This project integrates the Neo4j graph database to efficiently manage and store relational data, particularly focusing on storing and retrieving document contents and chat histories. Neo4j's powerful graph-based data structure allows for intuitive relationships between entities, making it ideal for use cases such as recommendation systems, question-answering pipelines, and context-aware conversational agents.

Key Features:

• Document Storage: Store and retrieve PDF contents as nodes in the graph, enabling rich querying capabilities.
• Chat History Management: Persist chat sessions and their message histories, allowing for easy retrieval and analysis of past interactions.
• Graph-Based Relationships: Leverage Neo4j's graph model to manage complex relationships between documents, sessions, and user interactions.
• To get started, ensure that you have the Neo4j Python driver

login into Neo4j Db

• Create new instance
• Store required credentials(url,username, password) inside environment variable.

Create Virtual environment

conda create -n langhainVenv python=3.10 -y

conda activate langhainVenv

pip install -r requirements.txt

Store Wikipedia page as Nodes and Relationship inside Neo4j DB

• app.py : User interface for initalize another variables and communaite with user.
• Configuration.py : It loads the environment variables.
• database.py : Initilize neo4j database driver and made connection.
• ResponseGenerator : split the input into documents for model undersatnding.
• DataRetriever : This class retrive data from Neo4j db. It is perform embedding, vector search and then finally perform similarity search to extract answer.
• LLMTransformer : This class will convert user entered wikipedia page in graph format, which futher store into Neo4j Db.
• WikipediaHandler.py : This class fetch data from Wikipedia page and then split into document.

Store user selected PDF file into Neo4j DB

• app.py : streamlit interface for user connectivity and calling subclasses.
• chat_manager : it stores session and message history.
• config.py : read all environment variables.
• document_handler.py : This class process the user selected pdf and store it into Neo4j db.
• embedding_store.py : This class perform embedding and vector store on pdf file.
• neo4j_handler.py : This class perform various operation related to Neo4j DB.
• rag_chain.py : implementaion of rag pipeline for data retriver.