Video Experimental Relacionado
Updated: Jan 7, 2026
07:50
A Metadata Extraction Approach for Clinical Case Reports to Enable Advanced Understanding of Biomedical Concepts
Published on: September 20, 2018
16.4K
Enfoque de aprendizaje de representación para la comprensión de documentos estructurados
Akkshita Trivedi1, Sandeep Khanna2, Santanu Chaudhury3
1Department of Computer Science and Engineering, Indian Institute of Technology Jodhpur, Jodhpur, India. trivedi.2@iitj.ac.in.
Scientific reports
|December 26, 2025
Resumen
Document Relationship Entity Embedding Learner (D-REEL) mejora la comprensión de documentos al modelar las relaciones semánticas entre texto, figuras y tablas. Este novedoso marco mejora la precisión de la correlación y el rendimiento de la extracción en diseños complejos.
Área de la Ciencia:
- Ciencias de la Computación
- Inteligencia Artificial
- Recuperación de Información
Sus antecedentes:
- Los métodos actuales de comprensión de documentos tienen dificultades con diseños complejos y conexiones lógicas entre elementos.
- Los enfoques existentes no logran capturar relaciones semánticas profundas en diversas estructuras de documentos.
Objetivo del estudio:
- Introducir un novedoso marco de aprendizaje de representación, Document Relationship Entity Embedding Learner (D-REEL), para modelar relaciones semánticas intrincadas dentro de los documentos.
- Mejorar la comprensión de documentos evaluando con precisión las correlaciones semánticas entre los campos del documento, independientemente de su disposición espacial.
Principales métodos:
- D-REEL genera candidatos de extracción y aprende representaciones vectoriales densas (embeddings) para cada uno.
- Compara embeddings para evaluar correlaciones semánticas y determinar relaciones de artículos.
- Combina información espacial con esquemas específicos del dominio para una extracción precisa y una puntuación de correlación.
- Propone la métrica Semantic Structural Congruence (SSC) utilizando la localización independiente de la ubicación.
Principales resultados:
- Mejoras significativas en la precisión de la correlación y el rendimiento de la extracción demostradas en conjuntos de datos públicos.
- Se logró un incremento promedio de mAP de 2-3% en la precisión de la correlación.
- Se obtuvo una mejora de SSC de casi el 10% para el conjunto de datos PRIMA, lo que indica una medición eficaz de las relaciones.
Conclusiones:
- D-REEL ofrece una solución robusta para comprender diseños de documentos complejos y relaciones semánticas.
- El marco permite una extracción precisa y una puntuación de correlación fiable, incluso con estructuras de documentos irregulares.
- La métrica SSC propuesta cuantifica eficazmente la congruencia estructural semántica en los documentos.
Palabras clave:
diseño de documentosextracción de informaciónaprendizaje de representacióncorrelación semánticaMás Videos Relacionados
05:47Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
1.2K
07:08Optimization of Synthetic Proteins: Identification of Interpositional Dependencies Indicating Structurally and/or Functionally Linked Residues
Published on: July 14, 2015
7.6K
Videos de Conceptos Relacionados
01:26Structuralism
3.1K
Structuralism, an early psychological theory developed by Wilhelm Wundt and his student Edward Bradford Titchener, sought to dissect the human mind into its most fundamental components. Wundt's groundbreaking work in his laboratory set the stage for Titchener to define structuralism's goal as cataloging the "atoms" of the mind—sensations, images, and feelings—akin to how chemists identify elements of matter.
Titchener's approach to structuralism was unique. He...
Titchener's approach to structuralism was unique. He...
3.1K
01:18Methods of Documentation I: Source-Oriented Records
1.7K
Source-oriented records, or SOR, are medical record-keeping organized by the data source. The SOR system was first developed in the mid-1900s to organize the growing patient data in hospitals and other healthcare facilities.
In an SOR, each discipline involved in patient care maintains a separate medical record section. This record-keeping method enables easy tracking of patient progress and ensures healthcare staff have access to up-to-date information.
Key Attributes include the following:
In an SOR, each discipline involved in patient care maintains a separate medical record section. This record-keeping method enables easy tracking of patient progress and ensures healthcare staff have access to up-to-date information.
Key Attributes include the following:
1.7K
01:30Introduction to Structures
1.6K
A structure is defined as a system of interconnected members designed to support or transfer forces and successfully withstand the loads acting on them. The internal forces of a structure can be determined by decomposing the structure and analyzing the free-body diagrams of the individual members or of a combination of members. This helps in understanding the structural elements' behavior and ensuring that the structure is stable and can withstand the subjected loads.
There are three main...
There are three main...
1.6K
02:34Structural Isomerism
21.4K
Isomerism in Complexes
Isomers are different chemical species that have the same chemical formula. Structural isomerism of coordination compounds can be divided into two subcategories, the linkage isomers and coordination-sphere isomers.
Linkage isomers occur when the coordination compound contains a ligand that can bind to the transition metal center through two different atoms. For example, the CN− ligand can bind through the carbon atom or through the nitrogen atom. Similarly, SCN− can...
Isomers are different chemical species that have the same chemical formula. Structural isomerism of coordination compounds can be divided into two subcategories, the linkage isomers and coordination-sphere isomers.
Linkage isomers occur when the coordination compound contains a ligand that can bind to the transition metal center through two different atoms. For example, the CN− ligand can bind through the carbon atom or through the nitrogen atom. Similarly, SCN− can...
21.4K
01:21Structural Properties and Dimensions of Lumber
359
Wood's structural properties derive from fibers aligned along the tree's length, contributing significantly to its mechanical strength. Wood exhibits up to twenty times greater tensile strength along these fibers compared to across them, and generally shows better performance under compression than tension. The length of fibers varies, with hardwoods having fibers around one twenty-fifth inch long and softwoods ranging from one-eighth to one-third inch.
The strength characteristics of...
The strength characteristics of...
359
01:28Structures of Carboxylic Acid Derivatives
3.7K
Structure of Carboxylic Acid Derivatives
Carboxylic acid derivatives contain an acyl group attached to a heteroatom such as chlorine, oxygen, or nitrogen. The carbonyl carbon and oxygen are both sp2-hybridized with an unhybridized p orbital.
The three sp2 orbitals of the carbonyl carbon form three σ bonds, one each with the carbonyl oxygen, the α carbon, and the heteroatom, whereas the other two sp2 orbitals of the carbonyl oxygen are occupied by the lone pairs. Further, the unhybridized p...
Carboxylic acid derivatives contain an acyl group attached to a heteroatom such as chlorine, oxygen, or nitrogen. The carbonyl carbon and oxygen are both sp2-hybridized with an unhybridized p orbital.
The three sp2 orbitals of the carbonyl carbon form three σ bonds, one each with the carbonyl oxygen, the α carbon, and the heteroatom, whereas the other two sp2 orbitals of the carbonyl oxygen are occupied by the lone pairs. Further, the unhybridized p...
3.7K