Update app.py

app.py

@@ -6,10 +6,10 @@ from langchain.embeddings import HuggingFaceEmbeddings
 from langchain.vectorstores import FAISS
 from langchain.llms import CTransformers
 from langchain.chains import ConversationalRetrievalChain
-from dl_hf_model import dl_hf_model
 from ctransformers import AutoModelForCausalLM
 from langchain_g4f import G4FLLM
 from g4f import Provider, models
+import spacy
 import requests
 # Define the path for generated embeddings
 DB_FAISS_PATH = 'vectorstore/db_faiss'
@@ -83,11 +83,26 @@ llm = load_llm()
 
 # Create a conversational chain
 chain = ConversationalRetrievalChain.from_llm(llm=llm, retriever=db.as_retriever())
+# Initialize spaCy with the Japanese model
+nlp = spacy.load("ja_core_news_sm")
 
 # Function for conversational chat
 def conversational_chat(query):
-    result = chain({"question": query, "chat_history": st.session_state['history']})
-    st.session_state['history'].append((query, result["answer"]))
+    result = None
+    similarity_score = 0
+    # Set a threshold for similarity (you can adjust this)
+    similarity_threshold = 0.8
+    while similarity_score <= similarity_threshold:
+        result = chain({"question": query, "chat_history": st.session_state['history']})
+        doc = nlp(result["answer"])
+    
+        # Define the Japanese phrase to match
+        target_phrase = query  # Replace with your desired Japanese phrase
+    
+        # Check for similarity
+        similarity_score = doc.similarity(nlp(target_phrase))
+        
+    st.session_state['history'].append((query, result["answer"]))   
     return result["answer"]
 
 # Initialize chat history