JP6918695B2 - Information processing equipment and programs - Google Patents

Description

The present invention relates to an information processing device and a program.

Software is known that analyzes a user's eating tendency by recording the name of the meal that the user eats and drinks and the image of the meal every day.

In such software, for example, statistical information (such as frequency of appearance) for each meal name in a plurality of users is generated and used for analysis processing of meal tendency.

Japanese Unexamined Patent Publication No. 2013-0106036

However, if the meal name is in the free input format, the meal name may be freely assigned to each user even if the meal is the same. For example, even if the same "stir-fried meat and vegetables" is used, there may be variations such as "stir-fried pork and vegetables", "stir-fried meat and Chinese-style vegetables", and "delicious stir-fried meat and vegetables". On the other hand, if the selection input format is used, the meal name that the user considers appropriate may not always be included in the options.

The present invention has been made in view of the above circumstances, and provides an information processing device capable of statistical processing related to a meal name while allowing a user to freely input a meal name. It is one of the purposes.

In addition, Patent Document 1 describes a processing example of integrating item character strings using a thesaurus.

The present invention for solving the problem of the above-mentioned conventional example is an information processing apparatus connected to a database storing a record including a meal name, and targets the meal name to be processed among the meal names. A means for dividing into words as a meal name, a generating means for generating a combined word by combining words selected from the divided words, and a means for subjecting the combined word as a normalized meal name to a predetermined process. It was decided to include it.

According to the present invention, it is possible to perform statistical processing related to a meal name while allowing the user to freely input a meal name.

It is a block diagram which shows the configuration example of the information processing apparatus which concerns on embodiment of this invention, and the connection example thereof. It is a functional block diagram which shows the example of the information processing apparatus which concerns on embodiment of this invention. It is explanatory drawing which shows the content example of the meal database which the information processing apparatus which concerns on embodiment of this invention accesses. It is explanatory drawing which shows the example of the word graph generated by the information processing apparatus which concerns on embodiment of this invention. It is a flowchart which shows the processing example of the information processing apparatus which concerns on embodiment of this invention.

Embodiments of the present invention will be described with reference to the drawings. As illustrated in FIG. 1, the information processing device 1 according to the embodiment of the present invention includes a control unit 11, a storage unit 12, and a communication unit 13. Further, the information processing device 1 is communicably connected to the user's terminal 2 via a communication means such as the Internet.

The control unit 11 of the information processing device 1 is a program control device such as a CPU, and operates according to a program stored in the storage unit 12. In the present embodiment, the control unit 11 records a meal name, a meal image, and the like in a meal database according to an instruction received from the user's terminal 2. Further, the control unit 11 sends the information recorded in the meal database to the terminal 2 in accordance with the instruction received from the user's terminal 2.

Further, in the present embodiment, the control unit 11 refers to a meal database storing a record including a meal name, and among the meal names recorded in the meal database, the meal name to be processed is selected as the target meal name. As a result, this target meal name is divided into words. The control unit 11 generates a combination word that combines words selected from the divided words, and uses the combination word as a normalized meal name for predetermined processing. The specific contents of the processing of the control unit 11 will be described later.

The storage unit 12 is a memory device, a disk device, or the like, and stores a program executed by the control unit 11. This program is provided, for example, stored in a computer-readable and non-transitory recording medium, and is stored in the storage unit 12. Further, in an example of the present embodiment, a meal database in which meal names, meal images, and the like are recorded is stored in the storage unit 12. The storage unit 12 also operates as a work memory of the control unit 11.

The communication unit 13 is a network interface or the like, and outputs information received from the terminal 2 via the network to the control unit 11. Further, the communication unit 13 sends information to the terminal 2 according to an instruction input from the control unit 11.

The terminal 2 is, for example, a smartphone or the like, which accepts the user's operation and identifies the user with information such as meal name information and a meal image (which may be taken by a camera or the like provided on the smartphone). It is sent to the information processing device 1 together with the information to be processed.

Next, the operation of the control unit 11 of the information processing device 1 of the present embodiment will be described. In an example of this embodiment, the control unit 11 functionally includes a meal information recording unit 31, a meal name normalization unit 32, and a statistical information generation unit 33, as illustrated in FIG. It is composed. Further, the meal name normalization unit 32 includes a meal name division unit 41 and a combination word generation unit 42.

