System and method for semantic parsing of digital documents using visual and textual features

The system addresses the limitations of current PDF parsing tools by categorizing and structuring document elements to recover hierarchical structures, enhancing information retrieval and semantic understanding in complex documents.

WO2026143433A1PCT designated stage Publication Date: 2026-07-09HONG KONG APPLIED SCI & TECH RES INST
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
HONG KONG APPLIED SCI & TECH RES INST
Filing Date
2024-12-31
Publication Date
2026-07-09

AI Technical Summary

Technical Problem

Current PDF parsing tools struggle with complex layouts and unconventional formats, failing to accurately process tables, lists, and charts, and lack the capability to recover hierarchical structures, which hampers effective information retrieval and semantic understanding.

Method used

A system and method for semantic parsing of digital documents using visual and textual features, comprising a user interface, document layout classification, semantic recovery, and text structuring modules, which categorize document elements, derive information from tables, lists, and charts, and generate human-readable hierarchical structures.

Benefits of technology

Enables accurate extraction and organization of complex document elements, recovering hierarchical structures for enhanced information retrieval and semantic understanding, improving operational workflows in fields like finance, legal, and healthcare.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2024144231_09072026_PF_FP_ABST
    Figure CN2024144231_09072026_PF_FP_ABST
Patent Text Reader

Abstract

A system for semantic parsing of an input digital document using visual and textual features is provided. The system includes a user interface, a document layout classification module, a semantic recovery module, and a text structuring module. The user interface enables users to upload the input digital document. The document layout classification module processes the document to categorize its elements based on page images and textual data, outputting layout information with tags and locations. The semantic recovery module uses this layout information to derive content, including tables, lists, and charts, and generates a hierarchical structure. The text structuring module organizes tokens based on the layout tags, groups text into sections by topic relevance, and handles page boundaries, producing another hierarchical structure.
Need to check novelty before this filing date? Find Prior Art