The post asks for ideas on a question, but the community members in the comments do not seem to fully understand the question. One community member mentions dealing with refresh when using the hierarchical node parser, and another community member asks about a more complex rag strategy for updating underlying nodes after documents have changed. However, there is no explicitly marked answer in the provided information.
