Skip to main content
Peer-Reviewed Research Database

The science behind the engine

Every compound-food link in CompKitchen comes from a peer-reviewed source. Search 38,144 flavor-relevant papers — part of a 44,648-paper indexed corpus — on aroma compounds, odor thresholds, GC-MS analyses, and sensory studies that power the pairing engine.

From the CompKitchen team
Original research
Featured Preprint · Zenodo · CC BY 4.0

Unsupervised Recovery of Food Taxonomy from Volatile Compound Profiles: A UMAP Analysis of 450 Ingredients

Without supervision or human-coded labels, a UMAP projection of 7,464 volatile compounds across 450 ingredients reproduces conventional food categories — and surfaces cross-category molecular bridges that traditional culinary intuition misses.

Johnson, M. C. (2026). Zenodo. DOI: 10.5281/zenodo.19719459

More original analyses in queue — Rychlik integration writeup and per-compound OAV audits.

44,648
Papers indexed
959
Ingredients cited
5,240
Compounds indexed
627,022
Paper–ingredient links
Get one surprising pairing every week — free.
2016 Molecules other

The Unusual Acid-Accumulating Behavior during Ripening of Cherimoya (Annona cherimola Mill.) is Linked to Changes in Transcription and Enzyme Activity Related to Citric and Malic Acid Metabolism

Indexación: Web of ScienceCherimoya (Annona cherimola Mill.) is a subtropical fruit characterized by a significant increase in organic acid levels during ripening, making it an int... Read full abstract on PubMed →

2016 Journal of chromatography. B, Analytical technologies in the biomedical and life sciences sensory panel

Characterization of odor-active compounds of various Chrysanthemum essential oils by gas chromatography-olfactometry, gas chromatography-mass spectrometry and their correlation with sensory attributes.

Volatiles of five kinds of Chrysanthemum essential oils with different manufactures were characterized by descriptive sensory analysis, gas chromatography-olfactometry (GC-O), gas... Read full abstract on PubMed →

All paper metadata — titles, authors, journals, years, DOIs — is public bibliographic information. Abstract snippets shown are truncated to fewer than 200 characters and link to the original source on PubMed or the publisher. We never reproduce full text.

Sources: PubMed (US National Library of Medicine), Semantic Scholar, Europe PMC, CORE. Papers are selected by targeted search queries for each ingredient and compound in our database.