experimental: LLMGraphTransformer - added relationship properties.

[istvan-nebulinq]

• Description:
  The generated relationships in the graph had no properties, but the Relationship class was properly defined with properties. This made it very difficult to transform conditional sentences into a graph. Adding properties to relationships can solve this issue elegantly.
  The changes expand on the existing LLMGraphTransformer implementation but add the possibility to define allowed relationship properties like this: LLMGraphTransformer(llm=llm, relationship_properties=["Condition", "Time"],)
• Issue:
  no issue found
• Dependencies:
  n/a
• Twitter handle:
  @IstvanSpace

-Quick Test

from dotenv import load_dotenv
import os
from langchain_community.graphs import Neo4jGraph
from langchain_experimental.graph_transformers import LLMGraphTransformer
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.documents import Document

load_dotenv()
os.environ["NEO4J_URI"] = os.getenv("NEO4J_URI")
os.environ["NEO4J_USERNAME"] = os.getenv("NEO4J_USERNAME")
os.environ["NEO4J_PASSWORD"] = os.getenv("NEO4J_PASSWORD")
graph = Neo4jGraph()
llm = ChatOpenAI(temperature=0, model_name="gpt-4o")
llm_transformer = LLMGraphTransformer(llm=llm)
#text = "Harry potter likes pies, but only if it rains outside"
text = "Jack has a dog named Max. Jack only walks Max if it is sunny outside."
documents = [Document(page_content=text)]
llm_transformer_props = LLMGraphTransformer(
llm=llm,
relationship_properties=["Condition"],
)
graph_documents_props = llm_transformer_props.convert_to_graph_documents(documents)
print(f"Nodes:{graph_documents_props[0].nodes}")
print(f"Relationships:{graph_documents_props[0].relationships}")
graph.add_graph_documents(graph_documents_props)

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Ignored Deployment

Name	Status	Preview	Comments	Updated (UTC)
langchain	⬜️ Ignored (Inspect)	Visit Preview		May 31, 2024 5:25pm

[baskaryan]

cc @tomasonjo

[tomasonjo]

LGTM... can you add node and relationship property description to docstring?

[tomasonjo]

@istvan-nebulinq can you please fix lint errors and I can add docstring later if needed

[istvan-nebulinq]

Yes. Will do asap.

[istvan-nebulinq]

@tomasonjo done. passed my local tests (now including lint). synced with main as well.

[tomasonjo]

@istvan-nebulinq thank you! Cc @baskaryan

[tomasonjo]

@istvan-nebulinq can you run

make format

[istvan-nebulinq]

make format

Executed the command in the root dir. Here is the output:
vscode ➜ /workspaces/langchain (graphtransformer_fix) $ make format
poetry run ruff format docs templates cookbook
1403 files left unchanged
poetry run ruff check --select I --fix docs templates cookbook
All checks passed!

[tomasonjo]

Try executing in libs/experimental V čet., 30. maj 2024, 12:56 je oseba istvan-nebulinq < ***@***.***> napisala:

…

make format Executed the command in the root dir. Here is the output: vscode ➜ /workspaces/langchain (graphtransformer_fix) $ make format poetry run ruff format docs templates cookbook 1403 files left unchanged poetry run ruff check --select I --fix docs templates cookbook All checks passed! — Reply to this email directly, view it on GitHub <#21856 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEYGGTIX3WOLVWNXTVUCUMDZE6AAHAVCNFSM6AAAAABH46WLAOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNBQG43DSOJYGI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[istvan-nebulinq]

Try executing in libs/experimental V čet., 30. maj 2024, 12:56 je oseba istvan-nebulinq < @.> napisala:
…
make format Executed the command in the root dir. Here is the output: vscode ➜ /workspaces/langchain (graphtransformer_fix) $ make format poetry run ruff format docs templates cookbook 1403 files left unchanged poetry run ruff check --select I --fix docs templates cookbook All checks passed! — Reply to this email directly, view it on GitHub <#21856 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEYGGTIX3WOLVWNXTVUCUMDZE6AAHAVCNFSM6AAAAABH46WLAOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNBQG43DSOJYGI . You are receiving this because you were mentioned.Message ID: @.>

it worked. I've pushed the changes in that one file.
FYI: it also modified a lot of other files. mainly added an empty second line. trace:

vscode ➜ /workspaces/langchain/libs/experimental (graphtransformer_fix) $ make format
poetry run ruff format .
47 files reformatted, 154 files left unchanged
poetry run ruff --select I --fix .
warning: ruff <path> is deprecated. Use ruff check <path> instead.
All checks passed!

vscode ➜ /workspaces/langchain/libs/experimental (graphtransformer_fix) $ make format
poetry run ruff format .
201 files left unchanged
poetry run ruff --select I --fix .
warning: ruff <path> is deprecated. Use ruff check <path> instead.
All checks passed!

[tomasonjo]

@istvan-nebulinq
One last thing... since we are using dynamic classes the linter is not happy...
Can you add # type: ignore at the end of line 423. You can see example in line 422

Sorry for the trouble, and then make sure to rerun the format again

[istvan-nebulinq]

@tomasonjo it's all good. thanks for your patience while I get up to speed with the flow.
I've written a unit test for this, however, it obviously needs an openai api key to make it work. do you have any policy around this situation?

[tomasonjo]

So those type of tests that require API keys or connection to database are placed under integration tests

[tomasonjo]

cc @baskaryan

[baskaryan]

So those type of tests that require API keys or connection to database are placed under integration tests

yep. that or ideally you mock out the model call and keep it a unit test, since its not openai behavior that this test should be checking