This dataset represents the data related to food recommender system. Two datasets are included in this dataset file. First includes the dataset related to the foods, ingredients, cuisines involved. Second, includes the dataset of the rating system for the recommendation system.

In 2024, some ** percent of Generation Z survey respondents in the United States and the United Kingdom stated that their most valuable social platform for food recommendations was TikTok. Instagram was the second most valuable platform for Generation Z.

Explore the culinary world with our extensive Food.com Recipes dataset. This dataset offers a rich collection of recipes sourced from Food.com, one of the largest and most trusted recipe platforms. Ideal for food enthusiasts, chefs, app developers, and data scientists, this dataset provides everything you need to create, analyze, and innovate in the kitchen.

The dataset includes detailed information such as recipe names, ingredients, step-by-step cooking instructions, preparation and cooking times, user ratings, and dietary preferences. Whether you're developing a new recipe app, conducting food-related research, or simply looking to explore new dishes, this dataset offers a wealth of information to help you achieve your goals.

Looking for additional data to power your food-related projects? Check out our Food & Beverage Data for access to a wide variety of datasets that can help you unlock new opportunities in the food and beverage industry.

With thousands of recipes covering a wide range of cuisines, meal types, and dietary requirements, this dataset is perfect for those looking to build recipe recommendation systems, nutritional analysis tools, or food blogs. Tap into the rich culinary diversity offered by Food.com and take your food-related projects to the next level.

ABSTRACT Going through several reviews could be laborious when this has to be done for multiple restaurants. One could instead read a graphical representation of what is great at the restaurant. Currently on Yelp, the food recommendations are only based on the total number of mentions of the food item in the reviews. Higher mentions, irrespective of the context, get an up-vote toward recommended items. Including context from reviews and tips could greatly improve the list of recommended items. In this project, we combine Named Entity Recognition and Sentiment Analysis of reviews. Based on the sentiment of the reviews we aim to suggest the best dishes of a restaurant or the best restaurant offering a dish. We have leveraged various feature engineering methods to produce state-of-the-art results. We established that if chosen, the appropriate feature vectors can significantly improve the classification performance. Fine-tuning BERT and bi-directional LSTM are producing better results than the machine learning models and if trained for more epochs can eventually prove to be the best classifier models. Keywords: Contextual Recommendation, Named Entity Recognition, BERT, LSTM, Count Vectors, TF-IDF

The dataset includes plan review submitted for new or remodeled construction with a food service or public swimming pool facility. Update Frequency: Daily

Context

I needed to record our grocery intake using a standard nomenclature that supports access and tabulation of nutrients consumed. That is, a nomenclature that spans retail grocery using UPC to the identifiers used to access nutrition labels. I wanted to know if our household's intake of nutrients complied with the bounds set forth in the 2015-2020 USDA Dietary Guidelines. And if not, what foods will most efficiently remedy dietary deficiencies/excesses. Professionally, I have been devising countermeasures to fight chronic disease and see food more so than medication as the key.

Content

Downloaded from the USDA FoodData Central database: https://fdc.nal.usda.gov/download-datasets.html, this constellation of tables is centered on the food table. That table identifies foods broadly classed as Standard like what you might find in the periphery of the store, or Branded which typically occupies the center aisles. Each Standard Reference food identifier points to exactly one food label, where each label (in the set of all labels) is a set of entity attribute value (EAV) triples comprised of FoodId NutrientId and Quantity per hectogram (i.e., 100g). Standard Reference and Branded Reference foods each occupy their own tables. Since categories for Standard Reference foods are coded, a table providing their descriptions are also provided. Note that categories for Branded Reference foods are in verbose text.

Layouts

For field definitions and table layouts see: Download_&_API_Field_Descriptions_April_2021.pdf in this distribution.

Derived tables

Syntax for tables derived from USDA FoodData Central: - Filenames are prefixed by user initials 'JCS_' - The remaining portion of the filename is a hyphenated list of its domain names. - FoodId in derived files were zero-left padded for proper collation and joining.

