Diagnosis system and program

Abstract

To improve the objectivity when an evaluation for a commodity is made based on user questionnaires, and to suppress a development to complaints from users even if the evaluation result is not good.SOLUTION: A diagnosis system includes: novel questionnaire obtaining means for obtaining questionnaire data 13a obtained by causing reply details from a user 3 to a questionnaire for the satisfaction level for a commodity 1 to be made in the form of data; an inferencing model 13c that has made machine-learning on, by what corresponds to the plurality of users, first questionnaire data that is antecedent data in the pieces of questionnaire data obtained in past from the plurality of users 2, second questionnaire data that is subsequent data from the first questionnaire, and determination data representing the determination result on whether or not the satisfaction level for the commodity in the second questionnaire is improved, for each user 2; and predicting means 13d for predicting, using the inferencing model 13c, whether the satisfaction level for the commodity in the novel questionnaire data 13a becomes worse or better.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a diagnostic system and a program. [Background technology]

[0002] Patent Document 1 discloses a technology related to an information processing system that outputs a health-related questionnaire to the residents of a house multiple times at predetermined intervals and acquires the response data entered by the residents, thereby evaluating whether living in the house will improve their health based on the residents' actual experiences. The questionnaire evaluation is conducted by researchers who are studying methods for evaluating the health of homes. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2024-2200 A Summary of the Invention [Problem to be solved by the invention]

[0004] In conventional information processing systems, researchers evaluate whether a home has a health improvement effect based on subjective questionnaires from residents, in accordance with their own research, and the results of the evaluation lack objectivity. Furthermore, if the evaluation results show that there is no hope for health improvement, no support can be provided to residents, which could lead to complaints from residents.

[0005] The present invention has been made in consideration of the above circumstances, and its purpose is to improve the objectivity when evaluating products based on user surveys, and to make it less likely that unfavorable evaluations will lead to complaints from users. [Means for solving the problem]

[0006] The invention described in claim 1 is a diagnostic system, as shown in, for example, FIGS. 1 to 5, a new questionnaire data acquisition means (operation reception unit 16) for newly acquiring questionnaire data 13a obtained by converting the answers of a user 3 (new user 3) to a questionnaire about satisfaction with product 1 (housing product 1); an inference model 13c that performs machine learning for each of a plurality of users 2 on the survey data 13a previously acquired from a plurality of users 2 (existing users 2), including first survey data obtained by converting a first survey A1 conducted earlier into data, second survey data obtained by converting a second survey A2 conducted later than the first survey into data, and judgment data indicating the results of judging whether the product satisfaction level in the second survey has worsened or improved compared to the product satisfaction level in the first survey A1; and The system is characterized by having a prediction means 13d (prediction program 13d) that predicts, using the inference model 13c, whether the satisfaction level of the product in the new questionnaire data 13a acquired by the new questionnaire data acquisition means will deteriorate or improve in the future.

[0007] According to the invention described in claim 1, the prediction means 13d predicts whether product satisfaction in new questionnaire data 13a acquired by the new questionnaire data acquisition means will deteriorate or improve using an inference model 13c machine-learned based on questionnaire data 13a previously acquired from multiple users. Therefore, even if the inference model 13c is generated based on the subjective opinions of past users 2, it is possible to statistically predict whether product satisfaction in the newly acquired questionnaire data will deteriorate or improve. Therefore, compared to a case where a researcher creates an inference model 13c based on their own research after looking at a questionnaire survey based on the subjective opinions of past users 2, the inference model 13c is a more objective diagnostic indicator. Therefore, it is possible to more objectively predict and diagnose whether user 3's evaluation of product 1 will deteriorate in the future. If such a prediction becomes possible, a support system for user 3 can be established even if user 3's evaluation of product 1 is unfavorable, making it less likely that user 3's evaluation of product 1 will lead to complaints.

[0008] The invention described in claim 2 is, for example, as shown in Figs. 2 to 4, in the diagnostic system described in claim 1, The first questionnaire A1 and the second questionnaire A1 have a plurality of questionnaire items with common content, The inference model 13c further performs machine learning of multiple levels of evaluation data for each of the questionnaire items for multiple users 2, The prediction means 13d is characterized in that, when it predicts that the satisfaction level of the product 1 in the new questionnaire data 13a will deteriorate in the future, it extracts characteristic questionnaire items from the plurality of questionnaire items that indicate that the evaluation data will deteriorate in the future.

[0009] According to the invention of claim 2, the inference model 13c performs machine learning on multiple levels of evaluation data for multiple questionnaire items with common content for each user 2 for multiple users, and therefore the prediction means 13d can predict whether the satisfaction level of the product in the new questionnaire data 13a acquired by the new questionnaire data acquisition means will deteriorate or improve in the future using the inference model 13c that has machine-learned the evaluations of the product 1 by past users 2. Then, when the prediction means 13d predicts that the satisfaction level of the product 1 in the new questionnaire data 13a will deteriorate in the future, it extracts characteristic questionnaire items from the multiple questionnaire items that indicate that the evaluation data will deteriorate in the future, so that a support system for the characteristic questionnaire items can be established, and the evaluation of the product 1 by the user 3 is less likely to lead to a complaint.

[0010] The invention described in claim 3 is, for example, as shown in Figs. 2 to 4, in the diagnostic system described in claim 2, The inference model 13c ranks the characteristic questionnaire items, The prediction means 13d is characterized in that, when it predicts that the satisfaction level for product 1 in the new questionnaire data 13a will deteriorate, it extracts a specific characteristic questionnaire item from among the plurality of characteristic questionnaire items according to the ranking.

