The present disclosure relates to the field of data processing. More particularly, the present disclosure relates to food item categorization.
The background description provided herein is for the purpose of generally presenting the context of the disclosure. Unless otherwise indicated herein, the materials described in this section are not prior art to the claims in this application and are not admitted to be prior art by inclusion in this section.
In monitoring a person's food and/or beverage intake it may be beneficial to categorize the food and/or beverages a user takes in. Historically, categorizing food and/or beverages may entail manual entry of any categories that may be applicable to the food and/or beverages. Such manual categorization may lead to the use of a limited number of broad food categories. In addition, the application of these broad food categories may need to be rigid due to the inability to efficiently re-categorize a large number of food and/or beverages when a change in food categories may occur.
Methods and systems for operating a food categorization engine are disclosed herein. In at least one embodiment, a food categorization engine is provided on a server and operating in a training mode is configured to receive a training food data set including a plurality of text descriptions of a plurality of food items and a corresponding association of each of the plurality of food items of the training food data set with one or more food categories of a plurality of food categories, wherein the training food data set is provided from a database of crowd-sourced data records of food items. A method includes extracting, by the food categorization engine, one or more features from each text description of the training food data set to generate a feature set for each food item of the training food data set. Thereafter, the method includes analyzing, by the food categorization engine, the feature set for each food item of the training food data set to determine sets of food categorization features that correlate to individual ones of the one or more food categories, wherein the sets of food categorization features are defined by a plurality of category vectors, each of the category vectors associated with one of the food categories. The method further includes standardizing a text description of a food item received from a user of a client device in communication with the server, the food item associated with a data record in the database. Additionally, the method includes analyzing, by the food categorization engine, one or more individual words and/or characters of the text description of the food item, wherein analyzing the one or more individual words and/or characters of the text description of the food item comprises generating a numerical vector representative thereof. The method also includes multiplying the numerical vector by a matrix defined by the plurality of category vectors, automatically selecting one or more food categories with which to associate the food item based at least in part on the multiplication, and amending the data record associated with the food item in the database to include the selected one or more food categories. Furthermore, the method includes selecting the amended data record based on the one or more food categories and transmitting data from the amended data record to the user for display on the client device.
In at least one embodiment, one or more non-transitory computer-readable media are disclosed having a plurality of instructions stored thereon. The instructions, in response to execution by a processor of a network-side computing device in communication with a crowd-sourced database of data records of food items, provide the computing device with a food item categorization engine operable to receive, in a training mode, a training food data set including a plurality of text descriptions of a plurality of food items and a corresponding association of each of the plurality of food items of the training food data set with one or more food categories of a plurality of food categories. The food categorization engine is further operable to extract one or more features from each text description of the training food data set to generate a feature set for each food item of the training food data set, and analyze the feature set for each food item of the training food data set to determine sets of food categorization features that correlate to individual ones of the one or more food categories, wherein the sets of food categorization features are defined by a plurality of category vectors, each of the category vectors associated with one of the food categories. The food categorization engine is further operable to receive, in an operational mode, a data record including a text description of a food item from a client-side application configured to log consumption of food items, wherein the received data record is selected by the user as being indicative of a consumed item, the data record being provided from the database of crowd-sourced data records of food items. Additionally, the food categorization engine is operable to standardize the text description of the food item included with the received data record, analyze one or more individual words and/or characters of the text description of the food item to create a numerical vector representative thereof, and automatically select one or more food categories with which to associate the received data record and associated food item based at least in part on a result of a comparison of the numerical vector to the plurality of category vectors associated with individual ones of the one or more food categories, wherein the comparison is accomplished by matrix-vector multiplication wherein the plurality of category vectors associated with individual ones of the one or more food categories define a matrix that is multiplied by the numerical vector. The food categorization engine is further operable to amend the data record associated with the food item in the database to include the automatically selected one or more food categories, and select the amended data record based on the one or more food categories.
