Skip to main content

Artificial Intelligence Glossary

This glossary compiles commonly used terms in artificial intelligence, data engineering, data science, and machine learning to help you better understand related concepts.

Quick Navigation​

A | B | C | D | E | F | I | K | L | M | N | P | R | S | T | V | W | X | Y

A​

AI Agent​

An AI Agent is an intelligent system with autonomous decision-making capabilities that can independently execute tasks in complex environments. It perceives the environment state, analyzes information, and makes decisions to achieve goals. Key characteristics of AI Agents include:

  • Environment Perception & Feedback: Perceives environmental changes in real time and adjusts behavioral strategies based on feedback
  • Autonomous Reasoning: Possesses an internal dialogue mechanism capable of logical reasoning and decision-making
  • Task Planning: Automatically generates and manages task lists and develops execution plans
  • Adaptive Learning: Continuously learns and optimizes during execution to improve task completion efficiency

AI Hallucination​

AI Hallucination refers to the phenomenon where AI systems generate content that appears plausible but is actually inaccurate or entirely fabricated. This is one of the primary challenges facing current large language models:

  • Training Data Limitations: The model may generate incorrect information when encountering questions not covered in training data
  • Overconfidence: The model displays excessively high confidence in uncertain answers
  • Contextual Understanding Bias: Incomplete understanding of complex or ambiguous contexts leads to incorrect reasoning
  • Blurred Knowledge Boundaries: The model cannot accurately identify the limits of its own knowledge and still provides answers when uncertain

AI Database​

An AI Database is a data storage and management system specifically optimized for artificial intelligence applications, featuring a unique design philosophy compared to traditional databases:

  • Vector Data Support: Native support for storage, indexing, and retrieval of high-dimensional vector data
  • Similarity Search Optimization: Built-in efficient similarity search algorithms supporting semantic retrieval
  • AI Model Integration: Provides seamless integration interfaces with machine learning frameworks
  • Large-Scale Processing Capability: Performance optimized for the large data volumes typical of AI applications

Authentication & Authorization​

Authentication and authorization form the foundation of information system security, ensuring that only legitimate users can access corresponding system resources:

  • Identity Verification Mechanism: Verifies user identity through username/password, biometrics, digital certificates, and other methods
  • Permission Control System: Role-based or attribute-based access control for precise user permission management
  • Security Token Management: Uses standard protocols like JWT and OAuth to manage user sessions and access tokens
  • Multi-Factor Authentication: Combines multiple authentication methods to enhance security, such as SMS verification codes and hardware tokens

API Gateway​

An API Gateway is a key component in microservices architecture, serving as a unified entry point for all client requests:

  • Unified Interface Management: Provides a single entry point, simplifying client-to-backend service interactions
  • Intelligent Route Forwarding: Routes traffic to appropriate backend services based on request characteristics
  • Load Balancing: Distributes requests across multiple service instances to improve system availability
  • Security Protection: Implements authentication, authorization, rate limiting, and attack prevention security policies

B​

Batch Processing​

Batch processing is a data processing mode that processes large volumes of data as a whole, suitable for scenarios where real-time requirements are not critical:

  • Bulk Data Processing: Combines multiple data items into batches, processing large volumes at once
  • Processing Efficiency Optimization: Significantly improves efficiency by reducing system calls and network transfers
  • Resource Utilization Optimization: Rationally allocates computing resources, avoiding frequent resource allocation and release
  • Cost Control: Reduces per-unit data processing costs through batch operations

Business Process​

A business process is a core enterprise operations concept, managing and optimizing business activities through systematic methods:

  • Process Orchestration: Defines the execution sequence and conditions of business activities to implement complex business logic
  • Automated Execution: Automatically executes repetitive tasks through workflow engines to improve efficiency
  • State Management: Tracks process execution state, supporting pause, resume, and rollback operations
  • Exception Handling: Establishes comprehensive exception handling mechanisms to ensure business continuity

C​

Context Learning​

Context learning is an important capability of modern AI systems, enabling models to adjust behavior based on specific situations:

  • Contextual Understanding: Analyzes current conversation or task context to understand user intent
  • Dynamic Adaptation: Adjusts response strategies and output formats based on context changes
  • Accuracy Enhancement: Leverages contextual information to improve response relevance and accuracy
  • Interaction Optimization: Provides more natural conversational experiences through context memory

