memgraph/memgraph

Features

• In-Memory Databases - Maintains the graph structure entirely in RAM for high-performance traversals and real-time analytical processing.
• Cypher Engines - A database engine that executes declarative graph queries using the Cypher language to retrieve and manipulate interconnected data.
• Graph Analytics - Provides native C++, Python, and CUDA implementations for high-performance graph analytics like PageRank.
• Agent Memory Storage - Provides a scalable graph backend for AI agents to store long-term memory.
• Knowledge Graphs - A scalable storage layer for AI agents that combines vector search and graph traversals to power retrieval augmented generation.
• GraphRAG Frameworks - Implements graph-based retrieval-augmented generation using vector and geospatial indexes with graph expansion.
• Graph Community Detection - Groups nodes into clusters based on connection density using hierarchical clustering algorithms.
• Dynamic Community Detection - Identifies clusters of related nodes in real-time using the LabelRankT algorithm.
• Transaction Lifecycle Control - Manages the transaction lifecycle by committing changes on success or rolling them back on failure.
• Centrality-Based Rankings - The product measures node importance using metrics like betweenness centrality, PageRank, or core numbering.
• Statistical Aggregators - Computes summary statistics such as averages, sums, and counts across query result sets.
• Database Backup Restoration - Recovers the full database state from durability files to perform system backups and restores.
• Database Connectivity - Establishes thread-safe connections with authentication to link applications to the graph database.
• Database Indexing - Builds label and label-property indices to accelerate node lookups and improve filtering performance.
• Graph Databases - A resilient data store that uses the Raft consensus protocol to ensure high availability and consistent data replication.
• Atomic Transactions - Groups multiple queries into a single atomic unit of work to ensure data consistency.
• Parallel Query Execution - Runs graph computations across multiple CPU cores to increase processing speed for analytical tasks.
• Schema Metadata Retrieval - The product retrieves a JSON representation of nodes, edges, constraints, and enums.
• Database Query Execution - Executes read and write transactions to retrieve or modify graph data using a query language.
• Execution Plan Analysis - Generates detailed query execution plans to identify and resolve performance bottlenecks.
• Database State Restoration - Recovers the database state by loading the latest snapshot and applying write-ahead logs.
• Graph Data Modifiers - The product modifies labels and properties on nodes and relationships including nested map values.
• Multi-Hop Traversals - Provides high-performance navigation of multi-hop paths through interconnected graph data.
• Graph Node Property Management - Enables adding, updating, or removing key-value pairs on nodes using identifiers or lists.
• Node Generation - The product generates nodes using labels and properties derived from procedure results or maps.
• Graph Object Instantiation - The product adds new nodes and relationships to the database using specified patterns and properties.
• Graph Querying - The product executes graph traversals and analytics using a declarative query language.
• Graph Relationship Modeling - The product establishes connections between nodes based on logical rules or shared properties.
• Relationship Establishment - The product establishes directed connections between nodes with specific types and associated properties.
• Graph Traversal - Enables the visiting of nodes by following edges to discover deep paths and connections.
• Traversal Constraints - Restricts path expansion by relationship type or direction to reduce execution time.
• Access Mode Management - Controls how multiple queries interact with storage using shared or exclusive access modes.
• Hybrid Search - Combines vector similarity with structured metadata and geospatial proximity to identify nodes.
• Hybrid Vector-Graph Databases - Combines semantic vector search with structural graph traversals to retrieve contextually relevant neighborhoods.
• Knowledge Graph Construction Tools - Transforms raw documents into connected knowledge graphs by extracting entities and generating embeddings.
• Index Creation - Builds label and property indexes to accelerate node matching and relationship creation.
• Graph Property Modifiers - Updates or removes attributes and labels on edges to manage connection metadata.
• Optional Pattern Matching - The product returns results for a pattern while filling missing elements with null values.
• Search Indexing - Accelerates retrieval of frequently queried properties by selectively creating indices on critical search fields.
• Dynamic Analytics Updates - The product recalculates node importance incrementally as the graph changes to maintain current analytics.
