Recipe recommendation system, recipe recommendation method and program

Abstract

This dish recommendation system includes: an acquiring unit that acquires, in advance, a plurality of items of event data obtained by classifying, by dish type information, identification information of dishes eaten by a user, the dish type information including taste sensation information for classifying the dishes by taste sensation, and meal time information; an extracting unit that performs time-series association analysis to calculate a plurality of characteristic indexes including at least one of a degree of support, a confidence level, and a lift value relating to the user's eating habits, by dish type information, on the basis of the plurality of items of event data, and extracts a specific characteristic index obtained by analyzing characteristics of the user's eating habits, from among the plurality of characteristic indexes, on the basis of the specific event data corresponding to specific dish type information input by the user; and a recommendation unit for outputting a dish candidate to be proposed to the user on the basis of the specific characteristic index.

Description

[Technical Field]

[0001] The present disclosure relates to a dish recommendation system, a dish recommendation method, and a program. [Background technology]

[0002] Conventionally, there have been proposed dish recommendation systems that suggest dishes based on a user's meal history. For example, Patent Literature 1 discloses a dish suggestion device that suggests the next dish to be eaten using eating patterns, preferences, meal amounts, etc. based on the user's past meal history. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-124701 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the device of Patent Document 1 automatically suggests dishes without confirming the user's desires, so there is a risk that dishes different from the ones the user wants to eat may be suggested.

[0005] The present disclosure aims to provide a dish recommendation system, a dish recommendation method, and a program that suggest appropriate dishes to a user. [Means for solving the problem]

[0006] The dish recommendation system according to the present disclosure includes an acquisition unit that acquires in advance multiple event data sorted by dish type information, including identification information of dishes eaten by a user, dish type information including taste sensory information that classifies dishes by taste, and mealtime information; an extraction unit that calculates, using time series association analysis, multiple characteristic indicators including at least one of support, confidence, and lift value related to the user's eating habits for each dish type information based on the multiple event data, and extracts a specific characteristic indicator that analyzes the characteristics of the user's eating habits from the multiple characteristic indicators based on specific event data corresponding to specific dish type information input by the user; and a recommendation unit that outputs dish candidates to be suggested to the user based on the specific characteristic indicator.

[0007] The dish recommendation method according to the present disclosure includes: The computer A plurality of event data items are acquired in advance, each of which includes identification information of the food eaten by the user, food type information including taste sensation information that classifies the food by taste sensation, and meal time information, and is classified for each food type information item. death By using time-series association analysis, a plurality of characteristic indices including at least one of support, confidence, and lift value related to the user's eating habits are calculated for each piece of cuisine type information based on a plurality of event data, and a specific characteristic indices that analyzes the characteristics of the user's eating habits are extracted from the plurality of characteristic indices based on specific event data corresponding to the specific cuisine type information input by the user. death , and outputs recipe suggestions to users based on specific characteristic indices. Also That is why.

[0008] The program disclosed herein causes a computer to perform the following steps: acquire in advance multiple event data sorted by cuisine type information, including identification information of the cuisine eaten by a user, cuisine type information including taste sensory information that classifies cuisines by taste, and mealtime information; calculate, using time series association analysis, multiple characteristic indices including at least one of support, confidence, and lift value related to the user's eating habits for each cuisine type information based on the multiple event data; extract a specific characteristic index that analyzes the characteristics of the user's eating habits from the multiple characteristic indices based on specific event data corresponding to specific cuisine type information input by the user; and output candidate cuisines to be suggested to the user based on the specific characteristic index. [Effects of the Invention]

[0009] According to the present disclosure, it is possible to suggest appropriate dishes to the user. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing an overview of one embodiment of the present disclosure. [Figure 2] FIG. 2 is a block diagram showing the functional configuration of a dish recommendation system according to an embodiment of the present disclosure. [Figure 3] FIG. 3 is a block diagram showing a hardware configuration of a dish recommendation system according to an embodiment of the present disclosure. [Figure 4] FIG. 4 is a flowchart illustrating the operation of one embodiment of the present disclosure. [Figure 5] FIG. 5 is a diagram showing dish type information determined based on ingredient information. [Figure 6] FIG. 6 is a diagram showing event data. [Figure 7] FIG. 7 shows analysis data obtained by time-series association analysis of user event data. [Figure 8] FIG. 8 is a diagram showing analytical data in which the characteristic indexes have been normalized. [Figure 9]FIG. 9 is a diagram showing how time-series association analysis is performed on event data of other users. [Figure 10] FIG. 10 is a diagram showing comparison data comparing the characteristic indexes of a user with other users. [Figure 11] FIG. 11 is a diagram showing proposal data including recipe candidates output from the recommendation unit. [Figure 12] FIG. 12 is a diagram showing the eating habit data output from the recommendation unit. DETAILED DESCRIPTION OF THE INVENTION

