DataX key terms

The system of decision rights, policies, and accountability that determines who may define, change, and use data assets, and how compliance with those rules is monitored across the organization.

A role accountable for the quality, definitions, and appropriate use of a specific data domain, acting as the bridge between business owners and technical teams who store and process the data.

The assignment of ultimate accountability for a data asset to a named person or function, including approval of access, definition changes, and decisions about retention or disposal.

Structured information describing other data, such as names, types, formats, owners, refresh schedules, and business meaning, used to discover, govern, and operate data assets reliably.

An inventory of data assets with searchable metadata, definitions, ownership, and usage information that helps analysts find trustworthy data and understand its context before use.

A curated list of agreed business terms with precise definitions, synonyms, and owners, used to align reporting and analytics so different teams interpret the same metric consistently.

Documentation of the technical structure of a dataset, listing each field, its data type, allowed values, constraints, and meaning, typically maintained close to the physical schema.

A recorded trace of how data moves and transforms from source systems through intermediate stages to final reports, supporting impact analysis, debugging, and regulatory audit.

The practice of labeling data by sensitivity and business criticality, such as public, internal, confidential, or restricted, so that handling, access, and retention controls can be applied proportionally.

Replacing or obscuring sensitive values with realistic substitutes so non-production users or analysts can work with data without seeing identifiable or confidential information.

Replacing direct identifiers with artificial keys so records can still be linked across datasets, while re-identification remains possible only with separately secured mapping information.

Irreversibly altering data so that individuals can no longer be identified, including by combining it with other datasets, which typically removes it from personal-data regulation scope.

The mechanisms that decide which users, services, or groups may read, write, or administer specific data resources, enforced through roles, policies, or attribute-based rules.

An access model where permissions are attached to job roles rather than individuals, so users inherit the rights of their assigned roles and changes are managed by editing role definitions.

A security principle requiring that each user or process receives only the minimum permissions needed for its task, and only for as long as that task requires them.

A high-level, technology-neutral representation of the main business entities and their relationships, used to agree scope and meaning before any physical design begins.

A detailed model of entities, attributes, keys, and relationships independent of any specific database product, capturing business rules while remaining implementable in several technologies.

A model tailored to a specific database engine, specifying tables, columns, data types, indexes, partitions, and constraints needed to actually create the schema.

Organizing tables to reduce redundancy and update anomalies by ensuring each fact is stored once, typically through progressive normal forms based on keys and dependencies.

Deliberately introducing redundancy, such as repeated descriptive columns or pre-joined tables, to reduce query joins and improve read performance in analytical workloads.

A dimensional design with a central fact table holding measures and foreign keys, surrounded by denormalized dimension tables that provide descriptive context for filtering and grouping.

The table in a dimensional model that stores quantitative measures at a defined grain, along with foreign keys linking to the dimensions that describe each measurement event.

A table holding the descriptive attributes used to filter, group, and label facts, such as customer, product, or date, usually with a surrogate key and a small number of columns.

A technique for preserving dimension history over time, using approaches such as overwriting values, adding new rows per change, or tracking prior values in additional columns.

A system-generated identifier with no business meaning, assigned to each dimension or entity row to keep joins stable even when natural business keys change or are reused.

The precise level of detail represented by one row in a fact table, such as one order line per day, which must be declared before measures and dimensions are chosen.

An integration pattern that extracts data from sources, transforms it in a separate processing engine, then loads the finished results into a target warehouse or data store.

An integration pattern that loads raw source data into a target platform first and applies transformations there, exploiting the target's scalable compute and keeping raw history available.

An automated sequence of steps that moves and transforms data between systems, including extraction, validation, transformation, and loading, scheduled or triggered by events.

A property of a pipeline step whereby running it repeatedly with the same input produces the same result, preventing duplicate rows or double-counted measures after retries.

Coordinating the order, dependencies, retries, and monitoring of many pipeline tasks so that downstream jobs start only when their upstream inputs have completed successfully.

A load operation that inserts a row when its key is absent and updates the existing row when the key is present, commonly used to keep targets synchronized with changing sources.

A technique for identifying and propagating only the inserts, updates, and deletes that occurred in a source since the last run, reducing load compared with full extracts.

A stored marker, such as a timestamp or sequence number, recording how far a pipeline has processed a source so the next run can resume without reprocessing everything.

The measurable aspects used to assess data fitness, typically including completeness, accuracy, consistency, timeliness, uniqueness, and validity, each with its own tests and thresholds.

Analyzing a dataset's actual values to reveal distributions, null rates, distinct counts, patterns, and outliers, establishing a baseline before cleansing or quality rules are designed.

Correcting or removing inaccurate, incomplete, duplicated, or malformed records through standardization, matching, deduplication, and imputation so downstream analysis is trustworthy.

Identifying records that refer to the same real-world entity and merging or removing the extras, often using fuzzy matching on names, addresses, or other identifying attributes.

An observation that lies far from the rest of its distribution, which may signal a data error to correct or a genuine rare event worth investigating rather than deleting.

Summarizing many detailed rows into fewer grouped values using functions such as sum, count, average, minimum, or maximum, usually combined with grouping on dimension attributes.

A numeric value in a fact table that can be aggregated, such as revenue, quantity, or duration, as opposed to the descriptive attributes stored in dimension tables.

A measure of how widely values spread around their mean, equal to the square root of the variance and expressed in the same units as the original data.

A statistic between minus one and one describing the strength and direction of a linear relationship between two variables, which does not by itself establish causation.

The pattern of how values of a variable are spread across their possible range, described by shape, center, spread, and the frequency of extreme or repeated values.

A centralized, integrated store optimized for analytical queries, holding structured data consolidated from many operational sources and organized for consistent historical reporting.

A storage repository that holds large volumes of raw data in native formats, including files and semi-structured records, with schema applied mainly when the data is read.

An architecture combining low-cost object storage with warehouse-style features such as schemas, transactions, and cataloging, so analytics and machine learning share one governed platform.

A focused subset of a warehouse serving one business area or team, offering simpler access and tighter scope than the full enterprise warehouse while drawing on shared conformed data.

Physically splitting a large table into segments by a column such as date or region, so queries can skip irrelevant segments and maintenance can target only affected parts.

An auxiliary data structure that speeds lookups and filtering on specified columns at the cost of extra storage and slower writes, because it must be maintained on every change.

A column or combination of columns whose values uniquely identify every row in a table and can never be null, enforcing entity integrity in the physical schema.

The documented origin and processing history of a data item, including which sources it came from, what transformations were applied, and when each step occurred.

A single-screen visual display of key metrics and trends, designed for quick monitoring and usually built on aggregated, pre-calculated data refreshed on a regular schedule.

A quantifiable measure chosen to track progress toward a specific business objective, with an agreed definition, target, and reporting cadence that stakeholders accept.