Estimation device, learning device, estimation method, learning method, and program

Abstract

To efficiently estimate the quality of group work without hearing subjectivity from participants of the group work.SOLUTION: An estimation device that estimates quality of group work from observation data in the group work in which a plurality of participants participate includes an acquisition unit that acquires observation data for each participant, an estimation unit that estimates subjective data by inputting the observation data for each participant to a learned model, and an evaluation unit that outputs the quality of the group work based on the subjective data estimated from the observation data.SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] The present invention relates to a technique for evaluating the quality of group work. [Background technology]

[0002] In recent years, the concept of employee well-being (the state or state of being in which a person feels "good") has become more widely known. Well-being cannot be achieved alone; relationships with team members and others are also important for well-being. For this reason, it is important to measure the state of well-being in an environment where a member works with other members. Without measurement, improvements cannot be made.

[0003] Group work is a place where a member works together with other members. When considering the well-being of the participating members, it is important to understand the quality (or even the state) of the group work. A conventional technique for understanding the quality of group work is disclosed in Non-Patent Document 1. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] SG Rogelberg et al., ""Not Another Meeting!" Are Meeting Time Demands Related to Employee Well-Being?," Journal of Applied Psychology 91(1):83-96, Jan. 2006 Summary of the Invention [Problem to be solved by the invention]

[0005] In conventional techniques such as Non-Patent Document 1, in order to grasp the quality of group work, participants must be asked to provide their subjective feedback using a questionnaire, which is time-consuming and inefficient. Furthermore, because participants must be asked to provide their subjective feedback, the quality of the group work could only be measured after it was finished.

[0006] The present invention has been made in consideration of the above points, and aims to provide a technology that makes it possible to efficiently estimate the quality of group work without asking the participants in the group work for their subjective opinions. [Means for solving the problem]

[0007] According to the disclosed technology, there is provided an estimation device that estimates the quality of group work in which a plurality of participants participate from observation data of the group work, the estimation device comprising: an acquisition unit that acquires observation data for each participant; an estimation unit that estimates subjective data by inputting the observation data for each participant into a trained model; an evaluation unit that outputs the quality of the group work based on the subjective data estimated from the observation data; An estimation device is provided, comprising: [Effects of the Invention]

[0008] The disclosed technology provides a technology that enables the quality of group work to be estimated efficiently without asking subjective opinions from the participants of the group work. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 illustrates an example of a system configuration. [Figure 2] FIG. 1 illustrates an example of a system configuration. [Figure 3] FIG. 1 illustrates an example of the configuration of a learning device 100. [Figure 4] 10 is a flowchart showing the operation procedure of the learning device 100. [Figure 5]FIG. 10 is a diagram illustrating an example of observation data. [Figure 6] FIG. 10 is a diagram illustrating an example of subjective data. [Figure 7] FIG. 10 is a diagram illustrating an example of learning data. [Figure 8] FIG. 2 is a diagram illustrating an example of the configuration of an estimation device 200. [Figure 9] 10 is a flowchart showing the operation procedure of the estimation device 200. [Figure 10] FIG. 10 is a diagram showing a display example. [Figure 11] FIG. 1 illustrates an example of the configuration of a learning device 100. [Figure 12] FIG. 2 is a diagram illustrating an example of the configuration of an estimation device 200. [Figure 13] FIG. 2 illustrates an example of a hardware configuration of the apparatus. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an embodiment of the present invention (the present embodiment) will be described with reference to the drawings. The embodiment described below is merely an example, and the embodiment to which the present invention is applied is not limited to the following embodiment.

[0011] An embodiment for estimating the quality of group work will be described below. Group work involves a group of multiple participants holding discussions (e.g., discussions to decide what to exhibit at an exhibition) and creating proposals. In this embodiment, "group work" is used in a broad sense, and any work in which multiple people talk and perform some kind of task is referred to as "group work." Regular meetings are also included in "group work."

[0012] (Outline of the embodiment) In this embodiment, the estimation device 200 described below estimates the quality of the group work (evaluation of the environment) rather than the quality of the individuals during the actual group work, using observation data obtained from the group work, without having individuals answer a questionnaire.

[0013] The above observation data is not limited to specific data, but as observation data, for example, all or any or several of the participants' "changes in facial expressions, amount of speech, content of speech" can be used.

