Response generation device, method, and program

Abstract

Provided is a conversion unit (34) that converts a negative expression included in a sentence recorded by a user into a positive expression, a calculation unit (36) that calculates points for each of a plurality of predetermined skills on the basis of the relationship between expressions included in the converted sentence and each of the plurality of predetermined skills, and a response generation unit (40) generates a response sentence for the converted sentence on the basis of the converted sentence and the points calculated for each of the plurality of skills.

Description

[Technical Field]

[0001] The present disclosure relates to a response generation device, a response generation method, and a response generation program. [Background technology]

[0002] Conventionally, a device has been proposed that provides information appropriate to the status of a poster who posted a message (for example, JP 2022-155034 A). This device acquires multiple messages posted by the poster, extracts multiple characteristic words that indicate the status of the poster from the multiple messages, and identifies changes in status that appear in the multiple messages posted by the poster by comparing the multiple characteristic words contained in the multiple messages. This device also creates a response message with content that corresponds to the identified changes. Summary of the Invention [Problem to be solved by the invention]

[0003] For example, members of the so-called Generation Z tend to feel guilty because they are unable to recognize the value of their actions, even when they are engaged in activities that are connected to their future or ideal selves. Also, even if they do something because they like it, if it does not lead to any beneficial results, they tend to feel guilty and feel like it is a waste of time.

[0004] In the above-mentioned conventional technology, when extracting information appropriate to the poster's status from the posted text and posting frequency, characteristic words are extracted, and changes in these words are associated with changes in the poster's status and a response based on a template is generated. Furthermore, in the conventional technology, response templates are determined in advance, and a template is selected based on the score to respond. In this case, a standard phrase is extracted from a phrase that indicates a certain characteristic, so a standard phrase related to the phrase is selected, which is a simple process. However, human emotions are complex, and multiple factors affect psychology. Furthermore, if there is no characteristic phrase, a comment may not be made.

[0005] The present disclosure has been made in consideration of the above points, and aims to provide a response generation device, method, and program that can generate response sentences that will increase the self-esteem and improve the motivation of users who feel guilty about their own words and actions. [Means for solving the problem]

[0006] The response generation device according to the present disclosure includes a conversion unit that converts negative expressions contained in a sentence recorded by a user into positive expressions, a calculation unit that calculates points for each of a plurality of predetermined skills based on the relationship between the expressions contained in the sentence converted by the conversion unit and each of the plurality of skills, and a response generation unit that generates a response sentence to the sentence converted by the conversion unit based on the sentence converted by the conversion unit and the points for each of the plurality of skills calculated by the calculation unit.

[0007] The response generation method according to the present disclosure is a response generation method executed by a response generation device including a conversion unit, a calculation unit, and a response generation unit, wherein the conversion unit converts negative expressions contained in a sentence recorded by a user into positive expressions, the calculation unit calculates points for each of a plurality of predetermined skills based on the relationship between the expressions contained in the sentence converted by the conversion unit and each of the plurality of predetermined skills, and the response generation unit generates a response sentence to the sentence converted by the conversion unit based on the sentence converted by the conversion unit and the points for each of the plurality of skills calculated by the calculation unit.

[0008] The response generation program of the present disclosure is a program for causing a computer to function as a conversion unit that converts negative expressions contained in a sentence recorded by a user into positive expressions, a calculation unit that calculates points for each of a plurality of predetermined skills based on the relationship between the expressions contained in the sentence converted by the conversion unit and each of the plurality of skills, and a response generation unit that generates a response sentence to the sentence converted by the conversion unit based on the sentence converted by the conversion unit and the points for each of the plurality of skills calculated by the calculation unit. [Effects of the Invention]

[0009] The response generation device, method, and program disclosed herein can generate a response that enhances the self-esteem and motivation of a user who feels guilty about their own words, actions, or thoughts. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a block diagram showing a schematic configuration of a response generation system. [Figure 2] FIG. 2 is a block diagram showing a hardware configuration of the response generation device. [Figure 3] FIG. 1 is a diagram illustrating an overview of services provided by a response generation system. [Figure 4] FIG. 10 is a diagram illustrating an example of skill items. [Figure 5] FIG. 2 is a block diagram illustrating an example of a functional configuration of a response generating device. [Figure 6] FIG. 10 is a diagram illustrating an example of a goal acceptance screen. [Figure 7] FIG. 10 is a diagram illustrating an example of a user DB. [Figure 8] FIG. 10 is a diagram illustrating an example of a posting screen. [Figure 9] FIG. 10 is a diagram illustrating an example of a diary DB. [Figure 10] FIG. 10 is a diagram illustrating an example of a conversion dictionary. [Figure 11] FIG. 10 is a diagram illustrating a response generation process. [Figure 12] FIG. 10 is a diagram for explaining conversion from a diary to a converted sentence. [Figure 13] FIG. 10 is a diagram illustrating an example of a skill correspondence DB. [Figure 14] FIG. 10 is a diagram illustrating an example of a point DB. [Figure 15] FIG. 10 is a diagram illustrating an example of generating a summary. [Figure 16] FIG. 10 is a diagram illustrating an example of a short-term feedback screen. [Figure 17]FIG. 10 is a diagram showing an example of a long-term feedback screen. [Figure 18] 10 is a flowchart showing the flow of an initial setting process. [Figure 19] 10 is a flowchart showing the flow of short-term response processing. [Figure 20] 10 is a flowchart showing the flow of long-term response processing. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, an example of an embodiment of the present disclosure will be described with reference to the drawings.

