Comment by mahin

It gets embeddings from open AI and compares them using cosine similarity. Stories are weighted by similarity, dissimilarity and HN score.