Skip to content

Python: How to get absolute path URL for citation reference content? #1889

@Pavan-Microsoft

Description

@Pavan-Microsoft

We are referring to the sample here:
azure_ai_with_azure_ai_search.py

Earlier, in Semantic Kernel, we used to get URLs like:

https://srch-kmgenaaaaaa.search.windows.net/indexes/trans_index/docs/45e88c13-dddd-4dc3-aaaa-69cfdb3ccccc__01?api-version=2024-07-01&$select=id,content

But now, after mapping the URL during agent creation with agent-framework, we’re getting values like instead of the actual absolute URL:

trans_45e88c03-d7e5-4dc3-89ce-69cfdb3e06_2024-12-08 15%3A00%3A00.json

Sample citation object we are currently getting with agent-framework:

{
  "type": "citation",
  "annotated_regions": [
    {
      "type": "text_span",
      "start_index": 728,
      "end_index": 740
    }
  ],
  "title": "trans_45e88c13-d7e5-4dc3-aaaa-69cfdb3d1e07__01",
  "url": "trans_45e88c13-d7e5-4dc3-aaaa-69cfdb3d1e07__2024-12-05 17%3A00%3A00.json"
}

We also have sample index fields like:

{
  "value": [
    {
      "@search.score": 0.79698914,
      "id": "45e88c13-d7e5-4dc3-aaaa-69cfdb3d1e07_01",
      "chunk_id": "45e88c03-d7e5-4dc3-aaaa-69cfdb3d1e07_01",
      "content": "Hi, This is sample content.",
      "url": "trans_45e88c13-d7e5-4dc3-aaaa-69cfdb3d1e07_2024-12-08 15%3A00%3A00.json",
      "title": "45e88c13-d7e5-4dc3-aaaa-69cfdb3d1e07_01",
      "contentVector": [
        -0.016070215,
        -0.009892279
      ]
    }
  ]
}

Could you please share any suggestions or insights on this change? Specifically, how can we obtain the absolute path URL for citation references similar to what Semantic Kernel provided earlier?

Package: agent-framework==1.0.0b251028

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

Status

Done

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions