shaken", which are inferred from the above low context information, are more valuable context information, and furthermore, "the stove is kept on high heat" and "the frying pan is shaken" are more valuable context information. "I'm making a stir-fry" and "I'm making a stir-fry with vegetables" are even more valuable high-context information. Since this high context information is generated by repeating analogical reasoning processing, it has a high degree of ambiguity, and it is desirable to present it in an ambiguous expression.

The server device SV of this embodiment generates weighted context information to be presented as a message to the user, paying attention to the above points.

(5) Generation of weighted context information When new context information is stored in the first or second context storage section 32, 33, the control section 1 of the server device SV controls the weighted context generation processing section 15. Under the control, in step S6, weighted context information generation processing is executed as follows.

FIG. 8 is a flowchart showing the processing procedure and processing contents of the weighted context generation process by the weighted context generation processing section 15.

That is, the weighted context generation processing unit 15 first reads newly stored context information from the first and second context storage units 32 and 33 in step 61. Then, in step S62, it is determined whether the read information is one or both of the first context information and the second context information. If either the first context information or the second context information has been read, the process moves to step S66, which will be described later.

On the other hand, if the result of the above determination is that both the first and second context information have been read, the weighted context generation processing unit 15 compares the first and second context information in step S63. Based on this comparison result, it is determined in step S64 whether the first context information and the second context information relate to the same behavior of the user US1. For example, if the first and second context information both relate to the operation of the cooking utensil ST, it is determined that the first and second context information at this time relate to the same action of the user US1.

Subsequently, in step S65, the weighted context generation processing unit 15 selects the one with the lower information value based on the information values of the first context information and the second context information. Here, the number of repetitions (the number of analogy steps) of analogy processing executed by the context decomposition processing unit 13 and the context analogy processing unit 14 when inferring the first and second context information, respectively, is the number of repetitions (number of analogy steps) of the information value of the context information. It can be used as an evaluation index. Therefore, the weighted context generation processing unit 15 compares the number of analogy steps when inferring the first and second context information, and selects the context information with the smaller number of analogy steps.

Next, in step S66, the weighted context generation processing unit 15 uses, for example, the number of analogy steps of the context information as the evaluation value, and compares this with a threshold value. As a result of this comparison, if the evaluation value is less than the threshold value, the process moves to step S67, and context information is generated by weighting the above-mentioned context information using an assertive expression. On the other hand, if the evaluation value is equal to or greater than the threshold value, the process moves to step S68, and context information is generated by weighting the above-mentioned context information using an ambiguous expression.

Here, as the above-mentioned assertive expression, for example, "-da" is used, and a assertive character with the same meaning as this character is added to the end of the context information. On the other hand, as ambiguous expressions, for example, "I think...", "I feel like...", and "It seems like..." are used, and these character strings are added to the end of the context information.

Note that as an evaluation index for the information value of the context information, for example, a context value distribution may be defined, an evaluation value is given to the context information according to this content value distribution, and this evaluation value is compared with a threshold value. . For example, the context value distribution is expressed as 0 to 1, an evaluation value is assigned to the context information within this range, and this evaluation value is compared with a threshold value of 0.5. Then, if the evaluation value is less than 0.5, "assertive expression" is selected to generate weighted context information, while if the evaluation value is 0.5 or more, "vague expression" is selected to generate weighted context information. generate.

Finally, the weighted context generation processing unit 15 stores the weighted context information generated by each of the above weighting processes in the weighted context storage unit 34 in association with the user ID in step S69.

(6) Message Distribution When the new weighted context information is stored in the weighted context storage unit 34, the control unit 1 of the server device SV, under the control of the message distribution processing unit 16, performs the above-mentioned New weighted context information is read from the weighted context storage unit 34. Then, based on the user ID, the read weighted context information is posted on a communication tool between the user terminal UT1 used by the corresponding user US1 and the user terminal UT2 with which the corresponding user US1 communicates. .

FIG. 10 shows an example of presentation of weighted context information generated based on low context information on a communication tool. As shown in this example, the low context information is presented using "assertive expressions" such as "-yo" and "there is...".

On the other hand, FIG. 11 shows an example of presentation of weighted context information generated based on high context information on a communication tool. As shown in this example, high context information is presented using "ambiguous expressions" such as "I wonder if..." and "It seems...".

(action/effect)
As described above, in one embodiment, in the server device SV equipped with a communication support function called a chatbot, agent, etc., detection data acquired from a sensor is first applied to the behavior of the user US1 included in the detection data. The context information is classified into detected data with high information value and detected data with low information value. Next, a plurality of pieces of low context information related to the behavior of the user US1 are decomposed and extracted from the classified high value detection data, and more valuable context information is inferred from each extracted low context information. On the other hand, from the detection data of low value, context information of higher value is inferred based on the context information included in the detection data. Then, the information value of the inferred context information is evaluated, and definitive expressions are used for context information whose evaluation value is lower than the threshold value, and ambiguous expressions are used for context information whose evaluation value is higher than the threshold value. The information is displayed on the communication tool so that the users US1 and US2 can view it.

Therefore, according to one embodiment, the first context information representing user behavior is decomposed and extracted from high-value detected data and then inferred, or the second context information is directly inferred from low-value detected data. Depending on the information value of the information, it is possible to use and present assertive and accurate expressions and ambiguous and interesting expressions. This makes it possible to effectively provide an opportunity for communication while ensuring the reliability of information.