Console​

A console is the core interface for system management, providing administrators with comprehensive monitoring and management capabilities:

  • Visual Management Interface: Offers an intuitive graphical interface that simplifies system management operations
  • Real-Time Monitoring Dashboard: Displays key system metrics with real-time monitoring and alerting support
  • Configuration Management Tools: Centralized management of system configurations with version control and rollback support
  • Operations Center: Integrates common operational tasks to improve efficiency

Containerization​

Containerization is the mainstream approach for modern application deployment, achieving standardized deployment through container technology:

  • Application Packaging: Packages applications and their dependencies into independent container images
  • Environment Consistency: Ensures consistent application behavior across different environments
  • Resource Isolation: Achieves process, network, and storage isolation through container technology
  • Elastic Scaling: Supports rapid startup and shutdown, enabling application elastic scaling

D​

Data Vectorization​

Data vectorization is the process of converting various data types into numerical vectors, forming the foundation for AI system data processing:

  • Multimodal Conversion: Supports vectorization of text, images, audio, and other data types
  • Semantic Preservation: Maintains semantic information and relationships during the conversion process
  • Similarity Computation: Vectorized data supports efficient similarity calculations
  • AI Model Compatibility: Generated vector formats are compatible with mainstream AI models

Distributed System​

A distributed system consists of multiple independent computing nodes that coordinate work through a network:

  • Multi-Node Collaboration: Multiple computing nodes work together to complete complex tasks
  • High Availability: Improves system availability through redundant design, avoiding single points of failure
  • Horizontal Scaling: Increases system processing capacity by adding more nodes
  • Fault Tolerance: Provides fault detection and recovery capabilities to ensure stable system operation

Dataset Management​

Dataset management is a critical component of AI projects, involving the full lifecycle management of data:

  • Data Collection & Storage: Establishes standardized data collection and storage processes
  • Data Preprocessing: Includes data cleaning, format conversion, feature extraction, and other operations
  • Quality Control: Establishes data quality assessment and monitoring mechanisms
  • Version Management: Tracks dataset change history with support for version rollback

Deployment Management​

Deployment management is a critical phase in the software delivery process, ensuring applications run stably and reliably:

  • Environment Configuration: Manages configuration differences across environments to ensure deployment consistency
  • Version Control: Tracks application versions with support for canary releases and blue-green deployments
  • Automated Deployment: Achieves automated deployment through CI/CD pipelines
  • Rollback Mechanism: Quickly rolls back to a stable version when deployment issues occur

E​

Embedding Vector​

An embedding vector is a technique for mapping discrete objects into continuous vector spaces, forming the foundation of modern NLP and recommendation systems:

  • Semantic Representation: Converts words, sentences, or documents into numerical vectors containing semantic information
  • Dimensionality Reduction: Transforms high-dimensional sparse data into low-dimensional dense vectors
  • Similarity Computation: Supports semantic similarity calculation through vector distance
  • Model Input: Provides standardized input format for deep learning models

Event-Driven​

Event-driven architecture is a loosely coupled system design pattern that uses events for inter-component communication:

  • Asynchronous Communication: Components communicate via events asynchronously, improving system responsiveness
  • Decoupled Design: Producers and consumers are decoupled, increasing system flexibility
  • Real-Time Response: Supports real-time event processing for rapid business change response
  • Extensibility: Easy to add new event handlers and extend system functionality

F​

Feature Engineering​

Feature engineering is a critical step in machine learning projects that directly impacts model performance:

  • Feature Extraction: Extracts meaningful features from raw data
  • Feature Transformation: Applies standardization, normalization, and other transformations to features
  • Feature Selection: Selects the most valuable feature subset for the model
  • Feature Construction: Constructs new features based on domain knowledge

Frontend Framework​

Frontend frameworks provide structured solutions for web application development:

  • Component-Based Development: Improves development efficiency through reusable components
  • State Management: Unified application state management that simplifies data flow
  • Routing Control: Manages page navigation for single-page applications
  • Performance Optimization: Provides virtual DOM, lazy loading, and other performance optimization mechanisms

