Słownik AI
Kompletny słownik sztucznej inteligencji
Entity Classification
Process of assigning a specific category (PER, LOC, ORG, MISC) to each named entity identified in the text, a crucial step in the NER pipeline.
IOB Tagging Scheme
Labeling convention using Inside (I), Outside (O), and Beginning (B) prefixes to delimit the boundaries of multi-word entities in NER tasks.
Neural Sequence Labeling
Neural network-based approach for solving sequence labeling problems, combining feature extraction and temporal dependency modeling.
Transfer Learning in NER
Technique involving the reuse of knowledge acquired from large databases to improve NER performance on specific domains with limited annotated data.
Span-based NER
Modern approach treating NER as a text span classification problem rather than token-by-token labeling, enabling better modeling of nested entities.
Character-level CNN
Convolutional neural network applied at the character level to capture morphological and orthographic information, complementary to word embeddings in NER.
CRF Layer
Final layer added to deep learning architectures to globally optimize the label sequence by considering valid transition constraints between states.
Entity Boundary Detection
Sub-task of NER consisting of precisely identifying the start and end positions of entities in the text, independently of their classification.