Unverified50% confidenceTradeoffExact time
Keyword search easily misses files that are semantically related but differently named, vector search has uneven embedding quality in the code domain, and LSP depends on correct project configuration to work.
1
Sources
50%
Confidence
Long-term
Relevance
9/21/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCode retrieval by programming agents typically relies on a combination of strategies including keyword/symbol search (e.g., grep/ripgrep), vector embedding similarity search, and symbol reference navigation via the Language Server Protocol (LSP).72% similarVerified关键词搜索基于精确字符匹配无法理解同义词，语义搜索依赖向量嵌入可捕捉语义关联，两者结合的混合检索效果优于单一方法70% similarUnverifiedTraditional search engines rely on keyword matching, requiring target pages to contain the user's input terms to be retrieved; semantic search, however, can find results even when the target page does not contain the exact search terms, as long as the semantics are similar.69% similarUnverifiedKeyword-based search algorithms perform reasonably well with small corpora but degrade as the content library crosses a critical size threshold69% similarUnverifiedTraditional search engines perform keyword matching based on inverted indexes and statistical methods such as TF-IDF, making it difficult to establish effective document relevance rankings when queries are too short or too vague.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/939128API
curl https://kongchang.com/api/v1/knowledge/claims/939128MCP
get_claim(id=939128)