In addition, if both the first context information and the second context information are obtained, it is determined whether these context information are information related to the same behavior of the user, and the information related to the same behavior is determined. In the case of information, the one with the lowest information value is selected, and based on the evaluation result of the information value, context information using assertive or ambiguous expressions is selectively generated and presented.

Therefore, if multiple pieces of context information representing the same behavior of the user US1 are obtained at the same time, a message will be generated and presented based on the context information with the lowest information value among them, and the message will be presented to the user US1. It is possible to present a message that more accurately approximates the behavior.

Furthermore, when inferring the behavior of the user US1 from multiple pieces of low context information extracted from the detected data, it is possible to infer from a single piece of low context information by making an analogy based on a combination of multiple pieces of low context information. Compared to the case, it becomes possible to more accurately infer the behavior of the user US1, and thereby it becomes possible to present an accurate message.

[Other embodiments]
In the embodiment described above, a case has been described in which the server device SV is provided with a communication support function using a chatbot or an agent, but it may be provided in any user terminal. In addition, the functional configuration, processing procedures, processing contents, etc. of the communication support device can be modified in various ways without departing from the gist of the present invention.

Although the embodiments of the present invention have been described above in detail, the above descriptions are merely illustrative of the present invention in all respects. It goes without saying that various improvements and modifications can be made without departing from the scope of the invention. That is, in implementing the present invention, specific configurations depending on the embodiments may be adopted as appropriate.

In short, the present invention is not limited to the above-described embodiments as they are, but can be implemented by modifying the constituent elements within the scope of the invention at the implementation stage. Moreover, various inventions can be formed by appropriately combining the plurality of components disclosed in the above embodiments. For example, some components may be deleted from all the components shown in the embodiments. Furthermore, components from different embodiments may be combined as appropriate.

SV...Server device UT1, UT2...User terminal NW...Network ST...Cooking equipment LG...Lighting equipment SS1, SS2...Sensor CM...Camera TR...Transmission/reception device 1...Control unit 2...Program storage unit 3...Data storage unit 4...Communication I/F
5... Bus 11... Detected data acquisition processing section 12... Context classification processing section 13... Context decomposition processing section 14... Context analogy processing section 15... Weighted context generation processing section 16... Message distribution processing section 31... Detected data storage section 32... First context storage unit 33...Second context storage unit 34...Weighted context storage unit

Claims (7)

A communication support device that supports communication between a first user and a second user via a network,
an acquisition processing unit that acquires detection data related to the behavior of the first user from the sensor;
The acquired detection data is divided into first detection data whose information value of context information related to the first user's behavior included in the detection data satisfies a preset condition, and second detection data which does not satisfy the condition. a classification processing unit that classifies the detected data into two types of detected data;
a decomposition processing unit that decomposes and extracts a plurality of first context information related to the behavior of the first user from the classified first detection data;
an analogy processing unit that infers second context information related to the behavior of the first user from context information included in the second detection data based on the classified second detection data;
Evaluate the information value of the plurality of first context information or the second context information, and based on the evaluation result, first weighted context information using an assertive expression and first weighted context information using an ambiguous expression. a generation processing unit that selectively generates second weighted context information;
A presentation processing unit that presents the generated first weighted context information or the second weighted context information in a state where the second user can view it. The generation processing unit is
If either the first context information or the second context information is obtained, evaluation of the information value of the obtained first context information or second context information selectively generating the first weighted context information and the second weighted context information based on the results;
The communication support device according to claim 1. The generation processing unit is
When both the first context information and the second context information are obtained, the obtained first context information and the second context information are the same as the first user's A processing unit that determines whether or not the information is related to an action, and if the information is related to the same action, selects the one with the lower information value between the first context information and the second context information. and,
selectively selecting the first weighted context information and the second weighted context information based on the evaluation result of the information value of the selected first context information or the second context information; The communication support device according to claim 1, comprising: a processing unit for generating. The generation processing unit is
If the third context information can be inferred from the combination of the plurality of first context information, the third context information using an assertive expression is based on the evaluation result of the information value of the third context information. selectively generating one weighted context information and the second weighted context information using an ambiguous expression;
The communication support device according to claim 1. The generation processing unit is
The communication support device according to claim 4, wherein the information value is evaluated based on the number of steps of analogy processing performed until the second context information or the third context information is obtained. A communication support method executed by an information processing device that supports communication between a first user and a second user via a network, the method comprising:
a step of acquiring detection data related to the behavior of the first user from a sensor;
The acquired detection data is divided into first detection data whose information value of context information related to the first user's behavior included in the detection data satisfies a preset condition, and second detection data which does not satisfy the condition. 2. A process of classifying the detected data into
A process of decomposing and extracting a plurality of first context information related to the behavior of the first user from the context information included in the second detection data based on the classified first detection data;
a step of inferring second context information related to the behavior of the first user from the classified second detection data;
Evaluate the information value of the plurality of first context information or the second context information, and based on the evaluation result, first weighted context information using an assertive expression and first weighted context information using an ambiguous expression. a step of selectively generating second weighted context information;
A communication support method comprising: presenting the generated first or second weighted context information in a state where the second user can view it. A program that causes a processor included in the communication support device to execute processing of each of the processing units included in the communication support device according to any one of claims 1 to 5.