[0014] The format of the group work is not limited to a specific format, and it can be conducted face-to-face by multiple people, or online by each person using a terminal such as a PC. In either case, it is sufficient that the observation data of each participant can be obtained separately from the observation data of other participants.

[0015] The estimation device 200 can estimate the quality of group work in real time. Furthermore, the estimation device 200 can use group characteristics in group work. By using group characteristics, individual identification is not required, and therefore a system can be realized that maintains individual privacy.

[0016] A learning device 100, which will be described later, learns a model to be used for estimation in the estimation device 200. The learning device 100 learns the model by using observed data and subjective data. The estimation device 200 receives only observed data as input and outputs an estimated quality of group work.

[0017] (Examples of data used for learning / estimation) Below are examples of data used for learning / estimation. Note that the data below is merely an example and is not limited to the data below.

[0018] For example, the following data can be used as subjective data (correct answer data used for learning) to gauge the quality of group work.

[0019] ·Psychological safety Effectiveness (whether the group work went well, whether the goal was achieved, etc.) -Evaluation indicators for group work (Self-as-We, etc.) Examples of observed data (data used for learning and estimation) include the following data:

[0020] Face sensor data (e.g., Action Unit) Emotion estimation Number of comments The configuration and operation of the system / device will be described below.

[0021] (System configuration example) An example of a system configuration for online group work is shown in Fig. 1. As shown in Fig. 1, this system has a configuration in which a terminal 10 used by each participant in the group work and an estimation device 200 are connected to a network 300 such as the Internet. Note that this configuration can also be used offline, and does not necessarily need to be via a network such as the Internet.

[0022] Each terminal 10 is, for example, a PC, a smartphone, a tablet, etc. Each terminal 10 is assumed to have functions necessary for carrying out group work as well as functions necessary for acquiring observation data (camera, microphone, various sensors, etc.).

[0023] The estimating device 200 collects observation data transmitted from each terminal 10, estimates the quality of the group work based on the observation data, and outputs the estimated quality. The estimating device 200 may be, for example, a terminal of an administrator (which may also be called a facilitator) who manages the group work, or a server or the like installed for that purpose. Any one of the multiple terminals 10 may have the functions of the estimating device 200. A terminal of an administrator (called an administrator terminal) that manages the group work may be provided separately from the estimating device 200, and the estimation results by the estimating device 200 may be displayed on the administrator terminal. The administrator terminal may also provide intervention to the participants of the group work via the estimating device 200, as described below.

[0024] Figure 2 shows an image of the system configuration when conducting face-to-face group work. As shown in Figure 2, there are multiple participants, and an estimation device 200 is provided. The estimation device 200 collects observation data from each participant and estimates the quality of the group work.

[0025] Hereinafter, examples of the configuration and operation of the learning device 100 and the estimation device 200 will be described in detail. In this embodiment, the learning device 100 and the estimation device 200 are separate devices, but the learning device 100 and the estimation device 200 may be the same device. That is, for example, the estimation device 200 may include the functions of the learning device 100. Also, the learning device 100 may include the functions of the estimation device 200.

[0026] (Learning device 100) First, the learning device 100 will be described. Figure 3 shows a configuration example of the learning device 100. As shown in Figure 3, the learning device 100 includes an observation data collection unit 110, an observation data conversion unit 120, a subjective data conversion unit 130, an observation data-subjective data combination unit 140, and a learning unit 150.

[0027] Note that the observation data collection unit 110 may be provided outside the learning device 100 (for example, near each participant). Also, "observation data collection unit 110 + observation data conversion unit 120" may be provided outside the learning device 100 (for example, near each participant). Even when the observation data collection unit 110 / observation data conversion unit 120 is provided outside the learning device 100, the learning device 100 includes an acquisition unit that acquires the data (observation data) obtained by the observation data collection unit 110 / observation data conversion unit 120.

[0028] Figure 4 is a flowchart showing the operation procedure of the learning device 100. The operations of each part in the learning device 100 will be described according to the procedure of the flowchart shown in Figure 4.

[0029] <S10(Step 101): Observation data collection> In S101, the Observation Data Collection Unit 110 collects a plurality of observation data from each participant in the group work. There is no limitation on the type of observation data. In the example of FIG. 3, as the observation data, video / audio data, LADAR (Laser Detection and Ranging) / LiDAR (Light Detection and Ranging, Laser Imaging Detection and Ranging) data, and various sensor data, etc. are collected. As the observation data, a wide variety of data can be used and is not limited to the example shown in FIG. 3.

