Fragrance to vector as scent technology

In order to understand human recognition on scent, it is required to learn sequencing odors and its expression. When using only distributed representations of words about odor and fragrance for scent recognition, they lose quantitative scales. This study is about a machine learning between textual and vector representation of scent. In this paper, scent vector and semi-supervised learning as a base of human expression for odors and fragrances are proposed. First, related researches are shown in section 1. In section 2 and 3, concept of scent technology using sensor data and machine learning is described.