Domains for derived tables from which individual column values are drawn: - SRCat: Standard Reference food category, zero-left padded to two places. - Applicable: A Boolean {YES,NO} indicating whether a standard reference food category applies to the current study. - CatDesc: The description of that food category. - FoodId: A unique and unchanging food identifier, zero-left padded to seven places. - Description: Text describing each Standard Reference food item. - FoodName: Text describing each Branded Reference food item. - BRCat: Branded Reference category expressed as verbose text. - UPC: Uniform Product Code that identifies a branded food item regardless of revision date. - BrandOwner: Corporate entity that owns that brand of foods. - SvgSizeHgs: Serving size of a branded food item in hectograms. - DateAvail: Revision date for a UPC's metadata or nutrition label. - VerCnt: Number of revisions that a food identified by its UPC code underwent. - NutrientId: Uniquely identifies a nutrient as a 4-digit number with cross-references stored in nutrient.csv - Per100g: Quantity of a nutrient per hectogram as a float but stored as plain text.

Acknowledgements

To my Chair, Prof. Taghi M. Khoshgoftaar, PhD.

Inspiration

Given a list of foods and when and how much of each were consumed: - aggregate quantities of each nutrient consumed, - norming the quantities of nutrients consumed to average daily calorie burn, and - comparing each nutrient consumed to the target range for that nutrient.

Nutrient-wise deficiencies and excesses relative to target ranges are fed to a recommender that identifies: - what foods most efficiently by weight remedies these deficiencies and excesses - while being foods most likely to be consumed.

Rankings presented by the recommender during one period are evaluated by foods consumed in the following period. What ranking is the most 'compact'? That is, what ranking has the most of its foods consumed among its top k foods?

ObjectiveSince large food portion sizes (PS) lead to overconsumption, our objective was to review PS recommendations for commonly consumed food groups reported in Food-Based Dietary Guidelines (FBDGs) globally and to assess variation in PS across countries and regions.MethodsConsumer-oriented FBDGs from the Food and Agriculture Organization (FAO) online repository were used to evaluate dietary recommendations, PS and number of portions for common food groups. Guidelines were classified for each group as qualitative, quantitative, or missing. A standardized approach was applied to convert PS recommendations given as household measures, cup equivalents, pieces and other measures into grams for cross comparison. Variation of recommended PS of common food groups within and across regions was examined.ResultsAmong 96 FBDGs, variations were found both across and within regions. At a regional level, the highest median PS recommendations were seen in Europe for Meat, Fish and Pulses, in the Near East for Dairy products, and in Africa for most grain-based foods. Recommendations for Fruits and Vegetables showed the highest consistency across FBDGs worldwide, whereas guidance on Meat, fish & eggs and Cooked cereals/grains showed discrepancies in the classification of foods into categories, as well as in the number of portions per day.DiscussionWhile some variation in PS recommendations across countries can be expected due to cultural and regional dietary practices, inconsistent definitions to refer to a portion and varied derivation methods may further produce discrepancies. Harmonizing development methods for FBDG could help establish more consistent reference portion sizes and therefore provide clearer guidance to consumers.

Dataset for a Food Ontology Integrating HUMMUS and Open Food Facts with Extended User Attributes

This dataset supports a food ontology that semantically integrates two existing resources: the HUMMUS knowledge graph (focused on recipes, users, and reviews) and Open Food Facts (OFF, focused on nutritional information for packaged foods). In addition to aligning entities and concepts across the two sources, the ontology introduces new user-specific attributes to enable more fine-grained modeling of food preferences, constraints, and behaviors. The resource is intended for use in research on personalized nutrition, food recommendation systems, and knowledge-based AI applications.

File explanation:

• off: Open Food Facts dataset with product name normalized (more or less 10GB)
• hummus: HUMMUS recipe with recipe name normalized (more or less 2GB)
• hummus_review: HUMMUS review with inferred info (more or less 0.5GB)
• hummus_member: HUMMUS member info with inferred info (more or less 0.5GB)
• merging_file: file needed to merge the ontologies (more or less 6GB)
• food_ontology_complete: the complete merged ontology file, merged with a threshold of 0.975 (more or less 150GB)
• food_ontology_complete_085: the complete merged ontology file, merged with a threshold of 0.85 (more or less 270GB)

Meal Recommendation App Market Outlook

According to our latest research, the global meal recommendation app market size in 2024 stands at USD 3.2 billion, reflecting robust adoption across consumer and professional segments. The market is poised for impressive expansion, registering a CAGR of 15.8% during the forecast period, and is projected to reach USD 10.2 billion by 2033. Key growth drivers include increasing health consciousness, technological advancements in artificial intelligence, and the widespread integration of personalized nutrition services. These factors are fundamentally reshaping how individuals and organizations approach dietary planning and nutrition management.

