KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Query Expansion
Technique consisting of enriching an initial query with additional relevant terms to improve information retrieval and broaden the search scope. This method aims to overcome vocabulary limitations and better capture user intent.
Query Rewriting
Process of syntactic and semantic transformation of an original query into one or more alternative queries optimized for search. This technique adapts the formulation to data structures and retrieval system capabilities.
Query Decomposition
Method consisting of dividing a complex query into several simpler, more targeted sub-queries to facilitate processing and improve result relevance. This approach better handles multi-intent queries and compound questions.
Query Intent Understanding
In-depth semantic analysis to identify the underlying user intent behind a query, whether informational, transactional, or navigational. This understanding guides query transformation for more relevant results.
Semantic Query Expansion
Query extension based on semantic and contextual relationships between terms rather than simple co-occurrences. This approach uses language models and ontologies to identify semantically related concepts.
Pseudo-Relevance Feedback
Automatic technique that uses initially retrieved documents as a relevant source to extend the original query, without requiring explicit user judgment. This iterative method progressively improves result relevance.
Query Augmentation
Process of enriching a query with contextual information, metadata, or external knowledge to improve its precision and scope. This technique combines original information with relevant complementary data.
Query Reformulation
Complete transformation of a query's structure and content to better align it with user expectations and system capabilities. This approach goes beyond simple expansion by fundamentally modifying the formulation.
Multi-Query Generation
Creation of multiple query variations from an initial query to explore different facets of the information need and increase the chances of retrieving relevant documents. This strategy diversifies the search approach angles.
Query Substitution
Replacement of specific terms in a query with synonyms, hypernyms, or terms more appropriate to the domain context. This technique optimizes the match between user vocabulary and that of the database.
Query Clarification
Interactive or automatic process aimed at resolving query ambiguities by adding details or requesting clarifications from the user. This approach ensures a better match between the query and the actual intent.
Query Personalization
Dynamic adaptation of queries based on each user's profile, history, and specific preferences to optimize result relevance. This personalization takes into account the individual search context.
Context-Aware Query Transformation
Modification of queries considering conversational, temporal, and situational context to maintain coherence and relevance in continuous interactions. This approach is essential in dialogue systems and serial searches.
Hybrid Query Expansion
Combination of multiple query expansion techniques (lexical, semantic, statistical) to leverage the strengths of each approach and mitigate their respective weaknesses. This hybrid strategy maximizes query enrichment effectiveness.
Query Embedding
Dense vector representation of a query in a continuous semantic space, allowing the capture of subtle relationships between concepts and enabling semantic similarity searches. This technique is fundamental in modern RAG systems.
Query Enrichment
Process of adding structured information and metadata to a query to improve its understanding by the system and refine search results. This enrichment includes entities, relationships, and contextual attributes.
Query Relaxation
Technique involving gradually relaxing the constraints of a query when no relevant results are found, by broadening the search criteria to maximize retrieval chances. This approach ensures better coverage.
Query Refinement
Iterative improvement of a query based on analysis of previous results and implicit or explicit feedback to converge toward increasingly relevant results. This refinement process optimizes search precision.
Query Translation
Conversion of a query from one language or format to another (natural to formal, or between natural languages) to adapt to the search system's capabilities or multilingual context. This translation preserves intent while optimizing execution.
Query Fusion
Intelligent combination of results from multiple variants or formulations of a query to produce a final result set that is more relevant and diverse. This fusion leverages the complementarity of different transformation approaches.