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

[0012] An overview of the present disclosure will be described. For example, as shown in FIG. 1, when a user's terminal 1 receives meal data (e.g., images of dishes) of the user's meals, the terminal 1 sequentially transmits the meal data to a server 2. Here, the server 2 is assumed to have a dish recommendation system of the present disclosure. When the dish recommendation system of the server 2 receives the meal data from the terminal 1, the system acquires, based on the meal data, identification information (e.g., name) of the dish eaten by the user (e.g., dish type information (e.g., light), and meal time information (e.g., date) of the dish eaten by the user, and pre-creates multiple pieces of event data by classifying this information by dish type information. The dish recommendation system then analyzes the characteristics of the user's eating habits based on the event data. However, analyzing a user's eating habits based on event data containing only dish identification information and meal time information results in analyzing eating habits that include various types of dishes, making it difficult to analyze eating habits based on dish type, such as eating habits for "light" dishes. Therefore, when a specific dish type information (e.g., light) that the user wants to eat is input to the terminal 1, the terminal 1 transmits the specific dish type information to the server 2. When the dish recommendation system of the server 2 receives specific dish type information from the terminal 1, it extracts characteristic indicators that analyze the characteristics of the user's eating habits based on the specific event data corresponding to the specific dish type information. The dish recommendation system then outputs dish candidates to be suggested to the user based on the extracted characteristic indicators and transmits the dish candidates to the terminal 1. As a result, the terminal 1 displays the dish candidates corresponding to the specific dish type information on the display unit.

[0013] Next, the configuration of the dish recommendation system according to the present disclosure will be described in detail. As shown in Fig. 2, the dish recommendation system 3 includes an acquisition unit 4, a storage unit 5, an extraction unit 6, and a recommendation unit 7. The acquisition unit 4 is connected to the storage unit 5, which is connected to the recommendation unit 7 via the extraction unit 6.

[0014] The acquisition unit 4 acquires multiple event data, each of which includes identification information of a dish eaten by a user, dish type information, and mealtime information, categorized by dish type information. Here, the dish may include a single dish (e.g., hamburger steak), a menu indicating multiple dishes (e.g., rice and hamburger steak), etc. Examples of the dish identification information include the name and identification number of the dish. The dish type information is information for classifying dishes by type, and may include taste sensation information for classifying dishes by taste sensation. Examples of the taste sensation information include texture information indicating the physical sensation of taste, and taste sensation information indicating the chemical sensation of taste. Specifically, the taste sensation information may include information indicating onomatopoeia that expresses the sensation of taste, i.e., information indicating onomatopoeic words or mimetic words. Examples of onomatopoeia include light, rich, firm, crunchy, fluffy, crispy, crunchy, and creamy. Here, the information indicating the onomatopoeia may be composed of, for example, at least a part of the character string of the onomatopoeia. Furthermore, the mealtime information is information related to the time of the meal, such as the time of the meal, date, time period (e.g., breakfast, lunch, dinner), etc.

[0015] For example, the acquisition unit 4 may receive meal data entered by the user from the terminal 1. Here, the meal data is data related to the dishes eaten by the user, and may include, for example, images (e.g., photos) of the dishes, menus, a food questionnaire (e.g., food preference information, avoided ingredient information, user attribute information, etc.), and dish type information. The food preference information is information related to the dishes preferred by the user, and may include, for example, the names of dishes or ingredients preferred by the user. The avoided ingredient information is information related to ingredients that the user avoids using in cooking, and may include, for example, ingredients the user dislikes or ingredients that cause allergic reactions. The user attribute information may include gender, whether the user is on a diet, whether the user has a chronic illness (e.g., high blood pressure), etc. Upon receiving the meal data, the acquisition unit 4 acquires identification information of the dishes eaten by the user, dish type information, and mealtime information based on the meal data, and creates multiple event data by classifying this information by dish type information.

[0016] The storage unit 5 stores the plurality of event data acquired by the acquisition unit 4. That is, the storage unit 5 stores the plurality of event data sorted by cuisine type information, with the identification information of the cuisine eaten by the user associated with mealtime information. The storage unit 5 also outputs, to the extraction unit 6, specific event data from the plurality of event data that corresponds to specific cuisine type information input by the user. For example, when the acquisition unit 4 receives from the terminal 1 specific cuisine type information input by the user to the terminal 1, the storage unit 5 may output the specific event data corresponding to the specific cuisine type information to the extraction unit 6. For example, when the user inputs specific taste sensory information "refreshing" as the specific cuisine type information, the storage unit 5 may output the event data corresponding to "refreshing" to the extraction unit 6.

[0017] The extraction unit 6 calculates, by time-series association analysis, multiple characteristic indices, including at least one of support, confidence, and lift value, related to the user's eating habits for each of the cuisine type information based on the multiple event data. The extraction unit 6 then extracts, from the multiple characteristic indices, a characteristic index that analyzes the characteristics of the user's eating habits based on specific event data input from the storage unit 5. Here, the characteristic of the eating habits indicates the tendency (e.g., rules or conditions) of the user's food selection over multiple meals, and may indicate, for example, the tendency of the user to select dishes when eating "light" dishes three times. Furthermore, the characteristic index indicates the characteristics of the eating habits, and may be calculated, for example, by comparing event data from multiple meals.

