ScentSee works by merging the art of fragrance with the precision of math. It does this by collecting information on notes and note types from various experts and publications online. Following that, it structures the data, separating the top, heart and base notes. Notes are then classified using a system similar to the Fragrance Wheel, although it is worth noting that our wheel has a bit more “spokes”. The classification phase assigns each fragrance its own signature, meaning how well-represented each of the note types is on every segment. While these signatures are an apparently weaker description of the perfume, they have the advantage of identifying similar perfumes, but which have different notes names (think “bergamot” and “lime”). Finally, in the comparison phase, the perfume signature and the actual notes are actually compared and matched on each segment, assigning match scores in the process. While there are more data science aspects to the matter, including the frequency analysis of the note spectrum, essentially the match score allows the user to quickly navigate similar perfumes. Our state-of-the-art algorithms perform tens of thousands of such comparisons in a split second, so that the best matches are retrieved instantaneously.