[0011] According to the invention described in claim 3, when the prediction means 13d predicts that the satisfaction level of the product 1 in the new questionnaire data 13a will deteriorate, it extracts a specific characteristic questionnaire item from among the multiple characteristic questionnaire items according to the ranking, so that an appropriate support system can be established in accordance with the extracted specific characteristic questionnaire item, and the evaluation of the product 1 by the user 3 is less likely to lead to a complaint.

[0012] The invention described in claim 4 is, for example, as shown in Figs. 2 to 4, in the diagnostic system described in claim 3, The method further includes a storage unit (13) in which comments are stored for each of the characteristic questionnaire items to prevent the evaluation of the product (1) from worsening in the future, The prediction means 13d is characterized in that, when extracting the specific characteristic questionnaire item, it associates the comment stored in the storage unit 13 with the specific characteristic questionnaire item.

[0013] According to the invention described in claim 4, when extracting a specific characteristic questionnaire item, the prediction means 13d associates it with comments stored in the memory unit 13, so that by checking comments to prevent the evaluation of product 1 from worsening in the future and establishing a support system in accordance with those comments, it becomes less likely that user 3's evaluation of product 1 will lead to a complaint.

[0014] The invention described in claim 5 is, for example, as shown in Figs. 1 to 5, in the diagnostic system described in claim 4, The system further includes a notification unit 14 (communication unit 14) that notifies external devices 5a and 6a of the prediction result by the prediction unit 13d, The notification means 14 is characterized in that it notifies the external devices 5a, 6a of the comments corresponding to the plurality of characteristic questionnaire items that are ranked highest among the plurality of characteristic questionnaire items that have been ranked.

[0015] According to the invention described in claim 5, the notification means 14 notifies the external devices 5a, 6a of comments corresponding to the highest ranked characteristic questionnaire items among the ranked characteristic questionnaire items, so that people who see the notification to the external devices 5a, 6a can prepare an appropriate support system in accordance with the comments, and the evaluation of the product 1 by the user 3 is less likely to lead to a complaint.

[0016] The invention of claim 6 is, for example, as shown in Figs. 2 to 4, in the diagnostic system of any one of claims 3 to 5, an inference model generation means 13b for performing machine learning on the first questionnaire data, the second questionnaire data, and the judgment data for a plurality of users 2 to generate an inference model 13c; The inference model generating means 13b is characterized in that it ranks the questionnaire items in order of the items that are characterized by the worsening judgment.

[0017] According to the invention described in claim 6, the inference model generation means 13b ranks the questionnaire items in order of their tendency to deteriorate, so that a support system can be established by focusing on questionnaire items that are prone to deterioration, making it less likely that the evaluation of product 1 by user 3 will lead to a complaint.

[0018] The invention described in claim 7 is, for example, as shown in FIGS. 1 to 5, a program, The computer 10 is capable of referencing an inference model 13c that has been machine-learned for multiple users 2 using survey data 13a that has been digitized from the responses of a user 3 to a survey about satisfaction with a housing product 1 that was previously obtained from multiple owners 2, including first survey data 13a that has been digitized from a first survey A1 that was conducted earlier, second survey data 13a that has been digitized from a second survey A2 that was conducted after the first survey A1, and judgment data that indicates the results of a judgment on whether satisfaction with the housing product 1 in the second survey A2 has worsened or improved compared to satisfaction with the product 1 in the first survey A1.The computer 10 is characterized in that it functions as a new survey data acquisition means that newly acquires survey data 13a that has been digitized from the responses of a user 3 to a survey about satisfaction with the product 1, and a prediction means 13d that uses the inference model 13c to predict whether satisfaction with the product 1 in the new survey data obtained by the new survey data acquisition means will deteriorate or improve in the future.

[0019] According to the invention described in claim 7, a computer 10 capable of referencing the inference model 13c functions as a new questionnaire data acquisition means for newly acquiring questionnaire data 13a, which is a digitalized version of User 3's responses to a questionnaire about satisfaction with Product 1, and a prediction means 13d for predicting, using the inference model 13c, whether satisfaction with Product 1 will deteriorate or improve in the new questionnaire data acquired by the new questionnaire data acquisition means. Even if the inference model 13c is generated based on the subjective opinions of User 2 in the past, it is possible to statistically predict whether product satisfaction will deteriorate or improve in the new questionnaire data acquired by the new questionnaire data acquisition means. Therefore, compared to, for example, a case in which a researcher creates the inference model 13c based on a questionnaire survey based on User 2's subjective opinions in the past based on their own research, the inference model 13c serves as a more objective diagnostic indicator. Therefore, it is possible to more objectively predict whether User 3's evaluation of Product 1 will deteriorate in the future. Such predictions enable a support system for User 3 to be established even when User 3's evaluation of Product 1 is unfavorable, making it less likely that User 3's evaluation of Product 1 will lead to complaints. [Effects of the Invention]

[0020] According to the present invention, it is possible to improve the objectivity of product evaluations based on user questionnaires, and to make it less likely that the evaluations will lead to complaints from users even if the evaluation results are unfavorable. [Brief explanation of the drawings]

[0021] [Figure 1] FIG. 1 is a diagram illustrating a configuration of a diagnostic system. [Figure 2] FIG. 2 is a block diagram showing the configuration of a diagnostic device. [Figure 3] A diagram showing an example of an inference model as a table. [Figure 4] 10 is a flowchart illustrating the procedure for generating an inference model. [Figure 5] 10 is a flowchart illustrating a procedure for diagnosing a first questionnaire of a new owner. DETAILED DESCRIPTION OF THE INVENTION

[0022] Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, although the embodiments described below are subject to various technically preferable limitations for carrying out the present invention, the technical scope of the present invention is not limited to the following embodiments and illustrated examples.