[0018] The recommendation unit 7 outputs dish candidates to be suggested to the user based on the characteristic indexes acquired by the extraction unit 6. In this case, the recommendation unit 7 may output dish candidates based on event data that meets at least one of the following conditions: a support level equal to or greater than a predetermined threshold, a confidence level equal to or greater than a predetermined threshold, and a lift value equal to or greater than a predetermined threshold. The recommendation unit 7 may also analyze the characteristics of other users' eating habits for each type of cuisine information based on multiple event data that includes identification information of dishes eaten by other users, cuisine type information, and mealtime information, sorted by cuisine type information. The recommendation unit 7 may then search for dish candidates to be suggested to the user based on the user's characteristic indexes calculated by the extraction unit 6 and the characteristic indexes of other users corresponding to specific cuisine type information.

[0019] FIG. 3 shows the hardware configuration of the dish recommendation system 3.

[0020] The dish recommendation system 3 includes a storage device 11, a processor 12, a user interface (UI) device 13, and a communication device 14, which are interconnected via a bus B.

[0021] In addition, the programs or instructions that realize the various functions and processes described below in the dish recommendation system 3 may be downloaded from any external device via a network, or may be provided from a removable storage medium such as a CD-ROM (Compact Disk-Read Only Memory) or flash memory.

[0022] Storage device 11 may be implemented by one or more non-transitory storage media, such as random access memory, flash memory, or a hard disk drive, and stores installed programs or instructions as well as files, data, etc. used in the execution of the programs or instructions.

[0023] The processor 12 may be realized by one or more central processing units (CPUs), which may be configured with one or more processor cores, graphics processing units (GPUs), processing circuitry, etc. The processor 12 executes various functions and processes of the dish recommendation system 3, which will be described later, in accordance with programs, instructions, data such as parameters required to execute the programs or instructions, etc., stored in the storage device 11.

[0024] The UI device 13 may be composed of input devices such as a keyboard, mouse, camera, microphone, etc., output devices such as a display, speaker, headset, printer, etc., and input / output devices such as a smartphone, tablet, touch panel, etc., and realizes an interface between the administrator and the dish recommendation system 3. For example, the administrator may operate the dish recommendation system 3 by operating a GUI (Graphical User Interface) displayed on a display or touch panel using a keyboard, mouse, etc.

[0025] It should be noted that the above-described hardware configuration is merely an example, and the dish recommendation system 3 according to the present disclosure may be realized by any other appropriate hardware configuration.

[0026] Next, the operation of this embodiment will be described with reference to the flowchart shown in FIG.

[0027] First, in step S1, the acquisition unit 4 of the dish recommendation system 3 shown in FIG. 2 acquires event data including identification information of the dish eaten by the user, dish type information, and meal time information. For example, the acquisition unit 4 may acquire the event data based on the user's meal data. For example, when the user inputs meal data, such as photographed images of the eaten dish, into terminal 1, terminal 1 sequentially transmits the meal data to server 2. Then, when the acquisition unit 4 of the dish recommendation system 3 in server 2 receives the user's meal data from terminal 1, it acquires identification information of the dish eaten by the user, dish type information, and meal time information based on the meal data. At this time, the acquisition unit 4 may further acquire ingredient information of the dish eaten by the user based on the meal data. Note that the ingredient information is information about ingredients used in the dish and may include information such as toppings (e.g., radish, carrot, etc.), seasonings (e.g., soy sauce, miso, etc.), and the amount of ingredients.

[0028] For example, the acquisition unit 4 may recognize the name (identification information) and ingredient information of the dish eaten by the user based on the photographed image of the dish received as meal data, and may estimate taste sensory information (dish type information) that classifies the dish by taste based on the ingredient information. For example, as shown in FIG. 5, even if the same dish is recognized as simmered chicken 1 to simmered chicken 5, the acquisition unit 4 may recognize the difference in the ingredient information used in the simmered chicken 1 to 5 and estimate the taste sensory information for each dish based on the ingredient information. For example, if the acquisition unit 4 recognizes the ingredient information of simmered chicken 1 as "yam" based on the photographed image of simmered chicken 1, it may estimate the taste sensory information as "fluffy" based on the texture of the yam. Note that although "daikon" is indicated as the ingredient information for both simmered chicken 4 and simmered chicken 5, they may be classified as "refreshing" and "mellow" based on other ingredients, respectively. In other words, the taste sensory information may be estimated comprehensively from the information on multiple ingredients used in the dishes. The acquisition unit 4 may also acquire mealtime information based on the image capture time attached to the meal data received from the terminal 1, for example.

[0029] In the above embodiment, the acquisition unit 4 acquires the identification information, dish type information, meal time information, and ingredient information of the dish eaten by the user based on the user's meal data, but the acquisition unit 4 may acquire such information as long as it can acquire such information. For example, the acquisition unit 4 may acquire the identification information, dish type information, meal time information, and ingredient information of the dish eaten by the user directly from the terminal 1. That is, the terminal 1 may accept input of the identification information, dish type information, meal time information, and ingredient information of the dish eaten by the user from the user, and transmit the dish identification information, dish type information, meal time information, and ingredient information input by the user to the server 2.

