Information processor, program and information processing method

Abstract

To provide an information processor, a program, and an information processing method capable of classifying a specific dish name even when a recipe is unknown.SOLUTION: An information processor (100, 200, 300, 400, 500, 600, 700) includes: a cluster generation part (130) which acquires information indicating a plurality of dish names and information indicating a seasoning included in each of the plurality of dish names, classifies the plurality of dish names according to the information indicating the seasoning, and generates a first number of a clusters according to the classification of the plurality of dish names; a data acquisition part (140) which acquires information indicating a specific dish name; and a related cluster estimation part (160) which estimates a cluster related to the specific dish name among the first number of clusters according to the information extracted from the specific dish name.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to an information processing device, a program, and an information processing method. [Background technology]

[0002] A cooking device that aims to create dishes that suit a user's preferences has been disclosed (see Patent Document 1). This cooking device acquires user preference information and generates recipe information based on the acquired preference information and the values ​​of cooking parameters that define the cooking method and are included in the acquired recipe information. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2015-206585 Summary of the Invention [Problem to be solved by the invention]

[0004] In order to manage information about multiple dishes using an information processing device, it is desirable to classify the multiple dishes by related dishes and seasonings. When classifying multiple dishes, if the recipes of each dish are known, it is possible to classify the target dishes based on, for example, the amounts of ingredients used. However, if the recipes of the dishes are unknown, it is difficult to know the relationships between this dish and other dishes and seasonings, making it difficult to classify these dishes.

[0005] The present disclosure has been made in consideration of the above-mentioned problems, and aims to provide an information processing device, program, and information processing method that can classify a specific dish name even if the recipe is unknown. [Means for solving the problem]

[0006] The information processing device according to the present disclosure is characterized by including a cluster generation unit that acquires information indicating a plurality of dish names and information indicating the seasonings contained in each of the plurality of dish names, classifies the plurality of dish names based on the information indicating the seasonings, and generates a first number of clusters based on the classification of the plurality of dish names; a data acquisition unit that acquires information indicating a specific dish name; and a related cluster estimation unit that estimates a cluster from the first number of clusters that is related to the specific dish name based on information extracted from the specific dish name. [Effects of the Invention]

[0007] According to the present disclosure, even if the recipe is unknown, it is possible to classify the name of a particular dish based on information indicating the seasonings. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram showing an example of a hardware configuration of an information processing device according to a first embodiment. [Figure 2] 1 is a block diagram showing a schematic configuration of an information processing device according to a first embodiment. [Figure 3] 4 is a flowchart showing processing performed by the information processing device according to the first embodiment. [Figure 4] FIG. 10 is a diagram showing an example of data created by a basic cluster constructor. [Figure 5] 5A, 5B, 5C, 5D, and 5E are diagrams showing some of the results of clustering performed by the basic cluster constructor. [Figure 6] FIG. 10 is a diagram showing the processing flow of a seasoning feature extraction unit. [Figure 7] FIG. 10 is a diagram showing an example of calculation of the importance of a seasoning. [Figure 8] FIG. 10 is a diagram showing the processing flow of a recipe feature extraction unit. [Figure 9] 10A and 10B are diagrams showing examples of feature extraction results of recipe words by a recipe feature extraction unit. [Figure 10] FIG. 10 is a flowchart showing the processing flow of a character type feature extraction unit. [Figure 11]6 is a diagram showing an example in which the character type feature extraction unit integrates the ratio calculation results for the basic clusters in FIG. 5 . [Figure 12] FIG. 10 is a diagram showing the processing flow of a dictionary creation unit. [Figure 13] FIG. 10 is a diagram showing a processing flow of a feature generating unit. [Figure 14] 10A and 10B are diagrams showing specific examples of statistics for each cluster calculated by a feature generating unit. [Figure 15] FIG. 15 is a diagram showing an example in which the basic cluster classification unit classifies basic clusters into three groups using the feature amounts shown in FIG. 14 . [Figure 16] FIG. 10 is a diagram showing the processing flow of an associated cluster estimation unit. [Figure 17] FIG. 10 is a diagram showing an example of a calculation result by an associated cluster estimation unit. [Figure 18] FIG. 10 is a diagram showing an example of a result of calculation of a relevance score by a related cluster estimation unit. [Figure 19] 19A is a diagram showing an example of the calculation result of the relevance score by the related cluster estimation unit, and FIGS. 19B and 19C are diagrams showing examples in which the basic cluster classification unit classifies basic clusters into three groups. FIG. [Figure 20] FIG. 10 is a diagram showing a processing flow of a preference estimation unit. [Figure 21] FIG. 10 is a diagram showing an example of visualization of the calculation result of the preference degree by the preference estimation unit. [Figure 22] FIG. 10 is a diagram showing an example of level classification by a preference estimation unit. [Figure 23] 22 is a diagram showing the result of level classification by a preference estimation unit for the preference degrees shown in FIG. 21. [Figure 24] 10A and 10B are diagrams showing the results of calculation of a user's preference degree by a preference estimation unit. [Figure 25] 23 is a diagram showing the results of the preference estimation unit classifying preference levels into five stages defined in FIG. 22. [Figure 26] A diagram showing the definition of the relevance score and preference level for each dish. [Figure 27] FIG. 10 is a block diagram showing a schematic configuration of an information processing device according to a second embodiment. [Figure 28] FIG. 10 is a diagram showing the processing flow of a dish adding unit. [Figure 29] 29A and 29B show examples of the results of estimation by the related cluster estimation unit, in which FIG. 29A shows the result of estimation of the cluster most related to the name of the dish eaten by user A, and FIG. 29B shows the result of estimation of the cluster most related to the name of the dish eaten by user B. [Figure 30] FIG. 10 is a diagram showing an example of a result of processing by a dish adding unit. [Figure 31] FIG. 11 is a block diagram showing a schematic configuration of an information processing device according to a third embodiment. [Figure 32] 10A and 10B are diagrams showing examples of preference estimation results obtained by a preference estimation unit multiple times. [Figure 33] FIG. 10 is a diagram showing a processing flow of a presentation information extraction unit. [Figure 34] 34A and 34B are diagrams showing examples of setting of an extracted theme by the presentation information extracting unit 370, and each shows a different setting example. [Figure 35] FIG. 10 is a block diagram showing a schematic configuration of an information processing device according to a fourth embodiment. [Figure 36] 10 is a flowchart showing an example of processing performed by an information processing device according to the fourth embodiment. [Figure 37] FIG. 13 is a diagram showing an example of information presented to a user terminal by a presentation unit of an information processing device according to the fourth embodiment. [Figure 38] FIG. 11 is a block diagram showing a schematic configuration of an information processing device according to a fifth embodiment. [Figure 39] 13 is a flowchart showing an example of processing performed by an information processing device according to the fifth embodiment. [Figure 40] FIG. 13 is a diagram showing an example of information presented to a user terminal by a presentation unit of the information processing device according to the fifth embodiment. [Figure 41] FIG. 13 is a block diagram showing a schematic configuration of an information processing device according to a sixth embodiment. [Figure 42] FIG. 20 is a diagram showing an example of information presented to a user terminal by a presentation unit of the information processing device according to the sixth embodiment. [Figure 43] FIG. 13 is a block diagram showing a schematic configuration of an information processing device according to a seventh embodiment. [Figure 44]FIG. 20 is a diagram showing an example of information presented to a user terminal by a presentation unit of the information processing device according to the seventh embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. First embodiment (with preference estimation) The first embodiment is an embodiment in which dishes are classified based on standard dishes and common seasonings, and the classification results are used to estimate a user's preferences for dishes from the name of any dish. FIG. 1 is a block diagram showing an example of a hardware configuration of an information processing device 100 according to the first embodiment. The information processing device 100 according to the first embodiment is configured by, for example, a computer having a processor 100a, a memory 100b, a communication device 100c, an input device 100d, and an output device 100e, which are connected to each other so as to be able to communicate with each other.

[0010] The processor 100a is an integrated circuit (IC) that performs processing and controls the operation of each unit of the information processing device 100. The processor 100a corresponds to, for example, a central processing unit (CPU), a demand side platform (DSP), a graphics processing unit (GPU), etc., and realizes the functions of each unit of the information processing device 100 by executing programs stored in memory. The memory 100b corresponds to, for example, a random access memory (RAM), a read only memory (ROM), a flash memory, a hard disk drive (HDD), or a solid state drive (SSD). The communication device 100c is a connection interface with an external device communicably connected to the information processing device 100, and includes a communication means such as a wired or wireless LAN (Local Area Network) or Bluetooth (registered trademark). The input device 100d corresponds to, for example, an input device such as a mouse, a keyboard, a touch panel, a microphone, etc. The input device 100d receives an input operation or voice from an operator via these input devices, and acquires information corresponding to the input operation or voice. The output device 100e corresponds to an output device such as a display such as an LCD (Liquid Crystal Display) or a speaker.

[0011] The information processing device is not limited to the above hardware configuration, and may be configured by a processing circuit which is dedicated hardware configured by, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, a system LSI (Large-Scale Integration), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field Programmable Gate Array), or may be configured by a combination of these hardware.

[0012] FIG. 2 is a block diagram showing a schematic configuration of the information processing device 100 according to the first embodiment. The storage unit 110 stores various programs executed by the information processing device 100, various data, and intermediate results of processing by the various programs. The basic cluster construction unit 120 constructs (generates) a basic cluster consisting of a plurality of clusters by clustering information indicating a plurality of pre-acquired dish names (standard dishes) and information indicating a plurality of seasoning names. In other words, the basic cluster construction unit 120 acquires information indicating a plurality of clusters that constitute the basic cluster by clustering the information indicating a plurality of pre-acquired dish names and information indicating a plurality of seasoning names to generate a basic cluster consisting of a plurality of clusters (a first number). In the following description, the plurality of clusters constructed by the basic cluster construction unit 120 are collectively referred to as "basic clusters." The basic cluster construction unit 120 also constitutes the cluster generation unit and cluster information acquisition unit in the first embodiment.

[0013] For example, the basic cluster construction unit 120 clusters popular dishes and seasonings using data in which the seasonings used in each dish are used as features, and defines the results as basic clusters. In other words, the basic cluster construction unit 120 acquires information indicating multiple dish names and information indicating the seasonings included in each of the multiple dish names, and generates a first number of clusters into which the multiple dish names are classified based on the information indicating the seasonings. The information processing device may be configured to acquire information indicating the basic cluster based on information stored in the storage unit or information from an external device communicably connected to the information processing device via a communication device. The basic cluster construction unit 120 is not limited to clustering both standard dishes and seasonings, but may be configured to cluster at least a plurality of dish names.

[0014] The basic cluster analysis unit 130 uses the basic clusters generated by the basic cluster construction unit 120 and data related to seasonings to extract features for each cluster, calculate the importance of words, and analyze the basic clusters. In the basic cluster analysis unit 130, the seasoning feature extraction unit 131 calculates the frequency of use of each seasoning for each cluster using the dishes classified into each cluster, and calculates the importance of seasoning words for each cluster. In the basic cluster analysis unit 130, the recipe feature extraction unit 132 extracts words related to recipes from the names of the dishes classified into each cluster, and extracts recipe feature information for each cluster. In the basic cluster analysis unit 130, the character type feature extraction unit 133 uses the dishes classified into each cluster of the basic cluster to calculate the ratio of character types (katakana, hiragana, kanji proper nouns, etc.) used in the names of dishes included in each cluster. In the basic cluster analysis unit 130, the dictionary creation unit 134 calculates the importance of words from the names of dishes classified into each cluster of the basic clusters.

[0015] The data acquiring unit 140 acquires the names of dishes to be subjected to preference analysis. For example, the data acquiring unit 140 acquires information indicating the names of specific dishes to be subjected to preference analysis through an input operation by the operator of the information processing device 100.

[0016] The dish classification unit 150 calculates the similarity between the names of dishes (the names of specific dishes to be analyzed) acquired by the data acquisition unit 140 and each dish classified into a basic cluster using information on the importance of words and the type of characters, and uses this as a feature to divide the basic cluster into a number of groups (clusters) smaller than the first number (a second number). The feature generating unit 151 in the dish classifying unit 150 generates statistics for each cluster as features for multiple types of similarity. In the dish categorization unit 150, the basic cluster categorization unit 152 divides the basic clusters into a smaller number of groups (clusters) using the features generated by the feature generation unit 151 in the dish categorization unit 150.