[0023] <Overview of the diagnostic system> 1 shows an overview of a diagnostic system that diagnoses whether an owner 3 (new owner 3)'s evaluation of a housing product 1 will worsen (whether it will worsen or improve). The new owner 3's evaluation of the housing product 1 is diagnosed by a diagnostic device 10 that acquires data from a questionnaire (first questionnaire A1) sent to the new owner 3.

[0024] In FIG. 1, reference numeral 1 indicates a housing product. This housing product 1 is a home sold as a product by a housing manufacturer HM, and in this embodiment refers to a newly built detached home. However, the housing product 1 is not limited to this, and may be a single dwelling unit in an apartment building, or the apartment building itself. Furthermore, instead of a newly built home, it may be a renovated property (a used home), or renovation work carried out on a home. In other words, there are no particular limitations as long as it can be treated as housing product 1. Furthermore, a questionnaire may be obtained by classifying the homes according to their attributes. Note that attributes include, in addition to whether the home is a detached home or an apartment building, various other attributes such as the construction area of ​​the home, the construction direction of the home, the number of floors of the home, a two-generation home, and the presence or absence of a private power generation device such as a solar power generation system.

[0025] The housing product 1 was purchased by multiple other owners 2 (existing owners 2) before the new owner 3 purchased the housing product 1. The housing manufacturer HM sent a first questionnaire A1 and a second questionnaire A2 to the multiple existing owners 2 who had already purchased the housing product 1 after a specified period of time. The multiple existing owners 2 then answered each item in the first questionnaire A1 and the second questionnaire A2 and sent them back to the housing manufacturer HM. The new owner 3 does not necessarily have to be a person who is purchasing the housing product 1 for the first time, and the same person may be included among multiple existing owners 2.

[0026] An input person 4 of the housing manufacturer HM uses an input device 4a such as a personal computer to input the answers to each item in the first questionnaire A1 and the second questionnaire A2 sent by multiple existing owners 2 into the diagnostic device 10 as questionnaire data 13a. The questions (questionnaire items) of the first questionnaire A1 and the second questionnaire A2 can be answered using multiple levels of evaluation, such as a five-point scale. The multiple-level evaluations for each questionnaire item are also converted into data one by one and input into the diagnostic device 10. The questionnaire data 13a includes identification information for identifying each owner and each housing product 1.

[0027] The diagnostic device 10 performs learning using the input questionnaire data 13a as learning data. Then, the diagnostic device 10 generates an inference model 13c for diagnosing whether the new owner 3's evaluation of the housing product 1 will deteriorate in the future, based on the content of the first questionnaire A1 completed by the new owner 3 about the housing product 1.

[0028] Furthermore, when the diagnostic device 10 acquires the first questionnaire A1 completed by the new owner 3 regarding the housing product 1, it compares the first questionnaire A1 with the inference model 13c. If the diagnostic result is unfavorable (if the condition is on a worsening trend), it notifies each customer representative accordingly.

[0029] Each customer representative includes a regular patrol representative 5 who periodically visits the owners 2 and 3 of the housing product 1, and an inspection representative 6. In addition, the notification destinations to which notifications are sent also include other notification destinations 7 (pre-set notification destinations 7) that have been set in advance. The regular patrol person 5 and the inspection person 6 use information terminals 5a, 6a such as personal computers and smartphones, and notifications from the diagnostic device 10 are sent by an appropriate method such as e-mail.

[0030] The customer representatives 5, 6 and the preset notification destination 7 who receive the notification can recognize that the first questionnaire A1 answered by the new owner 3 is on a worsening trend, and can prepare a support system for the new owner 3. Therefore, even if the evaluation result of the new owner 3 in the first questionnaire A1 is not good, it is possible to prevent a complaint from the new owner 3.

[0031] The input device 4a and the diagnostic device 10, the diagnostic device 10 and the information terminal 5a, the diagnostic device 10 and the information terminal 6a, and the diagnostic device 10 and the preset notification destination 7 are communicably connected to each other via a communication network. The communication network is not particularly limited, but may be, for example, the Internet, a LAN (Local Area Network), a WAN (Wide Area Network), or the like.

[0032] <Configuration of diagnostic equipment> 2 is a block diagram showing the configuration of a diagnostic device 10 of this embodiment, which may be a general PC (Personal Computer). The diagnostic device 10 includes a CPU (Central Processing Unit) 11 (processing unit), a RAM (Random Access Memory) 12, a storage unit 13, a communication unit 14, a display unit 15, an operation reception unit 16, etc. The diagnostic device 10 may also be a server.

[0033] The CPU 11 is a processor that performs arithmetic processing and controls the overall operation of the diagnostic device 10. The CPU 11 may have a single processor, or may have multiple processors that perform processing in parallel or independently depending on the application. The RAM 12 provides a working memory space for the CPU 11 and stores temporary data. The RAM 12 is, for example, a DRAM, but is not limited to this. The CPU 11 and RAM 12 function as a control unit that performs predetermined calculations and controls each unit by executing various programs stored in the storage unit 13. That is, the CPU 11 reads out the programs stored in the storage unit 13, expands them in the RAM 12, and executes the programs to perform various calculation processes.