[0030] Next, in step S2, the acquisition unit 4 creates a plurality of event data by classifying the acquired dish identification information, dish type information, and mealtime information for each dish type information, and stores the event data in the storage unit 5. At this time, the acquisition unit 4 may store a plurality of event data in the storage unit 5, in which at least one of the dish and ingredient information eaten by the user is associated with the dish type information. For example, as shown in FIG. 6, the acquisition unit 4 may create a plurality of event data 15a, 15b, etc., by classifying the dish name (identification information), taste sensory information (dish type information), and mealtime information for each taste sensory information such as "light," "heavy," etc. Specifically, when the acquisition unit 4 acquires data for April 1, it records the data in event data 15a based on the taste sensory information "light," and when the acquisition unit 4 acquires data for April 2, it records the data in event data 15b based on the taste sensory information "heavy." The acquisition unit 4 may also create multiple pieces of event data 15a, 15b... that associate the names of dishes eaten by the user with taste sensory information and ingredient information for each date of mealtime information, and store the multiple pieces of event data 15a, 15b... in the storage unit 5. In this way, the multiple pieces of event data 15a, 15b... are recorded so that data of the same taste sensory information is arranged in order of mealtime. The acquisition unit 4 updates the event data 15a, 15b... each time it receives the user's meal data, and creates multiple pieces of event data 15a, 15b... related to dishes eaten by the user in the past.

[0031] Meanwhile, terminal 1 may receive from the user a request for a proposal of a dish that the user is about to make. At this time, terminal 1 may also receive input of specific dish type information. For example, terminal 1 may receive specific taste sensation information that classifies dishes by taste sensation. At this time, terminal 1 may receive information indicating onomatopoeia that expresses the taste sensation as taste sensation information. When specific dish type information is input by the user, terminal 1 transmits the specific dish type information to server 2. Here, it is assumed that the user has input the onomatopoeia "refreshing" to terminal 1 as the specific taste sensation information.

[0032] In this way, the cuisine type information includes taste sensory information that classifies cuisines by taste sensation. For example, if cuisines are classified based on a food culture such as "French cuisine," there are various types of cuisines (for example, "light" cuisine and "rich" cuisine), making it difficult for the user to properly indicate the cuisine they desire. Therefore, by specifying the type of cuisine using taste sensory information, the user can properly specify the range of cuisines they desire. Furthermore, the taste sensory information includes information indicating onomatopoeia that express the sensation of taste. This allows the user to easily specify the taste sensory information.

[0033] When the dish recommendation system 3 of the server 2 receives the specific taste sensory information "refreshing" input by the user from the terminal 1, the extraction unit 6 selects specific event data 15a corresponding to the specific taste sensory information "refreshing" from the plurality of event data 15a, 15b, etc. stored in the storage unit 5 in step S3. At this time, the extraction unit 6 may select data from a predetermined period (e.g., 14 days) among the specific event data 15a. Then, in step S4, the extraction unit 6 may calculate a characteristic index indicating the user's eating habits by performing time-series association analysis on the extracted specific event data 15a. Examples of the characteristic index include a support factor, a confidence factor, and a lift value for a combination of events. For example, the extraction unit 6 may calculate a characteristic index including at least one of the support factor, the confidence factor, and the lift value. Note that the extraction unit 6 may automatically calculate a characteristic index for each dish type information based on the plurality of event data 15a, 15b, etc. stored in the storage unit 5 without receiving specific taste sensory information from the user.

[0034] Specifically, as shown in FIG. 6, the extraction unit 6 recognizes the names of dishes, taste sensory information, and ingredient information for each column (e.g., each day) of specific event data 15a for 14 days (a predetermined period) as event data 15a1, event data 15a2, and so on, and may calculate characteristic indexes based on combinations of two event data included in a preset analysis period (e.g., three days). For example, assuming April 10th as the reference date, the extraction unit 6 sets the analysis period to April 10th, April 8th (the day before), and April 5th (the day before), and calculates characteristic indexes for combinations of two event data included in the analysis period. For example, as shown in FIG. 7, the extraction unit 6 calculates characteristic indexes 17 for all combinations of event data 15z1 corresponding to event A before the reference date (an event on April 8 or April 5) and event data 15z2 corresponding to event B on the reference date (an event on April 10). Next, the extraction unit 6 shifts the reference date by one to April 8, sets April 8, April 5 (the date before), and April 3 (the date before), as the next analysis period, and similarly calculates the characteristic index 17 for each combination of two event data 15z1 and 15z2 included in the analysis period. In this way, the extraction unit 6 shifts the analysis period sequentially within 14 days (a predetermined period) and calculates the characteristic index 17 based on the combination of two event data 15z1 and 15z2 included in each analysis period. Then, the extraction unit 6 generates analysis data 16 for 14 days (a predetermined period) that includes event data 15z1 of event A, event data 15z2 of event B, and characteristic index 17.

[0035] Here, the support indicates, for example, the frequency of combinations of event data 15z1 of event A and event data 15z2 of event B, and may be calculated using the following formula (1). The confidence indicates, for example, the probability of eating a dish from event A one or two days before eating a dish from event B, and may be calculated using the following formula (2). The lift value indicates, for example, the degree of increase in the probability of eating a dish from event A one or two days before eating a dish from event B, and may be calculated based on the following formula (3). The total number of data items may be calculated, for example, by multiplying the number of pieces of dish identification information by the number of pieces of dish type information by the number of pieces of ingredient information by the number of analysis periods (e.g., 3 days).

