WhatToMake: A Semantic Web Application for Recipe Recommendation

Miao Qi, Yarden Neeman, Forest Eckhardt

Abstract: Semantic technology plays a significant role in the current Web development domain, and its collaboration with artificial intelligence techniques promotes the emergence of advanced query systems. These intelligent systems are able to provide more precise answers to users based on their queries. Food, as the essential substance to offer nutritional support to humans, is a frequently searched topic on the internet and many of the information related to food online is encoded through the semantically enriching markup in ontologies such as the FoodOn Ontology and BBC Food Ontology. However, most of these ontologies only focus on the food or recipes themselves without incorporating other dietary restriction information, such as food allergens and food glycemic indexes. And no ontology cares about the substitution of undesired food in recipes instead of simply ignoring them. In this paper, we will introduce a new system, WhatToMake, capable of recommending recipes to a user subject to the information about required ingredients, cooking time restrictions, recipe categories for meal and course, and various dietary needs of users. The most advanced feature of the system is to substitute one ingredient with another best-matched one through three different techniques that reduce the influence on taste of recipes caused by food restrictions.