[0034] The storage unit 13 is a non-volatile memory that stores the program 131 and various setting data. The non-volatile memory may include a hard disk drive (HDD). The storage unit 13 may be externally attached to the information processing device as a peripheral device. Alternatively, the storage unit 13 may be a network storage or a cloud server located on a network. The storage unit 13 also stores the above-mentioned questionnaire data 13a, inference model generation program 13b, inference model 13c, and prediction program 13d.

[0035] The communication unit 14 transmits and receives data to and from external devices in accordance with a communication standard, which may include various standards such as TCP / IP relating to a LAN (Local Area Network).

[0036] The display unit 15 has a display screen and displays information on the display screen under the control of the CPU 11. The display screen is, for example, a liquid crystal display screen, but is not limited to this. The display unit 15 may also have an LED lamp or the like for indicating the operating status, etc.

[0037] The operation reception unit 16 receives an input operation from the input device 4a by the input person 4 of the housing builder HM, and outputs an input signal to the CPU 11 according to the received content. The display unit 15 and the operation reception unit 16 may be configured separately from the information processing device 1 and may be attached and operated as peripheral devices.

[0038] <About the inference model> The inference model 13c is a trained model generated based on questionnaire data 13a including the content of responses to each item in a first questionnaire A1 and questionnaire data 13a including the content of responses to each item in a second questionnaire A2. More specifically, the inference model 13c is a trained model obtained by machine learning, for each user 2, on the following questionnaire data 13a (a group of questionnaires) previously acquired from multiple users 2: first questionnaire data 13a obtained by converting an earlier first questionnaire A1 into data; second questionnaire data 13a obtained by converting a second questionnaire A2 into data from a second questionnaire A2 conducted after the first questionnaire A1; and judgment data indicating the results of a judgment on whether the satisfaction level with the housing product 1 in the second questionnaire A2 has worsened or improved compared to the satisfaction level with the housing product 1 in the first questionnaire A1. The judgment data is derived by a judgment means, which is a function of a learning unit included in the diagnostic device 10, and is capable of comparing the earlier first questionnaire data with the later second questionnaire data to determine whether the product satisfaction level in the later second questionnaire data has worsened or improved compared to the earlier first questionnaire data. Such an inference model 13c is generated by executing the inference model generation program 13b at any time, such as once every few months, or when a certain amount of questionnaire data 13a has been accumulated.

[0039] Although the first questionnaire A1 and the second questionnaire A2 are not composed of questionnaire items (questions) with completely identical content, there are common questionnaire items that are correlated between the first questionnaire A1 and the second questionnaire A2, such as overall satisfaction, evaluation of sales, details related to defects, etc. Therefore, when the contents of the first questionnaire A1 and the second questionnaire A2 are accumulated as questionnaire data 13a, characteristic questionnaire items that are subject to a worsening trend will be identified, for example, as shown in Figure 3.

[0040] 3 is a diagram showing the inference model 13c as a table. The inference model 13c includes multiple questionnaire items in the questionnaire data 13a, ranks set for these multiple questionnaire items, and corresponding comments for each rank that are mastered for each questionnaire item.

[0041] The above multiple questionnaire items are questionnaire items listed in the first questionnaire A1 and the second questionnaire A2, and the inference model 13c may include all of the questionnaire items listed in the first questionnaire A1 and the second questionnaire A2, or may include only common questionnaire items that are correlated.

[0042] The above rankings are set for all of the common questionnaire items listed in the first questionnaire A1 and the second questionnaire A2. Because the rankings are set in this way, all of the common questionnaire items are weighted. In other words, there are questionnaire items with a high level of importance and those with a low level of importance. The importance is set in the order of questionnaire items that are characteristic of a worsening judgment in the questionnaire group acquired from existing owners 2 whose judgment results regarding satisfaction with product 1 indicated by the judgment data are a worsening judgment. In other words, for example, when the first questionnaire A1 and the second questionnaire A2 are collected from multiple (many) existing owners 2, the importance is set to the questionnaire items that appear most frequently among the questionnaire items that appear characteristically frequently in the questionnaire data 13a that indicate a worsening trend. Characteristic questionnaire items that are subject to a deterioration trend are items (questions) that are presumed to be the cause of deterioration in the questionnaire data 13a of a type that has deteriorated in evaluation as a result of accumulating questionnaire data 13a collected from multiple (large number) existing owners 2. In other words, at the stage of collecting the first questionnaire A1 and the second questionnaire A2 from multiple (large number) existing owners 2, questionnaire items that appear characteristically frequently in the questionnaire data 13a that shows a deterioration trend are extracted as characteristic questionnaire items that are subject to a deterioration trend. In this embodiment, among the multiple questionnaire items, the questionnaire item "Overall evaluation of sales representative in the sales-related questionnaire in Question 4" is assigned a rank of 1. Therefore, this questionnaire item is the most frequently occurring item characteristic of a worsening trend. In other words, this means that the overall evaluation of the sales representative has a significant impact on the deterioration of the evaluation of Housing Product 1 and, ultimately, of Housing Builder HM. In addition, the survey item "Defects after delivery in the survey regarding defects in question 10" was assigned a rank of 2. In addition, the survey item "Explanation of specifications and prices in the sales-related survey (Question 4)" is assigned a rank of 3. In addition, a rank of 4 was set for the survey item "Overall building evaluation in the satisfaction survey (Question 1)."

