KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Named Entity Recognition
Process of automatically identifying and classifying predefined entities such as person names, organizations, locations, or dates in unstructured text. This fundamental technique allows structuring information to facilitate its analysis and exploitation.
Relation Extraction
Automatic identification of semantic relationships between different named entities in a text, allowing the construction of structured knowledge graphs. This technique aims to understand logical connections between extracted information.
Text Classification
Process of automatically assigning a text to one or more predefined categories based on its semantic content. This technique uses supervised learning algorithms to efficiently organize and filter large volumes of textual data.
Attribute Extraction
Automatic identification of specific characteristics or properties associated with a named entity in a text. This technique allows enriching extracted entities with detailed and contextual information.
Temporal Entity Recognition
Process of identifying and normalizing temporal expressions such as dates, times, durations, and periods in a text. This technique is essential for understanding the chronology of events and their temporal context.
Event Extraction
Automatic identification of events triggered by specific actions and extraction of their participants, times, and locations. This technique allows structuring dynamic information and understanding complex scenarios described in texts.
Fact Extraction
Process of automatically identifying verifiable factual assertions presented as true in a text. This technique aims to extract objective and structured information to build reliable knowledge bases.
Sentiment Analysis
Automatic identification of opinions, emotions, and attitudes expressed in a text, usually classified as positive, negative, or neutral. This technique allows understanding subjectivity and human perspectives in textual data.
Keyword Extraction
Automatic process of identifying the most representative and relevant terms in a document or corpus. This technique allows for quickly summarizing the main content and facilitating indexing and information retrieval.
Automatic Text Summarization
Automatic generation of a condensed version of a text while preserving essential information and overall meaning. This technique uses extractive or abstractive methods to create coherent and relevant summaries.
Terminology Extraction
Automatic identification of technical terms and domain-specific expressions in a corpus of specialized texts. This technique helps to build glossaries and understand the expert vocabulary of a particular domain.
Pattern Extraction
Automatic identification of recurring linguistic structures or syntactic patterns that indicate specific types of information. This technique allows for discovering implicit rules to guide the extraction of similar information.
Extraction-based Learning
Machine learning methodology where models are trained using information automatically extracted from large unlabeled corpora. This approach reduces dependency on manually annotated data.
Semantic Annotation
Process of automatically enriching text with structured semantic metadata linking text segments to formal concepts. This technique allows for interconnecting textual content with existing knowledge bases.
Knowledge Extraction
Global process of discovering and automatically structuring knowledge from unstructured textual data. This technique combines multiple extraction methods to build exploitable formal representations.
Entity Normalization
Process of converting extracted entities to a canonical or standardized form to eliminate spelling variations and synonyms. This technique ensures the consistency and unification of extracted information.
Word Sense Disambiguation
Process of identifying the correct meaning of a polysemous word based on its context of use in the text. This technique is crucial for precise and unambiguous information extraction.
Candidate Sentence Extraction
Automatic identification of potential text segments containing relevant information for a specific extraction task. This technique aims to reduce the search space before detailed content analysis.
Information Filtering
Process of automatically selecting relevant documents or segments while rejecting irrelevant information according to predefined criteria. This technique allows focusing analysis on truly useful data.
Distant Supervision
Supervised learning technique where training labels are automatically generated by aligning structured knowledge sources with unlabeled texts. This method enables the creation of large training datasets with minimal human effort.