PX-25-A006 - Large Language Models for Investigating Co-Occurrences in Multi-Domain Indoor Environmental Quality ✓ Most Recent

Building occupants are continuously exposed to multiple indoor environmental stimuli, including thermal, visual, acoustic, and indoor air quality (IAQ) related factors. Historically, research into these stimuli has often addressed them in isolation. Thermal comfort studies, for example, extensively explored the effects of temperature and humidity on occupant satisfaction (De Dear and Brager 1998; Ma et al. 2021), while visual comfort research focused on lighting quality and glare-related issues (Carlucci et al. 2015). Similarly, studies on acoustic comfort assessed noise levels and their impact on cognitive performance (Jahncke et al. 2011), and and investigations into IAQ emphasized the health risks associated with pollutants such as CO2 and VOCs (Wargocki et al. 2000). Contrasting with this single-domain focus, multi-domain IEQ research explores the interplay between at least two different environmental domains, such as visual and thermal factors. Recent findings indicate that interactions across multiple domains can significantly influence occupant perceptions and overall well-being. For instance, it has been demonstrated that warm lighting can create a perception of increased warmth (Bellia et al. 2021), and that enhancements in IAQ coupled with increased noise from devices like portable air cleaner can lead to occupant dissatisfaction (Deng et al. 2023).

Our comprehensive understanding of the co-occurrence of IEQ problems and their collective impact on occupant dissatisfaction is still limited. This gap in knowledge remains limited regarding which IEQ domain issues frequently co-occur and how they collectively contribute to occupant dissatisfaction. Obtaining robust observational data from a large number of buildings is not only labor-intensive and costly but also complicated by the challenge of standardizing occupant feedback across varied contexts. However, recent advancements in large language models (LLMs) offer promising avenues for extracting insights from textual feedback provided by occupants (Ma et al. 2023; Zhang et al. 2024; Sadick and Chinazzo 2025). Leveraging LLMs to analyze extensive datasets of user-generated reviews offers a novel approach to systematically identify and understand patterns of IEQ issues, thereby enhancing our ability to address factors that affect occupant satisfaction in indoor environments.