In yet another embodiment, a server apparatus is disclosed comprising at least one transceiver, a storage apparatus configured to store a plurality of instructions of a food categorization engine, and one or more processors configured to execute the plurality of instructions of the food categorization engine. When executed the instructions cause the server apparatus to receive, in a training mode, a training food data set including a plurality of text descriptions of a plurality of food items and a corresponding association of each of the plurality of food items of the training food data set with one or more food categories of a plurality of food categories, the training food data set provided from a database of crowd-sourced data records of food items entered into the database by a plurality of users of a fitness management system. The instructions further cause the server apparatus to extract one or more features from each text description of the training food data set to generate a feature set for each food item of the training food data set, and analyze the feature set for each food item of the training food data set to determine sets of food categorization features that correlate to individual ones of the one or more food categories, wherein the sets of food categorization features are defined by a plurality of category vectors, each of the category vectors associated with one of the food categories. The instructions further cause the server apparatus to receive, in an operational mode, at the transceiver an input from a client device comprising a data record including a description of a food item, wherein the received data record is indicative of a consumed item, the data record being provided from the database of crowd-sourced data records. Additionally, the instructions cause the server apparatus to standardize the description of the food item to create a plurality of tokens representative of one or more individual words and/or characters thereof, convert the plurality tokens to a numerical vector, and compare the numerical vector to a plurality of sets of the plurality of category vectors, each set being associated to a respective one of a plurality of food categories, wherein said comparison is accomplished by matrix-vector multiplication wherein the plurality of category vectors define a matrix that is multiplied by the numerical vector. Based on the comparison, the instructions then cause the server apparatus to identify a one of the plurality of food categories associated to a one of the plurality of sets of the plurality of category vectors to which the numerical vector matches within a predetermined threshold, amend the data record to include the identified one of a plurality of food categories, the data record associated with the description of the food item in the database, and select the amended data record from the database based on the identified one of the plurality of food categories.
The above described features and advantages, as well as others, will become more readily apparent to those of ordinary skill in the art by reference to the following detailed description and accompanying drawings. While it would be desirable to provide a method and system for food categorization that provides one or more of these or other advantageous features as may be apparent to those reviewing this disclosure, the teachings disclosed herein extend to those embodiments which fall within the scope of the appended claims, regardless of whether they include or accomplish one or more of the advantages or features mentioned herein.
Disclosed embodiments include apparatuses, methods and storage media associated with food item categorization.
In the following detailed description, reference is made to the accompanying drawings which form a part hereof wherein like numerals designate like parts throughout, and in which is shown by way of illustration embodiments that may be practiced. It is to be understood that other embodiments may be utilized and structural or logical changes may be made without departing from the scope of the present disclosure. Therefore, the following detailed description is not to be taken in a limiting sense, and the scope of embodiments is defined by the appended claims and their equivalents.
Aspects of the disclosure are disclosed in the accompanying description. Alternate embodiments of the present disclosure and their equivalents may be devised without parting from the spirit or scope of the present disclosure. It should be noted that like elements disclosed below are indicated by like reference numbers in the drawings.
Various operations may be described as multiple discrete actions or operations in turn, in a manner that is most helpful in understanding the claimed subject matter. However, the order of description should not be construed as to imply that these operations are necessarily order dependent. In particular, these operations may not be performed in the order of presentation. Operations described may be performed in a different order than the described embodiment. Various additional operations may be performed and/or described operations may be omitted in additional embodiments.
For the purposes of the present disclosure, the phrase “A and/or B” means (A), (B), or (A and B). For the purposes of the present disclosure, the phrase “A, B, and/or C” means (A), (B), (C), (A and B), (A and C), (B and C), or (A, B and C).
The description may use the phrases “in an embodiment,” or “in embodiments,” which may each refer to one or more of the same or different embodiments. Furthermore, the terms “comprising,” “including,” “having,” and the like, as used with respect to embodiments of the present disclosure, are synonymous.
Referring now to
In some embodiments, the user may be able to further designate various food categories for these calories. For example, in some embodiments, the user may be able to designate a number of the user's budgeted calories that may be utilized towards various different food categories. These categories may be broad food categories (e.g., fruit, vegetable, dairy, grain, meat, etc.) or may be more granular (e.g., egg, cheese, bread, cereal, rice, pasta, poultry, beef, pork, bean, nuts, herb, spice, condiment, etc.). It will be appreciated that these categories are merely meant to be illustrative of possible food categories, food categories may be broader or more granular than the examples given above without departing from the scope of the present disclosure. In other embodiments, a user may be able to designate a nutritional source of the calories, for example, if a user would like to limit the user's caloric intake from fat, the user may designate a maximum budget of the user's calories from fat. As another example, if the user is looking to gain or maintain muscle mass the user may designate a minimum budget of the user's caloric intake that should come from proteins. It will be appreciated that these examples of fat and protein may also be included as different food categories, in addition to being the nutritional source of calories associated with various food items.