[0043] The above response comments are ranked responses mastered for each questionnaire item, and include advice for customer representatives 5 and 6, as well as measures to prevent the evaluation of housing product 1 and housing manufacturer HM from worsening in the future. Since the rank of a questionnaire item may change each time, the corresponding comment must also be changed for each rank. Therefore, the corresponding comments are mastered (stored in a database) for each questionnaire item, and if the rank setting is changed during the generation stage of the inference model 13c, the corresponding comment is also changed. Therefore, a large number of corresponding comments are stored in the database in the memory unit 13.

[0044] <Generating inference models> FIG. 4 is a flowchart illustrating the procedure of the inference model 13c. To generate the inference model 13c using the inference model generation program 13b, it is necessary to collect the first questionnaire A1 and the second questionnaire A2 from multiple existing owners 2 who purchased the housing product 1. That is, as described above, the data entry person 4 of the housing manufacturer HM uses the input device 4a to input the answers to each questionnaire item in the first questionnaire A1 and the second questionnaire A2 sent by the multiple existing owners 2 into the diagnostic device 10 as questionnaire data 13a, and the operation reception unit 16 of the diagnostic device 10 receives the input operation from the input device 4a by the data entry person 4 of the housing manufacturer HM and outputs an input signal according to the received content to the CPU 11.

[0045] More specifically, immediately after a certain owner (in this case, deemed to be a new owner 3) purchases a housing product 1, a first questionnaire A1 is sent to the owner. Then, the first questionnaire A1, in which the owner has filled in the answers to each questionnaire item, is sent to the housing manufacturer HM. Thereafter, the data entry person 4 at the housing manufacturer HM uses the input device 4a to input the contents of the first questionnaire A1 sent from the owner into the diagnostic device 10 (step S1). In other words, the diagnostic device 10 can acquire the first questionnaire data. At this time, the first questionnaire A1 of the owner who is regarded as the new owner 3 is assumed to have been diagnosed by the diagnostic device 10 as to whether the evaluation of the housing product 1 (and ultimately the evaluation of the housing manufacturer HM) will deteriorate in the future. Furthermore, the owner who purchased the housing product 1 is considered to be an existing owner 2 after the contents of the first questionnaire A1 are diagnosed by the diagnosis device 10.

[0046] Subsequently, after a certain period of time (for example, one year), a second questionnaire A2 is sent to the owner (at this time, the owner has changed from new owner 3 to existing owner 2). The second questionnaire A2, in which the owner has filled in the answers to each questionnaire item, is sent to the home builder HM. Thereafter, the data entry person 4 at the home builder HM uses the input device 4a to input the contents of the second questionnaire A2 sent from the owner into the diagnostic device 10 (step S2). In other words, the diagnostic device 10 can acquire the second questionnaire data.

[0047] Until the second questionnaire A2 is sent from the same owner, other owners also send second questionnaires A2, but the timing at which the inference model 13c is generated (regenerated) is arbitrary. For example, the inference model 13c is generated periodically, such as once every few months, or when a certain amount of questionnaire data 13a has been accumulated. In other words, the inference model generation program 13b is set to be executed when predetermined conditions are met, and it is confirmed each time whether the conditions are met (step S3). Then, if a predetermined condition is met, the inference model generation program 13b is executed (step S4). In short, the inference model 13c stored in the diagnostic device 10 periodically learns about the housing product 1, and the learning method is to compare the first questionnaire A1 and the second questionnaire A2 for the same owner. Therefore, while it is desirable that the first questionnaire A1 and the second questionnaire A2 for the same owner are completed, not all owners necessarily fill out the questionnaires, so the inference model generation program 13b is executed at a predetermined timing to learn about the housing product 1. Then, the inference model 13c is updated. Note that at this time, the responses of owners who have not completed the first questionnaire A1 and the second questionnaire A2 are not reflected in the inference model 13c.

[0048] When the inference model generation program 13b is executed, first, a second questionnaire A2 is extracted in which the evaluation of the housing product 1 has deteriorated. Then, the second questionnaire A2 is compared with the first questionnaire A1 of the same owner 2, and the questionnaire items that are thought to be the cause of the deterioration of the evaluation of the housing product 1 are extracted (step S5). To extract characteristic survey items that are the subject of such a deterioration trend, items that appear frequently in the survey data 13a that indicates a deterioration trend are selected from the survey data 13a collected from multiple (large number) existing owners 2.

[0049] Next, among the questionnaire data 13a collected from multiple (large number) existing owners 2, the questionnaire items that appear frequently and characteristically in the questionnaire data 13a showing a deterioration trend are ranked (step S6). In this embodiment, as described above, rank 1 is set for the questionnaire item "Overall evaluation of sales representative in the questionnaire regarding sales in question 4." Furthermore, rank 2 is set for the questionnaire item "Defects after delivery in the questionnaire regarding defects in question 10." Furthermore, rank 3 is set for the questionnaire item "Explanation of specifications and prices in the questionnaire regarding sales in question 4." Furthermore, rank 4 is set for the questionnaire item "Overall evaluation of the building in the questionnaire regarding satisfaction in question 1."

