Trust Assessment on Data Stream Imputation in IoT Environments

Abstract

In the era of internet of Things, stream data emitted by sensors may rise quality issues such as incompletenes caused mainly by sensors failure or transmission problems. It is therefore necessary to recover missing data because missing values can impact decision making. Within this landscape, trust on data imputation is a key issue for helping stakeholders involved in such process. In this paper, we address the problem related to the trustworthiness on imputed data streams in IoT environments. We propose here a method called CSIV (Confidence Score for Imputed Values) to assess trust by assigning a confidence score to imputed data. CSIV considers both trust score of non-missing values and neighboring sensors. We have evaluated CSIV on real datasets using accuracy and trustworthiness as evaluation metrics. Experiments show that CSIV is able to assign correctly a trust score to the imputed values.