I​

Index Optimization​

Index optimization is an important aspect of database performance tuning:

  • Query Acceleration: Significantly improves query speed through well-designed indexes
  • Storage Optimization: Optimizes index structures to reduce storage space
  • Maintenance Cost: Balances query performance and index maintenance costs
  • Concurrency Support: Supports index operations under high-concurrency access scenarios

Interface Documentation​

Interface documentation is an essential reference for API development and usage:

  • Specification Description: Detailed descriptions of API functionality, parameters, and return values
  • Code Examples: Provides usage examples in multiple programming languages
  • Error Handling: Explains possible error situations and handling methods
  • Version Management: Tracks API version changes and compatibility

K​

Knowledge Graph​

A knowledge graph is a structured knowledge representation method:

  • Entity-Relationship Modeling: Represents entities and their relationships in graph form
  • Semantic Reasoning: Supports reasoning queries based on graph structure
  • Knowledge Fusion: Integrates heterogeneous multi-source data to build unified knowledge bases
  • Intelligent Q&A: Provides knowledge support for intelligent question-answering systems

Kubernetes Orchestration​

Kubernetes is the de facto standard for container orchestration:

  • Container Management: Automates container deployment, scaling, and management
  • Service Discovery: Provides service registration and discovery mechanisms
  • Auto Scaling: Automatically adjusts service instance counts based on load
  • Load Balancing: Distributes traffic among multiple service instances

L​

Language Model​

Language models are core technology in natural language processing:

  • Text Understanding: Understands the grammar and semantics of natural language
  • Text Generation: Generates text that conforms to grammatical and semantic rules
  • Context Modeling: Models long-range dependencies in text
  • Multi-Task Learning: Supports multiple NLP tasks including translation, summarization, and Q&A

Log Management​

Log management is an essential component of system operations:

  • Log Collection: Unified collection of log information from distributed systems
  • Real-Time Monitoring: Real-time log analysis for timely anomaly detection
  • Alerting Mechanism: Triggers alert notifications based on log patterns
  • Performance Analysis: Analyzes system performance bottlenecks through logs

M​

Machine Learning​

Machine learning is a core branch of artificial intelligence:

  • Data-Driven: Automatically learns patterns and regularities from data
  • Pattern Recognition: Identifies complex patterns in data
  • Predictive Analytics: Predicts future trends based on historical data
  • Decision Support: Provides data-backed support for business decisions

Microservices​

Microservices is a distributed system architecture pattern:

  • Service Decomposition: Splits monolithic applications into independent microservices
  • Independent Deployment: Each service can be developed and deployed independently
  • Technology Diversity: Different services can use different technology stacks
  • Team Autonomy: Supports small teams independently managing specific services

N​

Neural Network​

Neural networks form the foundation of deep learning:

  • Bio-Inspired Design: Simulates the connection patterns of neurons in the human brain
  • Hierarchical Structure: Learns complex features through multi-layer networks
  • Nonlinear Mapping: Achieves nonlinear transformations through activation functions
  • End-to-End Learning: Supports end-to-end learning from raw input to final output

Network Communication​

Network communication is the foundation of distributed systems:

  • Protocol Standards: Follows standard protocols such as TCP/IP and HTTP
  • Data Transfer: Ensures reliable data transmission
  • Security Encryption: Ensures communication security through TLS/SSL and other technologies
  • Performance Optimization: Optimizes performance through connection pools, compression, and other techniques

P​

Prompt Engineering​

Prompt engineering is an important technique for optimizing AI model output:

  • Prompt Design: Designs effective prompts to guide model output
  • Format Control: Controls output format and structure through prompts
  • Context Management: Reasonably manages prompt contextual information
  • Effect Optimization: Iteratively optimizes prompt effectiveness

Performance Monitoring​

Performance monitoring is essential for ensuring stable system operation:

  • Metric Collection: Collects key metrics including CPU, memory, and network
  • Real-Time Monitoring: Displays system operating status in real time
  • Trend Analysis: Analyzes trends in performance metric changes
  • Capacity Planning: Plans capacity based on monitoring data

R​

Retrieval-Augmented Generation (RAG)​

RAG is an AI technique that combines retrieval and generation:

