Glossario IA
Il dizionario completo dell'Intelligenza Artificiale
Named Entity Recognition (NER)
A subtask of Natural Language Processing (NLP) that aims to identify and classify predefined entities such as people, organizations, or locations in unstructured text.
Entity Extraction
The process of identifying and isolating specific structured information (entities) from unstructured textual data to populate a knowledge base.
Entity Tagging
The action of associating semantic labels (tags) with entities extracted from a text, enabling their classification and use in question-answering systems.
Entity Dictionary
A database or structured list containing valid entities and their types, used as a reference for recognition and validation in a QA system.
Supervised Learning for NER
An approach where an NER model is trained on a manually annotated text corpus to learn to recognize and classify entities.
Sequence-to-Sequence Model (Seq2Seq)
A neural network architecture used for complex NER tasks, processing an input sequence (text) to produce an output sequence (entity labels).
Contextual Embeddings (ELMo, BERT)
Vector representations of words that capture their meaning based on the surrounding context, significantly improving the accuracy of extracting ambiguous entities.
Entity Normalization
The process of standardizing extracted entities (e.g., converting 'Tuesday', 'tue.', and 'Tuesday' to a canonical form) to ensure data consistency.
Entity Linking
Task of connecting a named entity mentioned in a text to a unique entry in a knowledge base (e.g., a DBpedia or Wikidata URI).
Annotated Corpus
Collection of texts where entities have been previously identified and labeled by humans, serving as ground truth for training and evaluating NER models.
False Positive in Extraction
Error where the system incorrectly identifies a text segment as a relevant entity, negatively impacting the precision of the question-answering system.
Extraction Pipeline
Sequential chain of modules (tokenization, NER, normalization, linking) that transform raw text into exploitable structured entities.
Knowledge-Based QA System
Type of question-answering system that finds answers by querying a structured knowledge base, populated by entity and relation extraction.
Hybrid NER
Approach combining rule-based methods (pattern matching) and machine learning models to leverage the precision of the former and the flexibility of the latter.
Entity Disambiguation
Task of resolving ambiguity when the same character string can refer to multiple distinct entities (e.g., 'Paris' the city vs. 'Paris' the myth).
Fine-Tuning for NER
Process of adapting a pre-trained language model (like BERT) on a specific corpus for a named entity recognition task.