Verified65% confidenceFactTime unknown
MemGAS uses a Gaussian Mixture Model (GMM) to divide historical memories into relevant and irrelevant sets, in order to determine which old memories are worth establishing associations with.
3
Sources
65%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
MemGAS: Multi-Granularity Memory Association for Precise AI Agent Recall
bilibiliAgent智能体深度研究院
Related Entities
Related Claims
Unverifiedclaude-mem可能采用了类似RAG（检索增强生成）的技术架构，将压缩后的历史上下文存储在向量数据库中进行语义检索68% similarUnverifiedThe mainstream approaches to dealing with the statelessness of large models fall into three categories: directly concatenating conversation history into the context window, encoding memories as embedding vectors in a vector database for retrieval, and having the model autonomously decide which information is worth writing to persistent storage and reading it back on demand.67% similarVerifiedVector memory databases convert historical conversation content into high-dimensional semantic vectors, and during new conversations dynamically retrieve relevant memory fragments via similarity search and inject them into the context window.67% similarUnverifiedThe memory system is implemented in three layers: curated records (current truth), a decision graph (recording the decision-making process), and complete history (all content, searchable).67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/20093API
curl https://kongchang.com/api/v1/knowledge/claims/20093MCP
get_claim(id=20093)