from backend.embeddings import get_embeddings
from backend.vector_store import VectorDB

db = VectorDB()

def recommend_similar(article_text, top_k=3):
    query_embed = get_embeddings([article_text])[0]
    return db.search(query_embed, k=top_k)