Clarification on weight and keywords fields in graph_data.json output

[sa9arr]

Hi LightRAG team,

First off, thanks for the great work on this project!

I’ve been experimenting with graph_visual_with_lightrag.py and noticed something a bit confusing in the output graph_data.json. Here’s a snippet from the file:

{
"source": "Responsible AI",
"target": "Data Privacy",
"weight": 0.0,
"description": "",
"keywords": "10.0",
"source_id": "Responsible AI practices include safeguarding Data Privacy to maintain ethical standards in AI systems."
}

I had a couple of questions about the fields:

Weight: All weight values are set to 0.0. What does the weight field represent in the graph context, and is this expected behavior?

Keywords: This field is currently populated with numerical values (e.g., "10.0", "9.0"). I would have expected a list of textual keywords here instead. Is this field being used differently than expected?

Just wanted to confirm if this is intended or if there might be a bug or misconfiguration somewhere.

Appreciate any clarification you can provide!

Thanks again!

[Sujen-Shrestha]

I have encountered the same issue.

[sa9arr]

I found the answer for this:
The original .graphml file contains the actual relevant information regarding the weights, keywords and other attributes of edges. While running the script examples/graph_visual_with_neo4j, it converts the .graphml file to the .json file and this is where the discrepancy arises and assigns the value of weights to Keywords and initializes weights to "0".
I think the LightRAG team should fix the conversion from .graphml to .json file.