[0017] The related cluster estimation unit 160 extracts word information from the dish names (the names of the dishes to be analyzed) acquired by the data acquisition unit 140, calculates a relevance score indicating the degree of relevance with the dish names to be analyzed for each cluster using the analysis results of the basic cluster analysis unit 130, and extracts clusters with high relevance scores.

[0018] The preference estimation unit 170 estimates the user's preference for cooking from highly related clusters.

[0019] Next, the operation of the information processing device 100 will be described. In the following, standard dishes and common seasonings will be referred to simply as "standard dishes and seasonings." The basic cluster constructing unit 120 clusters the standard dishes and seasonings using data in which the seasonings are used as feature quantities, and defines the results as basic clusters. The processing flow of the basic cluster constructor 120 is shown in FIG. In step S1030, the basic cluster constructing unit 120 arbitrarily selects a plurality of standard dishes and seasonings. A standard dish refers to a dish that is likely to be familiar to many people, such as gratin, curry, mapo tofu, meat and potato stew, etc. A standard dish is, for example, a dish that is positioned as a common home-cooked meal in a cuisine category such as Japanese, Chinese, or Italian, or a dish that is often found in many restaurants in that cuisine category. Common seasonings are those that many people are likely to know (those whose flavors can be imagined from the wording). For example, common seasonings include "thick sauce" and "sweet and sour."

[0020] In step S1030, the basic cluster constructing unit 120 arbitrarily selects a plurality of names of standard dishes (such as "gratin") and general seasoning terms (such as "thick sauce"). In step S1031, the basic cluster constructing unit 120 creates data with seasonings as feature quantities for the standard dishes and seasonings selected in step S1030. Specifically, for example, the names of seasonings are used as column names, and data is created indicating which seasonings are used for each standard dish and seasoning selected in step S1030.

[0021] An example (part) of the data created by the basic cluster constructor 120 in step S1030 is shown in Fig. 4. For seasonings that are used in each standard dish or seasoning, the value is set to "1," and for seasonings that are not used, the value is set to "0." This data is input into the information processing device 100, for example, manually or by voice by the operator of the information processing device 100. In step S1032, the basic cluster construction unit 120 clusters the standard dishes and seasonings using the data created in step S1031. Any clustering method may be used, such as hierarchical clustering or non-hierarchical clustering, and any known method may be used. In the first embodiment, all of the multiple clusters generated as a result of the clustering by the basic cluster construction unit 120 in step S1032 are collectively referred to as "basic clusters."

[0022] As an example of basic clusters, some of the results of clustering by the basic cluster construction unit 120 are shown in Figure 5. For simplicity, Figure 5 shows only some of the names of dishes classified into each cluster. For example, the cluster with cluster ID 1 (cluster 1) classifies dishes that mainly use curry as a seasoning, while cluster 2 classifies dishes that mainly use milk or butter as a seasoning. Cluster 3 classifies dishes that mainly use sake, soy sauce, and mirin.

[0023] The basic cluster construction unit 120 stores information about the generated basic cluster (cluster ID and description of the name of the popular dish and its seasoning) in the memory unit 110, and passes it to the basic cluster analysis unit 130, the dish classification unit 150, the related cluster estimation unit 160, and the preference estimation unit 170. In the following, for the standard dishes and seasonings classified into each cluster of the basic cluster, we will not distinguish between "standard dishes" and "seasonings" and will simply refer to both as "dishes."

[0024] The basic cluster analysis unit 130 uses the basic clusters generated by the basic cluster construction unit 120 and data on the seasonings used in each dish to extract features for each cluster and analyze the basic clusters. In the basic cluster analysis unit 130, the seasoning feature extraction unit 131 calculates the frequency of use of each seasoning for each cluster based on the seasonings used in the dishes classified into each cluster, and calculates the importance of seasoning words for each cluster.

[0025] The processing flow of the seasoning feature extraction unit 131 is shown in FIG. In step S1040, the seasoning feature extraction unit 131 calculates the use frequency Tf of seasoning i in cluster j using equation (1). Tf = (number of dishes in which seasoning i is used among the standard dishes and seasonings classified in cluster j) / (total number of times seasoning i is used for all standard dishes and seasonings classified in cluster j) Equation (1) In step S1041, the seasoning feature extraction unit 131 calculates the rarity of the seasoning i between clusters using equation (2). Idf = log(total number of clusters / (number of clusters in which seasoning i appears + 1)) ...Equation (2)

[0026] In step S1042, the seasoning feature extraction unit 131 uses the results calculated in steps S1040 and S1041 to calculate the importance of seasoning i in cluster j using equation (3). (Importance of seasoning i in cluster j) = Tf * Idf Equation (3)

[0027] A part of an example of the importance of each seasoning in each cluster is shown in Figure 7. In this way, the importance of a certain seasoning is calculated for each cluster. Hereinafter, this result will be referred to as the "importance of seasoning words." The seasoning feature extraction unit 131 stores the importance of the seasoning words in the storage unit 110 and passes it to the dish classification unit 150, the related cluster estimation unit 160, and the preference estimation unit 170. In the basic cluster analysis unit 130, the recipe feature extraction unit 132 extracts words related to recipes from the names of the dishes classified into each cluster, and extracts recipe feature information for each cluster.

[0028] The processing flow of the recipe feature extraction unit 132 is shown in FIG. In step S1050, the recipe feature extraction unit 132 defines words corresponding to recipes. The recipe feature extraction unit 132 defines words corresponding to recipes, for example, by acquiring a file that defines words corresponding to recipes. The file that defines the words is input to the information processing device 100, for example, manually or by voice input by the operator of the information processing device 100. For example, when considering three types of cooking methods, simmering, frying, and stir-frying, the recipe feature extraction unit 132 defines them as "simmered, fried, stir-fried." Other possible methods include "baking" and "steaming," and the recipe feature extraction unit 132 defines these in the same way.

[0029] In step S1051, the recipe feature extraction unit 132 performs morphological analysis on the names of dishes classified into each cluster of the basic cluster, checks whether they contain any recipe words defined in step S1050, and extracts the recipe words. For example, in the cluster with cluster ID 3 in Figure 5 (cluster 3), the dish "simmered hijiki" contains "simmered," which is defined as a recipe word, so the recipe feature extraction unit 132 extracts "simmered" as a recipe feature for cluster 3. In the same way, the recipe feature extraction unit 132 checks whether any recipe words exist for all dishes included in each cluster, and extracts the recipe words if any. In step S1052, the recipe feature extraction unit 132 combines the results of the recipe words extracted from each dish in step S1051 for each cluster to obtain the recipe word feature extraction results.

[0030] 9 shows an example of the result of feature extraction of recipe words by the recipe feature extraction unit 132. The format of the feature extraction result is arbitrary and is not limited to the format shown in FIG. The recipe feature extraction unit 132 stores the feature extraction results of the recipe words in the storage unit 110 and passes them to the dish categorization unit 150, the related cluster estimation unit 160, and the preference estimation unit 170.

[0031] In the basic cluster analysis unit 130, the character type feature extraction unit 133 performs morphological analysis on the dish names classified into each cluster of the basic clusters, and calculates the ratio of character types (katakana, hiragana, kanji proper nouns, etc.). The processing flow of the character type feature extraction unit 133 is shown in FIG. In step S1060, the character type feature extraction unit 133 performs morphological analysis on each dish name classified into each cluster of the basic cluster, and extracts words of target parts of speech. For example, the character type feature extraction unit 133 extracts words of parts of speech other than auxiliary verbs. In step S1061, the character type feature extraction unit 133 determines the character type (katakana, hiragana, kanji, etc.) of the words extracted in step S1060 for each dish, and calculates the ratio of each character type to the total number of words extracted as targets for each dish. For example, the ratio of katakana is calculated using equation (4). (Katakana ratio) = (number of words determined to be katakana for the target dish) / (total number of words targeted for the target dish) Equation (4)

[0032] For example, the name of a dish "gratin" has a target word "gratin" written in katakana, and therefore the katakana ratio is 1.0. For example, the name of the dish "Oyakodon" has two target words, "oyako" and "donburi," both of which are kanji characters, so the kanji ratio is 1.0. In this way, the character type feature extraction unit 133 calculates the ratio of each of katakana, hiragana, and kanji. Furthermore, the character type feature extraction unit 133 may be configured to add a condition for kanji or hiragana instead of the ratio of character types, and calculate, for example, the "ratio of proper nouns in kanji."

[0033] The types of ratios and calculation formulas calculated by the character type feature extraction unit 133 are not limited to those described above. The types of ratios and calculation formulas can be defined arbitrarily by combining detailed information such as the part of speech and character type (katakana, hiragana, kanji, etc.) extracted by the character type feature extraction unit 133, whether the word is a common noun or a proper noun, etc. For example, the character type feature extraction unit 133 may use "katakana ratio, hiragana or kanji ratio" instead of using the three types of "katakana ratio, hiragana ratio, kanji ratio." Furthermore, in the above operation example, the target parts of speech were described as being other than auxiliary verbs, but the target parts of speech can be any. In step S1062, the character type feature extraction unit 133 integrates the ratio calculation results of step S1061 for all the dishes in the basic cluster.

[0034] Fig. 11 shows an example in which the character type feature extraction unit 133 integrates the ratio calculation results for the basic clusters in Fig. 5. In Fig. 11, the types of ratios are "katakana ratio, hiragana ratio, and kanji ratio," but the types are not limited to this. It can also be "the ratio of katakana, the ratio of hiragana or kanji," and is arbitrary. The character type feature extraction unit 133 stores the output result (the result integrated in step S1062) in the storage unit 110 and passes it to the dish classification unit 150 and the related cluster estimation unit 160.

[0035] In the basic cluster analysis unit 130, the dictionary creation unit 134 performs morphological analysis on the names of dishes classified into each cluster of the basic clusters, and calculates the importance of the words. The processing flow of the dictionary creation unit 134 is shown in FIG. In step S1070, the dictionary creation unit 134 performs morphological analysis on the names of the dishes classified into the target cluster. In step S1071, the dictionary creation unit 134 calculates the importance of a word using the Term Frequency-Inverse Document Frequency (TF-IDF) method. The dictionary creation unit 134 calculates the importance of a word for each cluster by using the TF-IDF in the document as the TF-IDF in the cluster. TF is the proportion of times a word appears relative to the total number of times all words appear in all food names classified into the target cluster, and IDF is the degree to which a word appears only in some clusters rather than in all clusters. As a result, different importance values ​​are calculated for each word for each cluster. The dictionary creation unit 134 stores the calculation results (dictionary) of the importance of words in the storage unit 110 and passes them to the dish categorization unit 150 and the related cluster estimation unit 160.

[0036] The above is the operation of the pre-processing performed by the information processing device 100 prior to preference estimation.

[0037] Next, an operation of the information processing device 100 for estimating a user's food preferences from an arbitrary food name will be described. The data acquisition unit 140 acquires the names of dishes to be subjected to preference analysis. The data acquiring unit 140 acquires, as text information, for example, at least one of the "name(s) of the dishes eaten by the user" or the "name(s) of the favorite dishes" that are the subject of preference estimation. When it is desired to estimate preferences on a household basis, taking the preferences of all family members as a whole, the data acquisition unit 140 may, for example, define the names of dishes eaten by all family members over a specified period in a file and store the file in the storage unit 110, and the data acquisition unit 140 may acquire the names of dishes from the file. In this way, the name of a dish to be analyzed is not limited to one, but multiple names of dishes can be acquired.

[0038] Furthermore, the names of dishes to be analyzed do not have to be dishes eaten or favorite dishes, and can be any. For example, dishes can be selected arbitrarily from categories such as Japanese, Chinese, and Italian and defined in a file, and the data acquisition unit 140 can acquire the names of dishes from that file. The file defining the names of dishes can be created, for example, manually or by voice by the operator of the information processing device 100. Furthermore, instead of defining the names in a file, the names of dishes can be entered via a user interface such as a keyboard, and the data acquisition unit 140 can acquire them. The data acquisition unit 140 stores the acquired data on the dish names in the storage unit 110 and passes it to the dish categorization unit 150, the related cluster estimation unit 160, and the preference estimation unit 170.