[0012] Fig. 1 is a block diagram showing a schematic configuration of a response generation system 100 according to this embodiment. As shown in Fig. 1, the response generation system 100 includes a response generation device 10 and a plurality of user terminals 90. The response generation device 10 and each of the plurality of user terminals 90 are connected via a network. The response generation system 100 provides a response generation service via an application running on the user terminal 90.

[0013] The user terminal 90 is an information processing terminal used by a user who uses the services provided by the response generation system 100, and includes an information display function, an information input function, and a communication function. The user terminal 90 is realized, for example, by a personal computer, a tablet terminal, a smartphone, or the like.

[0014] 2 is a block diagram showing the hardware configuration of the response generation device 10. As shown in FIG. 2, the response generation device 10 includes a CPU (Central Processing Unit) 12, a memory 14, a storage device The system includes a storage device 16, an input device 18, an output device 20, a storage medium reader 22, and a communication I / F (Interface) 24. Each component is connected to each other via a bus 26 so as to be able to communicate with each other.

[0015] The storage device 16 stores a response generation program for executing a response generation process, including an initial setting process, a short-term response process, and a long-term response process, which will be described later. The CPU 12 is a central processing unit that executes various programs and controls each component. That is, the CPU 12 reads the program from the storage device 16 and executes the program using the memory 14 as a work area. The CPU 12 controls each component and performs various arithmetic processes in accordance with the program stored in the storage device 16.

[0016] The memory 14 is made up of RAM (Random Access Memory) and serves as a working area to temporarily store programs and data. The storage device 16 is made up of ROM (Read Only Memory), HDD (Hard Disk Drive), SSD (Solid State Drive), etc., and stores various programs including the operating system and various data.

[0017] The input device 18 is a device for performing various inputs, such as a keyboard or a mouse. The output device 20 is a device for outputting various information, such as a display or a printer. A touch panel display may be used as the output device 20 to function as the input device 18.

[0018] The storage medium reader 22 reads data stored in various storage media such as CD (Compact Disc)-ROM, DVD (Digital Versatile Disc)-ROM, Blu-ray Disc, and USB (Universal Serial Bus) memory, and writes data to the storage media.

[0019] The communication I / F 24 is an interface for communicating with other devices, and uses standards such as Ethernet (registered trademark), FDDI, or Wi-Fi (registered trademark), or a communication means using a contracted line.

[0020] Next, an overview of the services provided by the response generation system 100 will be described with reference to FIG.

[0021] An application for using the services provided by the response generation system 100 runs on the user terminal 90. Using this application, users can post free-form writing (hereinafter referred to as "diary") about their own actions, feelings, learning, comments from others, etc. at any time. The response generation device 10 converts the posted diary into positive content. The response generation device 10 also generates a summary that summarizes the diary posted in a first short-term period and converts the converted content into positive content.

[0022] The response generation device 10 also identifies skills related to the generated summary. Skills are abilities related to self-improvement, and examples of these include items such as those shown in FIG. 4. The response generation device 10 then returns short-term feedback to the user terminal 90 based on the generated summary, the identified skills, etc. Furthermore, the response generation device 10 analyzes the user's strengths and behavioral achievements from the identified skills for a second long-term period, and returns long-term feedback to the user terminal 90.

[0023] In the following description, the short-term first period is one day and the long-term second period is one month, but these periods are not limited to these.

[0024] Next, the functional configuration of the response generation device 10 according to this embodiment will be described.

[0025] Fig. 5 is a block diagram showing an example of the functional configuration of the response generation device 10. As shown in Fig. 5, the response generation device 10 includes, as its functional configuration, a setting unit 32, a conversion unit 34, a calculation unit 36, a summary generation unit 38, and a response generation unit 40. A predetermined storage area of ​​the response generation device 10 stores a user DB (database) 42, a diary DB 44, a conversion dictionary 46, a skill correspondence DB 48, a point DB 50, and a response generation model 52. Each functional configuration is realized by the CPU 12 reading out a response generation program stored in the storage device 16, expanding it into the memory 14, and executing it.

[0026] The setting unit 32 receives profile information of a user who uses an application provided by the response generating device 10. For example, the setting unit 32 displays a profile reception screen (not shown) on the user terminal 90 and receives the profile information from the user. The profile information includes, for example, the user's name, date of birth, gender, address, etc. When the setting unit 32 receives the profile information, it assigns a user ID, which is identification information of the user, and registers the profile information in the user DB 42.

