Product features

Updated at:

Document intelligence workflow

Document intelligence uses pre-trained multimodal models to extract structured data from unstructured documents, such as PDFs, Word files, Excel spreadsheets, and images. It relies on core technologies like hierarchical layout analysis, multimodal document recognition, and understanding of heterogeneous tables across multiple layouts to process long, multi-page documents and complex tables.

Document intelligence vs. text recognition

Document intelligence builds on text recognition by combining multiple technologies such as natural language processing (NLP), image processing, Electronic Document Parsing, and pre-trained document models to automatically process various unstructured and semi-structured documents, including PDFs, Word files, Excel files, and images. While text recognition is often limited to processing images with fixed layouts, document intelligence can handle various document formats with dynamic layouts and changing styles.

Use cases

Document intelligence supports a wide range of use cases. General scenarios include intelligent document and table parsing, data extraction, document comparison, and format conversion. Industry-specific applications include freight forwarding automation, financial risk management, bid announcement analysis, and legal case file processing. It also supports custom scenarios involving long documents, tables, and forms. Document intelligence is widely used in sectors such as government and enterprise administration, public services, legal, finance and taxation, and energy and power.

Product selection

Go to Capability Square to explore our general document intelligence and Document Self-learning products. You can use lightweight applications to upload sample documents and test if the results meet your needs. If the results fit your needs, you can integrate the service using the API or continue using the visual interface of the lightweight applications.

Supported formats

Supported formats vary by parsing edition:

  • Document Parsing (Large Model Edition):

    • Documents/Images: .pdf (supports scanned pages), .doc/.docx/.rtf/.docm, .pptx/.ppt, .xlsx/.xls, .png/.jpg/.jpeg/.bmp/.gif

    • Text/Webpages/E-books: .txt/.epub/.mobi, .md/.html

    • Audio/Video: .mp4/.mkv/.avi/.mov/.wmv, .mp3/.wav/.aac

  • Electronic Document Parsing:

    • Supported formats: .pdf (does not support scanned pages), .doc/.docx/.rtf/.docm, .pptx/.ppt, .xlsx/.xls, .txt/.epub/.mobi, .md/.html

  • Document Intelligence Parsing:

    • Supported formats: .png/.jpg/.jpeg/.bmp/.gif, .pdf (supports scanned pages), .doc/.docx