[0039] The dish classification unit 150 uses information on at least one of the word importance or character type for the names of dishes acquired by the data acquisition unit 140 (the names of dishes to be analyzed for preferences; hereinafter referred to as the names of dishes to be analyzed) to calculate the similarity with each dish classified into a basic cluster, and uses this as a feature to divide the basic cluster into fewer groups.

[0040] The feature generating unit 151 in the dish classifying unit 150 generates statistics for each cluster as features for multiple types of similarity. The processing flow of the feature generating unit 151 is shown in FIG. In step S1080, the feature generation unit 151 calculates the ratio of character types, such as katakana, hiragana, or kanji, for the name of the dish to be analyzed and the names of each dish classified into the basic cluster, and generates a feature vector based on the character types. The feature generation unit 151 performs morphological analysis on the dish name (the name of the dish to be analyzed) acquired by the data acquisition unit 140, calculates the ratio of character types such as katakana, hiragana, or kanji, and generates a feature vector based on the character types. The type of ratio and calculation formula used in step S1080 are the same as the type of ratio and calculation formula used by the character type feature extraction unit 133 of the basic cluster analysis unit 130. As described for the character type feature extraction unit 133 of the basic cluster analysis unit 130, the type of ratio and calculation formula are not limited to katakana, hiragana, and kanji, and can be arbitrarily determined in combination with detailed information such as the part of speech, the type of character (katakana, hiragana, kanji, etc.), whether the word is a common noun or a proper noun, etc.

[0041] 11, the feature generation unit 151 generates a feature vector having these three ratio types as components. When other ratio types are used, the feature generation unit 151 generates a feature vector having the used ratio type as a component. Furthermore, for each dish classified into a basic cluster, the character type feature extraction unit 133 of the basic cluster analysis unit 130 calculates the ratio, and the feature generation unit 151 uses this result to generate a feature vector for each dish whose components are values ​​corresponding to the type of ratio. In step S1081, the feature generation unit 151 calculates similarity based on character type using the result of step S1080 (feature vectors whose components are ratios based on character type for the dish to be analyzed and each dish classified into the basic cluster). The feature generation unit 151 can use cosine similarity or a known method as a formula for calculating similarity.

[0042] In step S1082, the feature generation unit 151 uses the dictionary created by the dictionary creation unit 134 of the basic cluster analysis unit 130 to generate feature vectors based on the importance of words for the name of the dish to be analyzed and for each dish name registered in the basic cluster. Specifically, the feature generation unit 151 performs morphological analysis on each dish name, and calculates the importance of the words that make up the dish name for each dish by using the importance (dictionary) calculated by the dictionary creation unit 134 of the basic cluster analysis unit 130, and then adds up the importance values ​​for all words that appear in each dish name. Because the importance created by the dictionary creation unit 134 differs for each cluster even for the same word, the feature generation unit 151 calculates the importance for one dish name as many times as there are basic clusters.

[0043] In the dictionary created by the dictionary creation unit 134, the importance of word i in cluster k is w ik If the number of words to be processed contained in the name of a dish for which importance is to be calculated (a certain dish name) is n, the sum of the importance of cluster k for the certain dish name is a k is calculated using equation (5). a k =Σ i=1 n w ik ...Equation (5)

[0044] Furthermore, the feature generating unit 151 calculates a for each cluster k using equation (5). k , and calculate the component P of the feature vector for cluster k. k is calculated using equation (6). P k =ak / (Σ k=1 Num a k )...Equation (6) In equation (6), Num is the total number of basic clusters. Using the above results, the feature generating unit 151 generates a feature vector based on the importance of words as follows: v=(P1, P2, . . . , P Num )

[0045] In step S1083, the feature generator 151 calculates the similarity based on the importance of words between the name of the dish to be analyzed and the names of the dishes classified into the basic cluster, using the feature vector v calculated in step S1082. The similarity may be calculated using cosine similarity or another known similarity index.

[0046] In step S1084, the feature generating unit 151 calculates statistics for each cluster for the similarity based on the character type and the similarity based on the importance of the word, and integrates the calculation results. Specifically, since each cluster of the basic clusters contains one or more dishes, the feature generation unit 151 calculates and integrates statistics (minimum, maximum, etc.) for each cluster for the similarity based on character type and the similarity based on word importance.

[0047] FIG. 14 shows a specific example of statistics for each cluster calculated by the feature generator 151. FIG. 14 shows an example in which the number of clusters in the basic cluster is Num=5. In FIG. 14, the similarity based on character type is represented as similarity 1, and the similarity based on word importance is represented as similarity 2. For example, "similarity 1_average value" means the average value for each cluster of similarity based on character type. The number 0.639 in the upper left of FIG. 14 (the value of the cell where the row with cluster ID 1 intersects with the column for similarity 1_average value) means the average value of similarity 1 (similarity based on character type) for the dishes included in cluster 1. 14, the average value, minimum value, and maximum value are used as the statistics, but the standard deviation, median value, etc. may also be used. The statistics calculated by the feature generating unit 151 are arbitrary.

[0048] In the dish classification unit 150, the basic cluster classification unit 152 classifies the basic clusters into a smaller number of groups using the features generated by the feature generation unit 151 in the dish classification unit 150. The basic cluster classification unit 152 can use known clustering methods such as hierarchical clustering and non-hierarchical clustering as the clustering method.

[0049] FIG. 15 shows an example in which the basic cluster classification unit 152 classifies the basic clusters into three groups using the feature amounts shown in FIG. In Fig. 15, the group ID indicates the ID of each group when each basic cluster is divided into a smaller number of groups. For example, in Fig. 15, clusters with basic cluster IDs of 2, 3, and 4 are grouped together as a group with group ID 2. The number of small groups (three in the case of FIG. 15) can be arbitrarily determined. The basic cluster classification unit 152 stores the results of classification by the dish classification unit 150 in the storage unit 110 and also passes them to the related cluster estimation unit 160.

[0050] The related cluster estimation unit 160 estimates a cluster related to the name of a dish to be analyzed for preferences from among multiple clusters in the basic cluster based on information extracted from the name of a dish to be analyzed for preferences. For example, the related cluster estimation unit 160 extracts word information from a specific dish name (the name of a dish to be analyzed for preferences) acquired by the data acquisition unit 140, compares the information with the basic cluster to calculate a relevance score, and extracts one or more clusters with a relatively high relevance score, which are considered to be clusters highly related to the dish to be analyzed. Specifically, the related cluster estimation unit 160 extracts word information from the specific dish name acquired by the data acquisition unit 140, compares the information with the basic cluster to calculate a relevance score, and extracts the cluster with the highest relevance score. Note that the extraction of multiple clusters including the cluster with the highest relevance score is not limited to this. The related cluster estimation unit 160 may be configured to extract one or more clusters including a cluster having a relevance score higher than other clusters. For example, the related cluster estimation unit 160 may be configured to extract multiple clusters including a cluster having the highest calculated relevance score, or may be configured to extract one of the multiple clusters having the highest calculated relevance scores. The processing flow of the related cluster estimation unit 160 is shown in FIG. In step S1090, the related cluster estimation unit 160 calculates the similarity of seasonings for each cluster. Specifically, the related cluster estimation unit 160 first performs morphological analysis on the name of the dish to be analyzed and extracts words related to seasoning from the dish name. Next, the related cluster estimation unit 160 uses the "importance of seasoning words" calculated by the seasoning feature extraction unit 131 of the basic cluster analysis unit 130 to calculate the total importance of all the extracted words related to seasoning, and sets this as the similarity.

[0051] For example, if the name of the dish to be analyzed is "Curried minced meat and eggplant," "curry" is extracted as a word related to seasoning. Since curry is the only word related to seasoning, the value of "importance of seasoning words" for the word curry is the similarity related to seasoning. The "importance of seasoning words" is calculated for each cluster by the importance of each word in the seasoning name. If the name of the dish being analyzed contains multiple words related to seasoning, the importance of all the words related to seasoning is calculated as the sum of the importance of all the words related to seasoning.

[0052] In step S1091, the related cluster estimation unit 160 calculates the similarity of the recipe between each cluster in the basic cluster and the name of the dish to be analyzed. Specifically, the related cluster estimation unit 160 first performs morphological analysis on the name of the dish to be analyzed, and extracts words corresponding to cooking methods from the dish name. The words corresponding to cooking methods are those defined in the cooking method feature extraction unit 132 of the basic cluster analysis unit 130. For example, if the name of the dish to be analyzed is "stir-fried minced meat and eggplant with curry," "stir-fried" is extracted as a word corresponding to a cooking method. Next, the related cluster estimation unit 160 compares the extracted recipe words with the feature extraction results (see FIG. 9 for an example) of recipe words extracted by the recipe feature extraction unit 132 of the basic cluster analysis unit 130, and calculates the similarity between each cluster and the recipe in the basic cluster. For example, if the extracted word is in a cluster that is included in the feature extraction results of recipe words, the similarity is set to 1, and if not, the similarity is set to 0. In the case of the feature extraction results of recipe words shown in FIG. 9, the similarity between clusters 3 and 4, in which "stir-fry" is extracted, is 1, and the similarities of the other clusters are 0.

[0053] In this way, the related cluster estimation unit 160 calculates the similarity of seasonings and cooking methods for each cluster of the dishes to be analyzed. An example of the calculation results by the related cluster estimation unit 160 is shown in FIG. In step S1092, the related cluster estimation unit 160 calculates a score (relatedness score) indicating the relatedness between the dish to be analyzed (the dish with the dish name acquired by the data acquisition unit 140) and each cluster of the basic cluster. The higher the relatedness score, the higher the relatedness between the dish classified into that cluster and the dish to be analyzed. The related cluster estimation unit 160 calculates the relatedness score using equation (7) by using the seasoning similarity calculated in step S1090, the cooking method similarity calculated in step S1091, and the statistics (maximum value, etc.) for each cluster of the similarity based on the character type and word importance calculated by the feature generation unit 151 of the cuisine classification unit 150. (Relevance score)=s(taste)*w1+s(method)*w2+Σ i=1 n s(word) i *x i +Σ i=1 m s(kind) i *y i ...Equation (7) s(taste): Similarity of flavor w1: Weight of flavor similarity s(method): Similarity of cooking method w2: Weight of recipe similarity s(word) i : Statistics for each cluster of similarity based on word importance (i = minimum, maximum, etc.) n: The number of statistics used to calculate the cluster-specific similarity statistics based on word importance. x i : Weight of the statistic i for each cluster of similarity based on the importance of words (i = minimum, maximum, etc.) s(kind) i : Statistics for each cluster of similarity based on character type i (i = minimum, maximum, etc.) y i : Weight of the statistical value i for each cluster of similarity based on character type (i = minimum, maximum, etc.) m: The number of statistics used to calculate the similarity statistics for each cluster based on character type In equation (7), the weight value can be set arbitrarily.

[0054] An example of the calculation result of the relevance score by the related cluster estimation unit 160 is shown in FIG. In step S1093, the related cluster estimation unit 160 compares the relevance score value calculated in step S1092 with the results of dividing the basic clusters into fewer groups by the basic cluster classification unit 152 of the cuisine classification unit 150, and extracts clusters from the basic clusters that are highly related to the cuisine being analyzed (the cuisine acquired by the data acquisition unit 140). Specifically, the related cluster estimation unit 160 first calculates a threshold value for extracting highly related clusters using equation (8), and extracts cluster IDs whose relatedness scores are equal to or greater than the threshold value. (threshold) = (maximum relevance score in all clusters) * (ratio) Equation (8) In the case of FIG. 18, the maximum value is 0.751, and if the ratio is 0.8, for example, the threshold value is 0.6. A cluster with a relevance score of 0.6 or more will be a cluster with a cluster ID of 1 (cluster 1).

[0055] Next, the related cluster estimation unit 160 compares the results with the results ( FIG. 15 ) of the basic clusters divided into groups by the basic cluster classification unit 152 of the cuisine classification unit 150. Cluster 1, extracted because its relevance score is equal to or greater than the threshold, does not form the same group as other clusters, and Cluster 1 is not in the same group as Cluster 5 (relevance score 0) and Cluster 2 (relevance score 0.044), which have low relevance scores. Therefore, if the cuisine being analyzed is "stir-fried minced meat and eggplant with curry," Cluster 1 is determined as the final result as the cluster that is highly related to this cuisine. As shown in FIG. 5, Cluster 1 is a cluster into which curry-flavored cuisine is classified, and Cluster 1 is extracted as a highly related cluster taking into consideration the curry seasoning.