[0050] Next, the response comments corresponding to each questionnaire item are mastered by rank (step S7). That is, when the first questionnaire A1 and the second questionnaire A2 are collected from multiple existing owners 2 and the inference model 13c is generated (regenerated), the questionnaire items that are thought to be the cause of the deterioration in the evaluation of the housing product 1 (characteristic questionnaire items that are subject to a deterioration trend) may change, or the rank may change even if such questionnaire items do not change. In such cases, the response comments (advice and countermeasures) made to the customer representatives 5 and 6 also need to be changed, so when the inference model generation program 13b is executed, the response comments are also reviewed and changed as necessary.

[0051] The inference model 13c is generated as described above. The inference model 13c generated in this manner is used to predict whether the evaluation of the housing product 1 (and ultimately the evaluation of the housing manufacturer HM) will deteriorate in the future when the first questionnaire A1 is sent from the new owner 3.

[0052] <About the diagnosis in the first questionnaire> Figure 5 is a flowchart explaining the procedure for diagnosing the first questionnaire A1 of the new owner 3 using the inference model 13c generated as described above. Whether the second questionnaire A2 will tend to deteriorate or not can be determined depending on how the new owner 3, who has just purchased the housing product 1, evaluates the housing product 1 in the first questionnaire A1. It is assumed that the inference model 13c has been generated in advance.

[0053] When making a diagnosis, first, the first questionnaire A1 answered by the new owner 3 is obtained (step S10). That is, the data entry person 4 of the housing manufacturer HM uses the input device 4a to input the answers to each questionnaire item in the first questionnaire A1 sent by the new owner 3 into the diagnostic device 10 as questionnaire data 13a, and the operation reception unit 16 of the diagnostic device 10 receives the input operation from the input device 4a by the data entry person 4 of the housing manufacturer HM and outputs an input signal according to the received content to the CPU 11.

[0054] Next, the first questionnaire A1 is diagnosed (step S11). That is, the CPU 11 reads out the prediction program 13d for diagnosing the first questionnaire A1 stored in the storage unit 13, loads it into the RAM 12, and executes the program 13d to start the predictive diagnosis of the first questionnaire A1. The CPU 11 functions as a prediction means in cooperation with the prediction program 13d. At this time, the prediction program 13d uses the inference model 13c stored in the storage unit 13 (step S12). Then, when the prediction program 13d is executed, the contents of the first questionnaire A1 answered by the new owner 3 are compared with the inference model 13c, and a predictive diagnosis is made as to whether the first questionnaire A1 is showing a worsening or improving trend.

[0055] As a result of executing the prediction program 13d (step S13), if the first questionnaire A1 is diagnosed as showing a worsening trend, the process proceeds to step S14, and if it is diagnosed as not showing a worsening trend (showing an improvement trend), the process proceeds to step S16. If a questionnaire item with a low rating in the first questionnaire A1 matches a questionnaire item with a high ranking in the inference model 13c, the first questionnaire A1 is determined to contain answers (questionnaire items) that are predicted to worsen in the future. In other words, the first questionnaire A1 obtained from the new owner 3 is diagnosed as being on a worsening trend.

[0056] If the first questionnaire A1 obtained from the new owner 3 is on a worsening trend, leaving it unattended will lead to a complaint, so it is necessary to notify the customer representatives 5 and 6 that the first questionnaire A1 is on a worsening trend and to provide comments on how to improve the evaluation. Therefore, first, from the corresponding comments included in the inference model 13c, the corresponding comments for the survey items that are ranked in the top three are extracted, and from the corresponding comments included in the inference model 13c, the corresponding comments for the survey items that received low ratings in the first survey A1 by the new owner 3 are extracted (step S14). In this embodiment, the top three survey items in the inference model 13c are, as described above, "Overall evaluation of sales representative in question 4" ranked 1, "Defects after delivery in question 10" ranked 2, and "Explanation of specifications and prices in question 4" ranked 3. Therefore, first, corresponding comments for these three survey items are extracted. Furthermore, response comments corresponding to the survey items that received low ratings in the first questionnaire A1 answered by the new owner 3 are also extracted from the inference model 13c. For example, if the survey item that received low ratings in the first questionnaire A1 answered by the new owner 3 was the survey item "Question 1: Overall Building Evaluation," the response comments for this item are extracted. Furthermore, if the survey item that received low ratings in the first questionnaire A1 answered by the new owner 3 overlaps with the top three survey items in the inference model 13c, additional response comments may be extracted, or a response comment corresponding to the survey item ranked 4 may be extracted. Furthermore, if there are multiple survey items that received low ratings in the first questionnaire A1 answered by the new owner 3, multiple response comments may be extracted.

[0057] Next, the customer representatives 5 and 6 (which may include the preset notification destination 7) are notified that the first survey A1 is on a worsening trend (step S15). The notification is recognized as an alert because it notifies that the first survey A1 is on a worsening trend. In this embodiment, the notification to the customer representatives 5 and 6 is sent by email as described above, but is not limited to this and can be changed as appropriate. Furthermore, the notification to the customer representatives 5 and 6 includes the above-mentioned response comments extracted in step S14, and the customer representatives 5 and 6 can check the response comments and respond to the customer.

[0058] If the first questionnaire A1 is diagnosed as not showing a worsening trend (showing an improvement trend) in step S13, the customer representatives 5 and 6 are not notified (step S16). Alternatively, they may be notified that the first questionnaire A1 is not showing a worsening trend.

[0059] In this way, the first questionnaire A1 of the new owner 3 can be diagnosed using the inference model 13c.

