Unverified50% confidenceConceptExact time
Recall@k is defined as: the number of expected text chunks that were retrieved divided by the total number of expected text chunks.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe Recall@k metric is defined as the proportion of the top k results that contain known relevant documents.79% similarUnverifiedDreaming reorganizes the index every day, which in essence adapts the index structure to the currently accumulated memory distribution; it is a latency-insensitive offline maintenance task decoupled from online inference.58% similarUnverifiedThe entire message history in the messages list is billed as input tokens on every call, the model cannot remember the previous call, and must receive the full context from scratch each time.57% similarUnverifiedThe Recall step uses semantic retrieval, keyword retrieval, relational graphs, and temporal retrieval in parallel, then merges and ranks the results to return relevant memories within a budget.57% similarUnverifiedThe key to maintaining the conversational context of a ReAct agent is to keep the message list outside the loop and append new content to the list each round; otherwise the agent will lose its memory and be unable to reason based on historical context.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1021492API
curl https://kongchang.com/api/v1/knowledge/claims/1021492MCP
get_claim(id=1021492)