embedsim 0.1.0

Release Notes - embedsim v0.1.0

A Python library for measuring semantic similarity and detecting outliers in text collections using
embeddings.

Features

Core Functionality:

• pairsim() - Compare two texts using cosine similarity of their embeddings
• groupsim() - Analyze text collections and identify outliers using centroid-based coherence scoring

Embedding Model Support:

• OpenAI models (openai-3-small, openai-3-large) via API
• Local sentence-transformer models (Jina v2, MiniLM, etc.) for privacy and offline use
• Configurable via function parameters or environment variables

Use Cases:

• Content moderation and off-topic detection
• Document clustering and outlier identification
• Quality assurance for generated content
• Search relevance scoring
• Duplicate detection