A primary growth factor in the meal recommendation app market is the surging consumer demand for personalized nutrition solutions. As global populations become more health-aware, there is a significant shift toward tailored dietary recommendations that accommodate individual preferences, allergies, and health conditions. The integration of AI and machine learning algorithms allows these apps to provide highly customized meal plans, which has led to increased user engagement and retention. Furthermore, the proliferation of wearable devices and health trackers has enabled seamless data synchronization, empowering apps to deliver real-time, actionable recommendations based on a user’s dietary habits, activity levels, and biometric data. This convergence of technology and health is fueling the market’s rapid growth trajectory.

Another essential driver is the rising prevalence of chronic diseases such as diabetes, obesity, and cardiovascular disorders, which necessitate ongoing dietary management. Meal recommendation apps have emerged as valuable tools for both patients and healthcare providers to monitor and guide nutritional intake. These platforms are increasingly being adopted by healthcare institutions and fitness centers to support disease management and preventive healthcare. Additionally, the COVID-19 pandemic has accentuated the importance of remote health monitoring and self-care, prompting a surge in downloads and subscriptions of meal recommendation apps. This trend is expected to persist as consumers prioritize digital health solutions for long-term wellness and disease prevention.

The market is also benefiting from the expanding ecosystem of partnerships among app developers, nutritionists, fitness professionals, and food retailers. Strategic collaborations are enabling the integration of meal recommendation apps with online grocery platforms and food delivery services, thereby enhancing user convenience and broadening monetization opportunities. The emergence of freemium and subscription-based models has further democratized access to premium features, while also providing sustainable revenue streams for app providers. As the competitive landscape intensifies, ongoing investments in research and development are fostering innovation, resulting in more intuitive user interfaces, advanced analytics, and comprehensive meal databases.

Regionally, North America dominates the meal recommendation app market, accounting for the largest revenue share in 2024. This is attributed to high smartphone penetration, advanced healthcare infrastructure, and a strong culture of digital health adoption. However, the Asia Pacific region is witnessing the fastest growth, driven by a burgeoning middle class, rising disposable income, and increasing awareness of nutrition and wellness. Europe follows closely, with robust regulatory support for digital health and a growing emphasis on preventive healthcare. The Middle East & Africa and Latin America are also experiencing steady growth, albeit from a smaller base, as digital transformation initiatives gain momentum across these regions.

Platform Analysis

The platform segment of the meal recommendation app market encompasses iOS, Android, and web-based applications, each catering to distinct user preferences and technologic

The Food and Food Categories (FFoCat) Dataset

The Food and Food Categories (FFoCat) Dataset contains 58.962 images of food annotated with the food label and the food categories of the Mediterranean Diet. It is one of the most complete datasets regarding the Mediterranean Diet as it is aligned with the standard AGROVOC and HeLiS ontologies and allows to study multitask learning problems in Computer Vision for food recognition and diet recommendation.

The dataset is already divided into the train and test folder. The file label.tsv contains the food labels, the file food_food_category_map.tsv contains the food labels with the corresponding food category labels. The following table compares the FFoCat dataset with previous datasets for food recognition.

This dataset has been published at the International Conference on Image Analysis and Processing (ICIAP - 2019). The source code for reproducing the experiments together with other information about the dataset is available here.

AGROVOC Alignment of Food Categories

The AGROVOC_alignment.tsv file contains the alignment of the food categories in the FFoCat dataset with AGROVOC, the standard ontology of the Food and Agriculture Organization (FAO) of the United Nations. This allows interoperability and linked open data navigation. Such alignment can be derived by querying HeLis, here we propose a shortcut.