[0030] <S102: Conversion of Observation Data> In S102, the Observation Data Conversion Unit 120 converts the observation data collected by the Observation Data Collection Unit 110 and saves the converted observation data. For example, the Observation Data Conversion Unit 120 calculates statistical quantities (average, variance, standard deviation, etc.) per unit time for the collected observation data for each participant. Also, the Observation Data Conversion Unit 120 performs a process of converting the data according to the time span of the subjective data.

[0031] It is also possible to process meaningful information (such as textifying the conversation content from the audio) from the observation data collected by the Observation Data Collection Unit 110 and use it as an input to the Observation Data Conversion Unit 120. The Observation Data Conversion Unit 120 may generate data in which important features of the conversation are vectorized using a method such as TF-IDF (Term Frequency-Inverse Document Frequency) from the input textified information.

[0032] <Specific Examples of Observation Data> Here, specific examples of the observation data will be described. FIG. 5 shows specific examples of the observation data. FIG. 5 is an example of the observation data of a specific participant. The Observation Data Collection Unit 110 acquires the same kind of data for all the participants in the group work. Note that the data shown in FIG. 5 may be the data collected by the Observation Data Collection Unit 110 or the data after conversion to statistical values by the Observation Data Conversion Unit 120.

[0033] The data in Fig. 5 is based on a scenario in which video and audio are captured during group work involving multiple participants. To obtain the various pieces of information shown in Fig. 5 from the video and audio, emotion estimation is performed using, for example, facial expression recognition technology.

[0034] <s103> In S103 of the flow in Fig. 4, the subjective data conversion unit 130 acquires the subjective data of each participant and converts it (aggregates it, etc.). Note that the timing of acquiring the subjective data is not limited to this timing (S103 of the flow in Fig. 4). The subjective data for learning is acquired, for example, by having participants fill out a questionnaire, and the contents of the questionnaire are input to the subjective data conversion unit 130.

[0035] <Examples of subjective data> Figure 6 shows a specific example of subjective data for a specific participant. As shown in Figure 6, data collected once every 30 minutes, for example, is used as subjective data. In addition, there are multiple questions for each item, such as psychological safety, effectiveness, and evaluation of the environment, and the average of the values ​​from multiple questions is used as subjective data. For example, a scale from 1 (not at all) to 7 (strongly agree) can be used.

[0036] <s104> 4, the observation data / subjective data combining unit 140 combines the observation data of each participant obtained by the observation data converting unit 120 (e.g., FIG. 5) with the subjective data of each participant obtained by the subjective data converting unit 130 (e.g., FIG. 6). The combined data becomes learning data to be input to the learning unit 150.

[0037] The data of each participant may be processed into group characteristics to evaluate the situation as a whole. For example, the characteristics of the situation and the situation of the group may be used as observational data, such as the fact that only one participant is feeling happy at the same time, while other participants do not.

[0038] <Examples of training data> A specific example of training data is shown in Figure 7. The training data shown in Figure 7 is data that combines statistical values ​​of observation data for a specific period with subjective data. In addition, data is aggregated for each participant and for each timing at which subjective data was obtained.

[0039] In the example shown in Figure 7, the "statistical values ​​for a specific period of observed data" are statistically calculated for the observed data in units where subjective data exists. In this example, the average is calculated for the first 30 minutes and the last 30 minutes, but it is not necessary to limit it to 30 minutes, and variance, standard deviation, etc. may also be used as a statistical quantity.

[0040] The subjective data in Figure 7 is the time-unit subjective data collected and aggregated for each participant. Note that the same format of data can also be used for estimation of statistical values ​​for a specific period of observed data.

[0041] <s105> 4, learning unit 150 receives data in which observation data and subjective data are combined as input and uses a machine learning algorithm to learn a model (for example, a model that uses an algorithm such as a neural network). Learning a model means, for example, adjusting (optimizing) the parameters of the model so that when observation data is input to the model, the model outputs correct subjective data.

[0042] The machine learning algorithm may be a supervised learning algorithm or an unsupervised learning algorithm, and for unsupervised learning, techniques such as deep learning, BERT, or Transformer may be used.

