Announcement generation device, announcement generation method, and recording medium

Abstract

Provided is an announcement generation device comprising: an acquisition means that acquires an announcement template which is for recommending products handled in a store, business type information which pertains to the store, and flyer information which pertains to the store; an announcement script generation means that, on the basis of the announcement template and the business type information, uses a language model to generate an announcement script for recommending products which are listed in a flyer; and an output means that outputs the announcement script.

Description

Announcement generating device, announcement generating method, and recording medium

[0001] The present disclosure relates to an announcement generating device, an announcement generating method, and a recording medium.

[0002] In stores, announcements are sometimes made over the in-store PA system to appeal to customers about the features of products and the like.

[0003] Japanese Patent Application Laid-Open No. 2003-144992 describes a message data output device that creates composite message data by combining product data and standard message data.

[0004] Japanese Patent Application Publication No. 11-039567

[0005] In a store, for example, announcement scripts may be changed depending on the store's business type or the products for which information is to be promoted to customers. However, it may not be easy to create announcement scripts that correspond to the store's business type or the products for which information is to be promoted to customers.

[0006] One of the objects of the present disclosure is to provide an announcement generation device and the like that can facilitate the generation of announcements.

[0007] An announcement generation device in one aspect of the present disclosure includes an acquisition means for acquiring an announcement template for recommending products handled in a store, store business type information, and store flyer information, an announcement script generation means for generating an announcement script for recommending products listed in the flyer using a language model based on the announcement template and the business type information, and an output means for outputting the announcement script.

[0008] An announcement generation method in one aspect of the present disclosure obtains an announcement template for recommending products handled by a store, store business type information, and store flyer information, generates an announcement script for recommending products listed in the flyer using a language model based on the announcement template and business type information, and outputs the announcement script.

[0009] An announcement generation program in one aspect of the present disclosure causes a computer to execute a process of acquiring an announcement template for recommending products handled by a store, store business type information, and store flyer information, generating an announcement script for recommending products listed in the flyer using a language model based on the announcement template and business type information, and outputting the announcement script.

[0010] Each program may be stored in a non-transitory computer-readable recording medium.

[0011] One example of the effect of the present disclosure is that it makes it possible to easily generate announcements.

[0012] 1 is a diagram showing an application example of an announcement generating device. FIG. 2 is a diagram showing an example of the configuration of a system including an announcement generating device. FIG. 3 is a diagram showing another example of the configuration of a system including an announcement generating device. FIG. 4 is a block diagram showing an example of the configuration of an announcement generating device. FIG. 5 is a diagram showing an example of a prompt. FIG. 6 is a diagram showing an example of a screen for inputting an announcement template, business type information, and flyer information. FIG. 7 is an example of a screen on which additional information can be input. FIG. 8 is a flowchart showing the operation of the announcement generating device. FIG. 9 is a block diagram showing an example of the configuration of the announcement generating device. FIG. 10 is a flowchart showing the operation of the announcement generating device. FIG. 11 is a diagram showing an example of the hardware configuration of the announcement generating device.

[0013] Embodiments of the present disclosure will be described in detail with reference to the drawings.

[0014] First, an application example of the announcement generation device will be described with reference to FIG. 1 . FIG. 1 is a diagram illustrating an application example of the announcement generation device. In the example illustrated in FIG. 1 , a customer visits a store. An announcement recommending products sold in the store to the customer is output from a speaker. Customers visiting the store can obtain information about the products recommended by the store through the announcement. The announcement generation device creates a script for the announcement recommending the products sold in the store. The announcement generation device also generates audio data of the script being read aloud. In the example illustrated in FIG. 1 , the announcement is made to customers inside the store, but the announcement may also be made to customers outside the store. An outside-store customer is, for example, a customer looking at products displayed on shelves outside the store. The outside-store customer may also be a customer who has visited the store and is about to enter the store, or a customer passing by outside the store. The outside-store customer can obtain information about the products recommended by the store through the announcement.

[0015] [First embodiment] An example of the configuration of a system including an announcement generation device 10 will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the configuration of a system including the announcement generation device 10. Referring to Fig. 2, the announcement generation device 10 is connected to a generation server 80 including a generation unit 801, an audio output device 90, a store terminal 91, and a database 92 via a wired or wireless network.

[0016] The audio output device 90 is a device that is installed in a store and outputs audio. For example, the audio of an announcement made by a store clerk is output from the audio output device 90. Also, audio data of an announcement script generated by the announcement generation device 10 being read aloud is output from the audio output device 90. The audio output device 90 is installed, for example, in a sales floor. Also, the audio output device 90 may be installed, for example, in a back room used by the clerk. The installation location of the audio output device 90 is not limited to these. The audio output device 90 may be installed in a location where a customer or a clerk can hear the announcement output from the audio output device 90. A specific example of the audio output device 90 is a speaker. The number of audio output devices 90 is not limited to one. There may be multiple audio output devices 90.

[0017] The store terminal 91 is a terminal used by a store clerk. For example, the clerk can operate the store terminal 91 when he or she wants to make an announcement. The clerk may also prepare an announcement script or audio data of the announcement script read out in advance. In this case, the clerk may operate the store terminal 91. The store terminal 91 is, for example, a smartphone, a personal computer, or a tablet terminal. The store terminal 91 is not limited to these. The store terminal 91 may be any terminal that can be used by the store clerk. The number of store terminals 91 is not limited to one. There may be multiple store terminals 91.

[0018] The database 92 stores product information about products handled by the store, announcement templates, and flyer information. The database 92 may also store sales information about the store. The database 92 may be included in the announcement generating device 10.

[0019] The generation server 80 is a server including a generation unit 801. The generation unit 801 generates, for example, an announcement script using a generative model. In other words, the generation server 80 is a server equipped with a generative model.

[0020] Here, the generation unit 801 may be provided with the announcement generation device 10. Therefore, with reference to Fig. 3, another example of the configuration of a system including the announcement generation device 10 will be described. Fig. 3 is a diagram showing another example of the configuration of a system including the announcement generation device 10. With reference to Fig. 3, the announcement generation device 10 includes the generation unit 801. The operation of the generation unit 801 will be described later.

[0021] The announcement generation device 10 inputs a prompt to a generation model used by the generation unit 801. Then, the announcement generation device 10 can acquire a generation result from the generation model used by the generation unit 801. The configurations other than the configuration related to the generation unit 801 are common to Figs. 2 and 3.

[0022] In the following description, the announcement generation device 10 may be connected to a generation server 80 including a generation unit 801 via a wired or wireless network. The announcement generation device 10 may also include the generation unit 801.

[0023] Next, the configuration of the announcement generation device 10 will be described with reference to Fig. 4. Fig. 4 is a block diagram showing an example of the configuration of the announcement generation device 10. The announcement generation device 10 includes an acquisition unit 101, an announcement script generation unit 102, and an output unit 103. The announcement generation device 10 may further include a prompt creation unit 104.

[0024] The acquisition unit 101 is one aspect of acquisition means that acquires announcement templates for recommending products handled in the store, store business type information, and store flyer information. The acquisition unit 101 acquires announcement templates to be made in the store to recommend products to customers. A store clerk can make an announcement by changing part of the announcement template depending on the recommended product or the situation in which the announcement is made. In other words, the announcement template includes a script for a portion that does not change depending on the recommended product or the situation in which the announcement is made. The announcement template also includes a script for a portion that changes depending on the recommended product or the situation in which the announcement is made. An example of a portion that does not change depending on the recommended product or the situation in which the announcement is made is a script for greeting or expressing gratitude to the customer. The acquisition unit 101 may acquire one type of announcement template or multiple types of announcement templates. A method for acquiring announcement templates by the acquisition unit 101 will be described later.

[0025] The acquisition unit 101 also acquires business type information of the store that will make the announcement. The store business type information is information about the business format of the store. The business format is a classification of stores based on the types of products the store handles or the sales methods. Examples of store business type information include department stores, supermarkets, electronics stores, and drugstores. The store business type information is not limited to these. The store business type information may be information about the business format in which announcements are made in the store to recommend products.

[0026] The store's business type information may also include information regarding the price range of the products sold. For example, even among stores operating under the business type of supermarkets, the price range of the products sold may differ depending on the supermarket. Specific examples of supermarkets that handle products in different price ranges include high-end supermarkets and discount supermarkets. Therefore, the store's business type information may include information regarding the price range of the products they handle, such as high-end supermarkets and discount supermarkets. The scripts of announcements made at a store may differ depending on the store's business type. For example, the language used in the announcement scripts may differ depending on the store's business type. Therefore, the store's business type information may include information regarding specific requests for the type of announcement script that the customer would like to see generated. One example of a specific request is for the announcement script to use polite language.

[0027] The acquisition unit 101 acquires flyer information of the store that will make the announcement. A flyer is a distribution material that lists information about products sold in the store. The flyer may be printed on paper and distributed, or may be distributed electronically. The store distributes flyers to customers, for example. The customers can understand information about the products sold in the store by checking the flyer. Here, the acquisition unit 101 only needs to be able to acquire information about products that may be listed in the flyer. In other words, the format of the information acquired by the acquisition unit 101 is not limited to flyers.

[0028] A store flyer contains product information about products sold at the store. Therefore, the flyer information acquired by the acquisition unit 101 includes product information about products sold at the store making the announcement. The product information may include, for example, the product name and product price. The product information may also include the sales company, best-before date, expiration date, sales quantity, sales target, sales season, sales start date, or sales end date.

[0029] Flyers may contain information about products that are on sale or special offer at the store. Therefore, the acquisition unit 101 may acquire store flyer information that includes product information about products that are on sale or special offer. Furthermore, store flyers may contain information about new products at the store. Therefore, the acquisition unit 101 may acquire store flyer information that includes product information about new products at the store. The information included in the flyer information is not limited to this. The store flyer information may also include, for example, information about events that are being held at the store. The store flyer information acquired by the acquisition unit 101 may be any information that the store wants to promote to customers. Furthermore, the flyer information acquired by the acquisition unit 101 may be product information about some of the products listed in the store flyer.

[0030] The flyer information may include information about the product listed in the flyer that is not listed in the flyer. An example will be described in which the flyer lists the product name, sales company, and product price among the information about the product listed in the flyer. Product information about the product may be stored, for example, in a list in database 92. In this case, the acquisition unit 101 can also acquire information other than the information listed in the flyer by referring to the list in database 92. In other words, the acquisition unit 101 can also acquire product information about the sales target and sales start date of the product that is not listed in the flyer.

[0031] The acquisition unit 101 may acquire flyer information as image data of the flyer. Alternatively, the acquisition unit 101 may acquire product information of products listed on the flyer as table-format data. Alternatively, the acquisition unit 101 may acquire product information of products listed on the flyer as text-format data. The format of the flyer information acquired by the acquisition unit 101 is not limited to these. A method for acquiring flyer information by the acquisition unit 101 will be described later.