Once such a fitness plan is developed, the user may utilize client side agent 122, in conjunction with FMA 112, to monitor the user's intake of foods and/or beverages. This may be accomplished by the user entering individual food and/or beverage items that the user consumes along with quantities of these food and/or beverage items consumed. In some instances the food and/or beverage item may already have a food record in food and beverages database 114. In such instances, the user may merely select the food and/or beverage item from a list of food and/or beverage items and then enter quantity consume. FMA agent 122, in conjunction with FMA 112, may then be configured to incorporate the consumption of the food and/or beverage item into the user's nutritional budget based upon a description of the food item in the food record of the food and beverages database 114 along with any food categories associated with the food item. Such a food record may include nutritional information, such as that discussed below. In some instances the food and/or beverage item may not have an existing record, or the user may not be able to find the appropriate food record for the food and/or beverage item. In such instances, the user may enter the information that the user knows about the food including a food identifier, brand name, country of origin, a description including any nutritional information the user knows, and any additional description the user would like to enter for the food.
In some embodiments, the user may also select one or more existing food categories and/or enter one or more new food categories with which to associate the food and/or beverage item when entering the above mentioned information. Food item categorization engine 124 may also be configured to identify one or more additional food categories with which to associate the food and/or beverage item through an automated selection of food categories by extracting features of the description and utilizing the features to identify the one or more additional food categories with which to associate the food and/or beverage item, a more detailed description of such a process is discussed further below. In embodiments, food item categorization engine 124 may be configured to utilize the one or more food categories selected and/or entered by the user to refine, or train, the automated selection of food categories based on feature extraction of the description, such as that described below. In other embodiments, the user may not enter any food categories with which to associate the food and/or beverage item and the food item categorization may be configured to automatically select one or more food and/or beverage categories by extracting features of the description and utilizing the features to identify the one or more food and/or beverage categories with which to associate the food and/or beverage item, a more detailed description of such a process is discussed further below. In some embodiments, any food and/or beverage categories associated with the food and/or beverage item may then be incorporated into the user's nutritional budget where the user has designated any food and/or beverage category restrictions. In some embodiments, the food and/or beverage categories may also be used, for example by FMA 112, to generate a report depicting the food and/or beverage categories from which the user has consumed any food and/or beverage items. Such a report may give the user a better idea of the user's eating and/or drinking habits.
In embodiments, server(s) 104, except for FMA 112 and databases 114-116, may be any one of a number of computer servers, real or virtual, known in the art, including, but not limited to, servers available from Dell Computing of Austin, Tex. or Hewlett Packard of Palo Alto, Calif. In embodiments, FMA 112 may include food item categorization engine 124, hereinafter referred to simply as “categorization engine 124.” Whereas food & beverages database 114 may include food and beverage items, including a description of the food and beverage items that may include nutrient information associated with the food and beverage items, as well as recipes and ingredient items. Examples of nutrient information may include, but are not limited to, amounts of protein; sugar; various fats, including trans-fat, saturated fat, etc.; sodium; carbohydrates, both complex and simple; calcium, various vitamins and/or calories per serving. In some embodiments, server(s) 104 may represent a cloud computing environment where each of FMA 112, categorization engine 124, food and beverages database 114, and user database 116 may reside, individually or in any combination, on servers, real or virtual, of the cloud computing environment.
In embodiments, client device 102 may be any one of a number of stationary or portable electronic devices known in the art, including, but not limited to, desktop computers (e.g. those available from Dell Computing of Austin, Tex.), smartphones, computing tablets, laptop computers, electronic readers, personal digital assistants, and so forth, such as Galaxy S4 from Samsung Electronics of Seoul, Korea, or iPad from Apple Computer of Cupertino, Calif. In embodiments, one or more portable computing devices 102 may be a wearable computing device, e.g., a smart watch, smart eyeglasses (e.g., Google Glass from Google of Mountain View, Calif.), and so forth. In embodiments, FMA agent 122 may be a web-based application configured to operate in a generic browser, such as Internet Explorer, available from Microsoft Corp., of Redmond, Wash., or Safari from Apple Computer of Cupertino, Calif. In other embodiments, FMA agent 122 may be a stand-alone client side application, e.g., a mobile app configured to run on a smartphone such as, for example, a mobile app available through the App Store from Apple Computer of Cupertino, Calif.
