Monitoring particulate matter (PM) in mine sites is challenging owing to the intermittent PM emission and dispersion over widespread areas at these locations. In these circumstances, multiple monitoring stations, including PM sensors, should be installed to monitor PM distributions throughout the site. However, this may result in high operation costs. Monitoring based on low-cost PM sensors is advantageous with respect to cost, though its poor data quality is problematic. Yet, big data processing techniques can be utilized to remedy this drawback. This study proposed a strategic solution to enhance the data reliability of low-cost PM monitoring through big data processing. The significance of an effective calibration was emphasized to overcome the poor quality of low-cost PM sensor data. In particular, this study focused on the usefulness of blind calibration for adjustment of background PM levels and the traditional linear regression-based calibration through direct comparison between low-cost and reference sensor data. The applicability of an ANN (Artificial Neural Network) algorithm was also considered for automatic decision making and implementation of the whole calibration, as well as for global optimization in nonlinear regression coefficients with a high complexity.