[0027] The setting unit 32 also accepts and sets skills that the user aims to achieve from among the multiple skills. For example, the setting unit 32 displays a goal acceptance screen 60 as shown in FIG. 6 on the user terminal 90 and accepts the user's selection of a goal and skills necessary to achieve the goal (hereinafter referred to as "target skills"). In the example of FIG. 6, the goal acceptance screen 60 includes a goal input area 62 for freely inputting the user's goal and a skill selection area 64 for selecting target skills related to the input goal from multiple predetermined skills. The user freely inputs abstract goals, such as the type of person the user wants to become, specific goals related to learning or work, goals related to personal fulfillment, etc., in text data into the goal input area 62. Setting a goal in this manner clarifies which skills should be selected as target skills from the multiple predetermined skills. In the skill selection area 64, the user selects one or more target skills from the multiple predetermined skills by, for example, checking a checkbox. The setting unit 32 registers the goals and target skills accepted on the goal acceptance screen 60 in the user DB 42.

[0028] 7 shows an example of the user DB 42. The user DB 42 stores profile information such as "user ID" and "name," as well as information such as set "goals" and "goal skills."

[0029] The conversion unit 34 accepts diaries posted by users. For example, the conversion unit 34 displays a posting screen 66 as shown in FIG. 8 on the user terminal 90, and accepts diary posts entered by users in a diary input area 68. The conversion unit 34 uses the conversion dictionary 46 to convert negative expressions (hereinafter referred to as "negative expressions") included in the accepted diary into positive expressions (hereinafter referred to as "positive expressions"). The conversion unit 34 registers the accepted diary and sentences in which negative expressions have been converted into positive expressions (hereinafter referred to as "converted sentences") in the diary DB 44.

[0030] An example of the diary DB 44 is shown in FIG. 9. In the example of FIG. 9, the diary DB 44 includes a table for each user ID. Each table stores a "diary," a "conversion sentence," and a "summary sentence" (described later) in association with the "posting date and time" when the diary was posted by the user identified by the user ID corresponding to the table. In addition, the conversion dictionary 46 stores, for example, a correspondence between "positive expressions" and "negative expressions," as shown in FIG. 10.

[0031] Specifically, when the conversion unit 34 receives a diary post such as that shown in FIG. 11A, it stores the received diary in a table in the diary DB 44 corresponding to the user ID of the corresponding user, in association with the posting date and time. The conversion unit 34 also performs morphological analysis on the received diary to extract content words. The conversion unit 34 searches for each extracted content word or a term similar to the content word from the "negative expression" in the conversion dictionary 46. A collocation of two or more content words that matches or is similar to a "negative expression" in the conversion dictionary 46 is also included in the search. Furthermore, a term similar to a content word may be, for example, converted into a word vector that semantically expresses the content word, and the term represented by the word vector may have a similarity to the word vector equal to or greater than a predetermined value. When a negative expression that matches or is similar to the extracted content word is searched for in the conversion dictionary 46, the conversion unit 34 acquires a "positive expression" corresponding to the searched negative expression. The conversion unit 34 converts the diary into a converted sentence as shown in FIG. 11(B) by replacing the negative expressions in the diary with the acquired positive expressions.

[0032] A more detailed explanation will be given using the example of post 1 in FIG. 11. The conversion unit 34 performs morphological analysis on a diary entry such as that shown in FIG. 12(A) and extracts content words as shown in FIG. 12(B). The conversion unit 34 searches for terms that match or are similar to each extracted content word and are registered as negative expressions in the conversion dictionary 46. In this case, if the preceding and following content words form a conjugation, the conversion unit 34 searches for the conjugation as a single content word. For example, the conversion dictionary 46 shown in FIG. 10 registers "gorogoro suru" (to roll around) as a negative expression. Therefore, the conversion unit 34 identifies the extracted content word "gorogoro shishita" (to roll around) as a negative expression and acquires the corresponding positive expression "to devote myself to resting" (to rest) from the conversion dictionary 46. If a negative expression that matches or is similar to the extracted content word is not registered in the conversion dictionary 46, the content word is not subject to conversion. In the example of FIG. 12(C), among the extracted content words, the underlined content words or their collocations are negative expressions.

[0033] The conversion unit 34 replaces the content word for which the corresponding positive expression has been acquired with the positive expression, as shown in (D) of Fig. 12. The conversion unit 34 constructs a sentence using the replaced positive expression and the content words not to be converted ("report" and "house" in the example of Fig. 12), thereby generating a converted sentence as shown in (E) of Fig. 12. Note that the positive expressions originally included in the diary before conversion are also included as they are in the converted sentence.

[0034] In the above example, a case where a diary is converted into a converted sentence using the conversion dictionary 46 has been described, but conversion may also be performed using a machine learning model. In this case, a plurality of pairs of a diary and a correct converted sentence into which the diary is converted are prepared, and these are used as training data to train a machine learning model such as a generation AI. Then, the conversion unit 34 may input the received diary into the machine learning model to obtain the converted sentence.

