Abstract

In recent years, machine learning (ML) based building analytics have been developed for diverse building services. Yet, the widespread sharing of building data, which underpins the establishment of ML models, is not a common practice in the buildings industry today. Clearly, there are privacy concerns. There are studies on protecting building data, e.g., to k -anonymize building data; yet these studies are computational methods. The root causes of why building operators are or are not willing to share data are unclear. In this paper, we study the problem of willingness to share building data. First, we justify our study by investigating the field to show that data sharing is indeed limited. Second, we examine the issue of the willingness to share building data from the perspective of a social science study. We observe that the intention to disclose (i.e., decision making on data sharing) is not only based on perceived risks , but also on perceived benefits. We leverage the privacy calculus theory and present a systematic study. We develop hypotheses, design a questionnaire, conduct a survey involving 95 building operators and service providers around the world, and analyze the results, wherein we quantify how various factors influence the willingness to share building data. We further enhance our results by a small scale interview. Third, we use trust, an important factors to the intention to disclose, to develop a trust model with differentiable trust levels. Such model provides building operators a mechanism to share data besides a 0-and-1 choice. We present a case study where we enhance an existing building data anonymization platform, PAD with the trust model. We show that the enhanced PAD has a substantially smaller computation workloads.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call