[0056] In the above example, in the groups generated by the basic cluster classification unit 152 of the cuisine classification unit 150, clusters with high relevance scores and clusters with low relevance scores are separated into separate groups, but another example is shown below. When the relevance score is as shown in FIG. 19A and the result of grouping the clusters by the basic cluster classification unit 152 of the cuisine classification unit 150 is as shown in FIG. 19B, for example, the following results are obtained.

[0057] First, when 0.8 is used as the ratio in equation (8), clusters with cluster IDs 3 and 4 are extracted as clusters with a relevance score of 0.66 or higher. When this result is compared with the grouping result of the basic cluster classification unit 152 (FIG. 19B), cluster ID 3 is not in the same group as the other clusters and has a high relevance score, so it is adopted as the final result. On the other hand, cluster 4 is classified into group 3, the same as cluster 5, and there is a large difference between the relevance score of cluster 4 (0.695) and the relevance score of cluster 5 (0.328). Therefore, cluster 4 is not adopted as the final result. From the above, only the cluster with cluster ID 3 is adopted as the final result as it is a highly relevant cluster.

[0058] On the other hand, for example, when the relevance score is as shown in Figure 19A and the result of grouping the clusters by the basic cluster classification unit 152 of the cuisine classification unit 150 is as shown in Figure 19C, the cluster with cluster ID 4 is classified into the same group (group 2) as the cluster with cluster ID 3, which has the highest relevance score, and there is little difference between this and cluster 3, which has the highest relevance score, so it is adopted as the final result. In other words, when the result of the basic cluster classification unit 152 is as shown in Figure 19C, the cluster with cluster ID 3 and the cluster with ID 4 are adopted as the final results as they are highly related clusters.

[0059] In this way, after extracting clusters with relevance scores equal to or greater than the threshold, the related cluster estimation unit 160 further refers to the classification results based on similarity generated by the basic cluster classification unit 152 of the cuisine classification unit 150 to adjust the final results to be regarded as highly related clusters. For example, the related cluster estimation unit 160 compares the cluster IDs extracted based on the threshold with the relevance score values ​​of other clusters in the group to which each cluster belongs (the group generated by the basic cluster classification unit 152), and adjusts whether or not to ultimately adopt the extracted cluster ID as a highly related cluster.

[0060] The above describes an example of operation when the dish name (the name of the dish to be analyzed) acquired by the data acquisition unit 140 is "stir-fried minced meat and eggplant with curry," but if there are multiple dish names acquired by the data acquisition unit 140, an estimation result of a highly related cluster can be obtained for each dish name. It is also possible to have only the top most related cluster instead of multiple highly related clusters. The related cluster estimation unit 160 stores the estimation results (highly related clusters and the values ​​of the related scores of those clusters) in the storage unit 110 and also passes them to the preference estimation unit 170.

[0061] The preference estimation unit 170 estimates the user's preference for the cuisine to be analyzed from the cluster(s) with high relevance and the value of the relevance score. The processing flow of the preference estimation unit 170 is shown in FIG. In step S1100, the preference estimation unit 170 obtains the extraction results of clusters that are highly relevant to a plurality of dishes. The "multiple dishes" are dishes to be used for preference estimation. For example, if you want to estimate a user's preferences for a week, you would use the names of the dishes eaten in a week, and if you want to estimate preferences for a month, you would use the names of the dishes eaten in a month. For each dish, the related cluster estimation unit 160 estimates a cluster with a high degree of association.

[0062] In step S1101, the preference estimation unit 170 totalizes the relevance score values ​​for each basic cluster using the results (cluster IDs and relevance scores) of clusters highly related to each dish extracted in step S1100. If multiple clusters are extracted for a certain dish in step S1100, the preference estimation unit 170 adds up the relevance score values ​​by weighting the multiple cluster IDs. In step S1102, the preference estimation unit 170 normalizes the values ​​calculated for each cluster in step S1101 to calculate the preference degree.

[0063] 21 shows an example of visualization of the calculation results of the preference degree by the preference estimation unit 170. An example in which a basic cluster is composed of five clusters is shown. In step S1103, the preference estimation unit 170 classifies the preference into levels using the preference degree in step S1102.

[0064] An example of level classification by the preference estimation unit 170 is shown in Fig. 22. In Fig. 22, "MAX" is the maximum value of the preference degree of all clusters. The preference estimation unit 170 calculates the ratio of the preference degree of each cluster to the maximum value of the preference degree of all clusters, and classifies it into five levels (love it, relatively love it, normal, not dislike it but not normal, dislike it).

[0065] The result of the level classification by the preference estimation unit 170 for the preference degrees shown in Fig. 21 is shown in Fig. 23. From this result and the dishes classified into each cluster ID, information indicating what kind of dishes the person likes and to what extent can be obtained. In this example, it is estimated that the cluster ID is 3, meaning that the person loves dishes seasoned with soy sauce, sake, mirin, and the like, which are common in Japanese cuisine, and the cluster ID is 2, meaning that the person also relatively likes dishes with creamy seasonings, such as gratin or cream stew.

[0066] Fig. 24 shows the result of the preference estimation unit 170 calculating the preference degree of a user different from the above. Fig. 25 shows the result of the preference estimation unit 170 classifying the preference levels into five stages defined in Fig. 22. In the case of this user, the basic cluster ID is 2, that is, it is estimated that the user loves dishes with creamy seasonings such as gratin or cream stew, and dislikes seasonings in the cluster with a basic cluster ID of 5 (Chinese seasonings such as chili bean paste).

[0067] In the above example of operation, the preference levels are divided based on the ratio of the preference degree of all clusters to the maximum value as shown in Fig. 22, but the levels may be divided based on the value of the preference degree without using the ratio to the maximum value. For example, a preference degree of 0.8 or more is "I love it," and a preference degree of 0.6 or more but less than 0.8 is "I like it relatively," etc. Furthermore, the number of levels is not limited to five, but can be any number of levels. Furthermore, even when using a ratio to the maximum value of the preference degree, ratio values ​​other than those shown in FIG. 22 may be used, and the numerical value of the ratio is arbitrary.

[0068] As described above, the information processing device 100 according to the first embodiment includes a cluster generation unit that acquires information indicating a plurality of dish names and information indicating the seasonings included in each of the plurality of dish names, classifies the plurality of dish names based on the information indicating the seasonings, and generates a first number of clusters by categorizing the plurality of dish names, a data acquisition unit that acquires information indicating a specific dish name, and a related cluster estimation unit that estimates a cluster related to the specific dish name from among the first number of clusters based on information extracted from the specific dish name. Thus, the information processing device 100 estimates a cluster related to the specific dish name from among the plurality of clusters into which the plurality of dish names have been classified based on word information extracted from the specific dish name, and is therefore able to classify the specific dish name based on its relevance to other dish names even if the recipe is unknown.

[0069] Furthermore, the information processing device 100 according to the first embodiment acquires information indicating a plurality of dish names and a plurality of seasoning names, and information indicating seasonings included in each of the plurality of dish names and the plurality of seasoning names, and generates a first number of clusters into which the plurality of dish names and the plurality of seasoning names are classified based on the information indicating the seasonings. Configured in this manner, the information processing device 100 estimates a cluster associated with a particular dish name from among the plurality of clusters into which the plurality of dish names and the plurality of seasoning names are classified, based on word information extracted from the particular dish name. Therefore, even if the recipe is unknown, the particular dish name can be classified based on its association with other dish names and other seasoning names.

[0070] Furthermore, the information processing device 100 according to the first embodiment estimates a cluster associated with a specific dish name from among the first number of clusters, based on information indicating the seasonings included in the multiple dish names and words related to the seasonings used in the specific dish name. With this configuration, the information processing device 100 estimates a cluster associated with a specific dish name from among the multiple clusters into which multiple dish names are classified, based on the seasonings included in the specific dish name, and therefore can estimate the association between a specific dish name and other dish names even if the recipe for the specific dish name is unknown.

[0071] Furthermore, the information processing device 100 according to the first embodiment estimates a cluster among the first number of clusters that is related to a specific dish name, based on information about characters used in the dish names and the specific dish name included in each of the first number of clusters. Configured in this way, the information processing device 100 can improve the accuracy of estimating the association between each cluster and the specific dish name by using information obtained from the dish names and the specific dish name included in each of the first number of clusters.

[0072] Furthermore, the information processing device 100 according to the first embodiment estimates a cluster among the first number of clusters that is related to a specific dish name, based on information about the dish names included in each of the first number of clusters and the words indicating the cooking methods used for the specific dish name. Configured in this way, the information processing device 100 can improve the accuracy of estimating the association between each cluster and a specific dish name by using information obtained from the dish names included in each of the first number of clusters and the specific dish name itself.

[0073] Furthermore, in the information processing device 100 according to embodiment 1, the related cluster estimation unit estimates clusters among the first number of clusters that are related to a specific dish name, based on information indicating the ratio of character types used in the dish names included in each of the first number of clusters and the specific dish name. Configured in this way, the information processing device 100 can improve the accuracy of estimating the relevance between each cluster and the specific dish name by using information obtained from the dish names included in each of the first number of clusters and the specific dish name itself.

[0074] Furthermore, the information processing device 100 according to the first embodiment estimates a cluster among the first number of clusters that is related to a specific dish name based on information indicating the importance of each word used in the dish names included in each of the first number of clusters and the specific dish name. Configured in this way, the information processing device 100 can prevent the association between each cluster and the specific dish name from being estimated based on words with low importance used in the dish names included in each of the first number of clusters and the specific dish name, thereby improving the accuracy of estimating the association between each cluster and the specific dish name.

[0075] Furthermore, the information processing device 100 according to the first embodiment includes a cuisine classification unit that classifies a first number of clusters generated by the cluster generation unit into a second number of clusters that is smaller than the first number, and the related cluster estimation unit calculates a relevance score indicating the degree of relevance with the specific cuisine name for each of the first number of clusters based on information extracted from the specific cuisine name, and the related cluster estimation unit estimates clusters from the first number of clusters that are related to the specific cuisine name based on the relevance score for each of the first number of clusters and the difference in relevance scores between the multiple clusters included in each of the second number of clusters classified by the cuisine classification unit. With this configuration, the information processing device 100 can estimate the relevance between each cluster and the specific cuisine name taking into account the relevance between the first number of clusters, thereby improving the accuracy of estimating the relevance between each cluster and the specific cuisine name.

[0076] Furthermore, in the information processing device 100 according to the first embodiment, the cuisine classification unit classifies the first number of clusters generated by the cluster generation unit into a second number of clusters that is less than the first number, based on information indicating seasonings. With this configuration, the information processing device 100 can estimate the association between each cluster and a specific cuisine name by taking into account the association between the seasonings used among the first number of clusters, thereby improving the accuracy of estimating the association between each cluster and a specific cuisine name.

[0077] Furthermore, in the information processing device 100 according to the first embodiment, the cuisine classification unit classifies the first number of clusters generated by the cluster generation unit into a second number of clusters that is less than the first number, based on information about characters used that is extracted from cuisine names included in each of the first number of clusters. With this configuration, the information processing device 100 can estimate the association between each cluster and a specific cuisine name by taking into account the association between the characters used among the first number of clusters, thereby improving the accuracy of estimating the association between each cluster and a specific cuisine name.

[0078] Furthermore, the information processing device 100 according to the first embodiment includes a preference estimation unit that estimates preferences for each cluster using the estimation results of clusters related to specific dish names estimated by the related cluster estimation unit, and estimates the user's preferences related to dishes. With this configuration, the information processing device 100 can estimate the user's preferences based on the association between each cluster and the specific dish name. Furthermore, by estimating the user's preferences, it becomes possible to provide the user with information that matches the user's preferences.

[0079] In the first embodiment, the preference estimation unit 170 normalizes the total value of the relevance scores for each cluster ID to obtain the preference degree, but the method for calculating the preference degree is not limited to this method and may be any method. The total value may be regarded as the preference level, or the total value may be classified into multiple levels and used as the preference level. In addition, the relevance score value and preference level for each dish can be defined, for example, as shown in Figure 26, and the number of each preference level can be tallied for each cluster in the basic cluster, and the preference for each cluster can be output based on the tallied value.

[0080] Embodiment 2 (no preference estimation, extraction of dishes highly related to a certain dish) The second embodiment is an embodiment in which, for a name of a dish to be analyzed, other dishes (there may be multiple dishes) that are highly related to that dish are extracted.

[0081] 27 is a block diagram showing a schematic configuration of an information processing device 200 according to Embodiment 2. The information processing device 200 according to Embodiment 2 differs from the configuration of the information processing device 100 according to Embodiment 1 in that it does not include a preference estimation unit 170 and includes a dish addition unit 270 and a related dish extraction unit 280. However, the processing contents and specific operations of the storage unit 110, basic cluster construction unit 120, basic cluster analysis unit 130, data acquisition unit 140, dish classification unit 150, and related cluster estimation unit 160 are the same as those in Embodiment 1. For this reason, the same components as those in Embodiment 1 are denoted by the same reference numerals and names, and descriptions thereof will be omitted.

[0082] The dish adding unit 270 associates dishes other than those registered in the basic cluster (dishes other than standard dishes and dishes used as seasonings) with each cluster ID of the basic cluster. In the second embodiment, the dish adding unit 270 constitutes an association information generating unit that generates information indicating the association between the name of the dish to be analyzed and any of the clusters in the basic cluster.

[0083] The processing flow of the dish adding unit 270 is shown in FIG. In step S2020, the dish adding unit 270 acquires the estimation result of the related cluster for the dish name. Specifically, since the related cluster estimation unit 160 has estimated a highly related cluster for the dish name (the name of the dish to be analyzed; for example, the name of a dish eaten by a certain user in a month) acquired by the data acquisition unit 140, the dish adding unit 270 acquires a result in which each dish name is associated with the ID of the highly related cluster.

[0084] 29A and 29B show examples of the estimation results by the related cluster estimation unit 160. For example, FIG. 29A shows the estimation result of the cluster that is most relevant to the name of the dish eaten by user A, and FIG. 29B shows the estimation result of the cluster that is most relevant to the name of the dish eaten by user B. Since the related cluster estimation unit 160 stores the estimation results in the storage unit 110, the dish adding unit 270 acquires this information from the storage unit 110. In step S2021, the dish adding unit 270 integrates the results acquired in step S2020 and organizes dish names for each cluster ID.

[0085] An example of the results of processing by the dish adding unit 270 is shown in Fig. 30. The integration format may be any format other than that shown in Fig. 30. In this way, the dish adding unit 270 associates dishes other than those registered in the basic cluster (dishes other than standard dishes and dishes used as seasonings) with each cluster ID of the basic cluster. The dish adding unit 270 stores the association results (basic cluster ID, dish name) in the storage unit 110 and passes them to the related dish extracting unit 280.

[0086] The related dish extraction unit 280 receives the name(s) of a new dish to be analyzed from the data acquisition unit 140, and extracts dishes that are highly related to the new dish(es). One possible method for extracting highly relevant dishes is to extract dishes that are classified into a cluster ID that is highly relevant. Dishes classified into a certain cluster ID include the following: The standard dishes and seasonings used in constructing the basic cluster in the basic cluster constructing unit 120 The dish adding unit 270 associates the dish with the ID of the basic cluster.

[0087] For example, if the name of the new dish to be analyzed is "stir-fried minced meat and eggplant with curry," and a cluster with cluster ID 1 (cluster 1) is extracted as a highly relevant cluster, and the dishes classified into cluster 1 based on the results of the dish addition unit 270 are as follows, then the following five dishes will be extracted as dishes highly relevant to "stir-fried minced meat and eggplant with curry." · Dishes used in constructing the basic cluster (standard dishes and seasonings) Curry rice, keema curry, taco rice -Foods added in Food Addition Section 270 Dish name N, Dish name a

[0088] The related dish extraction unit 280 may extract dishes from highly related cluster IDs in any manner, not just the above-described method. For the names of new dishes to be analyzed and each dish added by the dish adding unit 270, similarities may be calculated using the character type and / or the importance of words in the same manner as the processing contents and operations of the feature generating unit 151, and dishes to be extracted may be selected from the dishes classified into each cluster based on the similarities. The number of dishes to be extracted may be one or more, and is arbitrary. In this way, the related dish extraction unit 280 extracts one or more dishes that are highly related to the name of the new dish to be analyzed.

[0089] In the second embodiment, the dish adding unit 270 adds the estimated result of a cluster highly related to the name of a dish eaten by the user. However, the dish to be associated with the ID of a basic cluster is not limited to the dish eaten by the user and can be any dish. The dish name to be added is defined in a file, and the associated cluster estimation unit 160 reads the file and estimates the cluster highly related to each dish defined in the file. Using the estimated result, the dish adding unit 270 can associate dishes other than standard dishes and seasonings with each cluster of the basic cluster. In this way, the number of dishes associated with each cluster of the basic cluster can be increased.

[0090] As described above, the information processing device 200 according to embodiment 2 can increase the number of dishes associated with each cluster of the basic cluster for dishes other than those registered in the basic cluster by using the dish addition unit, and therefore can extract dishes related to the dish to be analyzed from any dishes, without being limited to only the dishes registered in the basic cluster.

[0091] Third embodiment (with preference estimation. Extraction of dishes based on preferences) The third embodiment is an embodiment in which preferences are estimated from the names of dishes to be analyzed, and information such as dishes to be presented to the user is extracted based on the preferences. FIG. 31 is a block diagram showing a schematic configuration of an information processing device according to the third embodiment. The information processing device 300 according to embodiment 3 differs from the configuration of the information processing device 100 according to embodiment 1 in that it includes a dish adding unit 270, a presentation information extracting unit 370, and a presentation unit 380, but the processing contents and specific operations of the storage unit 110, basic cluster constructing unit 120, basic cluster analyzing unit 130, data acquiring unit 140, dish categorizing unit 150, and related cluster estimating unit 160 are the same as those in embodiment 1. Furthermore, the processing contents and specific operations of the dish adding unit 270 are the same as those in embodiment 2. For this reason, the same components as those in embodiment 1 and embodiment 2 are denoted by the same reference numerals and names, and descriptions thereof will be omitted.

[0092] The preference estimation unit 170 according to the third embodiment estimates a user's preference for cooking from highly related clusters (there may be multiple clusters) and their related score values. The processing contents and specific operations of the preference estimation unit 170 according to the third embodiment are the same as those of the first embodiment, but the third embodiment describes an example of operation in which the preference estimation unit 170 operates multiple times in a predetermined period (for example, six months), and each time, the preference estimation result (the association between the ID of the basic cluster and the preference level) is recorded and accumulated in the storage unit 110 in association with the period.

[0093] FIG. 32 shows an example of a preference estimation result obtained by the preference estimation unit 170 according to the third embodiment multiple times. 32, the preference level is written in text, but it may be expressed in numerical form. The input format of the period may also be the period ID or the like. The format for recording and storing the associations with the preference estimation results over multiple times is arbitrary and is not limited to the format shown in FIG. The presentation information extraction unit 370 obtains the preference estimation results recorded and accumulated by the preference estimation unit 170 from the storage unit 110, and extracts and outputs information such as dishes to be presented to the user based on the preference estimation results. Note that in the third embodiment, the presentation information extraction unit 370 constitutes an output unit that outputs information related to dishes according to the user's preferences.

[0094] The processing flow of the presentation information extraction unit 370 is shown in FIG. In step S3030, the presentation information extraction unit 370 acquires preference estimation results for multiple past periods. For example, if six months have passed since the information processing device was started and preference estimation for a certain user has been performed every month, preference estimation results for each month (six times in total) have been accumulated, and the presentation information extraction unit 370 acquires these results.

[0095] In step S3031, the presentation information extraction unit 370 sets what preference level of dishes should be presented to the user (extraction theme of presentation targets) based on the preference estimation results for each period acquired in step S3030. Examples of extraction themes set by the presentation information extraction unit 370 are shown in FIGS. 34A and 34B. Figure 34A is an example setting that presents dishes that are highly preferred (favorite / relatively favorite) for all six months from April 2023 to September 2023, and is an example setting for extracting dishes that have been favorite for a long time. On the other hand, Figure 34B shows an example of settings for presenting dishes that had a below-average preference level from April to July, but whose preference level has recently increased (August and September). This corresponds to a dish that was previously eaten infrequently but has recently become more common, or a dish that was previously a food aversion and rarely eaten, but that the user has grown to like after trying it.

[0096] In step S3032, the presentation information extraction unit 370 extracts cluster IDs corresponding to the extracted theme for each period based on the theme set in step S3031, using the cluster IDs and preference levels estimated by the preference estimation unit 170, and extracts dishes that are highly related to the cluster IDs. The process and specific operations for extracting dishes that are highly related to a certain cluster ID are the same as those of the related dish extraction unit 280 in embodiment 2. It is also possible to extract related information such as images for the extracted dishes. The type and content of the related information are arbitrary.

[0097] In step S3033, the presentation information extraction unit 370 generates presentation reasons for the dishes based on the theme set in step S3031. For example, in the case of Fig. 34A, the presentation information extraction unit 370 generates a presentation reason such as "Presenting dishes that you have liked for the past six months." In the case of Fig. 34B, the presentation information extraction unit 370 generates a presentation reason such as "Presenting dishes that you liked only averagely six months ago but have liked more in the past two months." The format of the reasons for presentation is not limited to sentences, but may be bullet points. The expression and content are not limited to the above examples and are arbitrary. The number of dishes extracted by the presentation information extraction unit 370 may be one or more, and is arbitrary. The presentation information extraction unit 370 stores the processing result in the storage unit 110 and passes it to the presentation unit 380. In the third embodiment, the presentation information extraction unit 370 and the presentation unit 380 constitute an output unit that outputs information about dishes according to the user's preferences.

[0098] The presentation unit 380 outputs information to an external device communicatively connected to the information processing device 300. For example, when the presentation unit 380 receives information from the presentation information extraction unit 370, it presents (outputs) the received information using an output device such as a display or a speaker. For example, when the presentation unit 380 receives a dish and a reason for presentation from the presentation information extraction unit 370, it presents the extraction result of the dish and the reason for presentation on an output device such as a display or a speaker. The information presented by the presentation unit 380 may be video information of the dish, text information, audio information, or a combination of these.

[0099] In the third embodiment, an example of an operation in which the presentation information extracting unit 370 generates a presentation reason is shown, but a presentation reason may not be necessary, and only the extraction result of the dish may be presented.

[0100] Furthermore, in the third embodiment, an example of an operation in which the presentation unit 380 presents the name of a dish is shown, but the content to be presented is not limited to the name of a dish. It may be other information related to the name of a dish extracted by the presentation information extraction unit 370. Furthermore, the information is not limited to text, such as the name of a dish, but may be in other formats such as an image or sound.

[0101] The information processing device 300 according to the third embodiment can present information to the user in consideration of preferences for each period.

[0102] Furthermore, in Embodiment 3, the information extracted by the presentation information extraction unit is not limited to the above. For example, the presentation information extraction unit may be configured to extract or generate information indicating food items (cooked foods, semi-cooked foods, meal kits, retort foods, prepared meals, etc.) according to the user's preferences, information indicating restaurants that serve food items according to the user's preferences, information indicating grocery stores according to the user's preferences, information on food and drink according to the user's preferences, information on recipes for dishes according to the user's preferences, etc., to suggest to the user for use, based on the estimation result of the user's preferences estimated by preference estimation unit 170 and information stored in storage unit 110 (information on food items, restaurants, food stores, cooking recipes, date and time, calendar, climate, weather, etc.).

[0103] Furthermore, for example, the presentation information extraction unit may be configured to estimate nutrients that the user is lacking based on the name of the dish acquired by the data acquisition unit 140 and the recipe information stored in the memory unit 110, and to extract or generate information about food and drink that is in accordance with the user's preferences and that will supplement the user's lack of nutrients, such as information indicating food products that are in accordance with the user's preferences to supplement the user's lack of nutrients, information indicating restaurants that serve dishes that are in accordance with the user's preferences, information indicating retailers of food products that are in accordance with the user's preferences, and information regarding recipes for dishes that are in accordance with the user's preferences, based on the estimated results of the nutrients that the user is lacking, the estimated results of the user's preferences estimated by the preference estimation unit 170, and the information stored in the memory unit 110.

[0104] Furthermore, for example, the presentation information extraction unit may be configured to extract or generate information indicating restaurants that serve dishes that suit the user's preferences and information indicating stores that sell groceries that suit the user's preferences, to suggest to the user, based on information indicating a place name or region input by the user, the estimation results of the user's preferences estimated by the preference estimation unit 170, and the information stored in the memory unit 110.

[0105] Furthermore, for example, when extracting or generating information to be presented to the user, the presentation information extraction unit may also be configured to extract or generate information indicating a reason for presenting the information to the user. For example, when extracting information indicating the name of a dish to be presented to the user, the presentation information extraction unit may be configured to generate information indicating a reason for presenting the dish related to the name of the dish to be presented to the user based on information stored in the storage unit. Possible reasons for presenting a dish to the user include, for example, that the dish is a dish that suits the user's preferences, that it supplements nutrients that the user is lacking, that it prevents the user from consuming excessive nutrients, that it is a currently popular dish, that it is a dish that suits the climate or calendar at the time of presentation, etc. Possible forms for presenting the name of the dish and the reason for presenting it to the user include, for example, "Recommended dish: XX, user's preferences: XX, nutritional aspects: slightly low in dietary fiber / slightly high in fat," "The dish recommended for the user is XX. The reason is XX," etc.

[0106] The presentation information extraction unit may be configured to generate information indicating the reason for presenting the information to the user, for example, by using large-scale language models (LLMs). For example, the presentation information extraction unit may have a large-scale language model, provide the large-scale language model with information for generating information, such as an estimated result of a user's preferences, and cause the presentation unit to present the information generated by the large-scale language model. Note that the large-scale language model may be included in an external server communicably connected to the information processing device via a communication device.

[0107] As described above, the information processing device 300 according to the third embodiment includes an output unit that outputs information about dishes that match the user's preferences based on the estimation result by the preference estimation unit. With this configuration, the information processing device 300 can provide the user with information about dishes that match the user's preferences, such as the names of dishes that are recommended to the user, the names of restaurants that serve the dishes that are recommended to the user, and recipes for the dishes that are recommended to the user, based on the association between each cluster and a specific dish name.

[0108] In the first to third embodiments, the basic cluster construction unit 120 performs clustering using data in which the value of condiments used in each standard dish or seasoning is "1" and the value of condiments not used is "0", but the data used for clustering is not limited to data with values ​​of 0 or 1. The value can be any value. Furthermore, the information processing device may perform clustering using data in which any processing, such as dimension reduction or various conversion processing, has been performed on the feature quantities.

[0109] In the first to third embodiments, the clustering results obtained by the basic cluster constructor 120 are used as the basic clusters. However, the user may modify the clustering results and use the modified results as the basic clusters. The modification may involve, for example, moving some of the dishes classified in a certain cluster to another cluster (classifying them into a different cluster). The modification may be performed by editing the clustering result file, or by the user editing the clustering results via a user interface. The editing method is arbitrary. Furthermore, the basic cluster constructor 120 may generate a basic cluster consisting of multiple clusters as a result of processing. For example, the basic cluster constructor 120 may generate multiple clusters from a state where there are no clusters, or may generate a new cluster by adding it to an already generated basic cluster consisting of one or more clusters, or may update the information of an already generated basic cluster consisting of multiple clusters by adding new dishes and seasonings to the basic cluster.

[0110] In the first to third embodiments, an example of operation in which the basic cluster is made up of five clusters is shown, but the basic cluster may be made up of any number of clusters.

[0111] In the first to third embodiments, the dictionary creation unit 134 of the basic cluster analysis unit 130 calculates the importance of words using TF-IDF, but the method for calculating the importance of words may be a method other than TF-IDF, and any known method may be used.

[0112] In the first to third embodiments, the related cluster estimation unit 160 extracts words related to seasoning from the names of dishes, and calculates the sum of the importance values ​​for all the extracted words related to seasoning using the "importance of seasoning words" calculated by the seasoning feature extraction unit 131 of the basic cluster analysis unit 130, and uses this as the similarity, but the method for calculating the similarity related to seasoning is not limited to this and can be any method. For example, the similarity may be calculated by multiplying the sum of the importance values ​​by a weight.

[0113] In the first to third embodiments, the ratio is set to 0.8 in calculating the threshold for extracting highly related clusters in the related cluster estimation unit 160, but the value of the ratio is arbitrary. Also, although the threshold is determined using the ratio, the threshold may be input directly without using the ratio. Alternatively, a threshold may not be used, and only one cluster with the highest relevance score may be used as the final result.

[0114] In the first to third embodiments, the related cluster estimation unit 160 extracts candidates for highly related clusters based on the relevance score, and then compares the results with the results of grouping the basic clusters in the basic cluster classification unit 152 of the cuisine classification unit 150 to adjust the final result. However, it is also possible to determine highly related clusters (as the final result) based only on the relevance score. In this case, the device configuration is the same as that of the device in Fig. 2 except that the basic cluster classification unit 152 is removed.

[0115] As described above, by creating classification criteria (basic clusters) for dishes based on common dishes (standard dishes, common seasonings), the information processing device can estimate a user's preferences from the name of a dish even when there is no recipe information, which is a large amount of text information, or when recipe information is not used.

[0116] Embodiment 4 Next, an information processing device 400 according to embodiment 4 will be described with reference to Fig. 35 to Fig. 37. The information processing device 400 according to embodiment 4 differs from the information processing device 300 according to embodiment 3 in the configuration for performing analysis to clarify the nutrient intake status of the user and the configuration related to the content presented to the user, but the other configurations are the same, and the same names and symbols as those in embodiment 3 will be used and descriptions thereof will be omitted.

[0117] Fig. 35 is a block diagram showing a schematic configuration of an information processing device 400 according to embodiment 4. As shown in Fig. 35, the information processing device 400 according to embodiment 4 includes a storage unit 110, a basic cluster construction unit 120, a basic cluster analysis unit 130, a data acquisition unit 140, a dish classification unit 150, a related cluster estimation unit 160, a dish addition unit 270, a preference estimation unit 170, a nutritional analysis unit 460, a presentation information extraction unit 470, and a presentation unit 380, and is communicatively connected to a plurality of user terminals T1, T2, ..., TN. Note that the information processing device 400 may also be communicatively connected to the plurality of user terminals T1, T2, ..., TN via a communication network, server, base station, etc. (not shown).

[0118] The nutrition analysis unit 460 performs an analysis to clarify the user's nutrient intake status based on the information acquired by the data acquisition unit 140. For example, the nutrition analysis unit 460 analyzes information about dishes previously eaten by each user acquired by the data acquisition unit 140, and extracts nutrients that each user is deficient in from a plurality of preset nutrients. Furthermore, for example, the nutrition analysis unit 460 analyzes information about dishes previously eaten by each user acquired by the data acquisition unit 140, and extracts nutrients that each user is consuming in excess from a plurality of preset nutrients. For example, the nutrition analysis unit 460 compares the amount of nutrients contained in dishes previously eaten by the user with the recommended intake amounts of nutrients indicated in the "Dietary Reference Intakes for Japanese" published by the Ministry of Health, Labor and Welfare of Japan, acquired by the data acquisition unit 140, to extract nutrients that each user is consuming in excess from a plurality of preset nutrients. The nutritional analysis unit 460 may be configured to extract nutrients that each user is deficient in or ingesting in excess of, based on information other than the above-mentioned "Dietary Reference Intakes for Japanese" that indicates the amount of nutrients that each user is recommended to consume.

[0119] Furthermore, for example, the nutrition analysis unit 460 acquires information indicating the amount of nutrients contained in each of the plurality of dishes eaten by each user during a specific period in the past by analyzing information indicating the names of the plurality of dishes eaten by each user during a specific period in the past based on information regarding the amount of nutrients contained in each of the plurality of dishes acquired by the data acquisition unit 140, and extracts nutrients that each user is deficient in or has excessively consumed from the plurality of nutrients set in advance based on the acquired information indicating the amount of nutrients and the information indicating the amount of nutrients that each user is required to take, acquired by the data acquisition unit 140. Note that the nutrition analysis unit 460 may be configured to extract both the nutrients that each user is deficient in and the nutrients that each user has excessively consumed from the plurality of nutrients set in advance, or may be configured to extract either one of them.

[0120] The presentation information extraction unit 470 acquires the preference estimation results recorded and accumulated by the preference estimation unit 170 from the storage unit 110, and extracts and outputs information such as dishes to be presented to the user based on the preference estimation results and the analysis results of the nutritional analysis unit 460. For example, the presentation information extraction unit 470 acquires the preference estimation results recorded and accumulated by the preference estimation unit 170 from the storage unit 110, and outputs information on dishes that meet the preferences of each user and can supplement nutrients that are deficient in each user, together with information for identifying the user corresponding to the dishes. Also, for example, the presentation information extraction unit 470 acquires the preference estimation results recorded and accumulated by the preference estimation unit 170 from the storage unit 110, and outputs information on dishes that meet the preferences of each user and can reduce the intake of nutrients that each user is consuming in excess, together with information for identifying the user corresponding to the dishes. The presentation information extraction unit 470 may be configured to output both information on dishes that can supplement nutrients that are deficient in each user and information on dishes that can curb the intake of nutrients that each user is taking in excessively, or may be configured to output only one of them. Furthermore, the presentation information extraction unit 470 may be configured to output, together with the information on the extracted dishes, information on nutrients that can be supplemented by the extracted dishes, information on nutrients whose intake can be curbed by the extracted dishes, and information indicating the relationship between the extracted dishes and the preferences of each user.

[0121] FIG. 36 is a flowchart showing an example of processing performed by the information processing device 400 according to the fourth embodiment. For example, as shown in FIG. 36, when the information processing device 400 starts processing, the nutrition analysis unit 460 acquires information about nutrients in food (step S3601). In this processing, the nutrition analysis unit 460 acquires information indicating the amount of nutrients recommended for intake by a specific user and information indicating the amount of nutrients contained in dishes eaten by this user during a specific period in the past. After the nutrition analysis unit 460 performs the processing of step S3601, the nutrition analysis unit 460 extracts nutrients that are lacking in this user by analysis (step S3602). In this processing, the nutrition analysis unit 460 analyzes the information acquired in the processing of step S3601 to extract nutrients that are lacking in the specific user from a plurality of preset nutrients.

[0122] After the nutrition analysis unit 460 performs the process of step S3602, the presentation information extraction unit 470 extracts dishes that can supplement the nutrients that are deficient in the user (step S3603). In this process, the presentation information extraction unit 470 extracts dishes that can supplement the nutrients that are deficient in the specific user from the plurality of dishes based on information about the amounts of nutrients contained in each of the plurality of dishes acquired by the data acquisition unit 140, and outputs information about the extracted dishes. As described above, in the process of step S3602, the nutrition analysis unit 460 may extract nutrients that the user is consuming in excess instead of or in addition to the nutrients that are deficient in the user. In the process of step S3603, the presentation information extraction unit 470 may extract dishes that can suppress the user's intake of nutrients that are deficient in excess instead of or in addition to the dishes that can supplement the nutrients that are deficient in the user. The presentation unit 380 presents the information extracted by the presentation information extraction unit 470 to the user terminal corresponding to this user among the user terminals T1 to TN.

[0123] 37 is a diagram illustrating an example of information presented on a user terminal by the presentation unit 380 of the information processing device 400 according to Embodiment 4. As illustrated in FIG. 37, the user terminal displays, for example, images and names of dishes extracted by the presentation information extraction unit 470, amounts of nutrients contained in the dishes extracted by the presentation information extraction unit 470, information on nutrients that can be supplemented by the dishes extracted by the presentation information extraction unit 470, and information indicating a relationship between the dishes extracted by the presentation information extraction unit 470 and user preferences. For example, the information indicating the relationship between the dishes extracted by the presentation information extraction unit 470 and user preferences displays information indicating that the dishes match the user preferences, specifically, a string of characters such as "This dish is recommended for Mr. / Ms. X who likes stir-fries."

[0124] As described above, the information processing device 400 according to the fourth embodiment includes a nutritional analysis unit that acquires information about dishes consumed by a user during a specific period in the past, analyzes the information about the dishes consumed by the user, and extracts at least one of a nutrient that the user is deficient in and a nutrient that the user is consuming in excess from a plurality of preset nutrients, and outputs information about dishes to be recommended to the user based on the user's preferences estimated by the preference estimation unit and at least one of a nutrient that the user is deficient in and a nutrient that the user is consuming in excess from a plurality of preset nutrients. Thus configured, the information processing device 400 can provide the user with information about dishes that match the user's preferences and their nutrient intake status.