[0032] In addition to the announcement template, store business type information, and store flyer information, the acquisition unit 101 may also acquire additional information that assists in generating an announcement script suitable for the situation in which the announcement is made. In other words, the additional information is information that describes the situation of the store in which the announcement is made. The wording and words used in the announcement script to be made at the store may differ depending on the situation of the store in which the announcement is made. Therefore, the acquisition unit 101 may acquire additional information that describes the situation of the store in which the announcement is made.

[0033] An example of the additional information is sales information for each product listed in the flyer. The sales information is, for example, the number of units or sales amount of the product listed in the flyer that was actually sold within a predetermined period. The predetermined period may be any period that allows for understanding the sales situation. The predetermined period may be, for example, one day or one hour. In this case, the announcement script generation unit 102, which will be described later, can generate an announcement script using sales information for the current day. The predetermined period may also be a period of the past few days at the store. In this case, the announcement script generation unit 102 can generate an announcement script by referring to past sales trends.

[0034] The sales information is stored in, for example, a database 92. The acquisition unit 101 can then acquire sales information for products listed in a flyer from the database 92. The acquisition unit 101 may also acquire sales information for products not listed in a flyer. The sales information may be stored in a point of sales (POS) server of the store. In this case, the acquisition unit 101 may acquire the sales information for products listed in the flyer from the POS server. Here, the POS server and the announcement generating device 10 are connected via a wired or wireless network.

[0035] The sales information may include information on the sales plan or sales target for each product listed in the flyer. The sales plan information is the number of products or sales amount planned to be sold within a specified period. The sales target information is the number of products or sales amount that the store aimed to sell within a specified period. By the acquisition unit 101 acquiring the sales plan or sales target as product sales information, it becomes possible to compare the number of products sold or sales amount of products actually sold with the sales plan or sales target.

[0036] The additional information may also include demographic information about customers in the store where the announcement is made. An example of demographic information about customers is attribute information about customers in the store. The attribute information about customers is, for example, the gender or age of the customer. The attribute information may also include family structure or occupation.

[0037] The acquisition unit 101 may acquire attribute information of customers who use the store where the announcement is made. The attribute information of customers who use the store represents the attribute trends of customers who use the store. The acquisition unit 101 can acquire attribute information of customers who have previously shopped at the store where the announcement is made. The acquisition unit 101 can acquire customer demographic information from past purchase histories. For example, when a store checks out a customer, a store clerk may register the customer's attribute information. At this time, the customer attribute information registered by the clerk is stored, for example, in a database 92. In addition, the customer may register attribute information in advance. In addition, the acquisition unit 101 can acquire customer demographic information of customers who have registered as members of the store where the announcement is made. In addition, the customer may register attribute information as part of their membership registration. The attribute information registered by the customer is stored, for example, in a database 92. The acquisition unit 101 can acquire customer attribute information registered by the store clerk and stored in the database 92, and can also acquire attribute information registered by the customer and stored in the database 92.

[0038] The acquisition unit 101 can acquire customer attribute information from, for example, a camera installed in the store. The acquisition unit 101 acquires attribute information of customers present in the store by analyzing video images captured by the camera installed in the store. The analysis of the video images may be performed by the acquisition unit 101. Alternatively, the analysis of the video images may be performed by a device (not shown) connected to the announcement generation device 10 via a wired or wireless network. The analysis of the video images may be performed by the camera that captured the video images. Alternatively, the acquisition unit 101 may store the acquired attribute information of customers present in the store in, for example, a database 92. When generating an announcement script, the acquisition unit 101 can acquire attribute information of customers present in the store that was previously acquired from the database 92.

[0039] The acquisition unit 101 can acquire attribute information of customers who are in the store at the time the announcement is made. The acquisition unit 101 can acquire the customer attribute information from, for example, a camera installed in the store. The acquisition unit 101 acquires the attribute information of customers who are in the store by analyzing video images captured by the camera installed in the store. The acquisition unit 101 can also acquire the customer attribute information by, for example, referring to information registered in advance by the customer. A customer may have registered attribute information in advance by registering as a member, etc. The customer attribute information is stored in, for example, a database 92. Then, when a customer visits the store, the customer may, for example, check in to the store. At this time, the acquisition unit 101 can acquire the attribute information of the checked-in customer by referring to the database 92.

[0040] Furthermore, the customer demographic information may include information on customer traffic levels. Information on customer traffic levels is the number of customers in the store at a given time. The number of customers in the store may be measured, for example, by a gate installed in the store. The number of customers in the store may also be measured by analyzing video images captured by a camera installed in the store. The information on customer traffic levels may also be the store's congestion rate at a given time. The store's congestion rate may be the value obtained by dividing the number of customers in the store by the store's area, as described above. The store's congestion rate may also be the value obtained by dividing the number of customers in the store by the number of cash registers in the store. In other words, the store's congestion rate may be the number of customers that one cash register can potentially handle.

[0041] The information on the customer traffic may be a ratio to a pre-estimated number of customers. In other words, the ratio to the pre-estimated number of customers indicates whether the number of customers visiting the store is approximately the same as the pre-estimated number of customers, whether the number of customers visiting the store is greater than the pre-estimated number of customers, or whether the number of customers visiting the store is less than the pre-estimated number of customers. First, the number of customers in the store can be measured using the method described above. Then, information on the pre-estimated number of customers visiting the store is stored in, for example, the database 92. Therefore, the acquisition unit 101 can acquire the ratio to the pre-estimated number of customers using the number of customers in the store and the pre-estimated number of customers.

[0042] The information on customer traffic may be information on the results of a comparison with the number of past visitors. Specifically, the information on the results of a comparison with the number of past visitors is information that indicates whether the number of customers is higher or lower than usual. The number of past visitors is, for example, the average number of past visitors on the same day of the week at the same time. In other words, the information on the results of a comparison with the number of past visitors indicates whether the number of visitors is higher or lower than the average number of visitors on the same day of the week at the same time. The number of past visitors is not limited to the average number of past visitors on the same day of the week at the same time. The number of past visitors may also be the average number of past visitors on weekdays or on weekends and holidays at the same time.

[0043] The additional information may include weather information for the area around the store when the announcement is made. The weather information is information about the weather around the store when the announcement script is generated. For example, when an announcement is to be made immediately using the generated announcement script, current weather information may be necessary. This is because the current announcement may include weather-related content. The weather information may also be information about the weather forecast for the area around the store. For example, when generating a script for an announcement to be made in a few hours, a weather forecast for the next few hours may be necessary. Therefore, the acquisition unit 101 may acquire a weather forecast for the area around the store for up to a few hours. The weather information may also include information about the season when the announcement is made. For example, seasons when an announcement is to be made may include spring, summer, autumn, winter, and the rainy season, and may also include New Year's Day, Tanabata, Halloween, and Christmas.

[0044] The additional information may also include information about the time period during which the announcement will be made. A specific example of the information about the time period during which the announcement will be made is from 12:00 to 13:00. The information about the time period during which the announcement will be made may also include information about an event that will be held during that time period. For example, the information about the time period during which the announcement will be made may be from 16:00 to 17:00, when a limited-time sale is held. The announcement may, for example, recommend a product that is on limited-time sale. In this case, the announcement may be made to coincide with the time period during which the limited-time sale is held. Therefore, the acquisition unit 101 may acquire information about the time period during which the announcement will be made, including information about an event that will be held during the time period during which the announcement will be made. The information about the time period may also be the time at which the announcement will be made.

[0045] Next, a method in which the acquisition unit 101 acquires announcement templates for recommending products handled by a store, store business type information, and store flyer information will be described.

[0046] The acquisition unit 101 may acquire a prompt including an announcement template for recommending products handled by the store, store business type information, and store flyer information.

[0047] Here, the prompt will be described with reference to FIG. 5. FIG. 5 is a diagram showing an example of a prompt. Referring to FIG. 5, the prompt includes an instruction to generate an announcement script in the paragraph "# command sentence." The instruction also includes an instruction to use flyer information and a template to generate the announcement script. The instruction also includes store business type information. Referring to FIG. 5, the business type information is entered in the "<Store Business Type>" column.

[0048] The "#flyer information" paragraph may contain flyer information for the store. Information about the products featured in the flyer may be listed in bullet points, for example. Identification information for identifying image data of the flyer may also be listed. For example, image data of the flyer with the identification information attached thereto may be stored in database 92. The announcement script generation unit 102, which will be described later, may then refer to the image data stored in database 92 based on the identification information listed in the prompt, and use the image data to generate the announcement script.

[0049] The paragraph "# template" describes an announcement template. For example, as shown in FIG. 5, the announcement template may be described after "# template."

[0050] The prompt shown in FIG. 5 is an example. The format of the prompt is not limited to this. In the example prompt shown in FIG. 5, the generation instructions, flyer information, and template are written separately. However, the prompt may write the generation instructions, flyer information, and template together. The prompt may also be written in bulleted format, as in the example prompt shown in FIG. 5. The prompt may also be written in sentence format.

[0051] The prompt acquired by the acquisition unit 101 may include additional information. When the prompt includes additional information, the acquisition unit 101 acquires the additional information. For example, the prompt may include a paragraph titled "# Additional Information," and the additional information may be described in that paragraph. The paragraph may then state, for example, "Use sales information." When the prompt includes a request to use sales information as additional information, the acquisition unit 101 may acquire sales information from, for example, a database 92 in which the sales information is stored. The announcement script generation unit 102, described later, can then generate an announcement script using the sales information acquired by the acquisition unit 101. Furthermore, when weather information is used as additional information, the paragraph titled "# Additional Information" may include, for example, a request to "Use customer demographic information." The acquisition unit 101 may then acquire customer demographic information from, for example, a device that analyzes customer demographic information. The acquisition unit 101 may also acquire customer demographic information from the database 92 in which the customer demographic information is stored. The announcement script generating unit 102 (described later) can generate an announcement script using the customer demographic information acquired by the acquiring unit 101.

[0052] The prompt acquired by the acquisition unit 101 may be modifiable. For example, the acquisition unit 101 can accept a modified prompt on a screen on which the prompt can be modified. The announcement script generation unit 102, which will be described later, generates an announcement script based on the prompt acquired by the acquisition unit 101. Then, for example, the generated announcement script may need to be modified. For example, a store clerk may check the generated announcement script and want to modify the prompt. In such a case, the store clerk can modify the prompt. In other words, the acquisition unit 101 can accept a modified prompt.

[0053] Furthermore, the acquisition unit 101 may acquire the announcement template, business type information, and flyer information input on a screen on which the announcement template, business type information, and flyer information can be input. That is, the acquisition unit 101 acquires the input contents of the announcement template, business type information, and flyer information.

[0054] The acquiring unit 101 may acquire input contents regarding the additional information. That is, the store clerk may be able to input the additional information on a screen on which the announcement template, business type information, and flyer information can be input. The screen on which the announcement template, business type information, and flyer information can be input is displayed on the store terminal 91, for example.

