Semantic web-enabled outlier and missing value detection and replacement in smart buildings

Digital twins have the potential to leverage AI in buildings. The quality of AI algorithms is dependent on the quality of the input data and its preprocessing. This paper discusses the potential of using semantic web technologies in preprocessing tasks. After reviewing state-of-the-art initiatives in this field, a data integration method is introduced based on semantic web technologies. This integrated data is used in approaches to find outliers and missing values in time series data and in two semantic similarity-based imputation methods. The paper shows that semantic web technologies can enhance preprocessing tasks, by applying both explicit and implicit reasoning.