[0043] For learning, a model may be constructed that estimates each subjective data item individually from the observed data, or a model that estimates all the subjective data together. For example, a model that estimates psychological safety, a model that estimates effectiveness, and a model that estimates the environment evaluation may be constructed separately, or a model that estimates psychological safety, effectiveness, and the environment evaluation together may be constructed. A model may also be constructed that outputs the quality of group work (such as "good" in the display example described below) from the observed data.

[0044] Furthermore, rather than using only data from a certain moment, model learning can be performed to make estimates using data from the time the group work began until that time.

[0045] The trained model obtained by the training unit 150 is input to the estimation unit 230 in the estimation device 200, and the estimation unit 230 holds the trained model.

[0046] (Estimation device 200) Next, the estimation device 200 will be described. Fig. 8 shows an example configuration of the estimation device 200. As shown in Fig. 8, the estimation device 200 includes an observation data collection unit 210, an observation data conversion unit 220, an estimation unit 230, a shaping unit 250, an intervention means determination unit 240, a real-time intervention unit 260, and a display unit 270.

[0047] The observation data collection unit 210 may be provided outside the estimation device 200 (for example, at a location near each participant). Furthermore, the "observation data collection unit 210 + observation data conversion unit 220" may be provided outside the estimation device 200 (for example, at a location near each participant). Even when the observation data collection unit 210 / observation data conversion unit 220 are provided outside the estimation device 200, the estimation device 200 still includes an acquisition unit that acquires data (observation data) obtained by the observation data collection unit 210 / observation data conversion unit 220.

[0048] Furthermore, the estimation device 200 may not be provided with the intervention means determination unit 240 and the real-time intervention unit 260.

[0049] Fig. 9 is a flowchart showing the operation procedure of the estimation device 200. The operation of each unit in the estimation device 200 will be described along the procedure of the flowchart shown in Fig. 9. It is assumed that the estimation unit 230 holds a trained model.

[0050] <S201、S202> The processing of the observation data collection unit 210 and the observation data conversion unit 220 in S201 and S202 is the same as the processing of the observation data collection unit 110 and the observation data conversion unit 120 in S101 and S102 described in the learning device 100 section.

[0051] <s203> The estimation unit 230 receives observation data from the observation data conversion unit 220 in the same format as the statistical observation data format (e.g., the observation data format in FIG. 7) of the group (multiple participants) used by the learning unit 150. The estimation unit 230 inputs the observation data into a trained model and obtains subjective data corresponding to the observation data as output from the trained model. Note that the quality of group work (e.g., "good" in the display example described below) may also be obtained as output from the trained model.

[0052] Furthermore, the trained model can estimate subjective data not only using instantaneous data, but also using data over time, for example, from the start of group work to that time.

[0053] The trained model may, for example, take the observation data for each participant as input and estimate the subjective data for each participant. The trained model may also take the observation data for each participant as input and estimate the subjective data for all participants. The trained model may also take the statistical values ​​of the observation data for all participants obtained from the observation data for each participant as input and estimate the subjective data for all participants.

[0054] <s204> In S204, the shaping unit 250 evaluates (determines) the quality of the group work based on the value of the subjective data output from the estimation unit 230, and shapes the value of the subjective data to match the required output content. For example, if the quality of the group work is output as one of three levels, good, average (normal), or needs improvement, the value of the subjective data is converted (shaped) into one of good, average, or needs improvement.

[0055] The shaping unit 250 may evaluate the quality of the group work by using statistical values ​​of the subjective data of each participant (e.g., the average of all participants, etc.), or may output the values ​​of each subjective data obtained by the estimation unit 230 as is, or may perform other processing methods.

[0056] <s205> In S205, the display unit 270 has a function of displaying the shaping results by the shaping unit 250 on a display, a network-connected terminal, or the like. The display may be performed in real time or on demand in response to a request from a user (facilitator, etc.). It is also possible to simultaneously display the results of intervention by the intervention means determination unit 240, which will be described later. The shaping unit 250 and the display unit 270 may be collectively referred to as an evaluation unit.

[0057] <Display example> FIG. 10 shows a display example (UI (User Interface) image). In the example of FIG. 10, psychological safety, effectiveness, and environment evaluation are used as subjective data, and estimated results such as "good" are displayed for each. For example, with regard to psychological safety, the shaping unit 250 evaluates the environment as a whole by calculating a statistical value (e.g., average) of the "psychological safety value" of each participant participating in the group work obtained by the estimation unit 230, and makes a judgment such as "good" based on the statistical value. For example, the judgment is made by comparing the statistical value with a threshold value. Evaluation using a machine learning model or the like is also possible.