[0055] The following describes how announcement templates, business type information, and flyer information are input on a screen. For example, the store terminal 91 displays a screen on which the announcement templates, business type information, and flyer information can be input in a selective manner. For example, candidates for business type information are displayed on the screen, and the store clerk can select the business type information to use in generating the announcement script. The acquisition unit 101 can then acquire the selected business type information.

[0056] For example, candidates for flyer information are displayed on the screen of store terminal 91. Then, the store clerk can select flyer information to be used in generating the announcement script from the candidates for flyer information displayed on the screen. The flyer information is stored, for example, in database 92. Then, acquisition unit 101 can acquire the flyer information selected by the store clerk from database 92. At this time, the candidates for flyer information displayed on the screen may be flyer images or information that identifies the flyer images. The information that identifies the flyer images is, for example, a file name assigned to the image data.

[0057] Candidate announcement templates are displayed on the screen of the store terminal 91. The store clerk can then select an announcement template to be used to generate an announcement script from the candidate templates displayed on the screen. In other words, the acquisition unit 101 can acquire the announcement template selected by the store clerk. The candidate announcement templates displayed on the screen may be a preview of the announcement template or information identifying the announcement template. The information identifying the announcement template is, for example, a file name assigned to a document file of the announcement template.

[0058] The store terminal 91 may also display a screen on which the announcement template, business type information, and flyer information can be freely entered. The store clerk enters the announcement template, business type information, and flyer information, for example, using the store terminal 91. The acquisition unit 101 then accepts the entered information from the store terminal 91.

[0059] The input on the screen when the acquisition unit 101 acquires additional information is similar to the above. That is, for example, the store terminal 91 may display a screen that allows input of each piece of additional information in a selective manner. The store terminal 91 may also display a screen that allows input of each piece of additional information by freely writing it. The store clerk inputs the additional information using, for example, the store terminal 91. Then, the acquisition unit 101 can accept the input content from the store terminal 91.

[0060] 6 shows an example of a screen for inputting an announcement template, business type information, and flyer information. The store clerk can input each piece of information by selecting options or by writing freely.

[0061] The store clerk may also input an instruction for acquisition unit 101 to acquire flyer information or an announcement template from database 92. In other words, acquisition unit 101 may acquire, from database 92, flyer information or an announcement template instructed on the input screen.

[0062] 6, there is a Reference button next to the input field for flyer information. When the store clerk presses the Reference button, he or she can view, for example, a list of flyer information stored in database 92. The store clerk can then specify the flyer information to be used in generating the announcement script. At this time, acquisition unit 101 can acquire the specified flyer information from database 92.

[0063] 6, there is a Reference button next to the input field for the announcement template. When the store clerk presses the Reference button, he or she can view, for example, a list of announcement templates stored in the database 92. The store clerk can then specify the announcement template to be used in generating the announcement script. At this time, the acquisition unit 101 can acquire the specified announcement template from the database 92.

[0064] The example shown in FIG. 7 is an example of a screen on which additional information can be input. Referring to FIG. 7, sales information, information on the time period during which the announcement is made, weather information, and customer demographic information can be input as additional information. Referring to FIG. 7, the store clerk can check the additional information to be used in generating the announcement script. That is, the acquisition unit 101 acquires the checked additional information. For example, when sales information is checked, the acquisition unit 101 acquires the sales information from the database 92. The store clerk may also be able to specify a period for which sales information is to be acquired on the input screen. That is, the acquisition unit 101 acquires sales information for the specified period.

[0065] When the information on the time period for which the announcement will be made is checked on the input screen, the information on the time period for which the announcement will be made is used to generate the announcement script. When the information on the time period for which the announcement will be made is checked on the input screen, the time period including the input time may be used to generate the announcement script. For example, when 12:15 is input, the information on the time period for which the announcement will be made may be from 12:00 to 1:00 PM. Furthermore, the time period after the input time may be used to generate the announcement script. For example, when 12:15 is input, the information on the time period for which the announcement will be made may be from 1:00 PM to 2:00 PM. Furthermore, the store clerk may be able to specify the time period for which the announcement will be made on the input screen. In other words, the acquisition unit 101 may acquire the specified time period as the information on the time period for which the announcement will be made.

[0066] When the weather information is checked on the input screen, the acquisition unit 101 acquires the weather information. At this time, the acquisition unit 101 may acquire the weather information at the input time. The store clerk may also be able to specify the time at which the weather information is to be acquired on the input screen. In other words, the store clerk may be able to specify when the weather information is to be acquired. The acquisition unit 101 can acquire the weather information at the specified time.

[0067] When the customer demographic information box is checked on the input screen, the acquisition unit 101 acquires the customer demographic information. As described above, the customer demographic information includes attribute information of customers who use the store where the announcement is made, attribute information of customers who are in the store at the time the announcement is made, and information on the level of customer traffic. Therefore, the store clerk may be able to specify on the input screen which of the customer demographic information to acquire: attribute information of customers who use the store where the announcement is made, attribute information of customers who are in the store at the time the announcement is made, or information on the level of customer traffic. In other words, the acquisition unit 101 can acquire the specified customer demographic information.

[0068] The screens shown in Figures 6 and 7 are examples, and the display of the input screen is not limited to these. The input screen may be, for example, a screen on which a store clerk can input information to be used in generating an announcement. Furthermore, the information that can be input on the screen is not limited to the example of the screen in Figure 6. The information that can be input on the screen may be information that the announcement script generation unit 102 uses to generate an announcement script. Furthermore, the announcement template, business type information, flyer information, and additional information may be input on the same screen.

[0069] The input screens shown in Figures 6 and 7 are screens on which input content can be modified even after an announcement script has been generated by the announcement script generation unit 102 (described later). The announcement script generation unit 102 (described later) generates an announcement script based on the input content. The store clerk may then check the generated announcement script and wish to change the input content. In this case, the store clerk can modify the input content on the input screen. In other words, the acquisition unit 101 can accept the modified input content on the input screen. The announcement script generation unit 102 can then generate an announcement script using the modified input content.

[0070] 6 and 7 are referred to again. By selecting the "Prompt" tab, the store clerk can view a screen displaying a prompt created by the prompt creation unit 204 (described later). The store clerk can modify the prompt on the screen displaying the prompt created by the prompt creation unit 204. That is, the acquisition unit 101 may be able to acquire a modified prompt that has been modified on the screen displaying the prompt created by the prompt creation unit 104. Then, by selecting the "Announcement Script" tab, the store clerk can view a screen displaying an announcement script generated by the announcement script generation unit 102 (described later). The store clerk may view the generated announcement script and modify the input content. That is, the acquisition unit 101 can acquire the modified input content that has been modified on the input screen. The store clerk may also view the generated announcement script and modify the prompt. That is, the acquisition unit 101 can acquire a modified prompt that has been modified on the screen displaying the prompt.

[0071] The clerk may enter all of the additional information for which there is an input field on the input screen. The clerk may enter some of the additional information for which there is an input field on the input screen. In other words, the clerk may enter only the necessary additional information from the additional information for which there is an input field on the screen.

[0072] When the acquisition unit 101 receives a prompt, the announcement template, business type information, and flyer information may be reflected on the input screen based on the prompt. That is, the announcement template, business type information, and flyer information included in the prompt may be reflected on the input screen for the announcement template, business type information, and flyer information. For example, the business type information included in the prompt may be reflected in the store business type field of the input screen illustrated in FIG. 6 . Furthermore, the store clerk can modify the announcement template, business type information, or flyer information on the input screen reflecting the information included in the prompt. That is, the acquisition unit 101 can accept modifications to the announcement template, business type information, or flyer information input on the screen reflecting the information included in the prompt. Furthermore, the store clerk can additionally input announcement template, business type information, or flyer information on the input screen reflecting the information included in the prompt. That is, the acquisition unit 101 can accept additional input of announcement template, business type information, or flyer information input on the screen reflecting the information included in the prompt. Furthermore, when the acquisition unit 101 receives a prompt including additional information, the additional information may be reflected on the input screen based on the prompt.

[0073] When the acquiring unit 101 acquires the announcement template, business type information, and flyer information input on the input screen, the announcement generation device 10 may include a prompt creating unit 104 .

[0074] The prompt creation unit 104 is one aspect of a prompt generation means that creates a prompt including an announcement template, business type information, and flyer information. In other words, when the acquisition unit 101 acquires the announcement template, business type information, and flyer information input on the input screen, the prompt creation unit 104 creates a prompt based on the input content.

[0075] An example of a method for the prompt creation unit 104 to create a prompt will be described. For example, the prompt creation unit 104 has a prompt format that instructs the creation of an announcement. The prompt format may be, for example, a prompt format in which the generation instruction, the announcement template, and flyer information are listed in bullet points. The prompt format does not include specific details of the announcement template, business type information, and flyer information. Therefore, the prompt creation unit 104 writes the input content acquired by the acquisition unit 101 in a section of the prompt format that corresponds to the input content. As a result, the prompt creation unit 104 creates a prompt that instructs the creation of an announcement script from the prompt format.

[0076] A specific example will be used to explain how the prompt generator 104 generates a prompt. In the specific example, the prompt generated by the prompt generator 104 is the prompt shown in FIG. 5 as an example.

[0077] First, the acquisition unit 101 acquires business type information, "low-cost supermarket," on the input screen. At this time, the prompt creation unit 104 enters the business type information in the "<Store Type>" column in the prompt format. The acquisition unit 101 also acquires flyer information on the input screen. At this time, the prompt creation unit 104 enters the information acquired by the acquisition unit 101 in the "#flyer information" paragraph in the prompt format. Then, the acquisition unit 101 acquires an announcement template on the input screen. At this time, the prompt creation unit 104 enters the information acquired by the acquisition unit 101 in the "#announcement template" paragraph in the prompt format. Furthermore, if the acquisition unit 101 acquires additional information, the prompt creation unit 104 may enter the acquired additional information in the "#additional information" paragraph in the prompt format.

[0078] The input contents acquired by the acquisition unit 101 are examples and are not limited to these. The contents of the prompt created by the prompt creation unit 104 are examples and are not limited to these. Furthermore, the announcement template, business type information, and how to write flyer information are not limited to these. The prompt created by the prompt creation unit 104 may be written in a sentence format.

[0079] The method for generating a prompt by the prompt generator 104 is not limited to the above. The prompt generator 104 may generate a prompt using known sentence generation techniques. The prompt generator 104 may generate a prompt using, for example, a language model.

[0080] Next, the processing of the prompt creation unit 104 when the acquisition unit 101 acquires a prompt will be described. In this case, the prompt creation unit 104 may use the prompt acquired by the acquisition unit 101 as the prompt created by the prompt creation unit 104. Furthermore, when the acquisition unit 101 acquires a prompt, the announcement generation device 10 does not need to include the prompt creation unit 104.

[0081] When the announcement generation device 10 includes the prompt creation unit 104, the acquisition unit 101 may receive both an instruction to generate an announcement, flyer information, and an announcement template input to the input screen, and a prompt. The prompt creation unit 104 may also be included in the acquisition unit 101.