When the meal information recording unit 31 receives the information such as the meal name information and the meal image (which may be taken by a camera provided by the smartphone) as well as the information for identifying the user from the terminal 2, the acceptance is received. The date and time information is acquired from a calendar IC or the like (not shown). Then, as illustrated in FIG. 3, the meal information recording unit 31 associates the received information for identifying the user, the information on the meal name and the amount, and the information on the acquired date and time with each other (a record ( R) is stored in the meal database.

The meal name normalization unit 32 generates a normalized meal name by referring to the meal name information stored in the meal database. Specifically, the meal name division unit 41 of the meal name normalization unit 32 selects, among the meal names stored in the meal database, a meal name for which a normalized meal name has not yet been generated as a processing target. .. The meal name selected here is referred to as a target meal name for the sake of distinction below.

Then, the meal name division unit 41 divides the target meal name for each word. This division can be performed, for example, by referring to a dictionary prepared in advance and using a process of analyzing word boundaries for each word in the dictionary. As an example, the Kyoto text analysis toolkit KyTea can be used. As a specific example, according to this meal name division 41, "stir-fried pork and vegetables", "stir-fried meat and Chinese-style vegetables", and "delicious stir-fried meat and vegetables" are respectively.
"Pork""to""vegetables""no""stir-fried""things",
"Meat""to""Chinesestyle""no""vegetables""no""stir-fried""things",
"Meat""Vegetables""No""Delicious""Stir-fried""Things"
It is divided like this. Here, the words obtained by division are distinguished by quotation marks (brackets).

The combination word generation unit 42 searches the meal database for a meal name including any of the divided words obtained by the meal name partitioning unit 41. The meal name obtained by this search is referred to as a reference meal name for the sake of distinction below. The reference meal name may include the target meal name itself. It should be noted that, among the divided words, a predetermined stop word (particle or the like) may be removed, and the reference meal name may be searched using the word excluding the stop word as a key.

As an example, if the target meal name is "delicious stir-fried pork and vegetables" and the target meal name is divided into "pork", "to", "vegetables", "no", "delicious", "stir-fried", and "things". Search the meal database for meal names that include any of the stop words "to" and "no", "pork", "vegetables", "delicious", "stir-fried", and "things". The name of the meal found by is used as the reference meal name.

This allows, for example,
"Delicious stir-fried pork and vegetables" (target meal name itself)
"Sautéed pork" (including the word "pork")
"Stir-fried vegetables" (including the words "vegetables" and "stir-fried")
"Stir-fried pork and Chinese-style vegetables" (including the words "pork", "vegetables", and "stir-fried")
"Delicious protein drink" (including the word "delicious")
"Stir-fried meat and vegetables" (including the word "stir-fried")
...
Meal names such as are found as reference meal names.

The combination word generation unit 42 divides the reference meal name into words by the meal name dividing unit 41 in the same manner as the target meal name. Further, the combination word generation unit 42 generates a combination word for each of the reference meal names by using the information of the order of appearance in the reference meal name before the division of the divided words (the stop word may be excluded). do.

As an example, the combination word generation unit 42 connects the divided words for each of the reference meal names in the order of appearance in the reference meal name with directional edges (starting node S and ending node E). Is added. Hereinafter, it is called a word graph). In the above example, the reference meal name "delicious stir-fried pork and vegetables", which is the target meal name,
A word graph of S → "pork" → "vegetables" → "delicious" → "stir fry" → "things" → E is generated, and for reference meal names other than the target meal name,
"Sautéed pork": S → "pork" → "sautéed" → E
"Stir-fried vegetables": S → "Vegetables" → "Stir-fried" → E
"Stir-fried pork and Chinese-style vegetables": S->"pork"->"Chinese-style"->"vegetables"->"stir-fried"->"things"-> E
"Delicious protein drink": S → "Delicious" → "Protein" → "Beverage" → E
"Stir-fried meat and vegetables": S → "meat" → "vegetables" → "stir-fried" → E
Word graph is generated (Fig. 4).

ここで、freq(w)は、ノードｗの出現頻度（各参照食事名からノードｗが見出された頻度）であり、diff(ｓ，ｗ1，ｗ2）は、出現順を考慮した語間距離を表し、具体的には次の（２）式で表される値となる：

ただしpos（ｓ，ｗ）は、食事名ｓにおけるノードｗの出現順位（先頭からリンクを辿って何番目にノードｗが出現するかを表す値）であり、Σ（和）は、すべての参照食事名における和を演算することを意味する。The combination word generation unit 42 has a link source node (called an outlink node) wi of the edge (referred to as a processing target edge) and a link destination node (referred to as an outlink node) for each edge included in the graph generated for each meal name. Refer to wj (called an inlink node) and calculate the weight (cost) of the edge to be processed as follows (K. Fillipova, "Multi-sentence compression: Finding shortest paths in word graphs", Progress of the 23rd Use the sentence shortening cost of International Conference on Computational Linguistics 2010).

Here, freq (w) is the frequency of occurrence of the node w (frequency at which the node w is found from each reference meal name), and diff (s, w1, w2) is the inter-word distance considering the order of appearance. , And specifically, the value expressed by the following equation (2):

However, pos (s, w) is the order of appearance of the node w in the meal name s (a value indicating the number of the node w that appears following the link from the beginning), and Σ (sum) is all references. It means to calculate the sum in the meal name.

In the example given above,
When focusing on the edge of "pork" → "vegetables", the frequency of appearance of "pork" (wi) is "3", the frequency of appearance of "vegetables" (wj) is "4",
Diff (s, w1, w2) = 1 in "Delicious stir-fried pork and vegetables" (target meal name)
Diff (s, w1, w2) = 0 in "sautéed pork"
Diff (s, w1, w2) = 0 in "stir-fried vegetables"
Diff (s, w1, w2) = 2 in "stir-fried pork and Chinese-style vegetables"
Diff (s, w1, w2) = 0 in "delicious protein drink"
Diff (s, w1, w2) = 0 in "stir-fried meat and vegetables"
Therefore, the sum of the reciprocals of the diff (s, w1, w2) in this case is 1 + 1/2 = 1.5. Therefore, the weight (cost) of the edge of "pork" → "vegetable" is
(3 + 4) / (3 × 4 × 3/2) = 7/18 = 0.39 (rounded to the first decimal place)
Will be.

Others as well
About the node of "S" → "pork", (6 + 3) / (6 × 3 × 3) = 0.17
About the node of "vegetables" → "delicious", (4 + 2) / (4 × 2 × 1) = 0.75
For the "delicious" → "stir-fried" node, (2 + 4) / (2 x 4 x 1) = 0.75
About the node of "vegetables" → "stir-fried" (obtained from "stir-fried vegetables" etc.) / (4 x 4 x 5/2) = 0.2
...
And so on.

The combination word generation unit 42 obtains the sum of the weights related to the combination obtained by tracing the nodes of the words included in the target meal name in the order of the links among the weights of the edges calculated in this way. That is,
If S → "pork" → "vegetables" → "delicious" → "stir fry" → "things" → E, then 0.17 + 0.39 + 0.75 + 0.75 + 1.25 + 1.17 = 4.48
If S → "pork" → "delicious" → "stir fry" → E, then 0.17 + 1.67 + 0.75 + 0.17 = 2.76
If S → "pork" → "vegetables" → "stir fry" → E, then 0.17 + 0.39 + 0.2 + 0.17 = 0.93
...
Calculate like this. The combination word generation unit 42 finds the combination in which the sum of the weights is the smallest among the combinations to be calculated. In the case of the above example, the combination of S->"pork"->"vegetables"->"stir-fried"-> E is the minimum, so this combination is found.

Then, the combination word generation unit 42 connects the combinations found here in the order of the links to generate the combination word "fried pork and vegetables" as the normalized meal name, and associates this normalized meal name with the target meal name. And store it in the meal database.

The statistical information generation unit 33 performs statistical processing using the normalized meal name. Specifically, by processing to obtain a normalized meal name, for example, "delicious stir-fried pork and vegetables", "stir-fried pork and vegetables", "stir-fried pork and seasonal vegetables", etc. are all regular. If it is related to the stir-fried pork and vegetables, the statistical information generation unit 33 will use the normalized stir-fried pork and vegetables as a key instead of "stir-fried pork and vegetables". conduct. As a result, users can use any name such as "stir-fried pork and vegetables," "stir-fried pork and vegetables," "stir-fried pork and seasonal vegetables," and so on. Even if the same dish is expressed, all of them are aggregated using the normalized meal name "stir-fried pork and vegetables" as a key, so these records will be reflected in the statistical results.

The present embodiment basically has the above configuration and operates as follows. In the following description, information such as meal names for meals eaten and eaten is received from each of a plurality of users at least once, and the received information (records associated with information identifying users and meal name information). ) Is stored in the meal database in advance.

As shown in FIG. 5, the information processing device 1 of the present embodiment sequentially selects a meal name that is not associated with the normalized meal name from the meal names stored in the meal database as the target meal name. (S1), the selected target meal name is divided into words (S2). Then, a meal name (which may include the target meal name itself) including the word obtained by the division (the predetermined stop word may be excluded) is searched from the meal database, and the search result is referred to. Take out as a name (S3).

The information processing device 1 also divides the extracted reference meal name into words (S4). Then, the information processing device 1 generates a word graph in the order of appearance of the divided words (order included in the reference meal name) obtained in the processes S2 and S4, and sets the weight of the edge of the word graph to (1). , (S5) specified by Eq. (2). In the information processing device 1, among the links included in the word graph thus obtained, all the words of the nodes included in the link are included in the target meal name, and the sum of the weights is minimized from the node S to the node. Find the route to E (minimum cost path) (S6), connect the words of the nodes in the order of the links included in the minimum cost path to generate a combination word (S7), and use the combination word as the normalized meal name. , Record in the meal database in association with the target meal name (included in the record including the target meal name) (S8).

Here, the target meal name and the reference meal name are divided into words when the processes S2 and S4 are executed, but when the meal name is input, the meal name is divided into words and stored in the meal database. However, in the process S2 and the process S4, the divided words recorded in association with the target meal name or the meal name serving as the reference meal name may be read out.

Further, the information processing device 1 determines whether or not a meal name not selected as the target meal name exists in the meal database (S9), returns to the process S1 if any, and sets the unselected meal name as the target meal name. Select as to continue processing. Further, in the process S9, if the meal name not selected as the target meal name is not in the meal database, the process ends.

Hereinafter, the information processing device 1 executes predetermined processing such as statistical processing using the normalized meal name. As an example, the information processing device 1 includes information on the frequency of appearance for each normalized meal name associated with information that identifies the same user, information on the frequency of appearance for each normalized meal name in the meal database, and the like. To generate.

[Number of reference meal names]
In the above description, the number of reference meal names including the words obtained by dividing the target meal name is not limited, but in one example of the present embodiment, the reference meal name is used as, for example, the target meal name. Only a specified number of k may be searched in a more similar order or randomly. When selecting in a similar order here, the words obtained by dividing the target meal name and the words obtained by dividing the reference meal name are compared, the number of common words is counted, and the words are common. It is sufficient to take out k words (k types) in descending order of the number of words.

For example, when the words obtained by dividing the target meal name "delicious stir-fried pork and vegetables" are "pork", "vegetables", "delicious", "stir-fried", and "things" as described above, the order is similar. When k = 3 pieces (3 types) are to be taken out
"Delicious stir-fried pork and vegetables" (target meal name itself): Number of common words = 5
"Stir-fried pork and Chinese-style vegetables" (decomposed into "pork", "Chinese-style", "vegetables", "stir-fried", and "things"): Number of common words = 3
"Stir-fried meat and vegetables" (decomposed into "meat", "vegetables" and "stir-fried"): Number of common words = 2
"Stir-fried vegetables" (decomposed into "vegetables" and "stir-fried"): Number of common words = 2
"Sautéed pork" (decomposed into "pork" and "sautéed"): Number of common words = 1
"Delicious protein beverage" (decomposed into "delicious", "protein" and "beverage"): Number of common words = 1
Therefore, take out k = 3 types in descending order of the number of common words, for example, "delicious stir-fried pork and vegetables" (target meal name itself), "stir-fried pork and Chinese-style vegetables" , "Stir-fried meat and vegetables" are used as reference meal names.

For example, even if there are multiple records containing "stir-fried pork and Chinese-style vegetables" in the meal database, k = 3 types are taken out in a similar order, so "delicious stir-fried pork and vegetables". , "Stir-fried pork and Chinese-style vegetables" (even if there are multiple records, it is counted as one type), "Stir-fried meat and vegetables" are the reference meal names.

Furthermore, here, an example is made in which there are two meal names in which the number of common words is "2", and among these, "fried meat and vegetables" is used as a reference meal name. When selecting the meal names up to the top k in the number of common words as candidates when extracting, if the candidates exceed k (when there are multiple meal names with the same rank), among them The shortest meal name or the longest meal name may be selected, or may be randomly selected. Further, when there are a plurality of meal names having the same rank, the one having a large number of records including the meal name may be selected. In this example, for example, if there are 10 records including "stir-fried meat and vegetables" and 20 records including "stir-fried vegetables", "stir-fried vegetables" is selected.

The number k of the reference meal names selected here is determined empirically or experimentally.

[Generation of combination words considering word types]
In addition, considering that the meal name generally includes one or two words representing ingredients and one word representing a cooking method, the following processing may be performed. In this example, the meal name division unit 41 of the meal name normalization unit 32 has a word in the dictionary referred to when dividing the target meal name and the reference meal name, and whether or not the word represents a food material name. The information indicating that the word is used and the information indicating whether or not the word is a cooking method are recorded in association with each other. Specifically, "pork" is associated with information indicating that it is a food ingredient name and recorded, and "stir-fried" is associated with information indicating that it is a cooking method and recorded. In addition, for words such as "delicious" that are neither ingredients nor cooking methods, do not associate information indicating that they are the names of ingredients or information indicating that they are cooking methods.

Then, the meal name dividing unit 41 provides information indicating whether or not the word represents the ingredient name and information indicating whether or not the word represents the cooking method with respect to the word obtained by dividing the ingredient name. Associate and output. Specifically, when the ingredient name of "delicious stir-fried pork and vegetables" is divided,
"Pork": Ingredient name "To":-
"Vegetables": Ingredient name "no":-
"Delicious":-
"Stir-fry": Cooking method "Things":-
While dividing the words such as, the information indicating whether or not the word represents the name of the ingredient and the information indicating whether or not the word represents the cooking method are associated with each other. In addition, "-" indicates that neither the information indicating whether or not the word represents the name of these ingredients and the information indicating whether or not the word represents the cooking method are associated with each other. ..

Further, in this example, the combination word generation unit 42 generates a word graph based on the reference meal name, calculates the weight of each edge, and then includes N = 1 or 2 words included in the target meal name. The sum of the weights related to the combination of the word node representing the ingredient name and the node containing the word representing M = 1 cooking method can be obtained by tracing in the order of the links. If there is no word that represents the name of the ingredient, or if there is no word that represents the cooking method, the node of the word included in the target meal name (whether it is the name of the ingredient or cooking), as described above. The sum of the weights related to the combination that can be obtained by following the order of the links (regardless of the law) is calculated.

Specifically, the target meal name is "stir-fried burdock and beef with seasonal vegetables", and the meal name division section 41
"Seasonal vegetables": Ingredient name "Attached":-
"Gobo": Ingredient name "Beef": Ingredient name "Stir-fried": When separated from the cooking method (stop words are omitted), the combination word generator 42 is a word generated based on the reference meal name. From the graph
"Seasonal vegetables" → "burdock" → "stir-fried"
"Seasonal vegetables" → "Beef" → "Stir-fried"
"Burdock" → "Beef" → "Stir-fried"
"Seasonal vegetables" → "Stir-fried"
"Burdock" → "Stir-fried"
"Beef" → "Stir-fried"
Calculates the sum of the weights when tracing each node of. Then, the combination word generation unit 42 finds the combination in which the sum of the weights is the smallest among the combinations to be calculated.

Then, the combination word generation unit 42 connects the combinations found here in the order of the links to generate the combination word "fried pork and vegetables" as the normalized meal name, and associates this normalized meal name with the target meal name. And store it in the meal database. According to this example of the present embodiment, a relatively general food name including one or two words for the food material name and one word for the cooking method can be reconstructed.

The example here is an example. For example, the node may be selected so as to always include all the food name that appears, or the node may be selected so as to always include two words of the food name (in this case). If the ingredient name is less than one word, you can select the node regardless of whether it is the ingredient name or the cooking method). In this way, the number of words N of the ingredient name and the number of words M of the cooking method may be determined empirically. Both N and M are not limited to one natural number (natural number of 1 or more), and may include a plurality of natural numbers such as N = 2 or 3.

[Stop word set as a condition]
In the above description, the stop word is a word that is analyzed as a predetermined character string or a predetermined part of speech, such as a particle. However, the stop word is not limited to this in the present embodiment. For example, the stop word may include a symbol (parentheses, etc.). Further, as a stop word, it may be a word that satisfies a predetermined condition. The condition here may be described as a regular expression, for example. As an example, if "\ d" matches any one or more characters (including Chinese numerals),
\ d Serving (corresponds to "one serving", "two servings", etc.)
\ dgram (corresponds to "100 grams" etc.)
A stop word may be set as such.

According to this example, for example, "hamburger with cheese (150 grams)"
"(150 grams)" is removed as a stop word from meal names such as
"Cheese""Enter""Hamburg"
It will be divided into words and processed.