[0036] Support = Number of data showing the same combination of event data 15z1 of event A and event data 15z2 of event B ÷ Total number of data (1) Confidence = Number of data showing the same combination of event data 15z1 of event A and event data 15z2 of event B ÷ Number of data including event data 15z2 of event B (2) Lift value = number of data showing the same combination of event data 15z1 of event A and event data 15z2 of event B ÷ number of data including event data 15z2 of event B ÷ number of data including event data 15z1 of event A ÷ total number of data (3)

[0037] In this case, the extraction unit 6 may search the analysis data 16a based on the chronological order of the event data 15a1, 15a2, ... in the specific event data 15a, and may filter out data in an order that is not included in the specific event data 15a. For example, if event data 15z2 (corresponding to event B) is recorded next to event data 15z1 (corresponding to event A) in the specific event data 15a, the extraction unit 6 may filter out data in the analysis data 16a in which event A is event data 15z2 and event B is event data 15z1 (data in the reverse order from the specific event data 15a). This allows the specific event data 15a to be subjected to association analysis in chronological order.

[0038] In this way, the extraction unit 6 extracts characteristic indicators 17 that analyze the characteristics of the user's eating habits based on the specific event data 15a that corresponds to the specific cuisine type information "refreshing" entered by the user. At this time, since the information on the dishes eaten by the user in the multiple event data 15a, 15b, etc. is sorted by cuisine type information, the extraction unit 6 can easily extract the characteristics of the user's eating habits simply by analyzing the specific event data 15a that corresponds to the specific cuisine type information.

[0039] Furthermore, the extraction unit 6 analyzes the characteristics of the eating habits of the user P by performing time-series association analysis on the specific event data 15a, and outputs characteristic indices 17 such as support, confidence, and lift value. This allows the extraction unit 6 to accurately indicate the eating habits of the user P with the characteristic indices 17.

[0040] Furthermore, the plurality of event data 15a, 15b, etc. includes information on ingredients of the dishes eaten by user P. This allows the extraction unit 6 to more accurately analyze the characteristics of user P's eating habits based on the differences in the ingredient information.

[0041] Furthermore, the plurality of event data 15a, 15b, etc. are configured by associating at least one of the food and ingredient information eaten by the user P with the food type information, thereby enabling the extraction unit 6 to easily analyze the characteristics of the eating habits of the user P.

[0042] The dish type information also includes taste sensory information that classifies dishes by taste sensation. The extraction unit 6 performs analysis based on taste sensory information that significantly influences the user's eating habits, so it can accurately analyze the characteristics of user P's eating habits. The taste sensory information also includes information indicating onomatopoeia that express the taste sensation. The extraction unit 6 performs analysis based on onomatopoeia that accurately expresses taste sensory information, so it can more accurately analyze the characteristics of user P's eating habits.

[0043] The extraction unit 6 may further extract the characteristic indicators 17 using at least one of the user P's preference information and the avoided ingredient information. For example, the extraction unit 6 may search the analysis data 16 based on the user P's preference information and preferentially extract event data including characteristic indicators 17 in which dishes or ingredients corresponding to the preference information are registered. The extraction unit 6 may also search the analysis data 16 based on the user P's avoided ingredient information and extract event data including characteristic indicators 17 in which avoided ingredient information is not registered. In other words, the extraction unit 6 may delete event data in which avoided ingredient information is registered from the analysis data 16. The preference information and avoided ingredient information may be calculated based on the user P's dietary data or may be set by the user P. This allows the extraction unit 6 to appropriately extract the characteristic indicators 17 in accordance with the user P's preference information or avoided ingredient information.

[0044] The extraction unit 6 may also extract the characteristic index 17 by further using attribute information of the user P. For example, if the attribute information of the user P is set as being on a diet, the extraction unit 6 may extract event data of dishes with calories equal to or less than a predetermined value. Note that the attribute information may be set by the user P.

[0045] The extraction unit 6 may also extract the characteristic index 17 by further using the recipe browsing history. For example, when user P searches the web for a cooking recipe, terminal 1 may transmit the recipe browsing information to server 2. The extraction unit 6 may calculate preference information of user P based on the recipe browsing history transmitted from terminal 1. Then, the extraction unit 6 may preferentially extract event data including characteristic indexes 17 in which dishes or ingredients corresponding to the calculated preference information of user P are registered.

[0046] The extraction unit 6 may also normalize the characteristic indexes. For example, as shown in FIG. 8, the extraction unit 6 may create analysis data 16a in which the support, confidence, and lift values ​​of the characteristic indexes 17 are normalized and displayed as variable values. The variable names indicate combinations of event data 15z1 of event A and event data 15z2 of event B. Here, the variable names X, Y, and Z indicate the same combinations of event data 15z1 of event A and event data 15z2 of event B, separated into support (X1, X2...), confidence (Y1,...), and lift (Z1,...).

