On Approximately Searching for Similar Word Embeddings

摘要

We discuss an approximate similarity search for word embeddings, which is an operation to approximately find embeddings close to a given vector. We compared several metric-based search algorithms with hash-, tree, and graph-based indexing from different aspects. Our experimental results showed that a graph-based indexing exhibits robust performance and additionally provided useful information, e.g., vector normalization achieves an efficient search with cosine similarity.