SynDoc: A Hybrid Discriminative-Generative Framework for Enhancing Synthetic Domain-Adaptive Document Key Information Extraction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ding, Yihao, Han, Soyeon Caren, Jiang, Yanbei, Li, Yan, Li, Zechuan, Peng, Yifan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915519194464256
author Ding, Yihao
Han, Soyeon Caren
Jiang, Yanbei
Li, Yan
Li, Zechuan
Peng, Yifan
author_facet Ding, Yihao
Han, Soyeon Caren
Jiang, Yanbei
Li, Yan
Li, Zechuan
Peng, Yifan
contents Domain-specific Visually Rich Document Understanding (VRDU) presents significant challenges due to the complexity and sensitivity of documents in fields such as medicine, finance, and material science. Existing Large (Multimodal) Language Models (LLMs/MLLMs) achieve promising results but face limitations such as hallucinations, inadequate domain adaptation, and reliance on extensive fine-tuning datasets. This paper introduces SynDoc, a novel framework that combines discriminative and generative models to address these challenges. SynDoc employs a robust synthetic data generation workflow, using structural information extraction and domain-specific query generation to produce high-quality annotations. Through adaptive instruction tuning, SynDoc improves the discriminative model's ability to extract domain-specific knowledge. At the same time, a recursive inferencing mechanism iteratively refines the output of both models for stable and accurate predictions. This framework demonstrates scalable, efficient, and precise document understanding and bridges the gap between domain-specific adaptation and general world knowledge for document key information extraction tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2509_23273
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SynDoc: A Hybrid Discriminative-Generative Framework for Enhancing Synthetic Domain-Adaptive Document Key Information Extraction
Ding, Yihao
Han, Soyeon Caren
Jiang, Yanbei
Li, Yan
Li, Zechuan
Peng, Yifan
Computer Vision and Pattern Recognition
Domain-specific Visually Rich Document Understanding (VRDU) presents significant challenges due to the complexity and sensitivity of documents in fields such as medicine, finance, and material science. Existing Large (Multimodal) Language Models (LLMs/MLLMs) achieve promising results but face limitations such as hallucinations, inadequate domain adaptation, and reliance on extensive fine-tuning datasets. This paper introduces SynDoc, a novel framework that combines discriminative and generative models to address these challenges. SynDoc employs a robust synthetic data generation workflow, using structural information extraction and domain-specific query generation to produce high-quality annotations. Through adaptive instruction tuning, SynDoc improves the discriminative model's ability to extract domain-specific knowledge. At the same time, a recursive inferencing mechanism iteratively refines the output of both models for stable and accurate predictions. This framework demonstrates scalable, efficient, and precise document understanding and bridges the gap between domain-specific adaptation and general world knowledge for document key information extraction tasks.
title SynDoc: A Hybrid Discriminative-Generative Framework for Enhancing Synthetic Domain-Adaptive Document Key Information Extraction
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2509.23273