Citing FFoCat

If you use FFoCat in your research, please use the following BibTeX entry.

@inproceedings{DonadelloD19Ontology, author = {Ivan Donadello and Mauro Dragoni}, title = {Ontology-Driven Food Category Classification in Images}, booktitle = {{ICIAP} {(2)}}, series = {Lecture Notes in Computer Science}, volume = {11752}, pages = {607--617}, publisher = {Springer}, year = {2019} }

Dataset Card for "Amazon Food Reviews"

  Dataset Summary

This dataset consists of reviews of fine foods from amazon. The data span a period of more than 10 years, including all ~500,000 reviews up to October 2012. Reviews include product and user information, ratings, and a plain text review. It also includes reviews from all other Amazon categories.

  Supported Tasks and Leaderboards

This dataset can be used for numerous tasks like sentiment analysis, text… See the full description on the dataset page: https://huggingface.co/datasets/jhan21/amazon-food-reviews-dataset.

Diabetes, a chronic metabolic condition characterised by persistently high blood sugar levels, necessitates early detection to mitigate its risks. Inadequate dietary choices can contribute to various health complications, emphasising the importance of personalised nutrition interventions. However, real-time selection of diets tailored to individual nutritional needs is challenging because of the intricate nature of foods and the abundance of dietary sources. Because diabetes is a chronic condition, patients with this illness must choose a healthy diet. Patients with diabetes frequently need to visit their doctor and rely on expensive medications to manage their condition. It is challenging to purchase medication for chronic illnesses on a regular basis in underdeveloped nations. Motivated by this concept, we suggest a hybrid model that, rather than depending solely on medication to evade a visit to the doctor, can first anticipate diabetes and then suggest a diet and exercise regimen. This research proposes an optimized approach by harnessing machine learning classifiers, including Random Forest, Support Vector Machine, and XGBoost, to develop a robust framework for accurate diabetes prediction. The study addresses the difficulties in predicting diabetes precisely from limited labeled data and outliers in diabetes datasets. Furthermore, a thorough food and exercise recommender system is unveiled, offering individualized and health-conscious nutrition recommendations based on user preferences and medical information. Leveraging efficient learning and inference techniques, the study achieves a meager error rate of less than 30% using an extensive dataset comprising over 100 million user-rated foods. This research underscores the significance of integrating machine learning classifiers with personalized nutritional recommendations to enhance diabetes prediction and management. The proposed framework has substantial potential to facilitate early detection, provide tailored dietary guidance, and alleviate the economic burden associated with diabetes-related healthcare expenses.

The Food Code Reference System (FCRS) is a searchable database that provides access to FDA�s interpretative positions and responses to questions related to the FDA Food Code. Users of the FCRS can search the database using dropdown menus, keywords and date. This system is a searchable repository of food code interpretations. There are no exportable data sets. All reports are in PDF format.

These datasets contain peer-to-peer trades from various recommendation platforms.

Metadata includes

• peer-to-peer trades

• have and want lists

• image data (tradesy)

This is a takeout recommendation dataset (TRD) which contains a vast amount of meta information from Meituan Takeout app. We collect orders from 11 commercial districts in Beijing between March 1st and March 28th, 2021. The first three weeks of orders are as training, while the last week is used for testing to avoid data leakage. We briefly summarize each file as follows and for more details, please refer to README.md.

1. users.txt (attributes of all users)
2. pois.txt (attributes of all takeout restaurants)
3. spus.txt (attributes of all food)
4. orders_poi_session.txt (a sequence of restaurants clicked by user before ordering)
5. orders_spu_train.txt (order-food in training set)
6. orders_train.txt (order-restaurant in training set)
7. orders_test.txt (order-restaurant in training set)
8. orders_poi_test_label.txt (test labels of order-restaurant)
9. orders_spu_test_label.txt (test labels of order-food)

10 graph.bin (graph in DGL format)

graph.bin is build by above *.txt files, there is a vast amount of meta informarion on nodes and edges. just several codes can load this graph with 18,931,400 edges and 408,849 nodes:

from dgl import load_graphs #should install dgl
ds,_ = load_graphs("./graph.bin")
g = ds[0]
print(g)

The Marine Mammal Laboratory (MML) Food Habits Reference Collection, containing over 1000 specimens of cephalopod beaks and fish bones and otoliths, is used to identify undigested prey remains found in scats or stomachs of stranded or incidentally taken pinnipeds and cetaceans. Marine mammal food habits data are used in conjunction with satellite telemetry and dive records to better understand...

