Year: 2026
Type: preprint
Abstract: Choosing a meal often begins with a simple question: what can I make with the ingredients I already have? A recommendation is useful only if it fits those ingredients or clearly indicates what is missing. INGREDIA was built to support this task. It combines ingredient matching with text-based and practical signals to rank recipes from an Indian recipe collection. We evaluated it offline against a TF–IDF/KNN baseline using ranking, coverage, and feasibility measures. The results improve on the baseline under the evaluation setup used here, although some components need more testing. These results describe performance on the available dataset; they do not show that users are more satisfied or that household food waste is reduced.
Source: Research Square
Authors: C M Pranith, Saicharan S
Institution: SRM Institute of Science and Technology
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225782739