[0047] Next, the extraction unit 6 outputs the analysis data 16 including the extracted characteristic indicators 17 to the recommendation unit 7. Here, it is assumed that the extraction unit 6 has output the analysis data 16a to the recommendation unit 7. When the recommendation unit 7 receives the analysis data 16a, it compares the analysis data obtained by performing a time-series association analysis on the characteristics of the eating habits of other users with the user's analysis data 16a in step S5. Here, the recommendation unit 7 may analyze the characteristics of the eating habits of other users for each piece of cuisine type information, just like the user, based on a plurality of event data obtained by classifying identification information of dishes eaten by other users, cuisine type information, and mealtime information for each piece of cuisine type information, and may store the plurality of analysis data in advance in the storage unit.

[0048] For example, as shown in FIG. 9, the recommendation unit 7 acquires event data 18a, 18b, etc. of multiple other users P1, P2, etc. in advance. At this time, the recommendation unit 7 may create multiple event data for each of the other users P1, P2, etc., by classifying the dish identification information, dish type information, and mealtime information for each dish type information. For example, in the case of the other user P1, the recommendation unit 7 may create multiple event data 18a1, 18a2, etc., by classifying the dish identification information, taste sensation information, and mealtime information eaten by the other user P1 for each taste sensation information such as "light," "heavy," etc. Then, the recommendation unit 7 analyzes the characteristics of the eating habits of the other users for each taste sensation information such as "light," "heavy," etc., based on the respective event data 18a1, 18a2, etc. At this time, the recommendation unit 7 may calculate characteristic indicators 19 indicating the eating habits of other users by performing time-series association analysis on each of the event data 18a1, 18a2, etc. As a result, the recommendation unit 7 generates analysis data 21a1, 21a2, etc., each including event data 20z1 of event A, event data 20z2 of event B, and the characteristic indicator 19. The analysis data 21a1, 21a2, etc., are generated corresponding to the taste sensory information "refreshing," "rich," etc. Similarly, the recommendation unit 7 calculates characteristic indicators 19 indicating the eating habits of other users P2, P3, etc., and generates analysis data 21b, 21c, etc., each including event data 20z1 of event A, event data 20z2 of event B, and the characteristic indicator 19. The recommendation unit 7 stores analysis data 21a, 21b, etc. of other users P1, P2, etc. in advance in the storage unit.

[0049] If the variable names do not match among the other users P1, P2, ... (there is a missing variable name for a specific user), the recommendation unit 7 may complete the missing variable name for the specific user based on the average value of the variable values ​​of the other users P1, P2, ... for which the variable name exists. The recommendation unit 7 may also acquire the preference information, avoided ingredient information, attribute information, or recipe browsing history of user P from the extraction unit 6 and extract the characteristic index 19 using this information in the same way as the extraction unit 6. For example, between steps S5 and S6, the recommendation unit 7 may filter the analysis data 21a, 21b, ... based on the preference information, avoided ingredient information, attribute information, or recipe browsing history of user P.

[0050] When the analytical data 16a of a user indicating the taste sensory information "refreshing" is input from the extraction unit 6, the recommendation unit 7 selects analytical data 21a1, 21b1, ... corresponding to the taste sensory information "refreshing" from the analytical data 21a, 21b, ... of other users P1, P2, ... stored in the storage unit. Then, the recommendation unit 7 compares the analytical data 21a1, 21b1, ... corresponding to the taste sensory information "refreshing" with the user's analytical data 16a.

[0051] At this time, the recommendation unit 7 may search for dish candidates to suggest to the user from analysis data 21a1, 21b1, ... including eating habits corresponding to the specific taste sensory information "refreshing" of other users, based on a characteristic index 17 indicating eating habits corresponding to the specific taste sensory information "refreshing" input by the user. For example, as shown in FIG. 10, the recommendation unit 7 may use collaborative filtering to search for dish candidates to suggest to the user P by comparing a variable value 23 (characteristic index) of the user P with variable values ​​24 (characteristic indexes) of the other users P1, P2, ... in the comparison data 22. At this time, the recommendation unit 7 may calculate the similarity between the variable value 23 of the user P and the variable values ​​24 of the other users P1, P2, ... for all variables X1, X2, ..., Y1, ..., Z1, .... As an example, the recommendation unit 7 may calculate a cosine similarity based on the vector of variable values ​​23 of user P and the vectors of variable values ​​24 of other users P1, P2, .... The recommendation unit 7 then selects other users P1 and P2 whose similarity is a predetermined value or more, for example, 0.8 or more, as users whose eating habits are highly similar to user P's. Next, the recommendation unit 7 counts, for each variable X1, X2, ..., the number of variable values ​​24a that are within a predetermined value (e.g., a matching value) of the variable value 23 of user P from among the variable values ​​24 of other users P1 and P2 who have similar eating habits. The recommendation unit 7 then selects variables X2, Y1, and Z1 with the largest number of variable values ​​24a, and in step S6, outputs dish candidates based on the selected variables X2, Y1, and Z1. At this time, the recommendation unit 7 may preferentially select, from the selected variables X2, Y1, and Z1, a variable having an event A that matches the name of the dish that user P ate one or two meals before the reference date.