[Note: Integrated as part of FoodData Central, April 2019.] The database consists of several sets of data: food descriptions, nutrients, weights and measures, footnotes, and sources of data. The Nutrient Data file contains mean nutrient values per 100 g of the edible portion of food, along with fields to further describe the mean value. Information is provided on household measures for food items. Weights are given for edible material without refuse. Footnotes are provided for a few items where information about food description, weights and measures, or nutrient values could not be accommodated in existing fields. Data have been compiled from published and unpublished sources. Published data sources include the scientific literature. Unpublished data include those obtained from the food industry, other government agencies, and research conducted under contracts initiated by USDA’s Agricultural Research Service (ARS). Updated data have been published electronically on the USDA Nutrient Data Laboratory (NDL) web site since 1992. Standard Reference (SR) 28 includes composition data for all the food groups and nutrients published in the 21 volumes of "Agriculture Handbook 8" (US Department of Agriculture 1976-92), and its four supplements (US Department of Agriculture 1990-93), which superseded the 1963 edition (Watt and Merrill, 1963). SR28 supersedes all previous releases, including the printed versions, in the event of any differences. Attribution for photos: Photo 1: k7246-9 Copyright free, public domain photo by Scott Bauer Photo 2: k8234-2 Copyright free, public domain photo by Scott Bauer Resources in this dataset:Resource Title: READ ME - Documentation and User Guide - Composition of Foods Raw, Processed, Prepared - USDA National Nutrient Database for Standard Reference, Release 28. File Name: sr28_doc.pdfResource Software Recommended: Adobe Acrobat Reader,url: http://www.adobe.com/prodindex/acrobat/readstep.html Resource Title: ASCII (6.0Mb; ISO/IEC 8859-1). File Name: sr28asc.zipResource Description: Delimited file suitable for importing into many programs. The tables are organized in a relational format, and can be used with a relational database management system (RDBMS), which will allow you to form your own queries and generate custom reports.Resource Title: ACCESS (25.2Mb). File Name: sr28db.zipResource Description: This file contains the SR28 data imported into a Microsoft Access (2007 or later) database. It includes relationships between files and a few sample queries and reports.Resource Title: ASCII (Abbreviated; 1.1Mb; ISO/IEC 8859-1). File Name: sr28abbr.zipResource Description: Delimited file suitable for importing into many programs. This file contains data for all food items in SR28, but not all nutrient values--starch, fluoride, betaine, vitamin D2 and D3, added vitamin E, added vitamin B12, alcohol, caffeine, theobromine, phytosterols, individual amino acids, individual fatty acids, or individual sugars are not included. These data are presented per 100 grams, edible portion. Up to two household measures are also provided, allowing the user to calculate the values per household measure, if desired.Resource Title: Excel (Abbreviated; 2.9Mb). File Name: sr28abxl.zipResource Description: For use with Microsoft Excel (2007 or later), but can also be used by many other spreadsheet programs. This file contains data for all food items in SR28, but not all nutrient values--starch, fluoride, betaine, vitamin D2 and D3, added vitamin E, added vitamin B12, alcohol, caffeine, theobromine, phytosterols, individual amino acids, individual fatty acids, or individual sugars are not included. These data are presented per 100 grams, edible portion. Up to two household measures are also provided, allowing the user to calculate the values per household measure, if desired.Resource Software Recommended: Microsoft Excel,url: https://www.microsoft.com/ Resource Title: ASCII (Update Files; 1.1Mb; ISO/IEC 8859-1). File Name: sr28upd.zipResource Description: Update Files - Contains updates for those users who have loaded Release 27 into their own programs and wish to do their own updates. These files contain the updates between SR27 and SR28. Delimited file suitable for import into many programs.