  • Knowledge Retrieval: Retrieves relevant information from external knowledge bases
  • Context Enhancement: Uses retrieval results as generation context
  • Accuracy Improvement: Improves generated content accuracy through external knowledge
  • Real-Time Updates: Supports real-time knowledge base updates

Resource Scheduling​

Resource scheduling is a core challenge in distributed systems:

  • Resource Allocation: Rationally allocates computing, storage, and network resources
  • Load Balancing: Evenly distributes load across multiple nodes
  • Priority Management: Allocates resources based on task priority
  • Elastic Scaling: Dynamically adjusts resource allocation based on load

S​

Similarity search is an important information retrieval technique:

  • Vector Retrieval: Performs similarity searches in high-dimensional vector spaces
  • Semantic Matching: Based on semantic similarity rather than literal matching
  • Algorithm Optimization: Uses ANN algorithms to improve search efficiency
  • Large-Scale Support: Supports efficient retrieval across billions of vectors

System Administration​

System administration is a core function of IT operations:

  • User Management: Manages user accounts and permissions
  • Configuration Management: Maintains system configuration consistency
  • Resource Monitoring: Monitors system resource usage
  • Security Policies: Implements system security policies

Service Mesh​

A service mesh is the infrastructure layer for microservices architecture:

  • Service Communication: Manages network communication between services
  • Traffic Management: Controls traffic routing between services
  • Security Control: Implements security policies between services
  • Observability: Provides monitoring and tracing for service calls

T​

Text Vectorization​

Text vectorization is a foundational NLP technique:

  • Text Encoding: Converts text into numerical vector representations
  • Semantic Preservation: Preserves text semantics during the vectorization process
  • Dimension Control: Controls vector dimensions to balance performance and effectiveness
  • Model Compatibility: Generates vector formats compatible with downstream models

Task Management​

Task management is an important component of project management:

  • Task Planning: Develops task plans and schedules
  • Progress Tracking: Tracks task execution progress in real time
  • Resource Coordination: Coordinates human and material resources needed for tasks
  • Status Monitoring: Monitors task status and execution quality

V​

Vector Database​

A vector database is specifically designed for AI applications:

  • Vector Storage: Native support for high-dimensional vector data storage
  • Similarity Retrieval: Provides efficient vector similarity search
  • Index Optimization: Optimizes index structures for vector data
  • AI Integration: Deep integration with machine learning frameworks

Version Control​

Version control is a fundamental tool for software development:

  • Change Tracking: Tracks all code change history
  • Branch Management: Supports parallel development and feature branches
  • Collaborative Development: Supports multi-person collaborative development
  • Release Management: Manages software versions and release processes

W​

Workflow​

Workflows are core technology for business process automation:

  • Process Modeling: Models business processes in a graphical manner
  • Automated Execution: Automatically executes predefined business processes
  • State Management: Tracks process execution state and progress
  • Exception Handling: Handles exceptions during process execution

Workbench​

A workbench is an integrated development environment for developers:

  • Tool Integration: Integrates various development and debugging tools
  • Project Management: Manages development projects and codebases
  • Debug Support: Provides code debugging and performance analysis capabilities
  • Resource Monitoring: Monitors development environment resource usage

WebSocket​

WebSocket is a network protocol for real-time communication:

  • Full-Duplex Communication: Supports bidirectional communication between client and server
  • Low Latency: Lower latency compared to HTTP polling
  • Persistent Connection: Maintains persistent connections, reducing connection overhead
  • Real-Time Push: Supports server-initiated message pushing to clients

X​

Vector Index​

Vector indexing is a key technique for improving vector retrieval efficiency:

  • Retrieval Acceleration: Dramatically improves retrieval speed through index structures
  • Memory Optimization: Optimizes index memory usage efficiency
  • Precision Control: Finds the balance between retrieval speed and precision
  • Dynamic Updates: Supports dynamic index updates and maintenance

Y​

Semantic Understanding​

Semantic understanding is the core capability of AI systems for understanding human language:

  • Semantic Parsing: Understands the deep meaning and intent of text
  • Contextual Association: Establishes associative relationships between concepts in text
  • Intent Recognition: Identifies the user's true intent and needs
  • Sentiment Analysis: Analyzes emotional tendencies expressed in text