[0035] The calculation unit 36 ​​calculates points for each of a plurality of skills based on the relationship between the terms included in the converted sentence converted by the conversion unit 34 and each of a plurality of predetermined skills. Specifically, the calculation unit 36 ​​refers to the skill correspondence DB 48 and identifies skills corresponding to actions, feelings, etc. (hereinafter referred to as "actions, etc.") included in the converted sentence, as shown in (C) of FIG. 11. The skill correspondence DB 48 stores actions, etc. for improving each skill in association with each skill, as shown in, for example, FIG. 13. For example, the calculation unit 36 ​​divides the converted sentence into predetermined units such as sentence units or phrase units, searches the skill correspondence DB 48 for actions, etc. that match or are similar to the actions, etc. indicated by the divided parts, and identifies skills corresponding to the found actions, etc. The calculation unit 36 ​​calculates the number of times a skill is identified from the converted sentence as the points for each skill.

[0036] The calculation unit 36 ​​registers the calculated points in a points DB 50 such as that shown in Fig. 14. In the example of Fig. 14, the points DB 50 includes a table for each user ID. Each table stores, for a user indicated by the user ID corresponding to that table, a "skill" identified from a converted sentence obtained by converting the diary, in association with the "posting date and time" of the diary, and a "point" represented by the number of times that skill was identified from one converted sentence.

[0037] The summary generation unit 38 generates a summary, for example, as shown in FIG. 11 (D), based on one or more converted sentences converted by the conversion unit 34 from the diaries posted during a first short-term period (here, one day). The summary generation method may be any existing method, such as a method using a pre-trained machine learning model or a method based on predetermined rules, and therefore a detailed description thereof will be omitted here. The summary generation unit 38 may generate a summary directly from multiple converted sentences, or may generate a summary for one day by summarizing each converted sentence and then combining these summaries, as shown in FIG. 15. The summary generation unit 38 registers the generated summary in the diary DB 44.

[0038] The response generation unit 40 generates a response sentence that serves as a comment or advice regarding the daily sentence post, the converted sentence, the summarized sentence, and the points of each skill, based on the diary, the converted sentence, the summarized sentence, and the points of each skill. Specifically, the response generation unit 40 extracts or generates keywords to be input to the response generation model 52, based on the diary, the converted sentence, the summarized sentence, and the points of each skill.

[0039] More specifically, the response generation unit 40 refers to the point DB 50 and tallies up the points of each skill calculated from each converted sentence for one day. For example, assume that the points of each skill have been calculated for a user with user ID: 001, as shown in the point DB 50 of FIG. 14. In this case, the response generation unit 40 tallies up the points of each skill for one day, as shown in (E) of FIG.

[0040] The response generation unit 40 also extracts or generates keywords that can enhance the user's self-esteem and motivation based on the diary, the converted sentence, the summary, and the daily accumulated skill points. For example, the response generation unit 40 extracts words whose parts of speech are nouns contained in the diary as essential noun keywords. In the example of FIG. 11, for example, words such as "report," "part-time job," and "job offer" are extracted. Furthermore, for example, the response generation unit 40 extracts, as positive keywords, parts of the converted sentences that contain positive expressions converted from negative expressions in the original diary. In the example of FIG. 11, for example, "I didn't want to go, but I had no choice but to go" in post 2 is converted into a positive expression, resulting in "I was able to go (to my part-time job)" being extracted as a positive keyword. Furthermore, "I talked too much with the manager and ended up getting home late" in post 2 is converted into a positive expression, resulting in "I was able to talk (to the manager)" being extracted as a positive keyword. In addition, the phrase "I was anxious because I didn't get a job offer" in post 3 is converted into a positive expression, "I'll try my best to get one," which is extracted as a positive keyword.

[0041] In addition, the response generation unit 40 summarizes the behavior expressed as "was able to do something" in the converted sentence or the summary sentence, and generates a keyword expressing motivation. In the example of FIG. 11, for example, a keyword expressing motivation, such as "was able to go (to work)" or "was able to talk a lot" in the summary sentence is summarized to generate a keyword expressing motivation, such as "was able to have a good time (at work)." In addition, the response generation unit 40 generates a keyword for self-development in the converted sentence by combining a skill for which points have been calculated by the calculation unit 36 ​​with an expression related to self-development corresponding to that skill. In the example of FIG. 11, for example, a keyword for self-development is generated by combining the skill "self-discipline" worth two points with the expression "improvement" related to self-development. In addition, the response generation unit 40 generates a keyword for praise in the converted sentence by combining a skill for which points have been calculated by the calculation unit 36 ​​with a compliment. In the example of FIG. 11, for example, the keyword "sincerity·great" is generated as a compliment by combining the skill "sincerity" with two points and the compliment "great."

