Buffer Window Memory

​
Node Details

  • Name: bufferWindowMemory
  • Type: BufferWindowMemory
  • Version: 2.0
  • Category: Memory
  • Icon: memory.svg

​
Description

This node uses a window of size k to surface the last k back-and-forth exchanges to use as memory in a conversation. It’s useful for maintaining recent context without storing the entire conversation history.

​
Base Classes

  • BufferWindowMemory
  • (Additional base classes from langchain’s BufferWindowMemory)

​
Parameters

​
Inputs

​
Inputs

  1. Size (k)
    • Type: number
    • Default: 4
    • Description: The number of recent message pairs (back-and-forth exchanges) to keep in memory
    • Description: The number of recent message pairs (back-and-forth exchanges) to keep in memory
  2. Session Id (optional)
  3. Session Id (optional)
    • Type: string
    • Description: A unique identifier for the conversation session. If not specified, a random id will be used
    • Description: A unique identifier for the conversation session. If not specified, a random id will be used
  4. Memory Key
    • Type: string
    • Default: “chat_history”
    • Description: The key used to store and retrieve the chat history in the memory
    • Description: The key used to store and retrieve the chat history in the memory

​
Functionality

  1. Initialization:
    • Creates a BufferWindowMemoryExtended instance with the specified parameters.
    • Connects to a database using the provided DataSource.
  2. Message Retrieval:
    • Fetches messages from the database based on the session ID and chatflow ID.
    • Limits the number of messages based on the specified window size (k).
    • Can return messages as either IMessage or BaseMessage types.
  3. Message Storage:
    • Actual message storage is handled at the server level, not within this node.
  4. Memory Clearing:
    • Clearing of chat messages is also handled at the server level.

​
Use Cases

  • Maintaining recent context in chatbots or conversational AI systems.
  • Limiting the amount of historical data passed to language models to reduce token usage and improve relevance.
  • Providing a sliding window of conversation history for context-aware responses.

​
Integration

This node is designed to work within a larger system, likely a conversational AI framework. It interfaces with a database to persist and retrieve conversation history, making it suitable for applications that require stateful conversations across multiple interactions or sessions.