Unverified50% confidenceFactExact time
In LangChain, Embeddings, VectorStore and Retriever form the standard abstraction: documents are split and embedded into the store, and at query time the question is first embedded and then the Top-K relevant chunks are retrieved and spliced into the prompt for the LLM.
1
Sources
50%
Confidence
Long-term
Relevance
11/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe contextual retrieval technique involves prepending the document title and section path to each chunk before embedding, or using an LLM to generate a brief summary describing the chunk's position within the document.66% similarUnverifiedLlamaIndex's SentenceWindowNodeParser and LangChain's ParentDocumentRetriever implement the parent-child chunking / small-to-big retrieval pattern.66% similarUnverifiedElasticsearch基于倒排索引实现高速文本检索，其核心逻辑是词频统计和TF-IDF权重计算，无法理解语义关系或建立跨文档关联64% similarUnverifiedSemantic Memory stores conversation content in a vector database via Embedding and recalls it by semantic similarity at retrieval time; its pipeline is the same as RAG, except that the knowledge source is historical conversation rather than external documents.63% similarUnverifiedEmbedding Gemma 2 requires a short task prompt to be prepended to search queries, documents to be formatted as "title + body", and the templates used during training and inference must be completely consistent.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1122929API
curl https://kongchang.com/api/v1/knowledge/claims/1122929MCP
get_claim(id=1122929)