[0060] <About the effects> According to this embodiment, the following excellent effects are achieved. That is, the prediction program 13d predicts whether product satisfaction in new survey data 13a acquired by accepting input operations from the input device 4a will deteriorate or improve in the future using an inference model 13c machine-learned based on survey data 13a previously acquired from multiple users. Therefore, even if the inference model 13c is generated based on the subjective opinions of past users 2, it can predict from a statistical perspective whether product satisfaction in newly acquired survey data will deteriorate or improve in the future. Therefore, compared to, for example, a case in which a researcher creates an inference model 13c based on their own research after looking at a survey based on the subjective opinions of past users 2, the inference model 13c serves as a more objective diagnostic indicator. Therefore, it is possible to more objectively predict and diagnose whether user 3's evaluation of product 1 will deteriorate in the future. If such a prediction becomes possible, a support system for user 3 can be established even if user 3's evaluation of product 1 is unfavorable, making user 3's evaluation of product 1 less likely to lead to complaints.

[0061] Furthermore, since the inference model 13c has machine-learned the multi-level evaluation data for each of multiple questionnaire items with common content for each user 2, the prediction program 13d can use the inference model 13c that has machine-learned the evaluations of product 1 by past users 2 to predict whether the satisfaction level for the product in new questionnaire data 13a acquired by the new questionnaire data acquisition means will deteriorate or improve in the future. When the prediction program 13d predicts that the satisfaction level for product 1 in the new questionnaire data 13a will deteriorate in the future, it extracts characteristic questionnaire items from the multiple questionnaire items whose evaluation data indicates a tendency for the product to deteriorate in the future, thereby enabling a support system to be established for the characteristic questionnaire items, making it less likely that user 3's evaluation of product 1 will lead to a complaint.

[0062] Furthermore, when the prediction program 13d predicts that the satisfaction level for the product 1 in the new questionnaire data 13a will deteriorate, it extracts a specific characteristic questionnaire item from among the multiple characteristic questionnaire items according to the ranking, so that an appropriate support system can be established in accordance with the extracted specific characteristic questionnaire item, making it less likely that the evaluation of the product 1 by the user 3 will lead to a complaint.

[0063] Furthermore, when extracting a specific characteristic questionnaire item, the prediction program 13d associates it with comments stored in the memory unit 13, and by checking comments to prevent the evaluation of product 1 from worsening in the future and establishing a support system in accordance with those comments, it becomes less likely that user 3's evaluation of product 1 will lead to a complaint.

[0064] Furthermore, the notification means 14 notifies the external devices 5a, 6a of comments corresponding to the highest ranked characteristic questionnaire items among the ranked characteristic questionnaire items, so that people who see the notification sent to the external devices 5a, 6a can prepare an appropriate support system in accordance with the comments, making it less likely that the evaluation of the product 1 by the user 3 will lead to a complaint.

[0065] Furthermore, since the external devices 5a, 6a are information terminals 5a, 6a used by the customer service representatives 5, 6 of the seller of the housing product 1, if the new owner 3 gives a poor evaluation of the housing product 1, the customer service representatives 5, 6 can immediately provide support to the new owner 3. This makes it less likely that the evaluation of the housing product 1 by the new owner 3 will lead to a complaint from the new owner 3.

[0066] Furthermore, the inference model generation means 13b ranks the questionnaire items obtained from the user 2 whose judgment data indicates a worsening judgment regarding satisfaction with the product 1 in order of those that are characteristic of a worsening judgment, so that a support system can be established that focuses on the questionnaire items that are likely to worsen, making it less likely that the evaluation of the product 1 by the user 3 will lead to a complaint.

[0067] In recent years, there has also been a call to achieve the goals of the SDGs (Sustainable Development Goals), and various initiatives are being undertaken in the construction industry. In this embodiment, as described above, the points of dissatisfaction and concerns of the new owner 3 can be immediately grasped, and a support system can be established by the customer representatives 5 and 6, which makes it less likely that complaints will be made by the new owner 3. This makes it easier for the owner to feel attached to the house they purchased, which contributes to the creation of a city where people can continue to live, and ultimately contributes to the achievement of the SDGs (Goal 11).

[0068] [Modification] It should be noted that the embodiments to which the present invention can be applied are not limited to the above-described embodiments, and can be modified as appropriate without departing from the spirit of the present invention. Modifications will be described below. The following modifications may be combined as much as possible. In each of the following modifications, elements common to the above-described embodiments will be assigned the same reference numerals, and descriptions thereof will be omitted or simplified.

[0069] [Variation 1] In the above embodiment, the first questionnaire A1 and the second questionnaire A2 sent by the existing owner 2 and the first questionnaire A1 sent by the new owner 3 were input into the diagnostic device 10 as questionnaire data 13a by the data entry person 4 of the housing manufacturer HM using the input device 4a, and the operation reception unit 16 of the diagnostic device 10 received the input operation from the input device 4a by the data entry person 4 of the housing manufacturer HM and output an input signal according to the received content to the CPU 11. In contrast, in this modified example, the various questionnaires A1, A2 sent to the existing owner 2 and the new owner 3 are not on paper but are in the form of digital data. Furthermore, the destination of the digitalized questionnaires A1, A2 is not the input device 4a but the communication unit 14 of the diagnostic device 10. That is, the communication unit 14 of the diagnostic device 10 accepts the data of the various questionnaires A1, A2 (questionnaire data 13a) from the existing owner 2 and the new owner 3 and stores them in the storage unit 13. In this case, the information terminals used by the existing owner 2 and the new owner 3 and the diagnostic device 10 are connected to each other via a communication network so that they can communicate with each other.

