information extraction

This work addresses high-accuracy conversion of unstructured/semi-structured documents (e.g., contracts, academic papers, invoices) into structured, machine-readable data. Method: We systematically survey and empirically compare modular pipeline approaches against end-to-end multimodal large models, proposing a unified framework integrating OCR, layout analysis (LayoutParser), graph neural networks, vision-language models (VLMs), and specialized formula/table recognition. We identify and characterize core bottlenecks—layout understanding, dense text recognition, and cross-modal alignment—for the first time. Contribution/Results: We establish a comprehensive analytical framework covering methodology, challenges, and benchmarks, revealing >32% performance gaps of current SOTA on complex layouts (e.g., multi-column, nested tables). We propose a “dual-driven” evolution path emphasizing both data diversity and scale, and open-source a larger annotated dataset to significantly advance knowledge base construction and training-data generation for large models.

Low data quality in unstructured text (e.g., clinical notes) stems from semantic incompleteness, hindering reliable downstream analysis. Method: This paper proposes a data cleaning framework based on *updatable extraction views*, integrating rule-driven information extraction with formal modeling of document update semantics to construct verifiable and interpretable cleaning rules. It establishes, for the first time, sufficient conditions linking extraction program upgradability to cleaning effectiveness. Contribution/Results: Evaluated on real-world medical records, the framework significantly improves entity consistency and completeness, achieving a cleaning accuracy of 92.7%. It ensures verifiability and full traceability of the cleaning process—critical for high-stakes domains—and introduces a novel paradigm for unstructured data quality assurance that bridges theoretical rigor with engineering practicality.

Traditional approaches rely on manually designed annotation schemas and exhaustive document labeling, which are costly and difficult to scale. This work proposes an end-to-end framework leveraging large language models to automatically transform natural language research questions and raw text into structured databases, supported by an interactive interface that enables user-guided refinement. The method establishes, for the first time, a closed-loop pipeline from research questions to structured evidence, integrating expert feedback and domain-adaptation mechanisms. Evaluated in legal and computational biology domains, it significantly enhances the efficiency and accuracy of cross-domain information extraction. The system, along with its public web interface, has been open-sourced.

To address weak cross-domain generalization and poor schema adaptability in scientific paper metadata extraction, this paper proposes MeXtract—a family of lightweight language models (0.5B–3B parameters) fine-tuned from Qwen-2.5, incorporating schema-aware training and a transferable architecture. We introduce MOLE-Bench+, an extended benchmark specifically designed for metadata extraction, featuring new multi-domain and multi-format test subsets. Experiments demonstrate that MeXtract achieves state-of-the-art performance on MOLE, significantly outperforming existing methods, while maintaining strong generalization to unseen schemas and cross-domain settings. All code, data, and models are publicly released.

This work proposes DySECT, the first dynamic information extraction system that enables co-evolution of knowledge and extraction to address challenges such as the dynamic evolution of domain-specific terminology, lagging expert taxonomies, and difficulties in recognizing rare terms. DySECT continuously constructs a knowledge base by extracting triples using large language models, integrates probabilistic knowledge representation with graph-based reasoning to support autonomous knowledge expansion, and enhances the extraction model through prompt tuning, few-shot learning, or fine-tuning on synthetic data—establishing a closed-loop “extraction–knowledge” reinforcement mechanism. Experiments in dynamic domains including healthcare, legal, and human resources demonstrate that DySECT significantly improves both the accuracy and timeliness of information extraction, achieving continuous self-optimization of system capabilities.

Current large language models often suffer from insufficient factual coverage and fragmented relations when constructing knowledge graphs due to direct extraction approaches. This work proposes a question-answering–driven semantic scaffolding mechanism that explicitly models contextual dependencies and implicit relationships prior to triple extraction by generating 5W1H-guided question-answer pairs, thereby structuring and unfolding document semantics in a principled manner. By introducing question-answer pairs as an intermediate representation—a novel strategy to date—the method effectively mitigates the trade-off between coverage and connectivity. Evaluated on the MINE benchmark, the approach significantly improves fact retention and graph cohesion, maintaining high coherence even as the scale of the knowledge base expands substantially.

Automatically aligning unstructured text—such as clinical case notes—with diverse, heterogeneous database schemas remains challenging due to rigid schema-binding assumptions in existing approaches. Method: We propose a model-agnostic text structuring method grounded in an attribute grammar metamodel (MetaG), which constructs semantically enriched parse trees. Through iterative tree rewriting and extraction, the method jointly generates executable semantic schemas and their corresponding instances—without requiring prior knowledge of target data models. Contribution/Results: Our key innovation lies in decoupling text understanding from database schema design, enabling seamless adaptation to heterogeneous storage backends—including relational, graph, and document databases. Empirical evaluation on clinical texts demonstrates accurate generation of cross-model-compatible schemas and instances, significantly overcoming limitations of conventional schema-bound paradigms. The approach achieves robust schema generalization and instance fidelity across disparate target models, establishing a foundation for flexible, database-agnostic text-to-structure translation.

To address three core challenges in scientific literature information extraction—modeling long documents, understanding multimodal content, and standardizing fine-grained cross-paper information (especially under dynamically evolving data schemas)—this paper proposes SciEx, a modular, decoupled framework. SciEx explicitly separates PDF parsing, multimodal retrieval, LLM-driven extraction, and cross-document aggregation, enabling plug-and-play integration of diverse prompting strategies, foundation models, and inference mechanisms for rapid adaptation. Evaluated across three domain-specific datasets, SciEx achieves high accuracy and consistency in fine-grained information extraction. The study systematically identifies key strengths and bottlenecks of current LLM-based pipelines, offering an extensible and maintainable technical pathway for constructing scientific knowledge graphs that evolve with shifting data patterns and scholarly conventions.

This work addresses the absence of an end-to-end evaluation benchmark and semantics-aware assessment framework for structured information extraction from PDFs under enterprise-grade, complex JSON schemas. We introduce ExtractBench, the first open-source benchmark comprising 35 high-value economic-domain PDF documents, human-annotated JSON schemas, and 12,867 evaluable fields. It features a novel fine-grained evaluation framework that treats JSON schemas as executable specifications, enabling field-level differentiated scoring—including exact match, tolerance-based, and semantic equivalence—and explicitly distinguishing omissions from hallucinations. Experiments on leading large language models (e.g., GPT-5/5.2, Gemini-3, Claude 4.5) reveal significant performance degradation in broad-schema scenarios, with effective output rates dropping to 0% on a 369-field financial statement schema, underscoring the current models’ severe unreliability in complex structured extraction tasks.

Information extraction (IE) outputs often mismatch downstream database schemas, hindering direct integration. Method: This paper introduces TEXT2DB—a novel task requiring models to dynamically perform data completion, row insertion, and column expansion based on user instructions, document collections, and target database schemas. To address it, we propose OPAL, an agent framework operating via an Observe-Plan-Analyze closed-loop that orchestrates database interaction, code generation, IE model invocation, and pre-execution feedback analysis for end-to-end instruction understanding, schema alignment, and structured data population. Contribution/Results: Experiments demonstrate that OPAL accurately executes complex IE-database joint tasks across diverse database systems, significantly improving information-to-database deployment efficiency. The study further identifies critical challenges—including large-scale schema adaptation and model hallucination—highlighting open research directions for robust database-grounded IE.