In addition, the name of the place or the like may be set as a stop word. For example, the names themselves such as "(Z dining room)" and "(Restaurant A)" may be set as stop words, or using regular expressions ("*" matches any one or more characters. As a thing)
"(* Dining room)" (corresponds to "(Z dining room)", "(central dining room)", etc.)
"(Restaurant *)" (corresponds to "(Restaurant A)", "(At Restaurant B)", etc.)
A stop word may be set as such.

According to this example, for example, "Tempura set meal (Z dining room)"
"(Z dining room)" is removed as a stop word from meal names such as
"Tempura""Setmeal"
It will be divided into words and processed.

[Another example of processing]
In addition, as an example of processing using the normalized meal name, further, user information (for example, medical history information or the like is input to the user) whose appearance frequency (in order of appearance frequency) for each normalized food name is similar to each other is input. If you have received it, you may perform processing such as performing a correlation analysis between the tendency of the food you are eating and drinking and the tendency of lifestyle-related diseases, etc., with reference to the information).

Furthermore, by providing information on other users whose appearance frequency (in order of appearance frequency) for each normalized food name is similar to that of their own, information on other users with similar tastes can be obtained, or information on other users can be obtained. , The process of providing the medical history information of the other user may be performed.

[Effect of Embodiment]
According to this embodiment, since the meal names freely input by the user are aggregated as the normalized meal name, the statistical processing related to the meal name can be performed while allowing the user to freely input the meal name. It will be possible.

1 Information processing device, 2 terminals, 11 control unit, 12 storage unit, 13 communication unit, 31 meal information recording unit, 32 meal name normalization unit, 33 statistical information generation unit, 41 meal name division unit, 42 combination word generation unit ..

Claims (5)

An information processing device connected to a database that stores records including meal names.
A means for dividing a meal name to be processed among the meal names into words as a target meal name, and
A generation means for generating a combination word by combining words selected from the divided words, and
A means for performing a predetermined process using the combination word as a normalized meal name,
Only including,
The generation means
Among the meal names stored in the database, the meal name including the word obtained by dividing the target meal name is used as the reference meal name, and the word obtained by dividing the reference meal name is referred to as the reference meal name. A word graph connected by an edge is generated as a directed graph in the order of appearance in the inside, and a predetermined weight is set for the edge.
The sum of the weights related to the combinations of the words included in the target meal name can be traced in the order of the links in the word graph, and the combinations selected based on the sum of the combinations are connected in the order of the links. An information processing device that generates combination words. The information processing device according to claim 1.
The generation means sets the weight of the edge of the word graph as the frequency of appearance of the word related to the link source node and the word related to the link destination node in the reference meal name, and the order of appearance of the two words. Based on the predetermined inter-word distance using, the words related to the link source node and the words related to the link destination node appear in the order of appearance indicated by the edge so that the frequency becomes smaller.
The sum of the weights related to the combinations of the words included in the target meal name can be traced in the order of the links in the word graph, and the combination having the smallest sum is selected from the combinations, and the selected combination is selected. An information processing device that generates a combination of words connected in the order of links. The information processing device according to claim 1 or 2.
It is connected to a dictionary database that records a word in association with information indicating whether or not the word is a word representing an ingredient name and whether or not the word is a word representing a cooking method.
The generation means
From the divided words, N words (N is a predetermined natural number of 1 or more) and M words (M is 1 or more) are used to represent the name of the ingredient. An information processing device that attempts to select a default natural number) and, when selected, generates a combination word that combines the selected words. The information processing device according to any one of claims 1 to 3.
The predetermined processing is an information processing device including statistical calculation processing for meal names having a common normalized meal name. An information processing device connected to a database that stores records containing meal names,
A means for dividing a meal name to be processed among the meal names into words as a target meal name, and
A generation means for generating a combination word by combining words selected from the divided words, and
A means for performing a predetermined process using the combination word as a normalized meal name,
To function as,
When functioning as the generation means
Among the meal names stored in the database, the meal name including the word obtained by dividing the target meal name is used as the reference meal name, and the word obtained by dividing the reference meal name is referred to as the reference meal name. A word graph connected by an edge is generated as a directed graph in the order of appearance in the inside, and a predetermined weight is set for the edge.
The sum of the weights related to the combinations of the words included in the target meal name can be traced in the order of the links in the word graph, and the combinations selected based on the sum of the combinations are connected in the order of the links. A program that generates combinatorial words.

Images