Kumar, VimalBainbridge, DavidPhua, Thye Way2026-11-012026-11-012026https://hdl.handle.net/10289/18730Data provenance is essential for validating the origin, history, and trustworthiness of information. However, despite extensive research, provenance systems have seen limited mainstream adoption. This thesis argues that this gap stems from two fundamental deficiencies: (i) the lack of temporal awareness in existing models and (ii) the detachment of provenance metadata from the primary data. Current systems often treat versioning as symbolic markers rather than capturing the content of change, and they store metadata in centralised repositories that are prone to desynchronisation. To address these challenges, this research proposes a framework that integrates temporal data evolution with embedded provenance. The thesis posits that provenance must be integral to the data artefact itself to ensure portability and integrity. A prototype system was developed using a FUSE-based filesystem, employing delta encoding to capture fine-grained changes and a custom block-based format to embed provenance metadata and version histories directly within files. This architecture ensures that lineage information remains inseparable from the data it describes. The system was evaluated through microbenchmarks and a macrobenchmark involving a full Linux kernel compilation. The evaluation confirms that the delta-based storage model is efficient, achieving a sixty-fold reduction in space compared to full snapshot retention. Historical state reconstruction incurs linear overhead relative to version depth, remaining manageable for typical file lifespans. Although the prototype introduces a performance overhead of approximately 7.4% during intensive workloads, it successfully validates the feasibility of the approach. The findings demonstrate that integrating temporal awareness with embedded metadata significantly improves the accuracy and completeness of provenance tracking, offering a robust foundation for accountability in dynamic environments.enAll items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.data provenanceThesis