[Note: Integrated as part of FoodData Central, April 2019.] USDA's Food and Nutrient Database for Dietary Studies (FNDDS) is a database that is used to convert food and beverages consumed in What We Eat In America (WWEIA), National Health and Nutrition Examination Survey (NHANES) into gram amounts and to determine their nutrient values. Because FNDDS is used to generate the nutrient intake data files for WWEIA, NHANES, it is not required to estimate nutrient intakes from the survey. FNDDS is made available for researchers using WWEIA, NHANES to review the nutrient profiles for specific foods and beverages as well as their associated portions and recipes. Such detailed information makes it possible for researchers to conduct enhanced analysis of dietary intakes. FNDDS can also be used in other dietary studies to code foods/beverages and amounts eaten and to calculate the amounts of nutrients/food components in those items.
FNDDS is released every two-years in conjunction with the WWEIA, NHANES dietary data release. The FNDDS is available for free download from the FSRG website. Resources in this dataset:Resource Title: Website Pointer to Food and Nutrient Database for Dietary Studies. File Name: Web Page, url: https://www.ars.usda.gov/northeast-area/beltsville-md-bhnrc/beltsville-human-nutrition-research-center/food-surveys-research-group/docs/fndds/ USDA's Food and Nutrient Database for Dietary Studies (FNDDS) is a database that is used to convert food and beverages consumed in What We Eat In America (WWEIA), National Health and Nutrition Examination Survey (NHANES) into gram amounts and to determine their nutrient values.

📘 ABOUT

The Enhanced Zomato Dataset provides comprehensive information on restaurants, including user ratings, cuisine types, prices, and geographic details. This enhanced version of the popular Zomato dataset includes carefully cleaned data and newly engineered features to support advanced analytics, trend analysis, and machine learning applications.

It is especially valuable for data scientists, analysts, and machine learning practitioners seeking to build recommendation systems, price predictors, or restaurant review models.

✅ Key Features of the Dataset:

• ✅ Cleaned data with no missing values
• ✅ Contains restaurant, cuisine, city, and pricing information
• ✅ Includes user ratings for both dining and delivery experiences
• ✅ Features engineered columns such as popularity scores and price-per-vote ratios
• ✅ Ready for data visualization, machine learning, and business insights

This dataset is an excellent resource for exploring food industry patterns, building ML models, and performing customer behavior analysis.

📂 FILE INFORMATION

The dataset contains structured records of restaurant details, user ratings, pricing, and engineered features. It was compiled from a public Zomato dataset and enhanced through feature engineering and cleaning techniques.

• File Type: CSV
• Data Rows: 123,657
• Data Fields: Restaurant details, cuisine, city, prices, votes, ratings, and additional engineered features

📊 COLUMNS DESCRIPTION

description: This dataset provides cost of Food at Home at Four levels for the USDA Food Plans. The Food Plans represent a nutritious diet at four different cost levels. The nutritional bases of the Food Plans are the 1997- 2005 Dietary Reference Intakes, 2005 Dietary Guidelines for Americans, and 2005 MyPyramid food intake recommendations. In addition to cost, differences among plans are in specific foods and quantities of foods. Another basis of the Food Plans is that all meals and snacks are prepared at home. For specific foods and quantities of foods in the Food Plans, see Thrifty Food Plan, 2006 (2007) and The Low-Cost, Moderate-Cost, and Liberal Food Plans, 2007 (2007). All four Food Plans are based on 2001-02 data and updated to current dollars by using the Consumer Price Index for specific food items.; abstract: This dataset provides cost of Food at Home at Four levels for the USDA Food Plans. The Food Plans represent a nutritious diet at four different cost levels. The nutritional bases of the Food Plans are the 1997- 2005 Dietary Reference Intakes, 2005 Dietary Guidelines for Americans, and 2005 MyPyramid food intake recommendations. In addition to cost, differences among plans are in specific foods and quantities of foods. Another basis of the Food Plans is that all meals and snacks are prepared at home. For specific foods and quantities of foods in the Food Plans, see Thrifty Food Plan, 2006 (2007) and The Low-Cost, Moderate-Cost, and Liberal Food Plans, 2007 (2007). All four Food Plans are based on 2001-02 data and updated to current dollars by using the Consumer Price Index for specific food items.