RAG-Enhanced Large Language Models for Dynamic Content Expiration Prediction in Web Search

In commercial web search, aligning content freshness with user intent remains challenging due to the highly varied lifespans of information. Traditional industrial approaches rely on static time-window filtering, resulting in "one-size-fits-all" rankings where content may be chronologically recent b...