Knowledge-Driven Approach for Quality Assessment of HAR Data Sets: An Automated Tool

The quality of Human Activity Recognition (HAR) data sets can be affected by a number of factors, such as processes for governance, diversity of data collection protocols, information understanding, labelling accuracy, imbalanced data and missing or irrelevant information. These factors can be assessed by the construction of standard quality metrics and quality assessment frameworks and tools. This research outlines a knowledge-driven based approach for the qualitative assessment of HAR data sets, supported by an automated tool. Through the utilization of domain expertise, semantic modelling, and machine learning techniques, we established a set of quality metrics and criteria that effectively encapsulate the fundamental attributes of the HAR data set. The automated tool employs specific methodologies to evaluate quality, such as the Threshold-based technique and the Aggregative-based approach. The proposed research provides a novel method and tool that improves the evaluation of the quality of HAR data set, thereby facilitating more informed decisions and enhancements in the HAR domain.