In embodiments, client device 102 and server(s) 104 may be communicatively coupled with one another via communication link 106 over one or more wired and/or wireless, private and/or public networks, including the Internet. Client device 102 and server(s) 104 may be configured with the appropriate networking communication interfaces. An example of a wired communication interface may include, but is not limited to, Ethernet, while examples of wireless communication interfaces may include, but are not limited to, near field communication (NFC), Bluetooth, WiFi, 4G or 5G LTE. In between the communication interfaces of client devices 102 and server(s) 104 may be any number of gateways, routers, switches, based stations, and so forth.
For ease of description, hereinafter, including the claims, the term “food” will be used to mean “food and/or beverage,” unless the context clearly indicates otherwise.
At block 204, the categorization engine may analyze the description of the food item. In some embodiments, the analysis may include a data standardization process performed on the description. In embodiments this may be accomplished by the categorization engine tokenizing the description. As used herein, tokenizing may refer to the process of dividing the text of the description into smaller components, or tokens, to be processed. The tokens may be individual characters, words, groups of words (e.g., every three words), or groups of characters (e.g., every five characters).
Once the description has been tokenized it may be processed by the categorization engine to standardize the description. Such a standardization process may include the removal of stop words designated, for example, by a list of stop words contained within a table of the food and beverages database. The standardization may also include the removal of punctuation or special characters from the description. In addition, the standardization process may include: the removal of traditional stop words (e.g., a, the, etc.); the removal of numeric words from the food item description (e.g., one, two, three, etc.); removal of words related to quantities from the food item description (e.g., cup, ounce, etc.); and/or removal of food related stop words (e.g., calorie, breakfast, lunch, etc.). In some embodiments, the standardization may also include removal of “bad” keywords that may be defined, for example, in a table of bad keywords stored in the food and beverages database. Such bad keywords may be for example, profanities, vulgarities, etc. In still further embodiments, the standardization may include removal of non-English characters from the food item description and/or replacement of non-English characters with the closest English equivalent. For example, the ‘ñ’ jalapeno may be replaced by an English ‘n,’
Once the description has been standardized, the description may be processed to extract one or more features of the food item from the description. In embodiments, these features may be extracted by converting tokens of the description into semantic feature vectors. This may be accomplished, for example, by first constructing a big feature vector, where each member of the big feature vector may correlate with a word appearing in the food and beverages database. In such an embodiment, a semantic feature vector may be constructed out of the description. The semantic feature vector may be the size of the big feature vector; however, only those members that correlate with words in the description may be marked as non-zero values. For example, if the description were simply “chicken sandwich” then the semantic feature vector for this description may be represented as (0, 0, 1, 0, 0, . . . , 1, 0) where the first “1” represents “chicken” and the second “1” represents “sandwich.” The length of this semantic feature vector may be the same as that of the big feature vector discussed above, or one member per word contained in the food and beverages database. In addition, these 1s in a vector can be replaced by a data correlation scheme based on the relative importance of words. In some embodiments, this semantic feature vector may be normalized. For example, by squaring the numeric values of each token, summing the squared values, and then taking the square root of the sum to arrive at a normalized value. Such normalization may be beneficial to aid in giving food descriptions with varying lengths similar weight.
At block 206, the categorization engine may select one or more food categories based on a result of the analysis performed at block 204. In some embodiments, this selection may be based on the results of the feature extraction described above. In such embodiments, the extracted features of the description may be compared with a set of corresponding categorization features that have been determined to correlate with a food category. If a sufficient number of extracted features correspond with the categorization features of a food category, then that food category may be selected as one of the one or more food categories with which to associate the food item. In some embodiments, the above described process may continue until all food categories have been evaluated with respect to the extracted features and each food category that is determined to be within a predefined threshold of the extracted features may be selected as one of the one or more food categories. In other embodiments, only the category having the most categorization features in common with the extracted features may be selected as a food category with which to associate the food item. In other embodiments, this selection may be based on a mathematical combination and/or transformation between the extracted features and a semantic feature table. In such embodiments, the mathematical combination and/or transformation may be utilized to calculate the likelihood, or probability, that a food item belongs in each food category. For instance, for a certain food item, we may get features represented in a numerical vector such as, for example, x=(0.1, 4.1, . . . , 3.0) and the trained categorization engine may then utilize the semantic feature table, discussed further in reference to
At block 208, the categorization engine may establish an association of the food item with each of the one or more food categories. In some embodiments, such an association may be established through the addition of a unique identifier for each of the one or more categories to a category field, or table, associated with the food item in a food and beverages database (e.g., food and beverages database 114 of
Operation flow 300 may begin at block 302 where a labeled training food data set may be received by the categorization engine. Such a labeled training food data set may be received from a food and beverages database (e.g. food and beverages database 114 of
At block 304 the categorization engine may iterate through each of the food items and may extract a feature set of one or more features from the description of each food item in the labeled training food data set. The feature extraction process may be performed in a similar manner to that described above in reference to
At block 306, the feature set for each food item may be analyzed in conjunction with the label, or food categories, associated with the respective food item. As such, the categorization engine may be able to determine features that correlate with individual food categories. From this correlation the categorization engine may generate a distributional semantic model, such as a semantic feature table, that may be a data structure representation of the correlation between feature sets of each food item and the one or more categories associated with the respective food item.