[0082] The announcement script generation unit 102 is one aspect of announcement script generation means that generates an announcement script for recommending products listed in a flyer using a language model based on an announcement template and business type information. The announcement generation device 10 may include a generation unit 801. In this case, the announcement script generation unit 102 inputs a prompt to a generation model used by the generation unit 801. The generation unit 801 generates an announcement script using the generation model according to the announcement template, business type information, and flyer information included in the prompt. The generation unit 801 then outputs the generation result of the announcement script generated using the generation model to the announcement script generation unit 102. In other words, the announcement script generation unit 102 acquires the generation result of the announcement script generated by the generation model used by the generation unit 801.

[0083] The announcement generation device 10 may also be connected to a generation server 80 including a generation unit 801 via a wired or wireless network. The generation unit 801 generates an announcement script using a generation model. The generation unit 801 may also generate audio data for the announcement using the generation model. The announcement generation device 10 transmits a prompt to the generation server 80 including the generation unit 801. That is, the announcement generation device 10 inputs the prompt into the generation model used by the generation unit 801. Then, the announcement generation device 10 receives the generation result from the generation server 80 including the generation unit 801. That is, the announcement generation device 10 acquires the generation result based on the generation model used by the generation unit 801.

[0084] The generation of an announcement script by the generation unit 801 using a generative model will be described.

[0085] The generative model generates an announcement script for recommending a product included in the flyer information acquired by the acquisition unit 101.

[0086] The generative model generates an announcement script according to the store's business format information. As described above, business format information is information about the store's business format. The scripts of announcements made at a store may differ depending on the store's business format. For example, the product information included in the announcement script may differ depending on the store's business format. Furthermore, the wording used in the announcement script may also differ depending on the store's business format. Therefore, examples of announcement scripts according to business formats will be described.

[0087] For example, if a store is a department store, the opening part of the announcement script may be a sentence that politely expresses gratitude to customers for visiting the store. A specific example of a sentence that politely expresses gratitude to customers for visiting the store is, "Thank you very much for visiting Department Store A today." On the other hand, if a store is a supermarket, the opening part of the announcement script may be different from when the store is a department store. An example of the opening part of an announcement script used in a supermarket is, "Welcome, welcome. Thank you for visiting Supermarket B today." In this way, the wording used in the announcement script may differ depending on the store's business format.

[0088] We will also explain examples of how product information included in an announcement script varies depending on the store's business format. For example, if a store is a supermarket, the announcement script may emphasize the price or discount rate of a recommended product. An example of this emphasis is when the announcement script repeatedly states the price or discount rate of a recommended product. The announcement script may also include a sentence encouraging customers to purchase the recommended product due to its price or discount rate. An example of an announcement script for a supermarket is, "Product C is only 198 yen, 198 yen, a great deal!" On the other hand, if the store is a department store, the announcement script may not emphasize the product's price or discount rate. Furthermore, if the store is a department store, the announcement script may emphasize the product's name recognition or exclusivity. An example of an announcement script for a department store is, "Currently, Kyoto's popular cafe D Sabo is holding a special event on the first basement floor food floor. Please take advantage of this opportunity to stop by." Furthermore, if the store is a drugstore, the announcement script may emphasize the effects or ingredients of the products.

[0089] The above-mentioned announcement scripts are merely examples and are not limiting. Any announcement script may be used as long as it is suited to the store's business style. The above-mentioned examples of announcement scripts may also be included in the prompt as samples. The examples of announcement scripts in the following description may also be included in the prompt as samples.

[0090] The store's business type information may include information about the price range of the products sold. The scripts of announcements made at the store may differ depending on the price range of the products sold at the store. For example, the product information included in the announcement script may differ. Furthermore, the wording used in the announcement script may also differ depending on the store's business type.

[0091] An example will be described in which a store operates as a supermarket. When the store's business type information includes information about the price range of the products sold, examples of the store's business type include a high-end supermarket and a discount supermarket. In the case of a high-end supermarket, the announcement script may include information about the quality of the products recommended in the announcement. The announcement script may also use polite language. An example of an announcement script for a high-end supermarket is, "We have fresh saury in stock today. Saury is a fish in season right now. How about having saury for dinner tonight?" On the other hand, in the case of a discount supermarket, the announcement script may particularly emphasize the price or discount rate of the recommended product. The announcement script may also use energetic language that makes the customer feel familiar. An example of an announcement script for a supermarket is, "Come in, come in! Today we have a great deal on cabbages! 98 yen per head! 98 yen per head! A great bargain!"

[0092] Here, a method for determining which product to recommend in an announcement when multiple products are listed in a flyer will be described. Here, the number of products recommended in the announcement may be one or more.

[0093] The product to be recommended in the announcement may be determined, for example, by the announcement script generation unit 102. Then, the announcement script generation unit 102 may output a prompt including an instruction to generate an announcement script for the determined product to the generation unit.

[0094] When a flyer lists multiple products, the announcement script generation unit 102 may determine the product to be recommended in the announcement based on the flyer information. When the acquisition unit 101 acquires image data of the flyer, the announcement script generation unit 102 may determine the product written at the top of the flyer as the product to be recommended in the announcement. Furthermore, the announcement script generation unit 102 may determine the product whose product information occupies the largest proportion of the entire flyer as the product to be recommended in the announcement. Furthermore, the announcement script generation unit 102 may determine the product whose product photo is the largest in the flyer as the product to be recommended in the announcement.

[0095] When a flyer lists multiple products, the announcement script generation unit 102 can determine which product to recommend based on the business type information.The announcement script generation unit 102 can then generate an announcement script based on the business type information.The announcement script generation unit 102 can determine which product is recommended in the announcement if the product in the flyer is accompanied by a slogan about its low price.

[0096] When a flyer contains multiple products, the product to be recommended in the announcement may be determined by a generative model. In this case, the generative model may be a model that has been additionally trained to learn the correspondence between the flyer information or business type information and the products to be recommended.

[0097] In addition, when a flyer lists multiple products, the acquisition unit 101 may accept a selection of a product to be recommended in an announcement by a store clerk. At this time, the announcement script generation unit 102 may output a prompt including an instruction to generate an announcement script for the product whose selection has been accepted to the generation unit.

[0098] A case where the acquisition unit 101 acquires additional information will be described. In this case, if a flyer lists multiple products, the announcement script generation unit 102 may first determine the product to be recommended in the announcement. Then, the generative model can generate an announcement script for recommending the products listed in the flyer based on the additional information. In other words, the generative model can generate an announcement script including product information corresponding to the additional information. The generative model can also generate an announcement script including sentences corresponding to the additional information.

[0099] As described above, the additional information is information that assists in generating an announcement script appropriate for the situation in which the announcement is made, and is information that describes the situation of the store in which the announcement is made. The additional information includes, for example, sales information, customer demographic information, weather information, or information on the time period in which the announcement is made. Here, examples of announcement scripts corresponding to each type of additional information will be described.

[0100] First, an example in which the acquisition unit 101 acquires sales information as additional information will be described. The announcement script generation unit 102 can use the sales information to determine which products to recommend in the announcement from among the products included in the flyer information. For example, the announcement script generation unit 102 can determine, based on the sales information, products for which sales are desired to be increased as products to recommend in the announcement. For example, the announcement script generation unit 102 can determine, as products for which sales are desired to be increased, products with a low number of units sold or a low sales amount within a predetermined period. In other words, products with a low number of units sold or a low sales amount within a predetermined period are products for which sales are desired to be increased. Furthermore, the acquisition unit 101 may acquire, as sales information, sales plans or sales targets in addition to the number of units sold or the sales amount of the products. In this case, the announcement script generation unit 102 can determine, for example, products for which sales are not reaching the sales plan or the sales target as products to recommend in the announcement. In other words, products for which sales are not reaching the sales plan or the sales target are products for which sales are desired to be increased. Furthermore, the announcement script generation unit 102 can determine, for example, a product whose sales amount has not reached the sales plan or sales target as a product to be recommended in the announcement. In other words, a product whose sales amount has not reached the sales plan or sales target is a product for which sales should be increased.

[0101] The method of determining the products to be recommended in the announcement by the announcement script generation unit 102 is not limited to these. For example, the generative model may determine that the products to be recommended in the announcement are those with a large number of units sold within a predetermined period or those with a large sales amount. Furthermore, the determination of the products to be recommended in the announcement may be performed using the generative model, or the acquisition unit 101 may accept the selection of products by the store clerk.

[0102] The generative model can then generate an announcement script based on the sales information. For example, the generative model can generate an announcement script that emphasizes information about a product for which sales are desired to be increased. An announcement script that emphasizes information about a product for which sales are desired to be increased is, for example, an announcement script that emphasizes the information about the product for which sales are desired by repeating the information. An announcement script that emphasizes information about a product for which sales are desired to be increased may also be an announcement script that includes wording that emphasizes the information about the product for which sales are desired to be increased. An announcement script that emphasizes information about a product for which sales are desired to be increased may also be an announcement script that includes wording that encourages purchase of the product for which sales are desired to be increased. A specific example of an announcement script that emphasizes information about a product for which sales are desired to be increased is, "At our store, product F is only 300 yen! Only 300 yen! You can purchase it at our special price of 300 yen. Why not take one?" This specific example is an example of an announcement script that emphasizes the price of the product as product information. Information about a product for which sales are desired to be increased may also be emphasized, such as sales quantity or limited time availability. Furthermore, when multiple products are recommended in an announcement, the generative model can generate an announcement script that particularly emphasizes the product for which sales are desired to be increased. The announcement scripts based on the sales information described above are merely examples, and are not limited to these.

[0103] An example in which the acquisition unit 101 acquires customer demographic information as additional information will be described. First, the announcement script generation unit 102 can use the customer demographic information to determine which products to recommend in the announcement from among the products included in the flyer information. One example of customer demographic information is attribute information of customers who use the store for which the announcement is made. Another example of customer demographic information is attribute information of customers who are in the store at the time the announcement is made. Therefore, the announcement script generation unit 102 can determine products according to the attributes of customers who use the store for which the announcement is made as products to recommend in the announcement. For example, products targeted for sale to attributes with a large number of customers can be determined as products to recommend in the announcement. In other words, the announcement script generation unit 102 can generate an announcement script including information on products targeted for sale to attributes with a large number of customers. Specifically, if there are a large number of female customers using the store, the announcement script generation unit 102 can determine products recommended for women or products targeted for sale to women as products to recommend in the announcement. Additionally, if there are many elderly people, the announcement script generating unit 102 can determine that products targeted for sale to elderly people are the products to be recommended in the announcement.

[0104] The generative model can then generate an announcement script that corresponds to the customer demographic information. Examples of customer demographic information include attribute information of customers who use the store where the announcement is made and attribute information of customers who are in the store at the time the announcement is made. Therefore, the generative model can generate an announcement script that corresponds to the attributes of customers who use the store where the announcement is made. For example, the generative model can generate an announcement script that includes information that would interest customers with the attributes of using the store. For example, if the store has many business customers, information that would interest customers with the attributes of using the store could include the effects of a product that is effective against eye strain and swelling of the feet, or the feature of a product that a meal can be prepared in a short time. The generative model can also generate an announcement script that emphasizes information that would interest customers with the attributes of using the store.

