Intelligent Document Parsing

Our AI-powered ComIDP solution intelligently processes and comprehends content in documents and images, including reports, contracts, essays, and other standard files. It identifies and classifies elements, modules, and structures while preserving the original reading logic. By structuring unstructured and semi-structured data, it provides accurate data sources for downstream applications.

Currently supported parsed types include: text, paragraphs, images, standard tables, non-standard tables, tables of contents, ordered lists, titles, unordered lists, formulas, headers, footers, page numbers, image titles, image descriptions, references, table titles, table descriptions, code, and algorithms.

Home

From the homepage, you can directly access the intelligent document parsing feature. The homepage is the center of ComIDP's feature navigation, providing quick access to commonly used modules.

Single Document Parsing

Perform intelligent document parsing for a single document.

Select the file you want to parse. Document parsing will automatically begin after successful upload. You can switch between different result options to view the corresponding results. Currently supported options include: Markdown, JSON, TOC, Table, Image, and Text.

Batch Document Parsing

Batch parse up to 30 documents.
Click the Bulk button above to switch to batch parsing. Upload the file you want to parse, set the parsing parameters, click Start Parsing, and wait for a moment to see the results of the intelligent document parsing.

Introduction to Parsing JSON Fields

◆ success_count: Number of successfully converted pages
◆ total_count: Number of converted pages
◆ version (string): Version number

◆ pages (array): Page data container
├─ angle (integer): Text layout direction (0: horizontal / 90: vertical)
├─ page_id (number): Current page number
├─ image_id (string): Relative directory address of the image corresponding to the current page
├─ width (integer): Document page width (pixels)
├─ height (integer): Document page height (pixels)
├─ content (array): Basic data: Text lines, One of the images 
├─ id (integer): The reading order id of the current paragraph on the page. 
├─ score (integer): The score of the AI ​​layout analysis result, between 0 and 1, with values ​​closer to 1 representing greater credibility. 
├─ text (string): Text. 
├─ type (string): Type, paragraph (paragraph type, including text information such as body text, title, and formulas). 
├─ position (array): The coordinates of the four corner points of the table of contents area, in order: left-top, right-top, right-bottom, and left-bottom. 

◆ catalog (object): Directory tree structure 
│ └─ toc (array) 
│ ├─ pos (array): The coordinates of the four corner points of the table of contents area, in order: left-top, right-top, right-bottom, and left-bottom. 
│ ├─ paragraph_id (integer): The paragraph ID of the title
│ ├─ page_id (integer): The page number of the title (minimum page number is 1)
│ ├─ content (string): The title content
│ └─ type (string): Title type: catalogue, title, figure_title, table_title

◆ metrics (array): Page-level performance metrics
├─ page_image_width (integer): The current page rendering width (pixels)
├─ page_image_height (integer): The current page rendering height (pixels)
├─ page_id (number): The current page number
├─ angle (integer): The text layout direction (0: horizontal / 90: vertical)
└─ image_id (string): The relative directory address of the image corresponding to the current page (same as pages.image_id)

◆ detail (array): Pagination data container
├─ page_id (integer): The page number of the current paragraph
├─ paragraph_id (integer): The id of the current paragraph
├─ text (string): Text. Currently empty if type is image or table
├─ type (string): Type: paragraph (paragraph type, including text information such as body, title, and formulas), image (image type), or table (table type)
├─ image_url (string): Image URL
├─ position (array): The coordinates of the four corner points of the directory area, in order: left-top, right-top, right-bottom, and left-bottom.
├─ tags (array): Indicates whether there is special text in the paragraph. Types include formula and handwritten (not currently supported, empty array, reserved for future use).
│─ cells (array): Cell array, returned only when type is table
│ ├─ row_span (integer): Cell row span, defaults to 1
│ ├─ text (integer): Cell text content
│ ├─ type (integer): Cell represents a table cell
│ ├─ col (integer): Cell column number
│ ├─ col_span (integer): Cell column span, defaults to 1
│ ├─ position (integer): Coordinates of the four corner points of the cell, left-top, right-top, right-bottom, left-bottom.
│ └─ row (integer): Cell row number

Details

Type	Description
paragraph	Paragraph
figure	Figure
standard_table	Standard table
unstandard_table	Nonstandard table
ordered_list	Ordered list
catalogue	catalogue
title	Title
unordered_list	Unordered list
formula	Formula
header	Header
footer	Footer
page_number	Page number
figure_title	Figure title
figure_caption	Figure description
reference	References
table_title	Table title
table_caption	Table description
code	Code
algorithm	Algorithm