At block 404 the categorization engine may iterate through each of the food items and may extract a feature set of one or more features from the description of each food item in the unlabeled training food data set. The feature extraction process may begin by tokenizing the description of each food item, if not already performed in a standardization process. In some embodiments, the feature sets may be extracted by converting unique tokens of each description into numeric values. The numeric values may form a semantic feature vector for each token which may then be summed resulting in a vector sum representing the feature set for each description. In some embodiments, this vector sum may be normalized, as discussed in reference to
At block 406, the categorization engine may analyze the feature set of each food item to determine one or more possible food categories for each food item of the unlabeled training food data set. In embodiments, such an analysis may be based on a result of the feature extraction performed at block 404. In such embodiments, the extracted features of the description may be compared with a set of corresponding categorization features, such as those represented by a semantic feature table, as discussed above, that have been determined to correlate with a food category. If a sufficient number of extracted features correspond with the categorization features of a food category, then that food category may be selected as one of the one or more possible food categories with which to associate the food item. In some embodiments, the above described process may continue until all food categories have been evaluated with respect to the extracted features and each food category that is determined to be within a predefined threshold of the extracted features may be selected as one of the one or more possible food categories. In other embodiments, only the category having the most categorization features in common with the extracted features may be selected as a possible food category with which to associate the food item.
At block 408, the categorization engine may generate one or more questions associated with the one or more possible food categories to be sent to one or more users. Such questions may include binary, yes/no, questions. For example, a question for each of the one or more possible food categories may be whether a user thinks the food item belongs in each of the one or more possible food categories to which the user may either respond in the affirmative or the negative.
At block 410, the features that caused each food item to be associated with the one or more possible categories may be adjusted based on the answers the user provides to the one or more questions to fine tune the categorization. For example, where a particular feature set caused the categorization engine to determine that a food item belonged to a specific food category, a negative answer to the above discussed questions may decrease a weight, or priority, of one or more of the features in the feature set, while an affirmative answer may increase the weight, or priority, of the one or more features.
Referring now to
Each of these elements may perform its conventional functions known in the art. In particular, when used as server(s) 104, system memory 504 and mass storage devices 506 may be employed to store a working copy and a permanent copy of the programming instructions implementing the operations associated with fitness management application 112 and categorization engine 124 of
The permanent copy of the programming instructions may be placed into permanent storage devices 506 in the factory, or in the field, through, for example, a distribution medium (not shown), such as a compact disc (CD), or through communication interface 510 (from a distribution server (not shown)). That is, one or more distribution media having an implementation of the agent program may be employed to distribute the agent and program various computing devices.
The number, capability and/or capacity of these elements 510-512 may vary, depending on whether computer 500 is used as server(s) 104 or client device 102. When used as client device 102, computing device 500 may be a smartphone, computing tablet, e-reader, ultrabook, or laptop. Otherwise, the constitutions of elements 510-512 are known, and accordingly will not be further described.
It will be apparent to those skilled in the art that various modifications and variations can be made in the disclosed embodiments of the disclosed device and associated methods without departing from the spirit or scope of the disclosure. Thus, it is intended that the present disclosure covers the modifications and variations of the embodiments disclosed above provided that the modifications and variations come within the scope of any claims and their equivalents.
This document is a continuation of U.S. patent application Ser. No. 14/513,061, filed Oct. 13, 2014, the entire contents of which are incorporated by reference herein.
Number | Date | Country | |
---|---|---|---|
Parent | 14513061 | Oct 2014 | US |
Child | 16861901 | US |