Vector Search
Vector search is a technique that converts data such as text, images, and audio into high-dimensional numerical vectors and retrieves semantically related content by computing inter-vector similarity metrics such as cosine similarity or Euclidean distance. Unlike traditional keyword matching, vector search captures semantic relevance, supports fuzzy queries and cross-modal retrieval, and is widely applied in information retrieval, recommendation systems, question-answering systems, and large-scale semantic search.
Core Facts
Semantic search converts text into high-dimensional vectors and finds the most relevant conclusions by computing cosine similarity, allowing queries phrased in different ways to match semantically equivalent results.
向量数据库将文本内容通过嵌入模型（Embedding Model）转化为高维数值向量，检索时通过余弦相似度等算法找出语义最接近的内容
Vector retrieval captures semantic relevance rather than literal matching, which is its essential difference from keyword search.
Timeline (last 90 days)
"Immediate response" requires the recommender system to shift to an online learning or real-time inference architecture, needing to complete topic mapping, recompute ranking scores and instantly override old results within milliseconds to seconds, usually relying on vector retrieval and a ranking layer that can be updated in real time.
Solutions like engram usually need to layer symbolic indexing, project graphs, or hierarchical memory strategies on top of vector retrieval.
Agent memory retrieval usually adopts a hybrid strategy: vector retrieval captures semantic similarity, metadata filtering narrows the candidate set, and a reranking model performs the final ranking based on task context and preference weights.
Blue-green deployment in the vector retrieval scenario means that while the old index continues to serve, all vectors are fully built in another set of indexes under the new model identity, and after validation metrics are met the query routing is switched atomically, falling back to the old index if problems arise.
检索策略的效果高度依赖数据特征——在FAQ上BM25可能打败向量检索，在语义模糊的查询上则相反。
Traditional vector retrieval relies on cosine similarity or dot products to find the nearest entries in a high-dimensional vector space.
Traditional vector retrieval is a static geometric metric that only concerns itself with angles or distances between vectors and does not explicitly model the interaction forces between vectors.
In multi-tenant RAG scenarios, if vector retrieval does not filter by tenant at query time, cleverly worded questions may retrieve documents belonging to other customers, resulting in cross-tenant data leakage.
The rankings provided by vector retrieval are generally trustworthy, but their absolute similarity values are not, unless they have been calibrated.
Compared to vector retrieval, label-based retrieval has lower computational costs.
35 more timeline events
All Facts (20)
Semantic search converts text into high-dimensional vectors and finds the most relevant conclusions by computing cosine similarity, allowing queries phrased in different ways to match semantically equivalent results.
80%Verified向量数据库将文本内容通过嵌入模型（Embedding Model）转化为高维数值向量，检索时通过余弦相似度等算法找出语义最接近的内容
80%VerifiedVector retrieval captures semantic relevance rather than literal matching, which is its essential difference from keyword search.
75%VerifiedSemantically similar texts are also close in distance within vector space, which is the mathematical foundation of vector retrieval.
70%VerifiedCursor uses an embedding model to transform the entire codebase into a searchable vector space to enable semantic search.
70%VerifiedVector retrieval based on semantic similarity is good at finding content that 'sounds similar' but struggles to capture causal chains, temporal dependencies, and hierarchical structures between pieces of information.
65%VerifiedImproving the distinctiveness of Skill descriptions can increase the hit rate; the writing approach should shift from 'what this Skill is' to 'when to use it.'
65%Verified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
65%UnverifiedAgent框架通过工具调用、检索增强生成（RAG）、记忆管理系统、规划与反思机制来弥补大语言模型的固有限制
90%Unverified向量检索无法精准匹配无语义的业务编号和专业术语，这是技术本身的能力边界而非实现缺陷
85%UnverifiedSemantic search can match results even when the query terms differ from the words in the file, as long as the meanings are similar, which is fundamentally different from keyword-based full-text search.
60%UnverifiedWhen memory scales to hundreds of MB or even GB, a pure file-based approach may still need to be combined with indexing mechanisms and vector retrieval technologies to balance simplicity and performance
60%UnverifiedModern RAG systems widely adopt vector retrieval technology, converting text into high-dimensional vectors via embedding models and computing cosine similarity to find relevant documents.
60%Unverified"Immediate response" requires the recommender system to shift to an online learning or real-time inference architecture, needing to complete topic mapping, recompute ranking scores and instantly override old results within milliseconds to seconds, usually relying on vector retrieval and a ranking layer that can be updated in real time.
50%UnverifiedSolutions like engram usually need to layer symbolic indexing, project graphs, or hierarchical memory strategies on top of vector retrieval.
50%UnverifiedAgent memory retrieval usually adopts a hybrid strategy: vector retrieval captures semantic similarity, metadata filtering narrows the candidate set, and a reranking model performs the final ranking based on task context and preference weights.
50%UnverifiedBlue-green deployment in the vector retrieval scenario means that while the old index continues to serve, all vectors are fully built in another set of indexes under the new model identity, and after validation metrics are met the query routing is switched atomically, falling back to the old index if problems arise.
50%Unverified检索策略的效果高度依赖数据特征——在FAQ上BM25可能打败向量检索，在语义模糊的查询上则相反。
50%UnverifiedTraditional vector retrieval relies on cosine similarity or dot products to find the nearest entries in a high-dimensional vector space.
50%UnverifiedTraditional vector retrieval is a static geometric metric that only concerns itself with angles or distances between vectors and does not explicitly model the interaction forces between vectors.
50%