[0105] Another example of customer demographic information is information about the number of customers in a store. Information about customer demographic information refers to the number of customers in the store at a given time. Furthermore, the information about customer demographic information may be information about the ratio of the number of customers to a predetermined number of customers. Information about customer demographic information may also be information about the result of a comparison with the number of customers in the store in the past. A situation in which there are many customers in a store may mean that the store is crowded. A situation in which there are many customers in a store may mean, for example, when the number of customers in the store is equal to or greater than a predetermined number, when the number of customers exceeds a predetermined number, or when the number of customers exceeds the number of customers in the past. Therefore, the generative model can generate an announcement script that includes a sentence apologizing to customers for the congestion. An example of a sentence apologizing for the congestion is, "The store is currently crowded. We sincerely apologize for any inconvenience caused." Furthermore, when there are many customers in a store, the generative model can generate an announcement script that emphasizes important information from the product information of recommended products. The announcement scripts based on the above customer demographic information are merely examples and are not limited to these.

[0106] An example will be described in which the acquisition unit 101 acquires weather information as additional information. First, the announcement script generation unit 102 can use the weather information to determine which products to recommend in the announcement from among the products included in the flyer information. As described above, the weather information may include information on the season when the announcement is made. Therefore, the announcement script generation unit 102 can determine which products listed in the flyer are recommended for the season as the products to recommend in the announcement. Products recommended for the season are, for example, products whose sales season included in the product information corresponds to the season. Specifically, if the season when the announcement is made is the rainy season, the announcement script generation unit 102 can determine which products to recommend in the announcement are humidity-preventing products.

[0107] The generative model can then generate an announcement script that corresponds to the weather information. For example, if the weather around the store when the announcement is to be made is bad, the generative model can generate an announcement script that includes a sentence related to the bad weather. Bad weather includes, for example, rain, a typhoon, or a thunderstorm. An example of a sentence related to bad weather is, "Thank you very much for visiting our store today despite the bad weather." In other words, an announcement script that includes a sentence related to bad weather is an announcement script that expresses gratitude to customers who visited the store in bad weather. The sentence related to bad weather is not limited to this. Furthermore, the wording of the sentence related to bad weather may differ depending on the business type information of the store. In other words, the generative model can vary the wording of the sentence related to bad weather depending on the business type information. Announcement scripts that correspond to weather information are not limited to these.

[0108] An example will be described in which the acquisition unit 101 acquires information on the time period during which the announcement will be made as additional information. First, the announcement script generation unit 102 can use the time period information to determine which products will be recommended in the announcement from among the products included in the flyer information. For example, the announcement script generation unit 102 can determine which products are suitable for the time period during which the announcement will be made as the products to be recommended in the announcement. An example will be described in which the store is a supermarket. If the time period during which the announcement will be made is, for example, from 12:00 to 1:00 p.m., the generative model can determine which products are suitable for lunch from among the products listed in the flyer as the products to be recommended in the announcement. Furthermore, if the time period during which the announcement will be made is a time sale period, the announcement script generation unit 102 can determine which products will be sold during a time sale from among the products listed in the flyer as the products to be recommended in the announcement.

[0109] The generative model can then generate an announcement script that corresponds to the time period during which the announcement is made. For example, the generative model can generate an announcement script that includes information related to the time period during which the announcement is made. For example, if the time period during which the announcement is made is from 12:00 to 13:00, an announcement script that includes information related to the time period during which the announcement is made is an announcement script that includes the sentence "How about coming here for lunch?". Furthermore, if the time period during which the announcement is made is a time sale period, the generative model can generate an announcement script that emphasizes that the announcement is a time sale. Announcement scripts that correspond to the time period during which the announcement is made are not limited to these.

[0110] The generative model may be an existing model, or may be a model that has been trained to generate announcement scripts according to business type information, or may be a model that has been trained to generate announcement scripts according to each piece of additional information.

[0111] The generative model may also be a model that has been additionally trained on the advertising effect of a product through an announcement. In other words, the generative model may be additionally trained on the correspondence between the announcement scripts of past store announcements and the change in sales of the products recommended in the announcements after the announcements.

[0112] The generative model acquires sales information for a predetermined period before the announcement for the product recommended in the announcement. The generative model also acquires sales information for a predetermined period after the announcement for the product recommended in the announcement. The predetermined period before the announcement is, for example, one week before the announcement. The predetermined period after the announcement is, for example, one week after the announcement.

[0113] As described above, the sales information is the number of units or the sales amount of the products listed in the flyer that were actually sold within a predetermined period. The sales information is stored in, for example, the database 92.

[0114] The change in sales of a product recommended in an announcement after the announcement may reflect the advertising effectiveness of the announcement. For example, if sales of a product recommended in an announcement increase after the announcement, the advertising effectiveness of the announcement may be high. Therefore, the generative model is trained on the announcement script as an announcement script with high advertising effectiveness. On the other hand, if sales of a product recommended in an announcement remain unchanged or decrease after the announcement, the advertising effectiveness of the announcement may be low. Therefore, the generative model is trained on the announcement script as an announcement script with low advertising effectiveness. The comparison of sales before and after the announcement may be a comparison of the number of products sold or the sales amount within a predetermined period. Furthermore, the comparison of sales before and after the announcement may be, for example, a comparison of the average number of products sold or the average sales amount per day.

[0115] Furthermore, the promotional effectiveness of the announcement may be reflected in the results of comparing the sales of a product recommended in an announcement with a sales target or sales plan. Therefore, the generative model may be additionally trained using information on the sales target or sales plan of the product in addition to the sales information of the product recommended in the announcement. Information on the sales target or sales plan of the product is stored, for example, in a database 92. The information on the sales target or sales plan of the product includes the number of units or the sales amount of the product, which constitutes the sales target or sales plan of the product. If the sales of the product recommended in the announcement after the announcement reaches the sales target or sales plan of the product, the promotional effectiveness of the announcement may be high. Furthermore, if the sales of the product recommended in the announcement after the announcement exceed the sales target or sales plan of the product by a predetermined amount or more, the promotional effectiveness of the announcement may be high. Therefore, the generative model additionally trains the announcement script as one with high promotional effectiveness. On the other hand, if the sales of the product recommended in the announcement after the announcement are lower than the sales target or sales plan of the product, the promotional effectiveness of the announcement may be low. Furthermore, if the sales of the product recommended in the announcement after the announcement are lower than the sales target or sales plan for the product by a predetermined amount or more, the advertising effect of the announcement may be low. Therefore, the generative model additionally learns the announcement script as an announcement script with low advertising effect.

[0116] Here, the generative model that generates the announcement script is, for example, a language model. For example, GPT-2 (Generative Pre-Training-2) or GPT-3 can be used as the language model. These language models are also called large-scale language models. Further, as the language model, T5 (Text-to-Text Transfer Transformer), BERT (Bidirectional Encoder Representations from Transformers), RoBERTa (Robustly optimized BERT approach), or ELECTRA (Efficiently Learning an Encoder that Classifies Token Replacements Accurately) may be used. The language model is not limited to the above.

[0117] The output unit 103 is one aspect of output means that outputs an announcement script. The output unit 103 outputs the announcement script generated by the announcement script generation unit 102, for example, in a format that can be referenced by store staff. The output unit 103 outputs the announcement script to the store terminal 91. The store terminal 91 may then display the announcement script on its screen. The store staff can check the announcement script by looking at the screen of the store terminal 91. The output unit 103 may also store the announcement script in the database 92. The output unit 103 may also output the announcement script to a printing device (not shown). The printing device may then print the announcement script. The printing device and the announcement generation device 10 are connected via a wired or wireless network. The output unit 103 may also output the announcement script to a display device (not shown). The display device displays the announcement script on its screen. The display device is installed in a location used by store staff, such as a backroom of a store. The staff can check the announcement script by looking at the screen of the display device. The display device and the announcement generating device 10 are connected via a wired or wireless network.

[0118] 8, the operation of the announcement generating device 10 including the acquisition unit 101, the announcement script generating unit 102, and the output unit 103 will be described.

[0119] First, in step S101, the acquisition unit 101 acquires an announcement template for recommending products sold in the store, store business type information, and store flyer information. Next, in step S102, the announcement script generation unit 102 uses a language model to generate an announcement script for recommending products listed in the flyer, based on the announcement template and business type information. Furthermore, in step S103, the output unit 103 outputs the announcement script. Then, the announcement generation device 10 ends its operation.

[0120] The operation of the announcement generation device 10 when the announcement generation device 10 includes the prompt creation unit 104 will be described. After step S101, the prompt creation unit 104 creates a prompt. Then, in step S102, the announcement script creation unit 102 uses a language model to create an announcement script for recommending products listed in the flyer, based on the announcement template and business type information. Step S103 is as described above.

[0121] In this embodiment, the announcement generation device 10 includes an acquisition unit 101 that acquires announcement templates for recommending products sold in a store, store business type information, and store flyer information; an announcement script generation unit 102 that uses a language model to generate announcement scripts for recommending products featured in the flyer based on the announcement templates and the business type information; and an output unit 103 that outputs the announcement scripts. With this configuration, announcement scripts for announcements recommending products sold in a store may differ depending on the store's business type and products. Generating announcement scripts tailored to the store's business type and products may not be easy. However, the announcement generation device 10 makes it possible to easily generate announcement scripts tailored to the store's business type and products.

[0122] In this embodiment, the announcement generation device 10 further includes a prompt creation unit 104 that creates a prompt including an announcement template, business type information, and flyer information. The acquisition unit 101 acquires the announcement template, business type information, and flyer information input on a screen on which the announcement template, business type information, and flyer information can be input. A store clerk inputs the announcement template, business type information, and flyer information using, for example, a store terminal 91. The acquisition unit 101 then acquires the announcement template, business type information, and flyer information input from the store terminal 91. The prompt creation unit 104 creates a prompt including the announcement template, business type information, and flyer information based on the input content. This configuration allows even a person unfamiliar with prompt creation to generate announcements using a language model. Furthermore, it is possible to reduce the effort required to create an appropriate prompt.

[0123] In this embodiment, the announcement generation device 10 has an acquisition unit 101 that acquires a prompt including an announcement template, business type information, and flyer information, and an announcement script generation unit 102 that generates an announcement script based on the prompt. The acquisition unit 101 acquires the prompt itself. Then, the announcement script generation unit 102 generates an announcement script based on the prompt. By the acquisition unit 101 acquiring the prompt, the announcement generation device 10 can omit processing related to the prompt, such as creating the prompt. In other words, the announcement generation device 10 can reduce processing.