[0052] The method of searching for dish candidates to be suggested to the user by the recommendation unit 7 is not limited to collaborative filtering. For example, based on the characteristic index 17 indicating the eating habits corresponding to the specific taste sensory information "refreshing" input by the user, dish candidates to be suggested to the user may be searched for using statistical analysis, machine learning, deep learning, or the like from the analysis data 21a1, 21b1, ... including the eating habits of other users corresponding to the taste sensory information "refreshing."

[0053] For example, as shown in FIG. 11, when the dishes of the B event of variables X2, Y1, and Z1 are "boiled chicken," "meat and potatoes," and "pork belly and Chinese cabbage," the recommendation unit 7 may output those dishes as candidate dishes 25a to 25c. Furthermore, the recommendation unit 7 may output the dish of the A event of variables X2, Y1, and Z1, "marinated fish," as reference dishes 26a to 26c. In this case, if the dish eaten by user P two days before the reference date (e.g., two days before) is "marinated fish," the recommendation unit 7 may preferentially select the variables X2, Y1, and Z1 in which the dish of the A event is recorded as "marinated fish." Furthermore, the recommendation unit 7 may output condition-related information 27 related to the conditions under which variables X2, Y1, and Z1 were selected. As a result, proposal data 28 including candidate dishes 25a to 25c, reference dishes 26a to 26c, and condition-related information 27 is created.

[0054] In this way, the recommendation unit 7 outputs the dish candidates 25a to 25c to be suggested to the user P based on the characteristic indicators 17, such as the support level, the confidence level, and the lift value. Here, the characteristic indicators 17 are obtained by analyzing the characteristics of the eating habits of the user P based on the specific event data 15a corresponding to the specific dish type information input by the user P, among the plurality of event data 15a, 15b, etc., which are obtained by the extraction unit 6 sorting the identification information of the dishes eaten by the user P, the dish type information, and the meal time information by dish type information. This enables the recommendation unit 7 to suggest appropriate dishes to the user P.

[0055] The recommendation unit 7 also analyzes the characteristics of the eating habits of the other users P1, P2, ..., and searches for dish candidates 25a-25c based on the characteristic index 17 of user P and the characteristic indexes 19 of the other users P1, P2, .... In this way, the recommendation unit 7 outputs dish candidates 25a-25c based on the eating habits of the other users P1, P2, .... This allows the recommendation unit 7 to output various dish candidates 25a-25c other than those based on user P's eating history, and can prevent the recommendation unit 7 from repeatedly suggesting dishes that user P has eaten before.

[0056] Here, the recommendation unit 7 may output information other than the candidate dishes 25a-25c. For example, the recommendation unit 7 may output eating habit data indicating the relationship over time between event data 15z1 and event data 15z2, including dish identification information and dish type information, based on the analysis data 16a of user P output from the extraction unit 6. The recommendation unit 7 may also output eating habit data indicating the relationship over time between event data 20z1 and event data 20z2, including dish identification information and dish type information, based on the analysis data 21a1 and 21b1 of other users P1 and P2. For example, as shown in FIG. 12, the recommendation unit 7 may output eating habit data 32, based on the analysis data 16a of user P, indicating the relationship over time between event data 15z1 and event data 15z2, including dish names (identification information) and ingredient information, using nodes 30 and edges 31. Here, for example, the nodes 30 may be formed larger as the frequency or support of the event data 15z2 of event B is selected increases. Also, the edges 31 may be formed thicker as the lift value or confidence increases. In this case, the recommendation unit 7 may create the eating habit data 32 by limiting it to a predetermined number of edges 31 with the highest lift values ​​or confidence values ​​(for example, limiting it to the top 26 edges 31). Then, the recommendation unit 7 may select a path 33 that connects the nodes 30 over time in descending order of thickness for a predetermined number of meals (for example, four meals or four days), and display the path 33 by changing the display format.

[0057] In this way, the recommendation unit 7 outputs eating habit data 32 indicating the relationship over time between the event data 15z1 and 15z2 based on the characteristic index 17 indicating the characteristics of the eating habits of the user P. This visualizes the unconscious eating habits of the user P, allowing the user P to easily understand the intention behind the suggestion of the dish candidates 25a to 25c in the proposal data 28.

[0058] Next, the recommendation unit 7 transmits the created proposal data 28 to the terminal 1. At this time, the recommendation unit 7 may also transmit eating habit data 32 to the terminal 1. Upon receiving the proposal data 28 and the eating habit data 32, the terminal 1 displays the proposal data 28 and the eating habit data 32 on the display unit.

[0059] According to this embodiment, the extraction unit 6 acquires in advance a plurality of event data 15a, 15b, etc., which are obtained by classifying identification information of dishes eaten by user P, dish type information, and mealtime information by dish type information. When specific dish type information is input by user P, the extraction unit 6 extracts characteristic index 17 that analyzes the characteristics of user P's eating habits based on specific event data 15a corresponding to the specific dish type information. Then, the recommendation unit 7 outputs dish candidates 25a to 25c to be suggested to user P based on characteristic index 17. This allows the recommendation unit 7 to suggest appropriate dishes to user P.

