statistical regression where the dependent variable can take only two values, to estimate the probability that an observation with particular characteristics will fall into one of the categories
Discovered by embedding cosine similarity (sentence-transformers MiniLM, 384-dim).