[Bug]: <title>Incorrect weight assignment due to type mismatch in graph_extractor.py

[pan-2001]

Do you need to file an issue?

• [X] I have searched the existing issues and this bug is not already filed.
• [X] My model is hosted on OpenAI or Azure. If not, please look at the "model providers" issue and don't file a new one here.
• [X] I believe this is a legitimate bug, not just a question. If this is a question, please use the Discussions area.

Describe the bug

I've encountered a bug in the graphrag/index/graph/extractors/graph/graph_extractor.py file around line 253: https://github.com/microsoft/graphrag/blob/main/graphrag/index/graph/extractors/graph/graph_extractor.py#L253

weight = (
    float(record_attributes[-1])
    if isinstance(record_attributes[-1], numbers.Number)
    else 1.0
)

The problem arises because the LLM generates results as strings (str), while isinstance(xxx, numbers.Number) requires numeric types.

Here's a simple test case demonstrating the issue:

from numbers import Number

test_values = [1, 1.0, "1.0"]

def is_number(value):
    return isinstance(value, Number)

for value in test_values:
    print(f"{value}: is number={is_number(value)}, is float={isinstance(value, float)}")

# Expected output:
# 1: is number=True, is float=False
# 1.0: is number=True, is float=True
# "1.0": is number=False, is float=False

As shown, for string representations of numbers generated by the LLM, isinstance always returns False, leading to all weights being assigned a default value of 1.0.

I would appreciate it if this could be addressed and fixed in future versions. Thank you!

Steps to reproduce

During the index process.

python -m graphrag.index --root ./ragtest

Expected Behavior

The function isinstance should return True

GraphRAG Config Used

default, use openai key.

Logs and screenshots

Additional Information

• GraphRAG Version: 0.2.0 and 0.1.1