[0125] Fifth embodiment Next, an information processing device 500 according to embodiment 5 will be described with reference to Fig. 38 to Fig. 40. The information processing device 500 according to embodiment 5 differs from the information processing device 300 according to embodiment 3 in the configuration for performing analysis to clarify the nutrient intake status of the user and the configuration related to the content presented to the user, but the other configurations are the same, and the same names and symbols as those in embodiment 3 will be used and descriptions thereof will be omitted.

[0126] 38 is a block diagram showing a schematic configuration of an information processing device 500 according to embodiment 5. As shown in FIG. 38, the information processing device 500 according to embodiment 5 includes a storage unit 110, a basic cluster construction unit 120, a basic cluster analysis unit 130, a data acquisition unit 140, a dish classification unit 150, a related cluster estimation unit 160, a preference estimation unit 170, a related dish extraction unit 280, a command information generation unit 570, and a presentation unit 380, and is communicatively connected to an LLM (Large Language Models) server L1 and multiple user terminals T1, T2, . . . , TN. Note that the information processing device 500 may be communicatively connected to the LLM server L1 and multiple user terminals T1, T2, . . . , TN, which are external devices, via a communication network, server, base station, etc. (not shown), or may be communicatively connected to the multiple user terminals T1, T2, . . . , TN via the LLM server L1.

[0127] The instruction information generation unit 570 generates information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate information, based on the user preferences estimated by the preference estimation unit 170. For example, the instruction information generation unit 570 generates text information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate information about dishes that suit each user's preferences, based on each user's preferences estimated by the preference estimation unit 170. For example, the instruction information generation unit 570 generates text information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate recommended dishes that suit each user's preferences, characteristics of recommended dishes that suit each user's preferences, reasons why recommended dishes that suit each user's preferences are recommended, recipes for recommended dishes that suit each user's preferences, etc. In addition, when performing the above processing, the instruction information generation unit 570 may be configured to generate information to cause the LLM server L1 to generate information about food items (cooked items, semi-cooked items, meal kits, retort foods, prepared meals, etc.) instead of or in addition to information about the dishes.

[0128] The presentation unit 380 outputs information to external devices such as the LLM server L1 and user terminals T1 to TN that are communicably connected to the information processing device 500.

[0129] FIG. 39 is a flowchart showing an example of processing performed by the information processing device 500. For example, as shown in FIG. 39, when the information processing device 500 starts processing, the instruction information generation unit 570 generates information (instruction information) indicating the content of instructions to be given to the LLM server L1 (step S3901). For example, in this processing, the instruction information generation unit 570 generates text information such as "Please generate a delicious image of XX (the name of the dish that matches the user's taste)" as instruction information for the LLM server L1 based on the name of a dish that matches the taste of a specific user extracted by the related dish extraction unit 280. Also, for example, in this processing, the instruction information generation unit 570 generates text information such as "Please tell me a recipe that allows me to easily make XX (the name of the dish that matches the user's taste)" as instruction information for the LLM server L1 based on the name of a dish that matches the taste of a specific user estimated by the related dish extraction unit 280.

[0130] When the instruction information generation unit 570 performs the process of step S3902, the presentation unit 380 outputs the instruction information generated by the instruction information generation unit 570 to the LLM server L1 (step S3902). As a result, information about dishes based on the preferences of the specific user is generated by the LLM server L1.

[0131] When the presentation unit 380 performs the process of step S3903, the data acquisition unit 140 acquires information generated by the LLM server L1 (step S3903). For example, the data acquisition unit 140 acquires, from the LLM server L1, information about recommended dishes that match the preferences of a specific user, which information was generated by the LLM server L1. Specifically, the data acquisition unit 140 acquires, from the LLM server, an image (one or more images) of the name of a dish that matches the user's preferences and information about recipes (one or more images) that can be made quickly and easily, among recipes related to the user's preferences. When the data acquisition unit 140 acquires the information from the LLM server L1, the presentation unit 380 presents some or all of the acquired information on the user terminal corresponding to the user.

[0132] FIG. 40 is a diagram showing an example of information presented on a user terminal by the presentation unit 380 of the information processing device 500 according to embodiment 5. As shown in FIG. 40, the user terminal displays, for example, the name of the dish generated by the LLM server L1 and information indicating the characteristics of the dish generated by the LLM server L1. Note that the information processing device 500 may be configured to cause a generation AI server, which is an external device capable of generating text information and image information, to generate information instead of the LLM server L1. In such a case, the information processing device 500 may be configured to display an image showing the dish generated by the LLM server L1 in addition to a character string indicating the characteristics of the dish, as shown in FIG. 40.

[0133] As described above, the information processing device 500 according to the fifth embodiment includes an instruction information generation unit that generates information indicating instructions to the generation AI server for causing the generation AI server to generate information related to dishes based on the user's preferences estimated by the preference estimation unit, and an output unit that outputs information generated by the generation AI server based on the information generated by the instruction information generation unit. Thus configured, the information processing device 500 can provide the user with detailed dish information based on the user's preferences estimated by the preference estimation unit by utilizing the information generated by the generation AI.

[0134] Note that the information processing device 500 is not limited to the configuration described above, and may include, for example, one or more of the dish adding unit 270, the nutrition analysis unit 460, and the presentation information extraction unit 470 included in the information processing device 400 according to embodiment 4. For example, with this configuration, the information processing device can optimize the information presented to the user by having the instruction information generation unit 570 generate information indicating instructions to the LLM server L1 based on the information about the dishes added by the dish adding unit 270, having the instruction information generation unit 570 generate information indicating instructions to the LLM server L1 based on the analysis results by the nutrition analysis unit 460, and having the presentation information extraction unit 470 select and process the information obtained from the LLM server L1.

[0135] Specifically, when generating information indicating instructions to the LLM server L1 based on information about dishes added by the dish adding unit 270, the instruction information generating unit 570 can include information about the names of dishes that correspond to the user's preferences from among the dishes included in the basic cluster for the LLM server L1 and the dishes added by the dish adding unit 270 in the instruction information for the LLM server L1, thereby improving the accuracy of the information generated by the LLM server L1. Also, specifically, when generating information indicating instructions to the LLM server L1 based on the analysis results by the nutrition analysis unit 460, the instruction information generating unit 570 can include information indicating nutrients that a specific user is deficient in or has consumed in excess in the instruction information for the LLM server L1, thereby enabling the LLM server L1 to generate information that takes into account the nutrient intake status of this user. Specifically, when the amount of information obtained from the LLM server L1 is too large, the presentation information extraction unit 470 can select and discard the information to generate information to be presented to the user; when there is missing information in the information obtained from the LLM server L1, it can add information stored in the memory unit 110 to this information to generate information to be presented to the user; and it can change the layout of the information to present the information obtained from the LLM server L1 on the user terminal in an easy-to-read manner.

[0136] Sixth embodiment Next, an information processing device 600 according to the sixth embodiment will be described with reference to Fig. 41 and Fig. 42. The information processing device 600 according to the sixth embodiment differs from the information processing device 500 according to the fifth embodiment in the content of the instruction given to the LLM server by the instruction information generating unit, but other features are the same, and the same configuration as in the fifth embodiment will be given the same names and symbols as in the fifth embodiment, and description thereof will be omitted.

[0137] 41 is a block diagram showing a schematic configuration of an information processing device 600 according to embodiment 6. As shown in FIG. 41, the information processing device 600 according to embodiment 6 includes a storage unit 110, a basic cluster construction unit 120, a basic cluster analysis unit 130, a data acquisition unit 140, a cuisine classification unit 150, a related cluster estimation unit 160, a preference estimation unit 170, a command information generation unit 670, and a presentation unit 380, and is communicatively connected to an LLM server L1 and a plurality of user terminals T1, T2, . . . , TN. Note that the information processing device 600 may be communicatively connected to the LLM server L1 and the plurality of user terminals T1, T2, . . . , TN via a communication network, server, base station, etc. (not shown), or may be communicatively connected to the plurality of user terminals T1, T2, . . . , TN via the LLM server L1.

[0138] The instruction information generation unit 670 generates information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate information based on the user preferences estimated by the preference estimation unit 170. For example, the instruction information generation unit 670 generates text information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate information about restaurants that can provide dishes that suit each user's preferences, based on each user's preferences estimated by the preference estimation unit 170. For example, the instruction information generation unit 670 generates text information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate information about restaurants that can provide dishes that suit each user's preferences, such as names of dishes that suit each user's preferences, names of restaurants that can provide dishes that suit each user's preferences, business information about restaurants that can provide dishes that suit each user's preferences, contact information for restaurants that can provide dishes that suit each user's preferences, and addresses of restaurants that can provide dishes that suit each user's preferences. In addition, when performing the above processing, the instruction information generation unit 670 may be configured to generate information to cause the LLM server L1 to generate information regarding stores where food items (cooked items, semi-cooked items, meal kits, retort foods, prepared meals, etc.) according to the user's preferences can be purchased, instead of or in addition to information regarding the food.

[0139] FIG. 42 is a diagram illustrating an example of information presented on a user terminal by the presentation unit 380 of the information processing device 600 according to Embodiment 6. As illustrated in FIG. 42, the user terminal displays, for example, information generated by the LLM server L1, such as the names of restaurants that can provide dishes tailored to each user's preferences, the names of dishes tailored to each user's preferences, business information for restaurants that can provide dishes tailored to each user's preferences, and addresses of restaurants that can provide dishes tailored to each user's preferences. For example, the user terminal displays information indicating business hours, regular holidays, and other recommended dishes for restaurants that can provide dishes tailored to each user's preferences, as business information for restaurants that can provide dishes tailored to each user's preferences. Note that the information processing device 600 may be configured to cause a generation AI server capable of generating text information and image information to generate information, instead of the LLM server L1. In such a case, the information processing device 600 may be configured to display, in addition to a string of characters indicating information about the restaurant, images such as an image showing a dish according to each user's preferences, an image of the interior of a restaurant where the user can receive a dish according to the user's preferences, and an image of the exterior of a restaurant where the user can receive a dish according to the user's preferences, as shown in FIG. 42.

[0140] Furthermore, for example, the information processing device 600 can provide the user with more useful restaurant information by having the data acquisition unit 140 acquire information about the conditions under which the user selects a restaurant or a dish from the user terminal or an external device communicatively connected to the information processing device 600. Specifically, the instruction information generation unit 670 may be configured to generate information for causing the LLM server L1 to generate information about restaurants that meet the conditions under which the user selects a restaurant or a dish, based on the information about the conditions under which the user selects a restaurant or a dish acquired by the data acquisition unit 140. More specifically, information about the conditions under which the user selects a restaurant or a dish can be, for example, information about the location of the user terminal, the location to which the user plans to go, the location where the user is searching for a restaurant, the type of food the user desires to eat at the current or future time, the type of restaurant the user desires to eat at the current or future time, and the season, climate, weather, or temperature at the time the user desires to eat. Note that information about the season, climate, weather, or temperature at the time the user desires to eat can influence the user's choice of restaurant, and can therefore be considered information about the conditions under which the user selects a restaurant. The data acquisition unit 140 may be configured to acquire information regarding the conditions when a user selects a restaurant or dish as text information, sensing information from various sensors, audio information, or other information in various other forms.

[0141] As described above, the information processing device 600 according to the sixth embodiment generates information indicating instructions to the generation AI server for causing the generation AI server to generate information about restaurants that can provide dishes according to the user's preferences, based on the user's preferences estimated by the preference estimation unit. With this configuration, the information processing device 600 can reduce the effort required for the user to select a restaurant that can provide dishes according to the user's preferences, and can improve the user's satisfaction when eating and drinking at a restaurant.

[0142] Embodiment 7 Next, an information processing device 700 according to embodiment 7 will be described with reference to Fig. 43 and Fig. 44. The information processing device 700 according to embodiment 7 differs from the information processing device 600 according to embodiment 6 in that it includes a dish adding unit 270 and the contents of the instructions given to the LLM server by the instruction information generating unit, but other features are the same, and the same components as those in embodiment 6 are given the same names and symbols as those in embodiment 6 and descriptions thereof will be omitted.

