similarities

Why we included this project

Similarities is a Python library that handles similarity scoring and semantic search for both text and images, so teams that only need retrieval logic can skip wiring several libraries together. Text support spans lightweight lexical methods like BM25, TF-IDF, and SimHash up through SentenceBERT and CoSENT style embeddings, and the supported search backends (Faiss, ANNOY, HNSW) keep searches fast on large corpora. On the image side it wraps CLIP for text-to-image and image-to-image retrieval, with the same indexing approach applying to both domains. It also bundles a few practical extras: a CLI that builds indexes and serves a search endpoint, community detection for grouping related results, and paraphrase mining for finding near-duplicates. For projects that want controllable retrieval without standing up a separate vector database service, this is a convenient middle ground.