[0124] Furthermore, by having the acquisition unit 101 acquire a prompt, for example, it is possible to prevent an error when creating a prompt. An error when creating a prompt is, for example, an error in which part of the content acquired by the acquisition unit 101 is not included in the created prompt. If such an error occurs, it may be impossible to generate an appropriate announcement script. On the other hand, by having the acquisition unit 101 acquire a prompt, it is possible to prevent the above-mentioned error from occurring. In other words, the announcement generation device 10 can generate an announcement script that suits the situation in which the announcement is to be made.

[0125] In this embodiment, the announcement script generation unit 102 of the announcement generation device 10 generates an announcement script according to business type information. Which products listed in a flyer are recommended in the announcement may differ depending on the business type of the store making the announcement. In addition, the product information included in the announcement script may differ depending on the business type of the store making the announcement. Furthermore, the wording used in the announcement script may also differ depending on the business type of the store. Therefore, the announcement script generation unit 102 generates an announcement script according to the business type information that represents the business type of the store. With this configuration, the announcement generation device 10 can generate announcement scripts that effectively promote products.

[0126] For example, if a store is a department store and uses announcements in the same language as those used in discount supermarkets, some customers may feel uncomfortable. However, if the store is a department store, customers may be able to shop more comfortably if announcements are made in polite language. Therefore, by generating announcement scripts that match the store's business format, it is possible to improve customer satisfaction when using the store.

[0127] In this embodiment, the acquisition unit 101 of the announcement generation device 10 further acquires product sales information, and the announcement script generation unit 102 generates an announcement script based on the sales information. Stores sometimes make announcements to promote product information to customers. Therefore, recommending in an announcement, for example, a product listed in a flyer whose sales are lower than planned, can lead to increased sales of that product. In other words, the announcement script generation unit 102 determines the product to recommend in the announcement based on the sales information, thereby enabling increased sales of that product. Furthermore, the announcement script generation unit 102 generates an announcement script that emphasizes, for example, a listed product whose sales are lower than planned. The announcement script may include information other than product information. It may also include greetings to customers. Therefore, by emphasizing a product whose sales are desired to be increased, an announcement script that is effective for advertising can be generated.

[0128] In this embodiment, the acquisition unit 101 of the announcement generation device 10 further acquires customer demographic information about customers in the store, and the announcement script generation unit 102 generates an announcement script based on the customer demographic information. The customer demographic information includes, for example, attribute information about customers who use the store and attribute information about customers who are in the store at the time the announcement is made. Therefore, generating an announcement script based on the customer attribute information makes it possible to increase sales. The announcement script generation unit 102 generates an announcement script that includes information about listed products that are targeted for sale to attributes with a large number of customers, for example. By recommending products with a large number of target customers in an announcement, more customers are likely to purchase the products. In other words, generating an announcement script based on the customer demographic information makes it possible to increase sales.

[0129] The attribute information may also include information about the number of customers in the store. For example, when there are many customers in the store, the announcement script generation unit 102 can generate an announcement script that includes a sentence apologizing to the customers for the congestion. By including a sentence apologizing for the congestion in the announcement script, it is possible to reduce the discomfort customers feel about the congestion.

[0130] In this embodiment, the acquisition unit 101 of the announcement generation device 10 further acquires weather information about the area around the store when the announcement is made, and the announcement script generation unit 102 generates an announcement script based on the weather information. For example, if the weather information indicates that the weather around the store when the announcement is made will be bad, the announcement script generation unit 102 generates an announcement script that includes a sentence about the bad weather. Bad weather includes, for example, rain, a typhoon, or a thunderstorm. Including a sentence about bad weather in an announcement script may make customers feel that the store is considerate. In other words, by generating an announcement script based on the weather information, the announcement script generation unit 102 can generate an announcement script that leaves a favorable impression on customers.

[0131] In this embodiment, the announcement generation device 10 includes an acquisition unit 101 that acquires information regarding the time period during which the announcement will be made, and an announcement script generation unit 102 that generates an announcement script based on the time period during which the announcement will be made. The announcement script generation unit 102 can determine products suitable for the time period during which the announcement will be made as products to be recommended in the announcement. The announcement script generation unit 102 also generates an announcement script that includes an expression representing the time period. By generating an announcement script according to the time period during which the announcement will be made, an announcement script that is effective for advertising can be generated. An example of a time period during which the announcement will be made is a time period during which a limited-time sale is held. In this regard, the announcement script generation unit 102 can determine products to be sold in a limited-time sale as products to be recommended in the announcement. The announcement script generation unit 102 can also generate an announcement script that emphasizes the limited-time sale. By recommending products to be sold in a limited-time sale in an announcement, it is possible to increase sales of the products. Furthermore, by emphasizing the limited-time sale, it may be possible to notify more customers that a limited-time sale is being held. As a result, it is possible to increase sales.

[0132] [Second Embodiment] A second embodiment will be described in detail with reference to the drawings. Below, the description of the second embodiment will be omitted to the extent that it does not make the description of the present embodiment unclear.

[0133] The configuration of the announcement generation device 20 will be described with reference to Fig. 9. Fig. 9 is a block diagram showing an example of the configuration of the announcement generation device. The announcement generation device 20 includes an acquisition unit 201, an announcement script generation unit 202, an output unit 203, and a voice data generation unit 205. The announcement generation device 20 may also include a prompt creation unit 204. Here, the acquisition unit 201 corresponds to the acquisition unit 101, and the announcement script generation unit 202 corresponds to the announcement script generation unit 102. Furthermore, the output unit 203 corresponds to the output unit 103, and the prompt creation unit 104 corresponds to the prompt creation unit 204. Therefore, the announcement generation device 20 includes the voice data generation unit 205 in addition to the configuration of the announcement generation device 10.

[0134] Similarly to the acquisition unit 101, the acquisition unit 201 acquires an announcement template for recommending products handled by the store, store business type information, and store flyer information. Here, the acquisition unit 201 may acquire a prompt including an announcement template for recommending products handled by the store, store business type information, and store flyer information. Furthermore, the acquisition unit 201 may acquire the announcement template, business type information, and flyer information input on a screen on which the announcement template, business type information, and flyer information can be input. Here, the business type information may include information regarding a request for the type of voice in which voice data read aloud should be generated.

[0135] When the announcement generating device 20 includes the prompt generating unit 204, the prompt generating unit 204 generates a prompt including an announcement template, business type information, and flyer information.

[0136] The announcement script generating unit 202, like the announcement script generating unit 102, generates an announcement script for recommending products featured in a flyer using a language model based on an announcement template and business type information.

[0137] The voice data generating unit 205 is one aspect of voice data generating means for generating voice data for reading out a script of announcements. The voice data generating unit 205 generates voice data for reading out the script of announcements generated by the script of announcements generating unit 202. Furthermore, the voice data generating unit 205 can determine the type of voice for reading out the script of announcements and generate voice data for reading out the script of announcements in that type of voice.

[0138] The voice type refers to the way in which the announcement script is read aloud. The voice type includes, for example, the voice quality, the volume of the voice, or the reading speed. The voice quality includes male voice, female voice, low voice, high voice, calm voice, energetic voice, and the like. Examples of voice quality are not limited to these. Any voice type that can be set when generating voice data may be used.

[0139] The type of voice used to read the announcement script may be set in advance. In this case, the voice data generating unit 205 can determine to generate voice data in the preset voice type. Then, the voice data generating unit 205 generates voice data read in the preset voice type.

[0140] The type of voice used to read the announcement script may vary depending on the situation in which the announcement is made. That is, the voice data generating unit 205 can determine the type of voice appropriate to the situation in which the announcement is made. Then, it can generate voice data in which the announcement script is read in the determined voice type.

[0141] The situation in which the announcement is made is, for example, the business type of the store where the announcement is made. That is, the voice data generation unit 205 can generate voice data in which the announcement script is read aloud using a voice type corresponding to the business type information. The voice types corresponding to the business type information will be described. For example, if the business type of the store is a department store, the voice data generation unit 205 can generate voice data in which the announcement script is read aloud in a calm voice. Also, if the business type of the store is a supermarket, the voice data generation unit 205 can generate voice data in which the announcement script is read aloud in a loud and spirited voice. Furthermore, the voice data generation unit 205 may generate voice data read aloud in different voice types depending on whether the business type of the store is a discount supermarket or a high-end supermarket. The voice types corresponding to the business type information are not limited to these. The voice data generation unit 205 can generate voice data read aloud in an appropriate voice type depending on the business type of the store.

[0142] The situation in which the announcement is made may be, for example, the sales status of the product recommended in the announcement. In other words, the audio data generation unit 205 can generate audio data in which the announcement script is read aloud using a voice type corresponding to the sales information. The voice type corresponding to the sales information may be, for example, a voice type that emphasizes information about a product for which sales are desired to be increased. The audio data generation unit 205 can generate audio data in which the portion of the announcement script that contains information about the product for which sales are desired to be increased is read aloud in a loud and slow voice. Furthermore, the audio data generation unit 205 can generate audio data in which the portion of the announcement script that contains information about the product for which sales are desired to be increased is read aloud in a spirited voice. Here, an example of a product for which sales are desired to be increased is, as described above, a product with a small number of units sold or a low sales amount within a specified period. Another example of a product for which sales are desired to be increased is a product for which the number of units sold or the sales amount of the product has not reached the sales plan or sales target.

[0143] The situation in which the announcement is made refers to, for example, the situation of the customers of the store where the announcement is made. In other words, the voice data generation unit 205 can generate voice data in which the announcement script is read aloud using a voice type corresponding to the customer demographic information. As described above, the customer demographic information includes attribute information of the customers who use the store where the announcement is made. In this case, the voice type corresponding to the customer demographic information is a voice type that is well-received by the attributes of the customers who use the store where the announcement is made. For example, if the store where the announcement is made has a large number of female customers, the voice data generation unit 205 can generate voice data read in a voice that is well-received by women. Furthermore, if the store where the announcement is made has a large number of elderly customers, the voice data generation unit 205 can generate voice data read in a voice type that is easy for elderly people to hear. In other words, the voice data generation unit 205 can generate voice data read aloud in a loud and slow voice. Furthermore, the customer demographic information may include attribute information of customers who are in the store at the time the announcement is made. In this case, the voice type corresponding to the customer demographic information is a voice type that is well-received by the attributes of the customers who are in the store at the time the announcement is made. For example, if there are many families among the customers in the store when the announcement is made, the voice data generation unit 205 can generate voice data read in a voice that is popular with children.In addition, the voice data generation unit 205 may generate voice data read in a voice type that emphasizes information that is important to customers who use the store where the announcement is made or customers who are in the store when the announcement is made.

[0144] Furthermore, the customer demographic information may include information on the frequency of customer visits to the store. The information on the frequency of customer visits is the number of customers in the store at a given time. When there are many customers in the store, the voice data generation unit 205 can generate voice data that is read out loud and slowly. When there are many customers in the store, the store may be noisy. Therefore, the voice data generation unit 205 can generate voice data that is read out at a volume and speed that is easy for customers to hear. Here, when there are many customers in the store, for example, the number of customers in the store is equal to or greater than a predetermined number, exceeds a previously predicted number of customers, or is greater than the number of customers who have visited the store in the past.

