Get startedGet started for free

Extracting facts from conversation histories

Now to level-up your conversation history! You'll define pydantic classes for extracting facts from conversation histories in a structured way.

This will lay the brickwork for the next and final exercise of the course, where you'll actually perform the extraction.

The necessary pydantic classes have already been imported, and an llm has already been defined.

This exercise is part of the course

Graph RAG with LangChain and Neo4j

View Course

Exercise instructions

  • Define a pydantic ConversationFact class for extracting facts from a conversation; include object, subject, relationship, and session_id fields that match the descriptions provided.
  • Define a pydantic ConversationFacts class for creating lists of ConversationFact objects.
  • Bind the structured output format to the llm provided.

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

# Define the ConversationFact with the correct fields
class ____(____):
    """
    A class that holds the facts from a conversation in a format of object, subject, predicate.

    For example, if the conversation includes the fact that the user likes ice creamthe facts would be:
    - object: "Adam"
    - subject: "ice cream"
    - relationship: "LIKES"
    The class also includes a session ID to identify the conversation.
    """
    ____: str = Field(description="The session ID of the conversation.")
    ____: str = Field(description="The object of the fact. For example, 'Adam' ")
    ____: str = Field(description="The subject of the fact. For example, 'Ice cream'")
    ____: str = Field(description="The relationship between the object and the subject.  This should be a single word verb in upper case. For example, 'LIKES' or 'OWNS'")

# Define a ConversationFacts class for creating lists of ConversationFact objects
class ____(____):
    """A class that holds a list of ConversationFact objects."""
    facts: list[____] = Field(description="A list of ConversationFact objects.")

# Bind the output to the llm provided
llm_with_output = llm.____(____)
Edit and Run Code