Hypergraph Embedding Indexing for Efficient Dense Vector Retrieval
2026-08-24 • Information Retrieval
Information RetrievalArtificial Intelligence
AI summaryⓘ
The authors propose a new search method called Hypergraph Embedding Index (HEI) that organizes documents based on combinations of important features in their data, rather than treating each document as a single point. This approach allows for faster candidate selection like traditional search methods while keeping the benefits of deep data representations. They also show that using several different hypergraphs improves how many relevant documents are found without making the system too complex. Additionally, the authors identify a new way to measure how well an embedding works with this method, which can help improve search efficiency.
dense vector retrievalapproximate nearest neighbor (ANN)embeddinglatent dimensionshypergraphinverted indexsemantic searchcoordinate-inverted indexingactivation diversitycandidate generation
Authors
Kishore Konda
Abstract
Dense vector retrieval has become the foundation of modern semantic search, yet existing approximate nearest neighbor (ANN) indexes treat an embedding as an indivisible point in a high-dimensional space. In this work, we propose the Hypergraph Embedding Index (HEI), a framework that instead organizes documents according to combinations of highly activated latent embedding dimensions. This formulation enables inverted-index style candidate generation while preserving the semantic ranking capabilities of dense embeddings. We further demonstrate that constructing multiple complementary hypergraphs substantially improves retrieval coverage without the combinatorial growth associated with increasing the dimensionality of a single hypergraph. Finally, we establish that the statistical properties of embedding activations strongly influence coordinate-inverted indexing efficiency, introducing \emph{activation diversity} as a diagnostic metric governing embedding indexability in coordinate-inverted frameworks.