Mathematical analysis of the data about perfume collected on the Internet, shows how the unique combination of flavors found in different fragrances, contribute to the popularity and high ratings among consumers.
Vaiva Vasiliauskaite and Tim Evans from Imperial College London, UK, presented the results of their study, after examining about a thousand flavors to more than 10 000 perfumes.
Each perfume is a unique combination of various aromatic ingredients, oils and chemical molecules, which together form a harmonious flavor. Perfume is often described using the so-called notes, for example, the note of vanilla or musk, but the combinations of notes are called chords. Assuming that a certain perfume is popular largely thanks to its particular fragrance, the researchers seek to understand what is a popular smell. To achieve this, the structure of the perfume and its constituents are studied in accordance with the principles of the mathematical field known as complex network analysis.
Using this approach, scientists have identified what notes and chords are used more often, and identified the most popular and frequently used scents in perfumes with the highest rating. The researchers found that the most popular notes and most commonly used chords do not necessarily correspond to the highest ratings consumer. For example, the combination of Jasmine and mint has made a significant contribution to the increase in rating, but were underrepresented in the studied fragrances.
The researchers also determined which notes are able to give chords a special expressiveness. They found that notes with high popularity, such as the musk and vanilla, as a rule, increased the chord, and floral notes. Further analysis showed that the major fashion brands produce many of the most successful perfumes, but the popularity of the product seems to be nothing to do with its price or with the time of its release.
The results show that complex network analysis can be a useful tool for perfume companies to explore new combinations, increasing the chances of product success.
The authors of the project say: “We have studied data about spirits and descriptors of odor as a complex network, to get an idea about how the composition of notes, called chords, affect the success of fragrances. We were able to determine the chords that are usually present in spirits, receiving the highest ratings from customers, for example, the chord of lavender, oak moss and geranium. Our work represents a Foundation which will be a tool for the perfumers in the study of multidimensional space of flavors. It can serve as a starting point for statistical analysis of fragrances based on simple properties and customer feedback, and also to provide the technology for the systematic selection of ingredients and to act as an artificial “nose”.