Semantic Search Nodes
Semantic Search Nodes are specialized network participants that maintain distributed indexes of Memory Shards and enable efficient discovery of relevant vectors through semantic similarity matching.
How Semantic Search Works
Unlike traditional keyword-based search, semantic search operates in high-dimensional vector space. When a consumer agent queries the network, the search nodes:
Convert Query to Vector: The query text is embedded using the same model standard as the stored Memory Shards.
Similarity Calculation: The node computes cosine similarity between the query vector and all indexed Memory Shards.
Ranking & Retrieval: Results are ranked by relevance score, and the top matches are returned to the consumer.
Node Architecture
Semantic Search Nodes utilize advanced indexing structures such as:
HNSW (Hierarchical Navigable Small World): Enables sub-linear search time even with millions of vectors.
FAISS (Facebook AI Similarity Search): GPU-accelerated similarity search for high-throughput scenarios.
Distributed Indexing: Nodes maintain sharded indexes to handle the scale of the global Memory Shard registry.
Incentive Model
Search nodes are rewarded in protocol assets for:
Successfully routing queries to completed transactions
Maintaining high uptime and low latency
Contributing to the distributed index infrastructure
This creates a competitive market for search services, ensuring fast and reliable discovery of Memory Shards.
Last updated