[0058] Regarding the quality of the group work, the shaping unit 250 synthesizes (for example, calculates the sum of) the statistical values ​​of psychological safety, effectiveness, and evaluation of the environment, and makes a judgment such as "good" based on the synthesized value. The "intervention means" shown in Figure 10 will be described later.

[0059] <About the intervention> In this embodiment, the intervention means determination unit 240 and the real-time intervention unit 260 enable, for example, an administrator who manages group work to intervene in the group work.

[0060] The intervention means determination unit 240 determines an intervention means (e.g., instructing a change in the speech ratio) as necessary. The determination method may be rule-based, or may use intervention means recommendations that utilize machine learning or the like with various conditions as input.

[0061] The real-time intervention unit 260 has a function of supporting intervention in real time based on the intervention means determined by the intervention means determination unit 240. For example, the real-time intervention unit 260 makes it possible to introduce the intervention means to a manager or the like in real time via the display unit 270.

[0062] For example, when the intervention means determination unit 240 detects a member who rarely smiles during a meeting based on the observation data obtained from the observation data conversion unit 220 or the subjective data obtained from the estimation unit 230, it determines that the intervention means is to implement an intervention to make that member laugh. The real-time intervention unit 260 displays the intervention means (making a specific member laugh) in real time via the display unit 270, for example, on the terminal of the manager of the group work.

[0063] (Other configuration examples) Learning device 100 may have the configuration shown in Fig. 11. Learning device 100 shown in Fig. 11 includes an acquisition unit 160 that acquires observation data and subjective data in group work, and a learning unit 170 that uses the observation data and the subjective data to learn a model. Acquisition unit 160 may include, for example, "an observation data collection unit 110, an observation data conversion unit 120, a subjective data conversion unit 130, and an observation data / subjective data combination unit 140."

[0064] The estimating device 200 may have the configuration shown in Fig. 12. The estimating device 200 shown in Fig. 12 includes an acquiring unit 310 that acquires observation data for each participant, an estimating unit 320 that acquires subjective data by inputting the observation data for each participant into a trained model, and an evaluating unit 330 that outputs the quality of group work based on the subjective data.

[0065] The acquisition unit 310 may include, for example, an observation data collection unit 210 and an observation data conversion unit 220. The evaluation unit 330 may include, for example, a formatting unit 250 and a display unit 270.

[0066] (Example of hardware configuration) Any of the devices described in this embodiment (learning device 100, inference device 200) can be realized, for example, by causing a computer to execute a program. This computer may be a physical computer or a virtual machine on the cloud.

[0067] That is, the device can be realized by executing a program corresponding to the processing performed by the device using hardware resources such as a CPU and memory built into a computer. The program can be recorded on a computer-readable recording medium (such as a portable memory) and stored or distributed. The program can also be provided via a network such as the Internet or email.

[0068] Fig. 13 is a diagram showing an example of the hardware configuration of the computer. The computer in Fig. 13 includes a drive device 1000, an auxiliary storage device 1002, a memory device 1003, a CPU 1004, an interface device 1005, a display device 1006, an input device 1007, an output device 1008, and the like, all of which are interconnected via a bus B. The computer may further include a GPU.

[0069] A program for realizing processing on the computer is provided by a recording medium 1001 such as a CD-ROM or a memory card. When the recording medium 1001 storing the program is set in the drive device 1000, the program is installed from the recording medium 1001 to the auxiliary storage device 1002 via the drive device 1000. However, the program does not necessarily have to be installed from the recording medium 1001, but may be downloaded from another computer via a network. The auxiliary storage device 1002 stores the installed program as well as necessary files, data, etc.

[0070] The memory device 1003 reads and stores a program from the auxiliary storage device 1002 when an instruction to start the program is received. The CPU 1004 realizes the functions related to the device in accordance with the program stored in the memory device 1003. The interface device 1005 is used as an interface for connecting to a network, etc. The display device 1006 displays a GUI (Graphical User Interface) or the like according to the program. The input device 1007 is composed of a keyboard, mouse, buttons, a touch panel, etc., and is used to input various operation instructions. The output device 1008 outputs the results of calculations.

[0071] (Effects of the embodiment) As described above, the technology according to the present embodiment makes it possible to efficiently estimate the quality of group work without asking subjective opinions from the participants of the group work.

