DataScent
info

With DataScent, behaviour and the sensory power of perfume are fused via Artificial Intelligence.

It began in March 2018 at the Royal College of Art as a speculative digital afterlife project using the freely available IBM Watson personality analysis API.

One of the outputs of the analysis is a chart of the subject’s “Big Five personality traits” also known as the OCEAN model (see Sunburst Chart Visualization). Volatility grouping created by W.A. Poucher, which lists fragrant materials in order of volatility and groups them under respective evaporation coefficients (perfume notes) that range from 1 to 100, is used.

The percentage outputs of personality traits are interpreted within the context of evaporation, meaning a trait marked 90% would classify as a 90 on the 1-100 scale. Notes that can be sensed within varying time frames after the application of a perfume are traditionally classified within a “fragrance pyramid” consisting of top/head notes, middle/heart notes, and base notes. The higher the percentage of a personality trait, the longer the evaporation process. Percentages are translated to whole numbers and ratios are created and applied to volatility grouping as follows:

Top Notes: 1 to 14 (most volatile)
Middle Notes: 15 to 60
Base Notes: 61 to 100 (least volatile)

A long list of fragrances such as amber, blackberry, sandalwood, lemon, rose, etc. are used to create different combinations of perfumes.