Italian gastronomich recipes dataset

USAGE LICENSE<br> Creative Commons Attribution 4.0 International Public License FILE CONTENTS<br> dataset<br> |_ foods folder containing files about foods dataset (starting data get from giallozafferano.it)<br> | |_ CSV folder containing files in .csv format<br> | | |_ categories.csv <br> | | |_ foodDataset.xlsx each of other .csv files derived from sheet of this excel file<br> | | |_ ingredients.csv <br> | | |_ ingredientsClasses.csv <br> | | |_ ingredientsMetaclasses.csv <br> | | |_ preparations.csv <br> | | |_ recipes.csv <br> | |_ TXT folder containing files in .txt format<br> | |_ scorpored values folder containing values in textData scorpored by type of data<br> | | |_ category-cost-difficulty.txt<br> | | |_ ingredients.txt<br> | | |_ names.txt<br> | | |_ preparations.txt<br> | | |_ preparationTime.txt<br> | |_ textData.txt .txt version of dataset/foods/CSV/foodDataset.xlsx file<br> |_ survey_answers folder containing the files about the results of the surveys on the food preferences of the dataset<br> | |_ sorts folder containing the results of three questions of the survey in which the users were asked to sort some foods by preference<br> | | |_ sort1.csv each of the three file csv contain the survey id, and then the food ordered by the user (each row represent the answer of a user)<br> | | |_ sort2.csv<br> | | |_ sort3.csv<br> | |_ answers.csv results of the surveys (TRANSLATION NOTE: survey ID =&gt; ID-sondaggio; user ID =&gt; ID-utente; answer ID =&gt; ID-risposta)<br> | |_ labels.txt labels of the samples in samples.txt <br> | |_ samples.txt couples of food extracted from the sorts in dataset/survey_answers/sorts/ translating the preference sorts in the form of pairwise comparison *<br> |_ readme.txt * Each sample has the form &lt;IDfood1, IDfood2&gt; and has label 1 if IDfood1 is preferred over IDfood2, -1 if IDfood2 is preferred over IDfood1, 0 if there is indifference relationship over IDfood1 and IDfood2<br> The preference sorts have been translated in two phases:<br> 1- search of inconsistencies among the three sorts in order to get the "certain" preference relationships<br> 2- search of transitivity of preferences among the "certain" couples, giving indifference relationship to those for which no transitivity has been found. RECIPES DATASET DESCRIPTION<br> the description refer to dataset/foods/CSV/foodDataset.xlsx<br> Name italian name of the recipe<br> ID ID associated to the recipe<br> Link link of where the food data has been get<br> Category Name mame of the category (Starter, Complete Meal, First Course, Second Course, Savoury Cake)<br> Category ID ID associated to the category.<br> Cost cost indicator of the food, discrete interval from 1 to 5<br> Difficulty difficuly indicator of the food, discrete interval from 1 to 4<br> Preparation Time integer positive number that indicates preparation time of the food expressed in minutes<br> Ingredient english name of an ingredient of the recipe<br> Ingredient ID ID associated to the ingredient.<br> W weight that the ingredient has in the composition of the interested recipe<br> NOTE: the last three columns repeats for 18 times, leaving empty spaces when the recipe has no ingredients other than those already entered<br> Preparation english name of a preparation performed on the recipe<br> Preparation ID ID associated to the preparation.<br> W weight that the preparation has in the composition of the interested recipe<br> NOTE: the last three columns repeats for 5 times similarly to ingredients in other sheet of the file are reportet all the ingredients, divided in classes and metaclasses, preparations and categories NOTE: in dataset/foods/TXT/textData.txt ingredients and preparation has been vectorized as follow:<br> - each element of the ingredient vector represent the weight of the ingredient class in the recipe. The weight of an ingredient class in a recipe is collected by sum up the weight of the ingredients owned by that particular ingredient class in the recipe.<br> - each element of the preparation vector represent the weight of the preparation in the recipe. # PREFERENCES DATASET DESCRIPTION<br> In the file *dataset/survey_answer/samples.txt* are reported 54 user's orderings in the form of *pairwise comparison*. Thus for each row correspond the ordering of a user. The ordering is written in the form of pairwise comparison, so each element of the ordering are paired with all others (avoiding simmetries). <br> For instance, given the recipes as their ID: 1;2;3 becomes: 1,2;1,3;2,3 The file is written following the *.csv* format.<br> In the file *dataset/survey_answer/lables.txt* are written the correspongin labels of the couples. HOW TO CITE:<br> D.Fossemò, F.Mignosi, L.Raggioli, M.Spezialetti, F.A.D'Asaro. Using Inductive Logic Programming to globally approximate Neural Networks for preference learning: challenges and preliminary results. BEWARE-22, co-located with AIxIA 2022, November 28-December 2, 2022, University of Udine, Udine, Italy.