[0042] The response generation unit 40 generates a response sentence using a response generation model 52, which is a machine learning model such as a generation AI that has been trained in advance to generate a response sentence based on the input keywords. Specifically, the response generation unit 40 inputs the extracted or generated keywords into the response generation model 52, and generates, for example, a response sentence as shown in (F) of FIG.

[0043] The response generation unit 40 may refer to the user DB 42 and use the target skill selected by the user as the skill when generating the self-improvement keywords and the praise keywords. The response generation unit 40 may also extract or generate two or more keywords using skills, such as the self-improvement keywords and the praise keywords. This generates a response using keywords corresponding to at least two or more skills. For example, if only a keyword using the skill "ambition" is extracted or generated as a skill-based keyword, a response such as "You're wonderful for being positive and working hard" is generated. On the other hand, if keywords using the two skills "ambition" and "leadership" are extracted or generated, a response such as "It's wonderful to see you working hard and staying positive, but if you share your efforts with everyone on your team, it will inspire your teammates to work hard too" can be generated.

[0044] Furthermore, the response generation model 52 may be a model trained using, instead of keywords, a diary, a converted sentence, a summary, skill points, and a comment that is the correct answer as training data. In this case, the response generation model 52 is configured to generate a response sentence that is a comment when a diary, a converted sentence, a summary, and skill points are input. In this case, the response generation model 52 may be configured so that the above-mentioned keywords are extracted or generated within the response generation model 52. Furthermore, the generation of a response sentence is not limited to using the response generation model 52, and may also be generated by assembling the extracted or generated keywords based on predetermined rules.

[0045] The response generation unit 40 transmits the daily feedback including the summary generated by the summary generation unit 38, information on the points for each skill for one day, and the generated response sentence to the user terminal 90. On the user terminal 90, for example, a short-term feedback screen 70 as shown in FIG. 16 is displayed. In the example of FIG. 16, the short-term feedback screen 70 includes a summary display area 72 in which the summary for one day is displayed, a skill display area 74 in which the skills for which points have been calculated are displayed, and a response sentence display area 76 in which the generated response sentence is displayed.

[0046] Furthermore, the response generation unit 40 generates a response sentence based on the change in points for each of a plurality of skills over a long-term second period (here, one month), which is longer than a short-term first period (here, one day). Specifically, the response generation unit 40 identifies skills whose points are increasing over a one-month period. The response generation unit 40 may identify two or more skills, including the target skill selected by the user, from among the skills whose points are increasing. The response generation unit 40 generates a response sentence according to the increase in points for the identified skills. For example, if the points for the skill "ambition" are constantly increasing over the course of a month, the response generation unit 40 generates a response sentence such as "It's great that you always make constant efforts. Your efforts will always be rewarded." Alternatively, if the points are increasing sharply near the end of the month, the response generation unit 40 generates a response sentence such as "You worked hard towards your goal. If you can keep going, it will be a great help."

[0047] The response generation unit 40 may prepare response sentence templates according to the skill and point increase trend in advance, and generate a response sentence using a template selected according to the identified skill and point increase trend. Similarly to the above-described daily response sentence, the response generation unit 40 may generate a response sentence using the response generation model 52. The response generation model 52 used here may be a machine learning model such as a generation AI that is trained in advance to generate a response sentence when one month's worth of points for each skill are input.

[0048] Furthermore, the response generation unit 40 may generate a response sentence including at least one suggestion of a job type, industry, and lifestyle to which the person may be adapted, based on the skill whose points have increased over the past month. In this case, the response generation unit 40 determines in advance the relationship between the improvement of a skill and the job type, industry, lifestyle, etc. to which the person may be adapted. The response generation unit 40 may then identify the skill whose points have increased, and generate a suggestion of a job type, industry, lifestyle, etc. to which the person may be adapted, based on the identified skill and this relationship.

[0049] Furthermore, the response generation unit 40 may generate a response sentence that includes content based on the monthly tally of the points for each skill and content that takes into account the characteristics of the corresponding month in a year. For example, suppose the points for the skill "self-reflection" in March were higher than those in February. In this case, the response generation unit 40 may generate a response sentence for March such as, "This month, my self-reflection has increased compared to last month. This month was a month of a lot of self-reflection, but that reflection has given me the motivation to move forward." Also, suppose the points for the skill "ambition" increased in April. In this case, the response generation unit 40 may generate a response sentence for April such as, "Ambition is increasing. It's wonderful that you were able to welcome the new year with a positive attitude. Let's work hard next month with this mindset," taking into account the year in April.

[0050] In addition, when including comments comparing points on a monthly basis, in order to smooth out the difference in the number of posts per month, you can use a ratio expressed as the sum of the points for each skill in that month to the sum of the points for all skills in that month.

[0051] When the response generation unit 40 generates multiple response sentences, it compiles these response sentences into a final response sentence and transmits one-month feedback including the generated response sentences to the user terminal 90. The user terminal 90 displays, for example, a long-term feedback screen 78 as shown in FIG. 17 . In the example of FIG. 17 , the long-term feedback screen 78 includes a response sentence display area 80 in which the generated response sentences are displayed. The response generation unit 40 may also include in the one-month feedback whether the points of the target skill selected by the user have increased based on the points of each skill for one month. In this case, as shown in FIG. 17 , the long-term feedback screen 78 may include a target skill improvement degree display area 82. In the example of FIG. 17 , the improvement degree of the target skill is represented by the number of upward arrows (↑). The improvement degree may be calculated based on the increasing trend of points for each skill over the month, a comparison with points in the previous month, or the like.

[0052] Next, the operation of the response generation system 100 according to this embodiment will be described.

[0053] Fig. 18 is a flowchart showing the flow of initial setting processing executed by the CPU 14 of the response generation device 10. Fig. 19 is a flowchart showing the flow of short-term response processing executed by the CPU 14 of the response generation device 10. Fig. 20 is a flowchart showing the flow of long-term response processing executed by the CPU 14 of the response generation device 10. The CPU 14 reads the response generation program from the storage device 16, expands it in the memory 14, and executes it, causing the CPU 14 to function as each functional component of the response generation device 10 and executes each of the initial setting processing, short-term response processing, and long-term response processing. The response generation processing including the initial setting processing, short-term response processing, and long-term response processing is an example of a response generation method of the present disclosure.

[0054] First, the initial setting process will be described with reference to FIG.

[0055] In step S10, the setting unit 32 displays a profile reception screen on the user terminal 90 and receives profile information from the user. The setting unit 32 then assigns a user ID and registers the profile information in the user DB .

[0056] Next, in step S12, the setting unit 32 displays the goal acceptance screen 60 on the user terminal 90, sets the goal entered in the goal input area 62 on the goal acceptance screen 60, and registers it in the user DB 42. Next, in step S14, the setting unit 32 sets the skill selected in the skill selection area 64 on the goal acceptance screen 60 as the user's goal skill, registers it in the user DB 42, and the initial setting process ends.

[0057] Next, the short-term response process will be described with reference to FIG.

[0058] In step S20, the conversion unit 34 displays the posting screen 66 on the user terminal 90, accepts the diary post entered by the user in the diary input area 68, and registers it in the diary DB 44. Next, in step S22, the conversion unit 34 converts negative expressions contained in the accepted diary into positive expressions and registers the converted converted sentences in the diary DB 44.

[0059] Next, in step S24, the calculation unit 36 ​​determines whether or not one day, which is the first short-term period, has ended. If one day has not ended, the process returns to step S20, and if one day has ended, the process proceeds to step S26.

[0060] In step S26, the calculation unit 36 ​​refers to the skill correspondence DB 48 to identify skills corresponding to the actions, etc. included in the converted sentences for one day, and calculates the number of times a skill is identified from the converted sentences as points for each skill.The calculation unit 36 ​​then registers the points calculated for the identified skills in the point DB 50.Next, in step S28, the summary generation unit 38 generates a summary sentence for one day based on one or more converted sentences converted from the diary for one day by the conversion unit 34, and registers the summary sentence in the diary DB 44.

[0061] Next, in step S30, the response generation unit 40 refers to the point DB 50 and totals the skill points for one day from the points for each skill calculated from each converted sentence for one day. The response generation unit 40 also extracts or generates keywords that can increase the user's self-esteem and improve their motivation based on the diary, converted sentences, summarized sentences, and the totaled skill points for one day. The response generation unit 40 then inputs the extracted or generated keywords into the response generation model 52 to generate a response sentence.

[0062] Next, in step S32, the response generator 40 transmits the summary, information on the points for each skill for one day, and one day's feedback including the generated response to the user terminal 90, and the short-term response process ends. As a result, the short-term feedback screen 70 is displayed on the user terminal 90, and the user can check the feedback for one day's posts.

[0063] Next, the long-term response process will be described with reference to FIG.

[0064] In step S40, the response generation unit 40 determines whether or not the long-term second period of one month has ended. If the one month has not ended, the determination in this step is repeated, and if the one month has ended, the process proceeds to step S42.

[0065] In step S42, the response generation unit 40 identifies skills whose points are increasing over a one-month period. Next, in step S44, the response generation unit 40 generates a response sentence according to the point increase trend of the identified skill, such as whether the increase in points is constant or concentrated at the end of the month.

[0066] Next, in step S46, the response generation unit 40 generates a response sentence including at least one suggestion of a job type, industry, and lifestyle to which the employee may be potentially adapted, based on the skills whose points have increased over the past month. Next, in step S48, the response generation unit 40 generates a response sentence including, for example, a comparison with the points from the previous month, based on the results of tallying the points for each skill on a monthly basis. The response generation unit 40 also generates a response sentence including content that takes into account the characteristics of the relevant month in the year.

