DataTerminology.comNEW
Also known as: Knowledge Discovery in Databases
The process and field of Knowledge Discovery in Databases encompassing data mining.
DESCRIPTION
KDD refers to the end-to-end process of discovering useful knowledge from data, including selection, preprocessing, transformation, data mining, and interpretation/evaluation. It emphasizes validity, novelty, usefulness, and understandability of patterns. The term predates the widespread use of “data science” and remains foundational. Modern practice often packages KDD steps within data science workflows and MLOps pipelines.
ALTERNATIVE DEFINITIONS
Wikipedia
Knowledge discovery in databases (KDD) is the process of discovering useful knowledge from a collection of data.
Highlights KDD as a full process beyond just mining.
AI Magazine (AAAI)
KDD is the nontrivial process of identifying valid, novel, potentially useful, and ultimately understandable patterns in data.
Canonical criteria: validity, novelty, usefulness, understandability.
RELATED TERMS
Data Mining Pattern Recognition Feature Reduction Model Evaluation Extract-Transform-Load Data Preprocessing Exploratory Data Analysis Machine Learning Knowledge Representation
COMPETING TERMS
KNOWLEDGE AREAS
Data Science Data Management AI
TARGET AUDIENCE
Data Scientist Analytics Engineer BI Specialist Data Analyst Data Management Consultant
EXTERNAL SOURCES
- Wikipedia - Knowledge discovery in databases
- ACM SIGKDD - Community site and conference
- AI Magazine (AAAI) - Foundational KDD paper
TOP QUESTIONS
What steps are in the KDD process?
Typical steps include selection, preprocessing, transformation, data mining, interpretation, and evaluation. Iteration is common as understanding improves.
Category: Definition
How does KDD differ from data mining?
Data mining is one step in KDD focused on pattern extraction. KDD is the broader end-to-end process including preparation and interpretation.
Category: Misunderstanding
Is KDD still relevant today?
Yes; KDD’s structured approach underpins modern data science and MLOps practices. Its principles guide reproducible, value-driven discovery.
Category: Relevancy
What are common pitfalls in KDD?
Poor data preparation, overfitting, ignoring domain context, and weak evaluation criteria lead to spurious patterns and limited business value.
Category: Mistakes
What makes a discovered pattern useful in KDD?
Patterns should be valid, novel, actionable, and understandable, aligning with business objectives and ethical constraints.
Category: Requirements
Where to start learning about KDD?
Read the AI Magazine KDD paper and explore ACM SIGKDD resources. Then practice with open datasets using structured KDD steps.
Category: Learning
What is currently a hot topic around KDD?
Automated feature engineering, causal discovery, and responsible AI practices integrate KDD principles into modern pipelines.
Category: Topic
How do KDD outcomes integrate with BI?
KDD-derived models and metrics feed dashboards and data products, improving decision-making and predictive analytics.
Category: Management
What evaluation criteria are used in KDD?
Statistical significance, cross-validation, domain plausibility, and cost/benefit analysis underpin robust KDD evaluation.
Category: Validation