backend/semantic_search.py

import logging
from pathlib import Path
import time

import lancedb
from sentence_transformers import SentenceTransformer


# Setting up the logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Start the timer for loading the QdrantDocumentStore
start_time = time.perf_counter()

proj_dir = Path(__file__).parents[1]

# Log the time taken to load the QdrantDocumentStore
db = lancedb.connect(proj_dir/"lancedb")
tbl = db.open_table('arabic-wiki')
lancedb_loading_time = time.perf_counter() - start_time
logger.info(f"Time taken to load LanceDB: {lancedb_loading_time:.6f} seconds")

# Start the timer for loading the EmbeddingRetriever
start_time = time.perf_counter()

name="sentence-transformers/paraphrase-multilingual-minilm-l12-v2"
st_model = SentenceTransformer(name, device='cuda')

# used for both training and querying
def embed_func(query):
    return st_model.encode(query)

def vector_search(query_vector, top_k):
    return tbl.search(query_vector).limit(top_k).to_list()

def retriever(query, top_k=3):
    query_vector = embed_func(query)
    documents = vector_search(query_vector, top_k)
    return documents


# Log the time taken to load the EmbeddingRetriever
retriever_loading_time = time.perf_counter() - start_time
logger.info(f"Time taken to load EmbeddingRetriever: {retriever_loading_time:.6f} seconds")