Unexpectedly Deep Section Nesting on Second Page – Bug or Expected Behavior in Azure Document Intelligence Multi-page Processing?

[kazunaritakeichi]

When processing a multi-page file (see attachment) with Azure Document Intelligence, the resulting layout shows that the section on the second page is nested much deeper than expected.

Is this simply a detection failure?

Or is this expected behavior when handling multi-page documents—meaning that post-processing is required to adjust layout consistency across pages?

I'd appreciate any clarification on whether this is a bug or something that needs to be handled on the client side.

$ for line in result.content.splitlines():
$   if '#' in line:
$     print(line)
# This is title
## 1. Text
## 2. Page Objects
### 2.1 Table
### 2.2. Figure
## 3. Others
## This is title
### 1. Text
### 2. Page Objects
#### 2.1 Table
#### 2.2. Figure
### 3. Others

$ for i, paragraph in enumerate(result.paragraphs):
$   if paragraph.get('role') in ['title', 'sectionHeading']:
$     print(paragraph['role'], paragraph['content'])
title This is title
sectionHeading 1. Text
sectionHeading 2. Page Objects
sectionHeading 2.1 Table
sectionHeading 2.2. Figure
sectionHeading 3. Others
title This is title
sectionHeading 1. Text
sectionHeading 2. Page Objects
sectionHeading 2.1 Table
sectionHeading 2.2. Figure
sectionHeading 3. Others

[mccoyp]

Hi @kazunaritakeichi, thank you for opening an issue. I'll tag the appropriate folks so we can get back to you as soon as possible.

[github-actions]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @bojunehsu @vkurpad.