[0145] The situation in which the announcement is made may be, for example, the type of product recommended in the announcement. In other words, the voice data generation unit 205 can generate voice data in which the announcement script is read aloud in a voice type corresponding to the product recommended in the announcement. The voice data read aloud in a voice type corresponding to the product recommended in the announcement may be, for example, voice data read in a calm voice when the product recommended in the announcement is a luxury item. The voice data read aloud in a voice type corresponding to the product recommended in the announcement may be, for example, voice data read in a spirited and loud voice when the product recommended in the announcement is a bargain item. In addition, the voice data generation unit 205 can generate voice data that emphasizes important information among the information about the product recommended in the announcement. For example, when the product recommended in the announcement is a bargain item, the voice data generation unit 205 may generate voice data read aloud in a voice type that emphasizes the price of the product.

[0146] The situation in which the announcement is made refers to, for example, when the announcement is to be made. In other words, the voice data generation unit 205 can generate voice data in which a scripted announcement is read aloud using a voice type that corresponds to the time period in which the announcement is to be made. An example of information about the time period in which the announcement is to be made is information about an event that will be held during that time period. An example of an event that will be held during the time period in which the announcement is to be made is a limited-time sale. Therefore, the voice data generation unit 205 can generate voice data in which a scripted announcement is read aloud in a spirited voice for an announcement during the time period in which a limited-time sale is to be held.

[0147] Examples of voice data read in a voice type appropriate for the situation in which the announcement is made are not limited to those described above. The voice data generating unit 205 can generate voice data read in a voice type appropriate for the situation in which the announcement is made.

[0148] A method for generating voice data will now be described. The voice data generation unit 205 may determine the type of voice to read the announcement script. The voice data generation unit 205 can determine the type of voice to read the announcement script, for example, by using a table that associates the situation in which the announcement is to be made with the type of voice in advance.

[0149] The voice data generating unit 205 then generates voice data in the voice type determined using a table in which the situation in which the announcement is made is associated with the voice type. The voice data generating unit 205 can generate voice data using known technology.

[0150] Furthermore, the voice data generation unit 205 may generate voice data using a generation model. That is, the voice data generation unit 205 can use the generation model to determine the type of voice that will read out the announcement script, and generate voice data that reads out the announcement script in that type of voice.

[0151] Here, the generation model used by the voice data generation unit 205 may be an existing model. Furthermore, the generation model used by the voice data generation unit 205 may be a generation model that has learned the type of voice corresponding to the situation in which the announcement is made. The generation model may have learned the correspondence between the situation in which the announcement is made and the type of voice corresponding to the situation, as described above.

[0152] The announcement script generator 202 and the voice data generator 205 may be configured in the same way. That is, the generation of the announcement script and the generation of the voice data for reading out the announcement script may be performed by the same configuration.

[0153] The announcement script generation unit 202 may also generate an instruction for the type of voice for the voice data to be generated by the voice data generation unit 205. That is, the announcement script generation unit 202 generates an announcement script and an instruction for the type of voice to read the announcement script. The voice data generation unit 205 may then generate voice data to be read aloud in accordance with the type of voice for which the instruction was generated by the announcement script generation unit 202. In this case, the generation model may be a generation model that has learned the type of voice according to the situation in which the announcement is made.

[0154] The output unit 203 outputs the audio data generated by the audio data generation unit 205. This is one aspect of a method for outputting an announcement script. The output unit 203 may output the audio data to the audio output device 90. Then, the audio output device 90 outputs the audio data. That is, the audio output device 90 plays the audio data. By outputting the audio data from the audio output device 90, an announcement can be made in the store. Then, by listening to the announcement, customers can understand information about the product recommended in the announcement. The output unit 203 may also output the audio data to the database 92. By storing the audio data in the database 92, the store clerk can use the audio data when making an announcement. That is, the store clerk can make an announcement in the store using the audio data stored in the database 92 via the audio output device 90. The output unit 203 may also output the audio data to the store terminal 91. For example, the store terminal 91 can output the audio data. That is, by playing the audio data on the store terminal 91, the store clerk can check the audio data.

[0155] The operation of the announcement generating device 20 will be described with reference to Fig. 10. The announcement generating device 20 includes an acquisition unit 201, an announcement script generating unit 202, an output unit 203, and an audio data generating unit 205. Fig. 10 is a flowchart showing the operation of the announcement generating device 20.

[0156] First, in step S201, the acquisition unit 201 acquires an announcement template for recommending products sold in the store, store business type information, and store flyer information. Next, in step S202, the announcement script generation unit 202 generates an announcement script for recommending products listed in the flyer using a language model based on the announcement template and business type information. Furthermore, in step S203, the voice data generation unit 205 generates voice data of the announcement script read aloud. Then, in step S204, the output unit 203 outputs the announcement script and the voice data. Then, the announcement generation device 20 ends its operation.

[0157] The operation of the announcement generation device 20 when the announcement generation device 20 includes the prompt creation unit 204 will be described. After step S201, the prompt creation unit 204 creates a prompt. Then, in step S202, the announcement script creation unit 202 uses a language model to create an announcement script for recommending products listed in a flyer, based on the announcement template and business type information. Steps S203 and S204 are as described above.

[0158] In this embodiment, the announcement generation device 20 further includes a voice data generation unit 205 that generates voice data for reading a scripted announcement, and the output unit 203 outputs the voice data. In stores, store clerks may make announcements by reading scripted announcements. However, in the announcement generation device 20, the announcement script generation unit 202 generates the scripted announcements and the voice data generation unit 205 generates the voice data, thereby eliminating the need for a store clerk to read a scripted announcement and make the announcement. Furthermore, if a store clerk makes multiple announcements, the store clerk may be unable to perform other tasks. However, the configuration of the announcement generation device 20 prevents the clerk's work from being interrupted by the announcement. In other words, the configuration of the announcement generation device 20 enables work efficiency to be improved. Furthermore, the store can reduce labor costs for the clerks who make announcements. Furthermore, the voice data generation unit 205 generates voice data, reducing individual differences between the clerks who read the announcements. For example, when a store clerk makes an announcement by reading a script, the reading style and ease of audibility may differ depending on the store clerk. However, by generating voice data using the voice data generation unit 205, individual differences in reading style and ease of audibility among store clerks can be reduced. Therefore, the announcement generation device 20 can make announcements of a consistent quality.

[0159] In this embodiment, the announcement generation device 20 has a voice data generation unit 205 that generates voice data in which a scripted announcement is read aloud using a voice type corresponding to the business type information. Store business type information is information related to the business format of the store. The type of voice used in announcements made within the store may differ depending on the business format of the store. Examples of the voice type include voice quality, volume, and speed. If the voice type corresponding to the business format is not used, customers may feel uncomfortable or annoyed. Therefore, by generating voice data in which a scripted announcement is read aloud using a voice type corresponding to the business type information, customers may be able to shop more comfortably. In other words, customer satisfaction can be improved.

[0160] In this embodiment, the acquisition unit 201 of the announcement generation device 20 further acquires sales information for the listed products, and the voice data generation unit 205 generates voice data in which an announcement script is read aloud using a voice type corresponding to the sales information. In a store, an announcement may be made to encourage customers to purchase a product. For example, an announcement may be made to recommend a product for which the store wishes to increase sales. In this case, the voice data generation unit 205 can generate voice data that emphasizes information about the product for which the store wishes to increase sales. By emphasizing product information in the announcement, the possibility of customers purchasing the product may increase. Therefore, the store can increase sales.

[0161] In this embodiment, the announcement generation device 20 further acquires customer demographic information about customers in the store through the acquisition unit 201, and the voice data generation unit 205 generates voice data for reading an announcement script using a voice type corresponding to the customer demographic information. The customer demographic information is, for example, attribute information about customers who use the store where the announcement is made or attribute information about customers who are in the store at the time the announcement is made. The voice type corresponding to the customer demographic information is, for example, a voice type that matches the customers who will be listening to the announcement. By generating voice data read using a voice type that matches the customers who will be listening to the announcement, customers may find the announcement easier to listen to. For example, if there are many elderly customers in the store, the voice data generation unit 205 can generate voice data read using a voice type that is easy for elderly people to hear. In other words, by generating voice data read using a voice type corresponding to the customer demographic information, it is possible to generate an announcement that leaves a favorable impression on customers. As a result, customer satisfaction can be improved. Furthermore, for stores, announcements can be generated that make it easy for customers to understand the information about the products recommended in the announcements, meaning that announcements can be generated with high advertising effectiveness, which can lead to increased sales of the products recommended in the announcements.

[0162] In this embodiment, the announcement generation device 20 includes an acquisition unit 201 that acquires information about the time period during which the announcement will be made, and an audio data generation unit 205 that generates audio data in which a scripted announcement is read aloud using a voice type corresponding to the time period during which the announcement will be made. The audio data generation unit 205 can generate audio data in which the scripted announcement is read aloud using a spirited voice or a calm voice depending on the time period during which the announcement will be made. For example, if the time period during which the announcement will be made is a limited-time sale, the audio data generation unit 205 can generate audio data read aloud using a spirited voice to liven up the store. As a result, customers who were unaware of the limited-time sale or who were not interested in the limited-time sale may become interested in the limited-time sale. Furthermore, customers who had planned to purchase products during the limited-time sale may end up purchasing more products than planned. In other words, the announcement generation device 20 can generate announcements with high advertising effectiveness. As a result, an increase in store sales can be expected.

[0163] In this embodiment, the announcement generation device 20 generates voice data in which the voice data generation unit 205 reads out an announcement script in a voice type corresponding to the published information. For example, if the product recommended in the announcement is a luxury item, the voice data generation unit 205 can generate voice data read out in a calm voice. In this case, even if the product recommended in the announcement is a luxury item, if the announcement is made in a loud voice, for example, it may not stimulate customers' desire to purchase. However, if the announcement is made in a calm voice, customers may become interested in the product recommended in the announcement. Therefore, the announcement generation device 20 can generate announcements with high advertising effectiveness.

[0164] 11 is a diagram illustrating an example of a hardware configuration of the announcement generation device 30 according to the present disclosure. The announcement generation device 30 is realized by a computer. The announcement generation device 30 is an example in which the announcement generation device 10 and the announcement generation device 20 are realized by a computer.

[0165] The announcement generating device 30 includes a processor 301, a ROM (Read Only Memory) 302, a RAM (Random Access Memory) 303, a storage device 304 such as a hard disk for storing programs, an input / output interface 305 for inputting and outputting data, and a communication interface 306 for network connection. Each component is connected via a bus 307.

[0166] The processor 301 runs an operating system to control the entire computer. Examples of the processor 301 include a CPU (Central Processing Unit), a DSP (Digital Signal Processor), and a GPU (Graphics Processing Unit). The processor 301 loads programs stored in, for example, the ROM 302 or the storage device 304. The processor 301 then executes each process coded in the program. The processor 301 may execute processes or instructions in the illustrated flowchart based on the program.

