Semantic Enrichment and Visualisation of HAR Data: Ontology Development from Unstructured Data Sets Metadata

Human Activity Recognition (HAR) is crucial for a wide range of applications, including healthcare monitoring and smart home systems. Despite significant advancements, existing HAR systems often face challenges related to the semantic interpretation and meaningful visualization of the data set metadata and its quality. This research presents a novel approach to address these challenges through ontology-based semantic enrichment and visualisation of unstructured metadata to facilitate better understanding and analysis of HAR data. Text mining techniques were utilised for extracting information from diverse file formats to develop an ontology. The extracted information was pre-processed and cleaned using removal of punctuation, lower casing, removal of stop words, tokenisation, lemmatisation, and removal of non-alphabetic tokens. The processed tokens were compared to a predefined key-value pairs data dictionary to develop the ontology via matched tokens. The proposed approach was demonstrated to achieve the effectiveness through a case study utilising the metadata from the Opportunity dataset. The effectiveness of the ontology is evaluated and validated using a quantitative approach involving metrics such as coverage, consistency, and precision. The results demonstrate that the proposed approach not only enhances the semantic complexity of HAR data but also provides useful visual representations, hence improving the understanding and decision-making in HAR applications.