Characterizing Office Smellscape Perception through Participatory Approaches and Natural Language Processing (NLP) Techniques

Understanding how individuals experience and perceive the indoor olfactory environment, or smellscape, is essential for achieving indoor air quality (IAQ) that is positively perceived by occupants. Improving the indoor smellscape requires effective methods to assess it, particularly in terms of perceptual dimensions that capture subjective experiences in occupied buildings. This study aims to identify key descriptors of office smellscape perception and establish future research directions towards the development of a measurement model able to evaluate olfactory perception. To achieve this, a set of 118 English-language descriptors, originally developed for assessing other sensory domains, was tested for its applicability to indoor smellscapes using an expert-based participatory approach, resulting in the selection of 37 attributes. An online survey was then conducted with 200 English-speaking participants, ensuring a gender-balanced sample. Participants completed the questionnaire while in their offices, describing their emotional responses to the perceived olfactory environment. This led to the identification of an additional 190 attributes. Natural Language Processing (NLP) techniques were employed to cluster the combined set of 227 attributes (37 expert-validated and 190 participant-derived), resulting in 80 distinct descriptors that capture people’s emotional and sensory responses to office olfactory environments. The most frequently used attributes included "familiar" (50%), "clean" (20%), "stale" (20%), "fresh" (15%), "musty" (13%), "warm" (13%), "sterile" (12%), and "artificial" (11%). Offering a comprehensive understanding of our affective responses to indoor smellscapes, this study lays the groundwork for a structured model of smellscape perception. This model will provide assessment scales based on key perceptual dimensions (e.g., for post-occupancy evaluations) and indices to guide the design phase.