[0060] In this embodiment, the recommendation unit 7 outputs the dish candidates 25a-25c based on the eating habits of other users P1, P2, ..., but is not limited to this as long as it can output the dish candidates 25a-25c. For example, the recommendation unit 7 may select the dish candidates 25a-25c from the analysis data 16a indicating the eating habits of user P based on the characteristic index 17. For example, the recommendation unit 7 may select the dish candidates 25a-25c in descending order of the value of the characteristic index 17 in the analysis data 16a.

[0061] In addition, in this embodiment, the dish recommendation system 3 is disposed in the server 2, but is not limited to this as long as it can output dish candidates. For example, the dish recommendation system 3 may be disposed in the terminal 1.

[0062] In addition, although the present disclosure has been described using names of Japanese dishes or ingredients in the present embodiment, the present disclosure is not limited thereto. For example, the names of dishes or ingredients may be changed to suit the food culture of the country in which the dish recommendation system, dish recommendation method, and program are used.

[0063] Although specific examples of the present disclosure have been described in detail above, these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and alterations of the specific examples exemplified above.

[0064] This application is based on a Japanese patent application (Patent Application No. 2023-051620) filed on March 28, 2023, the contents of which are incorporated herein by reference. [Industrial Applicability]

[0065] The dish recommendation system according to the present disclosure can be used as a system that suggests dish candidates to a user. [Explanation of symbols]

[0066] 1 device 2 Server 3. Food recommendation system 4 Acquisition part 5 Storage section 6 Extraction part 7 Recommendation Department 11 Storage device 12 processors 13 UI device 14. Communications equipment 15a, 15b Event data 16,16a Analysis Data 17 Characteristic indicators 18a, 18b Event data 21a, 21b Analysis data 22 Comparative Data 23 Variable Values 24 Variable Values 25a~25c Culinary Candidates 26a~26c Reference dishes 27 Condition-related information 28 Proposal Data 30 nodes 31 Edge 32 Eating Habits Data 33 Routes P User P1,P2 Other users

Claims

1. an acquisition unit that acquires in advance a plurality of event data items, each of which includes identification information of a dish eaten by a user, dish type information including taste sensation information that classifies the dish by taste sensation, and meal time information, and which is classified for each of the dish type information items; an extraction unit that calculates, by time-series association analysis, a plurality of characteristic indices including at least one of a support factor, a confidence factor, and a lift value related to the user's eating habits for each of the plurality of pieces of cuisine type information based on the plurality of pieces of event data, and extracts, from the plurality of characteristic indices, a specific characteristic indices that analyzes the characteristics of the user's eating habits based on specific event data corresponding to the specific cuisine type information input by the user; a recommendation unit that outputs dish candidates to be suggested to the user based on the specific characteristic index; A food recommendation system with

2. The food recommendation system according to claim 1 , wherein the taste sensation information includes information indicating onomatopoeia that expresses the sensation of taste.

3. The dish recommendation system according to claim 1 , wherein the plurality of event data includes information on ingredients of dishes eaten by the user.

4. The dish recommendation system according to claim 3 , wherein the plurality of event data are configured by associating at least one of the dish identification information and the ingredient information with the dish type information.

5. The dish recommendation system according to claim 1 , wherein the extraction unit further extracts the specific characteristic index by using at least one of preference information and avoided ingredient information of the user.

6. The dish recommendation system according to claim 1 , wherein the recommendation unit analyzes characteristics of eating habits of other users and searches for the dish candidates based on the specific characteristic index of the user and the characteristic index of the other users.

7. The cooking recommendation system according to claim 1 , wherein the recommendation unit further outputs eating habit data indicating a temporal relationship of the specific event data using nodes and edges based on the specific characteristic index.

8. A computer comprising: Acquire in advance a plurality of event data items, each of which includes identification information of a dish eaten by a user, dish type information including taste sensation information that classifies the dish by taste sensation, and meal time information, and classifies the dish type information items; calculating, based on the plurality of event data, a plurality of characteristic indices including at least one of a support factor, a confidence factor, and a lift value related to the user's eating habits for each of the plurality of cuisine type information items, and extracting, from the plurality of characteristic indices, a specific characteristic indices that analyzes the characteristics of the user's eating habits, based on specific event data corresponding to the specific cuisine type information input by the user; outputting dish candidates to be suggested to the user based on the specific characteristic index; How to recommend dishes.

9. Acquire in advance a plurality of event data items, each of which includes identification information of a dish eaten by a user, dish type information including taste sensation information that classifies the dish by taste sensation, and meal time information, and which are classified for each of the dish type information items; calculating, by time-series association analysis, a plurality of characteristic indices including at least one of support, confidence, and lift value for each of the plurality of pieces of cuisine type information based on the plurality of event data, and extracting a specific characteristic indices that analyzes characteristics of the user's cuisine type from the plurality of characteristic indices based on specific event data corresponding to the specific cuisine type information input by the user; outputting dish candidates to the user based on the specific characteristic indicators; A program that causes a computer to execute the following.

Images