Resistive Random-Access Memory (RRAM) is a promising non-volatile memory technology due to its high-speed operation, low power consumption, and scalability. Since the early study on oxide-based RRAM, significant advancements have been made in understanding resistive switching mechanisms, improving material compositions, and integrating RRAM for neuromorphic computing applications. This paper revisits the resistive switching and conduction mechanisms in oxide-based RRAM and updates previous study findings with modern developments in interface engineering, multi-layer stacking, and novel material innovations such as graphene oxide and high-k dielectrics. Key advancements in conduction models, including filamentary conduction, Poole-Frenkel emission, and trap-assisted tunneling, are discussed, supported by recent experimental and theoretical findings. Furthermore, existing experimental data are reanalyzed using modern insights, and potential applications of RRAM in artificial intelligence accelerators and edge computing are proposed. The results highlight the improved endurance, retention, and switching dynamics of oxide-based RRAM, making it a viable candidate for next-generation memory solutions.