The Web has become a major conduit to information repositories of all kinds. Web caches are employed to store web views to provide an immediate response to recurring queries. However, the accuracy of the replicates in web caches encounters challenges due to the dynamicity of web data. We are thus developing and evaluating a web caching system equipped with an efficient refresh strategy. With the assistance of a novel index structure - the Aggregation Path Index (APIX), we built Argos, a web caching system based on the GMD XQL query engine. Argos achieves a high degree of self-maintenance by diagnosing irrelevant data update cases. It hence greatly improves the refresh performance of the materialized web view. We also report preliminary experimental results assessing the performance of Argos compared to the state-of-the-art solution in the literature.