[0143] 43 is a block diagram showing a schematic configuration of an information processing device 700 according to embodiment 7. As shown in FIG. 43, the information processing device 700 according to embodiment 7 includes a storage unit 110, a basic cluster construction unit 120, a basic cluster analysis unit 130, a data acquisition unit 140, a dish classification unit 150, a related cluster estimation unit 160, a dish addition unit 270, a preference estimation unit 170, a command information generation unit 770, and a presentation unit 380, and is communicatively connected to an LLM server L1 and a plurality of user terminals T1, T2, . . . , TN. Note that the information processing device 700 may be communicatively connected to the LLM server L1 and the plurality of user terminals T1, T2, . . . , TN via a communication network, server, base station, etc. (not shown), or may be communicatively connected to the plurality of user terminals T1, T2, . . . , TN via the LLM server L1.

[0144] The instruction information generation unit 770 generates information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate information, based on the user preferences estimated by the preference estimation unit 170 and information on dishes eaten in the past by each user acquired by the data acquisition unit 140. For example, the instruction information generation unit 770 generates text information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate information on dishes according to each user's preferences and their nutrient intake status, based on the user's preferences estimated by the preference estimation unit 170 and information indicating the names of dishes eaten by each user during a specific period in the past acquired by the data acquisition unit 140. For example, the instruction information generation unit 770 generates text information indicating instructions to the LLM server L1 for causing the LLM server L1 to generate information on dishes that are recommended to each user, based on each user's preferences and taking into account nutrients that each user is deficient in and nutrients that each user is excessively consuming. Specifically, the instruction information generation unit 770 generates text information indicating instructions to the LLM server L1 to cause the LLM server L1 to generate information indicating the name of a dish recommended for the user to eat or drink, the reason for recommending that the user eat or drink that dish, and the nutrients the user is lacking and the nutrients the user is consuming in excess, in accordance with the preferences of a specific user and taking into consideration the nutrients the user is lacking and the nutrients the user is consuming in excess.

[0145] More specifically, the instruction information generation unit 770 generates text information such as "Please tell me what dishes are recommended for an adult male who ate a grilled saury set meal yesterday morning, tonkotsu ramen yesterday lunchtime, and curry rice last night, and who prefers soy sauce and mirin seasonings, as well as information on nutrients that this person is deficient in and ingesting in excess, and the reason why the dishes are recommended," as instruction information for the LLM server L1, based on the preferences of the specific user estimated by the preference estimation unit 170 and information on dishes that each user has previously eaten and drunk, acquired by the data acquisition unit 140. Note that the instruction information generation unit 770 may be configured to generate, in addition to the above information, instruction information for causing the LLM server L1 to generate other information on the dishes recommended for the specific user estimated by the preference estimation unit 170, such as information on restaurants that can provide the dishes recommended for the specific user estimated by the preference estimation unit 170, as instruction information for the LLM server L1.

[0146] FIG. 44 is a diagram showing an example of information presented to a user terminal by the presentation unit 380 of the information processing device 700 according to embodiment 7. As shown in FIG. 44, the user terminal displays, for example, information generated by the LLM server L1, indicating the names of dishes recommended to the user and the reasons for recommending the dishes. Note that the information processing device 700 may be configured to cause a generation AI server capable of generating text information and image information to generate information, instead of the LLM server L1. In such a case, the information processing device 700 may be configured to display an image showing the recommended dish in addition to a character string indicating information about the recommended dish, as shown in FIG. 44.

[0147] The instruction information generation unit 770 may be configured to generate instruction information for the LLM server L1 based on the preferences of a specific user estimated by the preference estimation unit 170, information about dishes previously eaten by each user acquired by the data acquisition unit 140, and dishes added by the dish addition unit 270. By generating instruction information for the LLM server L1 based on the dishes added by the dish addition unit 270, the instruction information generation unit 770 can present the user with information about dishes that match the user's preferences with higher accuracy. When performing the above processing, the instruction information generation unit 670 may be configured to generate information for causing the LLM server L1 to generate information about food items (cooked dishes, semi-cooked dishes, meal kits, retort foods, prepared meals, etc.) that match the user's preferences, instead of or in addition to information about dishes.

[0148] As described above, the information processing device 700 according to the seventh embodiment generates information indicating instructions to the generation AI server for causing the generation AI server to generate information about dishes recommended to the user and information about reasons for recommending the dishes to the user, based on the user's preferences estimated by the preference estimation unit. With this configuration, the information processing device 700 can reduce the effort required for the user to select dishes to eat or drink. Furthermore, for example, when presenting dishes to the user based on the user's nutrient intake status, the information processing device 700 can present the user with dishes that are in line with the user's preferences and that can maintain or improve the user's health.

[0149] In any of the above-described embodiments, part of the configuration and functions of the information processing device may be provided by another device that is connected to the information processing device so as to be able to communicate with it, such as an external server.

[0150] In any of the above-described embodiments, part of the configuration and functions of the information processing device may be provided in another device that is communicably connected to the information processing device via a communication device.

[0151] In addition, the present disclosure allows for free combination of the respective embodiments, modification of any of the components of the respective embodiments, or omission of any of the components of the respective embodiments. [Explanation of symbols]

[0152] 100 Information processing device, 100a Processor, 100b Memory, 100c Communication device, 100d Input device, 100e Output device, 110 Storage unit, 120 Basic cluster construction unit (cluster generation unit, cluster information acquisition unit), 130 Basic cluster analysis unit, 131 Seasoning feature extraction unit, 132 Cooking method feature extraction unit, 133 Character type feature extraction unit, 134 Dictionary creation unit, 140 Data acquisition unit, 150 Cuisine classification unit, 151 Feature generation unit, 152 Basic cluster classification unit, 160 Related cluster estimation unit, 170 Preference estimation unit, 200 Information processing device, 270 Cuisine addition unit (related information generation unit), 280 Related cuisine extraction unit, 300 Information processing device, 370 Presentation information extraction unit (output unit), 380 Presentation unit (output unit), 400 Information processing device, 460 Nutrition analysis unit, 470 presentation information extraction unit, 500 information processing device, 570 instruction information generation unit, 600 information processing device, 670 instruction information generation unit, 700 information processing device, 770 instruction information generation unit.

Claims

1. a cluster generating unit that acquires information indicating a plurality of dish names and information indicating condiments included in each of the plurality of dish names, classifies the plurality of dish names based on the information indicating the condiments, and generates a first number of clusters based on the classification of the plurality of dish names; a data acquisition unit that acquires information indicating a specific dish name; a related cluster estimation unit that estimates a cluster related to the specific dish name from among the first number of clusters based on information extracted from the specific dish name.

1. An information processing device comprising:

2. The cluster generation unit acquires information indicating the plurality of dish names and the plurality of seasoning names and information indicating the seasonings included in each of the plurality of dish names and the plurality of seasoning names, and generates the first number of clusters into which the plurality of dish names and the plurality of seasoning names are classified based on the information indicating the seasonings.

2. The information processing apparatus according to claim 1, wherein:

3. The related cluster estimation unit estimates a cluster related to the specific dish name from among the first number of clusters based on the information indicating the seasoning and words related to the seasoning used in the specific dish name.

2. The information processing apparatus according to claim 1, wherein:

4. The related cluster estimation unit estimates a cluster related to the specific dish name from among the first number of clusters based on information about dish names included in each of the first number of clusters and characters used in the specific dish name.

2. The information processing apparatus according to claim 1, wherein:

5. The related cluster estimation unit estimates a cluster related to a specific dish name from among the first number of clusters based on information about dish names included in each of the first number of clusters and words indicating cooking methods used for the specific dish name.

5. The information processing apparatus according to claim 4.

6. The related cluster estimation unit estimates a cluster related to the specific dish name from among the first number of clusters based on information indicating the dish names included in each of the first number of clusters and the ratio of character types used in the specific dish name.

5. The information processing apparatus according to claim 4.

7. The related cluster estimation unit estimates a cluster related to the specific dish name from among the first number of clusters based on information indicating the importance of each word used in the specific dish name and the dish names included in each of the first number of clusters.

5. The information processing apparatus according to claim 4.

8. a cuisine classifier that classifies the first number of clusters generated by the cluster generator into a second number of clusters that is smaller than the first number; the related cluster estimation unit calculates a relatedness score indicating a degree of relatedness between each of the first number of clusters and the specific dish name based on information extracted from the specific dish name; The related cluster estimation unit estimates a cluster related to the specific dish name from among the first number of clusters based on the relevance scores of each of the first number of clusters and the difference in relevance scores between a plurality of clusters included in each of the second number of clusters classified by the dish classification unit.

2. The information processing apparatus according to claim 1, wherein:

9. The dish classification unit classifies the first number of clusters generated by the cluster generation unit into a second number of clusters that is smaller than the first number, based on the information indicating the seasonings.

9. The information processing apparatus according to claim 8,

10. The dish classification unit classifies the first number of clusters generated by the cluster generation unit into a second number of clusters that is smaller than the first number, based on information about characters used in the dish names included in each of the first number of clusters.

9. The information processing apparatus according to claim 8,

11. a preference estimation unit that estimates a user's preference for each cluster using the estimation result of the cluster related to the specific dish name estimated by the related cluster estimation unit, and estimates the user's preference for dishes; 11. The information processing apparatus according to claim 1, wherein the information processing apparatus is a computer.

12. an output unit that outputs information about dishes according to the user's preferences based on the estimation result by the preference estimation unit; 12. The information processing apparatus according to claim 11.

13. an associated information generation unit that generates information indicating an association with any one of the first number of clusters estimated to be associated with the specific dish name based on an estimation result by the associated cluster estimation unit; a storage unit that stores the specific dish name and any one of the first number of clusters in association with each other based on the information generated by the association information generation unit.

2. The information processing apparatus according to claim 1, wherein:

14. a nutritional analysis unit that acquires information about a dish consumed by a user and analyzes the information about the dish consumed by the user to extract at least one of nutrients that the user is deficient in and nutrients that the user is consuming in excess; The output unit outputs information about a dish to be suggested to the user based on the preference of the user estimated by the preference estimation unit and at least one of a nutrient that the user is deficient in and a nutrient that the user is consuming in excess.

13. The information processing apparatus according to claim 12.

15. an instruction information generation unit that generates first instruction information indicating an instruction content for generating information about dishes based on the user's preference estimated by the preference estimation unit; The output unit outputs information generated by an external device based on the first instruction information.

13. The information processing apparatus according to claim 12.

16. an instruction information generation unit that generates second instruction information indicating instruction content for generating information about restaurants that can provide dishes based on the user's preferences, based on the user's preferences estimated by the preference estimation unit; The output unit outputs information generated by an external device based on the second instruction information.

13. The information processing apparatus according to claim 12.

17. an instruction information generation unit that generates third instruction information indicating instruction content for generating information on dishes recommended to the user and information on reasons for recommending the dishes, based on the user's preference estimated by the preference estimation unit; The output unit outputs information generated by an external device based on the third instruction information.

13. The information processing apparatus according to claim 12.

18. On the computer, a process of acquiring information indicating a plurality of dish names and information indicating condiments included in each of the plurality of dish names, classifying the plurality of dish names based on the information indicating the condiments, and generating a first number of clusters based on the classification of the plurality of dish names; A process of obtaining information indicating a specific dish name; and a process of estimating a cluster associated with the specific dish name from among the first number of clusters based on information extracted from the specific dish name.

19. An information processing method performed by an information processing device including a cluster generation unit, a data acquisition unit, and a related cluster estimation unit, the cluster generation unit acquires information indicating a plurality of dish names and information indicating condiments included in each of the plurality of dish names, classifies the plurality of dish names based on the information indicating the condiments, and generates a first number of clusters based on the classification of the plurality of dish names; The data acquisition unit acquires information indicating a specific dish name; the related cluster estimation unit estimating a cluster related to the specific dish name from among the first number of clusters based on information extracted from the specific dish name.

1. An information processing method comprising:

20. a cluster information acquisition unit that acquires information indicating a plurality of clusters into which a plurality of dish names are classified; a data acquisition unit that acquires information indicating a specific dish name; and a related cluster estimation unit that estimates a cluster among the plurality of clusters that is related to the specific dish name based on feature amounts of the plurality of clusters that are based on word information extracted from the dish names included in each of the plurality of clusters, word information extracted from the specific dish name, and information indicating seasonings used in the dishes of the dish names included in each of the plurality of clusters.

1. An information processing device comprising:

Images