[0067] Next, in step S50, the response generation unit 40 calculates the degree of improvement of the target skill by comparing the point increase trend of the target skill over the month with the points of the previous month and the previous month. Next, in step S52, the response generation unit 40 transmits a final response sentence summarizing the generated response sentences and one-month feedback including the degree of improvement of the target skill to the user terminal 90, and the long-term response processing ends. As a result, the long-term feedback screen 78 is displayed on the user terminal 90, and the user can check the feedback for one month's worth of posts.

[0068] In the above long-term response process, the response sentences generated in steps S44 to S48 are compiled into a final response sentence, but it is not necessary to generate all of the response sentences. The final response sentence may be a compilation of the response sentences generated by executing at least one of steps S44 to S48.

[0069] As described above, according to the response generation system of this embodiment, the response generation device converts negative expressions contained in a diary posted by a user into positive expressions. The response generation device also calculates points for each of a plurality of predetermined skills based on the relationship between the terms contained in the converted converted sentence and each of the plurality of skills. The response generation device then generates a response to the diary based on the diary, the converted sentence, and the points for each of the plurality of skills. This makes it possible to generate a response that increases the self-esteem and motivation of a user who feels guilty about their own words, actions, or thoughts, compared to when a response is generated based on characteristic words contained in the diary.

[0070] <Variation 1> In Modification 1, the contents of a diary posted by a user are subjected to sentiment analysis, and each word included in the diary is determined to be positive or negative. Then, pairs of words determined to be positive or similar to such words and their corresponding negative expressions are added to the conversion dictionary 46 used in the above embodiment. This makes it easier for the conversion unit 34 to convert negative expressions into positive expressions that the user himself or herself is likely to use.

[0071] In addition, if conversion is performed using a machine learning model rather than conversion dictionary 46, pairs of words determined to be positive from the diary or words similar to those words and corresponding negative expressions for those words can be added to the training data and the machine learning model can be retrained.

[0072] <Variation 2> For example, users of Generation Z tend to place importance on "being themselves" and "what kind of person they are." Therefore, in Modification 2, the system learns the writing tendencies of each user based on frequently occurring words contained in the sentences entered by the user, and generates a response sentence based on the user's tendencies based on the learning results. Specifically, the system extracts words that frequently appear in the diary entries posted by the user and in the goals they set, includes these words as keywords, and trains the response generation model 52 so that responses are generated using these words or positive expressions similar to these words. This makes it possible to generate responses that refer to the user's tendencies and values.

[0073] <Variation 3> In the above embodiment, the case where all of the response generation processing is executed by the response generation device 10 has been described, but in Modification 3, all or part of the functional units of the response generation device 10 are provided in an application running on the user terminal 90. As a result, all or part of the response generation processing is executed by the application on the user terminal 90.

[0074] When all or part of the response generation process is executed on the user terminal 90 side, a storage device provided inside the user terminal 90 or an external storage device is used.

[0075] Furthermore, the response generation process executed by the CPU by reading software (programs) in the above embodiments may be executed by various processors other than the CPU. In this case, the processor may be a device whose circuit configuration can be changed after manufacturing, such as an FPGA (Field-Programmable Gate Array). Examples of such processors include dedicated electrical circuits, which are processors having a circuit configuration specifically designed to perform specific processing, such as programmable logic devices (PLDs) and application-specific integrated circuits (ASICs). A DSP (Digital Signal Processor), which is capable of parallel processing of not only addition but also multiplication, may also be used as a processor other than a CPU. The response generation process may be performed by one of these various processors, or by a combination of two or more processors of the same or different types (e.g., multiple FPGAs, or a combination of a CPU and an FPGA). The hardware structure of these various processors is, more specifically, an electrical circuit that combines circuit elements such as semiconductor devices.

[0076] In addition, in each of the above embodiments, the response generation program is described as being pre-stored (installed) in a storage device, but this is not limiting. The program may be provided in a form recorded on a recording medium such as a CD-ROM, a DVD-ROM (Digital Versatile Disc Read Only Memory), or a USB (Universal Serial Bus) memory. The program may also be downloaded from an external device via a network.

[0077] The following supplementary items are disclosed:

[0078] (Additional note 1) a conversion unit that converts negative expressions contained in sentences recorded by a user into positive expressions; a calculation unit that calculates points for each of a plurality of predetermined skills based on a relationship between an expression included in the sentence converted by the conversion unit and each of the plurality of predetermined skills; a response generation unit that generates a response sentence to the sentence converted by the conversion unit based on the sentence converted by the conversion unit and the points of each of the plurality of skills calculated by the calculation unit; A response generation device comprising:

[0079] (Additional note 2) a summary generation unit that generates a summary of one or more sentences that are recorded during a predetermined first period and converted by the conversion unit; the response generation unit generates the response sentence by further using the summary sentence generated by the summary generation unit. Item 1. A response generation device according to item 1.

[0080] (Additional note 3) The response generation device described in Appendix 2 generates the response sentence including, from among the terms contained in the sentence converted by the conversion unit or the summary sentence, an expression corresponding to the skill to which points have been added by the calculation unit, and a positive expression corresponding to the expression.