[0070] This modification eliminates the need for the data entry staff 4 to input responses to the various questionnaires A1 and A2, thereby contributing to a reduction in labor costs and labor on the part of the housing manufacturer HM. It is also preferable because it reduces the costs of mailing the various questionnaires A1 and A2.

[0071] [Variation 2] In the above embodiment, the object of evaluation assessed by the assessment device 10 is a house-related item that has been sold, but it is not limited to this and may be a tangible product including high-value items such as automobiles, or an intangible product including various insurances, IT, medical (health checkups), and various services that are sold or provided. Also, a user may be referred to as a purchaser if they are a person who has purchased a product.

[0072] [Variation 3] In the above embodiment, when ranking characteristic questionnaire items, the questionnaire items were ranked in order of their tendency to worsen in a group of questionnaires acquired from users 2 whose judgment results regarding product 1 satisfaction indicated by the judgment data indicate a worsening judgment. In contrast, in this modified example, when generating the inference model 13c by executing the inference model generation program, the questionnaire items are ranked in order of their tendency to worsen in a group of questionnaires including users 2 whose judgment results regarding product 1 satisfaction indicated by the judgment data indicate an improvement judgment. In other words, the inference model 13c can be generated by machine learning even for questionnaires from users 2 whose satisfaction with product 1 is predicted to improve in the future. As a result, the inference model generation means 13b ranks the questionnaire items in order of their tendency to worsen, regardless of whether user 2's judgment results regarding product 1 satisfaction indicated by the judgment data indicate a worsening judgment or an improvement judgment. This allows for a support system that focuses on questionnaire items that are prone to worsening, making it less likely that user 3's evaluation of product 1 will lead to complaints. [Explanation of symbols]

[0073] HM Housing Manufacturer A1 First Questionnaire A2 Second Questionnaire 1 Housing products 2 Existing owners 3 New Owner 4. Data entry person 4a Input device 5. Regular patrol officer 5a Information terminal 6. Inspector 6a Information terminal 7 Pre-configured notification destinations 10 Diagnostic equipment 11 CPU 12 RAM 13 Storage section 13a Survey data 13b Inference model generator 13c Inference Model 13d forecasting program 14 Communications Department 15 Display section 16 Operation reception section

Claims

1. new survey data acquisition means for newly acquiring survey data that is a digitalized version of user responses to a survey about product satisfaction; an inference model obtained by machine learning for each of a plurality of users on the survey data previously acquired from the plurality of users, the first survey data being a compilation of a first survey conducted earlier, the second survey data being a compilation of a second survey conducted later than the first survey, and judgment data showing the results of judging whether the product satisfaction level in the second survey has worsened or improved compared to the product satisfaction level in the first survey; A diagnostic system characterized by comprising a prediction means for predicting, using the inference model, whether product satisfaction in the new survey data acquired by the new survey data acquisition means will deteriorate or improve in the future.

2. 10. The diagnostic system of claim 1, the first questionnaire and the second questionnaire have a plurality of questionnaire items with common content; The inference model further performs machine learning of multiple levels of evaluation data for each of the questionnaire items for multiple users, A diagnostic system characterized in that the prediction means, when predicting that product satisfaction in new questionnaire data will deteriorate in the future, extracts characteristic questionnaire items from the plurality of questionnaire items that indicate that the evaluation data will deteriorate in the future.

3. 3. The diagnostic system according to claim 2, the inference model ranks the characteristic questionnaire items; The diagnostic system is characterized in that the prediction means extracts specific characteristic questionnaire items from the plurality of characteristic questionnaire items in accordance with the ranking when it predicts that product satisfaction in new questionnaire data will deteriorate.

4. 4. The diagnostic system according to claim 3, a storage unit that stores comments for each of the plurality of questionnaire items to prevent the degree of satisfaction with the product from decreasing in the future; A diagnostic system characterized in that the prediction means, when extracting a specific characteristic questionnaire item, associates the comment stored in the storage unit with the characteristic questionnaire item.

5. 5. The diagnostic system according to claim 4, a notification unit that notifies an external device of a prediction result by the prediction unit; A diagnostic system characterized in that the notification means notifies the external device of the comments corresponding to the plurality of characteristic questionnaire items that are ranked highest among the plurality of characteristic questionnaire items that have been ranked.

6. 6. The diagnostic system according to claim 3, an inference model generation means for performing machine learning on the first questionnaire data, the second questionnaire data, and the judgment data for each user for a plurality of users to generate an inference model; A diagnostic system characterized in that the inference model generation means performs the ranking in order of questionnaire items that are characteristic of the worsening judgment.

7. a computer that can refer to an inference model that has been machine-learned for each of a plurality of users using, for each user, first questionnaire data that is data of a first questionnaire that was conducted earlier, second questionnaire data that is data of a second questionnaire that was conducted later than the first questionnaire, and judgment data that indicates the results of a judgment as to whether the product satisfaction level in the second questionnaire has worsened or improved compared to the product satisfaction level in the first questionnaire, among questionnaire data that has been previously obtained from a plurality of users and has been converted into data of the contents of responses to questionnaires about product satisfaction levels, new survey data acquisition means for newly acquiring survey data that is a digitalized version of user responses to a survey about product satisfaction; A program characterized by functioning as a prediction means that uses the inference model to predict whether product satisfaction in the new survey data acquired by the new survey data acquisition means will deteriorate or improve in the future.

Images