The invention provides a note association recommendation method and
system based on abstract retrieval and LLM analysis, and relates to the technical field of
artificial intelligence and
natural language processing, and the method comprises the steps: receiving a user request, including a current note, and generating an abstract for the current note; performing vector retrieval based on the abstract, and obtaining a related note identifier
list from a historical note
library; generating a
unique identifier based on the current note content and the related note identifier
list; querying the cache by using the
unique identifier, and returning a recommendation result if the cache is hit; if not, using the
unique identifier to acquire the distributed lock, if acquisition succeeds, calling a large
language model (LLM) to deeply analyze the current note and the note corresponding to the related note identifier
list, generating a recommendation result, returning the recommendation result to the user, and releasing the distributed lock; and if the acquisition fails, executing the asynchronous retry process. According to the method, mechanisms such as double
fingerprint caching, a distributed lock and
exponential backoff retry are introduced, so that the
cache hit rate and accuracy are improved, and efficient recommendation of associated notes is realized.