[0072] The following additional notes are provided regarding the above-described embodiments.

[0073] <Additional Notes> (Additional note 1) An estimation device that estimates the quality of group work from observation data of the group work in which a plurality of participants participate, comprising: an acquisition unit that acquires observation data for each participant; an estimation unit that estimates subjective data by inputting the observation data for each participant into a trained model; an evaluation unit that outputs the quality of the group work based on the subjective data estimated from the observation data; An estimation device comprising: (Additional note 2) The acquisition unit collects observation data of each participant, calculates statistics of the collected observation data, and inputs the statistics to the estimation unit. Item 1. The estimation device according to item 1. (Additional note 3) The evaluation unit determines the quality of the group work based on the statistical values ​​of the subjective data for each participant. Item 1. The estimation device according to item 1. (Additional note 4) A learning device for learning a model that estimates subjective data from observed data in group work involving a plurality of participants, comprising: an acquisition unit that acquires the observation data and the subjective data in the group work; a learning unit that learns the model using the observation data and the subjective data; A learning device comprising: (Additional note 5) An estimation method executed by an estimation device for estimating quality of group work in which a plurality of participants participate, based on observation data of the group work, comprising: acquiring observation data for each participant; estimating subjective data by inputting the observed data for each participant into a trained model; a step of outputting the quality of the group work based on the subjective data estimated from the observation data; An estimation method comprising: (Additional note 6) A learning method executed by a learning device for learning a model that estimates subjective data from observed data in group work involving a plurality of participants, comprising: a step of acquiring the observation data and the subjective data in the group work; training the model using the observed data and the subjective data; A learning method that includes: (Additional note 7) A non-transitory storage medium storing a program for causing a computer to function as each unit in the estimation device described in any one of appended claims 1 to 3. (Additional note 8) A non-transitory storage medium storing a program for causing a computer to function as each part of the learning device described in appendix 4.

[0074] Although the present embodiment has been described above, the present invention is not limited to such a specific embodiment, and various modifications and changes are possible within the scope of the gist of the present invention described in the claims. [Explanation of symbols]

[0075] 10 devices 100 Learning Device 110 Observation Data Collection Department 120 Observation data conversion unit 130 Subjective Data Conversion Unit 140 Observation data and subjective data combination section 150, 170 Learning Department 160, 310 Acquisition Department 200 Estimation device 210 Observation Data Collection Department 220 Observation Data Conversion Unit 230 Estimation Department 240 Intervention Means Determination Unit 250 Plastic surgery department 260 Real-time Intervention Section 270 Display section 300 Network 320 Estimation Department 330 Evaluation Department 1000 Drive Device 1001 Recording media 1002 Auxiliary storage 1003 Memory device 1004 CPU 1005 Interface device 1006 Display device 1007 Input Device 1008 Output Device

Claims

1. An estimation device that estimates the quality of group work from observation data of the group work in which a plurality of participants participate, comprising: an acquisition unit that acquires observation data for each participant; an estimation unit that estimates subjective data by inputting the observation data for each participant into a trained model; an evaluation unit that outputs the quality of the group work based on the subjective data estimated from the observation data; An estimation device comprising:

2. The acquisition unit collects observation data of each participant, calculates statistics of the collected observation data, and inputs the statistics to the estimation unit. The estimation device according to claim 1 .

3. The evaluation unit determines the quality of the group work based on the statistical values ​​of the subjective data for each participant. The estimation device according to claim 1 .

4. A learning device for learning a model that estimates subjective data from observed data in group work involving a plurality of participants, comprising: an acquisition unit that acquires the observation data and the subjective data in the group work; a learning unit that learns the model using the observation data and the subjective data; A learning device comprising:

5. An estimation method executed by an estimation device for estimating quality of group work in which a plurality of participants participate, based on observation data of the group work, comprising: acquiring observation data for each participant; estimating subjective data by inputting the observed data for each participant into a trained model; a step of outputting the quality of the group work based on the subjective data estimated from the observation data; An estimation method comprising:

6. A learning method executed by a learning device for learning a model that estimates subjective data from observed data in group work involving a plurality of participants, comprising: a step of acquiring the observation data and the subjective data in the group work; training the model using the observed data and the subjective data; A learning method that includes:

7. A program for causing a computer to function as each unit of the estimation device according to any one of claims 1 to 3.

8. A program for causing a computer to function as each unit in the learning device according to claim 4.

Images