[0167] The ROM 302 stores application programs, programs according to the embodiments, etc. The RAM 303 is used as a work area for the processor 301.

[0168] The storage device 304 may be, for example, a semiconductor memory such as a flash memory, a hard disk driver (HDD), etc. The storage device 304 stores, for example, an operating system (OS) program, application programs, programs according to each embodiment, etc.

[0169] The input / output interface 305 is connected to peripheral devices (not shown) via a wired network or a wireless network.

[0170] The communication interface 306 is connected to a communication network (not shown) such as a LAN (Local Network) or a WAN (Wide Area Network) via a wireless or wired network. The communication network may be configured with multiple communication networks. This allows the computer to be connected to external devices via the communication network. The announcement generating device 30 may have components other than those shown in FIG. 11. For example, the announcement generating device 30 may have a drive device or the like. For example, the processor 301 may be attached to a drive device or the like and read programs and data stored in a non-transitory tangible recording medium into the RAM 303.

[0171] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure.

[0172] Some or all of the above-described embodiments can be described as, but are not limited to, the following supplementary notes.

[0173] (Supplementary Note 1) An announcement generation device comprising: an acquisition means for acquiring an announcement template for recommending products sold in a store, business type information of the store, and flyer information of the store; an announcement script generation means for generating an announcement script for recommending products listed in the flyer using a language model based on the announcement template and the business type information; and an output means for outputting the announcement script.

[0174] (Supplementary Note 2) The announcement generation device according to Supplementary Note 1 further comprises a prompt creation means for creating a prompt including the announcement template, the business type information, and the flyer information, and the acquisition means acquires the announcement template, the business type information, and the flyer information input on a screen on which the announcement template, the business type information, and the flyer information can be input.

[0175] (Supplementary Note 3) The announcement generation device according to Supplementary Note 1, wherein the acquisition means acquires a prompt including the announcement template, the business type information, and the flyer information, and the announcement script generation means generates the announcement script based on the prompt.

[0176] (Supplementary Note 4) The announcement generating device according to any one of Supplementary Notes 1 to 3, wherein the announcement script generating means generates the announcement script according to the business type information.

[0177] (Supplementary Note 5) The announcement generating device according to any one of Supplementary Notes 1 to 4, wherein the acquiring means further acquires sales information of the product, and the announcement script generating means generates the announcement script based on the sales information.

[0178] (Supplementary Note 6) The announcement generating device according to Supplementary Note 5, wherein the announcement script generating means generates the announcement script that emphasizes the listed product whose sales are lower than planned.

[0179] (Supplementary Note 7) The announcement generating device according to any one of Supplementary Notes 1 to 6, wherein the acquisition means further acquires customer demographic information of customers present in the store, and the announcement script generation means generates the announcement script based on the customer demographic information.

[0180] (Supplementary Note 8) The announcement generating device according to Supplementary Note 7, wherein the announcement script generating means generates the announcement script including information on the listed product targeted for sale to the attribute with a large number of customers.

[0181] (Supplementary Note 9) The announcement generating device according to any one of Supplementary Notes 1 to 8, wherein the acquisition means further acquires weather information about the area around the store when the announcement is made, and the announcement script generation means generates the announcement script based on the weather information.

[0182] (Supplementary Note 10) The announcement generation device according to Supplementary Note 9, wherein the announcement script generation means generates the announcement script including a sentence about bad weather if the weather information indicates that the weather around the store at the time the announcement is to be made is bad weather.

[0183] (Supplementary Note 11) The announcement generation device according to any one of Supplementary Notes 1 to 10, wherein the acquisition means acquires information relating to a time period during which the announcement will be made, and the announcement script generation means generates the announcement script based on the time period during which the announcement will be made.

[0184] (Supplementary Note 12) The announcement generating device according to Supplementary Note 11, wherein the announcement script generating means generates the announcement script including an expression indicating the time period.

[0185] (Supplementary Note 13) The announcement generating device according to any one of Supplementary Notes 1 to 12, further comprising: a voice data generating unit that generates voice data of the announcement manuscript being read aloud; and the output unit that outputs the voice data.

[0186] (Supplementary Note 14) The announcement generating device according to Supplementary Note 13, wherein the voice data generating means generates the voice data of the announcement script read out aloud according to a type of voice corresponding to the business type information.

[0187] (Appendix 15) The announcement generating device according to appendix 13 or 14, wherein the acquiring means further acquires sales information of the listed products, and the voice data generating means generates the voice data of the announcement script read out using a voice type corresponding to the sales information.

[0188] (Appendix 16) The announcement generating device described in any one of Appendices 13 to 15, wherein the acquisition means further acquires customer demographic information of customers present in the store, and the voice data generation means generates the voice data of the announcement script read out using a voice type corresponding to the customer demographic information.

[0189] (Supplementary Note 17) The announcement generating device according to any one of Supplementary Notes 13 to 16, wherein the acquiring means acquires information regarding the time period during which the announcement will be made, and the voice data generating means generates the voice data of the announcement script read out using a voice type corresponding to the time period during which the announcement will be made.

[0190] (Supplementary Note 18) The announcement generating device according to any one of Supplementary Notes 13 to 17, wherein the voice data generating means generates the voice data of the announcement script read out using a voice type corresponding to the listed product.

[0191] (Supplementary Note 19) An announcement generation method comprising: acquiring an announcement template for recommending products sold in a store, business type information of the store, and flyer information of the store; generating an announcement script for recommending products featured in the flyer using a language model based on the announcement template and the business type information; and outputting the announcement script.

[0192] (Appendix 20) A recording medium storing a program that causes a computer to execute the following processes: acquiring an announcement template for recommending products sold in a store, business type information of the store, and flyer information of the store; generating an announcement script for recommending products featured in the flyer using a language model based on the announcement template and the business type information; and outputting the announcement script.

[0193] (Appendix 21) An announcement creation program that causes a computer to perform the following processes: acquire an announcement template for recommending products sold in a store, business type information of the store, and flyer information of the store; generate an announcement script for recommending products listed in the flyer using a language model based on the announcement template and the business type information; and output the announcement script.

[0194] Some or all of the configurations described in Supplementary Notes 2-18, which are dependent on Supplementary Note 1, may also be dependent on Supplementary Notes 19-21 in the same dependency relationship as Supplementary Note 2-18. Not limited to Supplementary Notes 1, 19-21, but also to various hardware, software, various recording devices for recording software, or systems, some or all of the configurations described as Supplements may be made dependent on each other within the scope of the above-mentioned embodiments.

[0195] 10, 20, 30 Announcement generating device 101, 201 Acquisition unit 102, 202 Announcement script generating unit 103, 203 Output unit 104, 204 Prompt creating unit 205 Voice data generating unit 301 Processor 302 ROM 303 RAM 304 Storage device 305 Input / output interface 306 Communication interface 307 Bus 80 Generation server 801 Generation unit 90 Voice output device 91 Store terminal 92 Database

Claims

1. An announcement generation device comprising: an acquisition means for acquiring an announcement template for recommending products handled in a store, the business type information of the store, and the flyer information of the store; an announcement manuscript generation means for generating an announcement manuscript for recommending products to be listed in the flyer using a language model based on the announcement template and the business type information; and an output means for outputting the announcement manuscript.

2. The announcement generation device according to claim 1, further comprising a prompt creation means for creating a prompt including the announcement template, the business type information, and the flyer information, wherein the acquisition means acquires the announcement template, the business type information, and the flyer information input on a screen capable of inputting the announcement template, the business type information, and the flyer information.

3. The announcement generation device according to claim 1, wherein the acquisition means acquires a prompt including the announcement template, the business type information, and the flyer information, and the announcement manuscript generation means generates the announcement manuscript based on the prompt.

4. The announcement generation device according to any one of claims 1 to 3, wherein the announcement manuscript generation means generates the announcement manuscript according to the business type information.

5. The announcement generation device according to any one of claims 1 to 4, wherein the acquisition means further acquires sales information of the product, and the announcement manuscript generation means generates the announcement manuscript based on the sales information.

6. The announcement generation device according to claim 5, wherein the announcement manuscript generation means generates the announcement manuscript that emphasizes the listed products whose sales are lower than the plan.

7. The announcement generation device according to any one of claims 1 to 6, wherein the acquisition means further acquires customer layer information of customers in the store, and the announcement manuscript generation means generates the announcement manuscript based on the customer layer information.

8. The announcement generation device according to claim 7, wherein the announcement manuscript generation means generates the announcement manuscript including information on the listed products targeted for sale of attributes with a large number of customers.

9. The acquisition means further acquires weather information around the store when the announcement is made, and the announcement manuscript generation means generates the announcement manuscript based on the weather information. The announcement generation device according to any one of claims 1 to 8.

10. When the weather around the store at the time of making the announcement is bad weather in the weather information, the announcement manuscript generation means generates the announcement manuscript including a sentence regarding bad weather. The announcement generation device according to claim 9.

11. The acquisition means acquires information regarding the time zone when the announcement is made, and the announcement manuscript generation means generates the announcement manuscript based on the time zone when the announcement is made. The announcement generation device according to any one of claims 1 to 10.

12. The announcement manuscript generation means generates the announcement manuscript including an expression representing the time zone. The announcement generation device according to claim 11.

13. The announcement generation device according to any one of claims 1 to 12 further includes voice data generation means for generating voice data for reading out the announcement manuscript, and the output means outputs the voice data.

14. The voice data generation means generates the voice data for reading out the announcement manuscript according to the type of voice corresponding to the business type information. The announcement generation device according to claim 13.

15. The acquisition means further acquires sales information of the advertised product, and the voice data generation means generates the voice data for reading out the announcement manuscript according to the type of voice corresponding to the sales information. The announcement generation device according to claim 13 or 14.

16. The acquisition means further acquires customer layer information of customers in the store, and the voice data generation means generates the voice data for reading out the announcement manuscript according to the type of voice corresponding to the customer layer information. The announcement generation device according to any one of claims 13 to 15.

17. The acquisition means acquires information regarding the time zone when the announcement is made, and the voice data generation means generates the voice data for reading out the announcement manuscript according to the type of voice corresponding to the time zone when the announcement is made. The announcement generation device according to any one of claims 13 to 16.

18. The voice data generation means generates the voice data obtained by reading out the announcement manuscript according to the type of voice corresponding to the listed product. The announcement generation device according to any one of claims 13 to 17.

19. An announcement generation method that acquires an announcement template for recommending products handled in a store, the business format information of the store, and the flyer information of the store, generates an announcement manuscript for recommending the listed products of the flyer using a language model based on the announcement template and the business format information, and outputs the announcement manuscript.

20. A recording medium storing a program that causes a computer to execute a process of acquiring an announcement template for recommending products handled in a store, the business format information of the store, and the flyer information of the store, generating an announcement manuscript for recommending the listed products of the flyer using a language model based on the announcement template and the business format information, and outputting the announcement manuscript.

Images