[0081] (Additional note 4) The response generation device described in Appendix 3, wherein the response generation unit generates the response sentence using expressions corresponding to at least two or more skills from among the terms contained in the summary sentence when there are two or more skills to which points have been added by the calculation unit.

[0082] (Additional note 5) The response generation device according to any one of claims 2 to 4, wherein the response generation unit generates the response sentence based on changes in points of each of the plurality of skills during a second period that is longer than the first period.

[0083] (Additional note 6) The response generation device according to claim 5, wherein the response generation unit generates the response sentence according to the trend of point increases for skills whose points are increasing during the second period.

[0084] (Additional note 7) The response generation device described in Appendix 5 or Appendix 6, wherein the response generation unit generates the response sentence including at least one suggestion of a job type, industry, and lifestyle that may be applicable based on the skills whose points have increased during the second period.

[0085] (Additional note 8) a setting unit for setting a target skill of the user from among the plurality of skills; the response generation unit generates the response sentence related to a skill that the user is aiming for. The response generating device according to any one of supplementary items 1 to 7.

[0086] (Additional note 9) The response generation device according to any one of appendix 1 to appendix 8, wherein the conversion unit converts the sentence recorded by the user based on a correspondence relationship between positive expressions and negative expressions, and adds the positive expressions contained in the sentence recorded by the user to the correspondence relationship.

[0087] (Additional note 10) The response generation device according to any one of appendix 1 to appendix 9, wherein the response generation unit learns the tendencies of each user's writing based on frequently occurring words contained in sentences recorded by the user, and generates the response sentence based on the tendencies of the user based on the learning results.

Claims

1. a conversion unit that converts negative expressions contained in sentences recorded by a user into positive expressions; a calculation unit that calculates points for each of a plurality of predetermined skills based on a relationship between an expression included in the sentence converted by the conversion unit and each of the plurality of predetermined skills; a response generation unit that generates a response sentence to the sentence converted by the conversion unit based on the sentence converted by the conversion unit and the points of each of the plurality of skills calculated by the calculation unit; A response generation device comprising:

2. a summary generation unit that generates a summary of one or more sentences that are recorded during a predetermined first period and converted by the conversion unit; the response generation unit generates the response sentence by further using the summary sentence generated by the summary generation unit. The response generating device according to claim 1 .

3. The response generation device according to claim 2, wherein the response generation unit generates the response sentence including, from among the terms contained in the sentence converted by the conversion unit or the summary sentence, an expression corresponding to the skill to which points have been added by the calculation unit, and a positive expression corresponding to the expression.

4. The response generation device described in claim 3, wherein when there are two or more skills to which points have been added by the calculation unit, the response generation unit generates the response sentence using expressions corresponding to at least two or more skills from among the terms contained in the summary sentence.

5. A response generation device described in any one of claims 2 to 4, wherein the response generation unit generates the response sentence based on changes in points of each of the plurality of skills during a second period longer than the first period.

6. The response generation device according to claim 5 , wherein the response generation unit generates the response sentence according to a trend of increase in points for a skill whose points have increased during the second period.

7. The response generation device described in claim 5, wherein the response generation unit generates the response sentence including a suggestion of at least one of a job type, industry, and lifestyle that may be suitable based on the skills whose points have increased during the second period.

8. a setting unit for setting a target skill of the user from among the plurality of skills; the response generation unit generates the response sentence related to a skill that the user is aiming for. The response generating device according to any one of claims 1 to 4.

9. The response generation device according to any one of claims 1 to 4, wherein the conversion unit converts the sentence recorded by the user based on a correspondence between positive expressions and negative expressions, and adds the positive expressions contained in the sentence recorded by the user to the correspondence.

10. The response generation device according to any one of claims 1 to 4, wherein the response generation unit learns the tendencies of each user's writing based on frequently occurring words contained in the sentences recorded by the user, and generates the response sentence based on the user's tendencies based on the learning results.

11. A response generation method executed by a response generation device including a conversion unit, a calculation unit, and a response generation unit, the conversion unit converts negative expressions contained in sentences recorded by a user into positive expressions; the calculation unit calculates points for each of a plurality of skills based on a relationship between an expression included in the sentence converted by the conversion unit and each of a plurality of predetermined skills; the response generation unit generates a response sentence to the sentence converted by the conversion unit based on the sentence converted by the conversion unit and the points for each of the plurality of skills calculated by the calculation unit. Response generation method.

12. Computer, a conversion unit that converts negative expressions contained in sentences recorded by a user into positive expressions; a calculation unit that calculates points for each of a plurality of predetermined skills based on a relationship between an expression included in the sentence converted by the conversion unit and each of the plurality of predetermined skills; and a response generation unit that generates a response sentence to the sentence converted by the conversion unit based on the sentence converted by the conversion unit and the points for each of the plurality of skills calculated by the calculation unit; A response generator to act as a

Images