• Real-Time Analytics - A system for performing dynamic community detection and centrality measurements on streaming data using native C++ and CUDA.
• Semantic Search - Uses vector similarity to find relevant information based on meaning and context.
• Subquery Support - The product runs a nested query within a larger flow to process data in isolated scopes.
• Isolation Level Configuration - Adjusts transaction isolation levels to balance data consistency and system performance across sessions.
• Traversal Depth Constraints - The product constrains the maximum number of hops during traversal to prevent unnecessary scanning.
• Vector Search - Locates nodes by finding entities most similar to an embedded query using high-dimensional vector spaces.
• Vector Similarity Search - Identifies nearest neighbor nodes using distance metrics to find semantically similar entities.
• Write-Ahead Logging - Ensures durability and recovery by recording all database modifications via write-ahead logging.
• Database High Availability - Implements a Raft-based consensus protocol to ensure continuous service availability through automated failover.
• Raft Consensus Implementations - Coordinates multiple database instances using the Raft consensus protocol to ensure high availability and failover.
• Graph Analytics Implementations - Executes complex graph analytics and centrality measures using compiled native C++ code for maximum efficiency.
• Structural Pattern Matching - Retrieves nodes and relationships that match structural patterns, labels, or paths.
• Shortest Path Algorithms - The product calculates the most efficient route between two nodes to minimize hops or costs.
• Deep Path Traversals - The product performs deep path traversals using BFS, DFS, and shortest path algorithms.
• Shortest Path Algorithms - The product calculates the path with the lowest total cost using Dijkstra's algorithm.
• Degree Centrality - The product measures node importance by counting incoming and outgoing relationships.
• AI Agent State Coordination - Guides autonomous agents to execute specific retrieval pipelines to answer complex questions.
• Contextual Information Retrieval - Combines vector search with graph traversal to gather relationship-rich neighborhoods for AI grounding.
• Entity Extraction Pipelines - The product identifies entities and relationships within unstructured text to construct a knowledge graph.
• Graph-Based Retrieval Augmentation - Combines vector-based semantic search with graph traversal to pull connected nodes for model prompts.
• Streaming Centrality Updates - The product updates node importance in streaming scenarios based on local changes instead of full recalculations.
• Semantic Modeling - The product represents domain knowledge using labeled property graphs to enable semantic reasoning.
• Document Knowledge Extraction - Converts unstructured text into structured knowledge graphs by extracting entities and relationships using language models.
• Link Prediction - Predicts the likelihood of new edges forming between vertices based on graph features.
• Model Training - Trains neural networks for link prediction using inductive or transductive setups with graph attention layers.
• Projected Pattern Extraction - Creates lists of projected values by matching graph patterns within a single expression.
• Recommendation Scoring and Ranking - The product ranks the most likely nodes to connect to a source by calculating edge scores.
• Similarity Graph Construction - The product establishes graph edges between nodes based on calculated similarity scores to represent proximity.
• Node Classification - Implements node classification to predict labels for individual nodes using neighbor and structural analysis.
• Spatio-Temporal Graph Neural Networks - Processes continuous-time dynamic graphs using neural networks to capture evolving structural patterns.
• LLM Completion Interfaces - Sends graph-related text to language models and receives completions within queries.
• In-Query Model Execution - Triggers large language model completions directly from within a database query.
• Text Embedding Generators - Computes vector representations for node properties using local models or remote APIs.
• Graph-Based Cluster Summarization - Synthesizes information across the entire graph to identify emergent themes or missing data.
• Dynamic Embeddings - Generates evolving vector representations of vertices by processing streamed edge events in real-time.
• Database Integrations - A database that exports graph structures to neural networks for node classification and link prediction tasks.
• Inductive Learning - The product aggregates neighbor representations to learn inductive patterns applicable to new datasets.
• Graph - Generates low-dimensional feature representations using biased random walks to preserve graph neighborhoods.
• Spatial Analysis - Stores coordinates to calculate geospatial distances and determine if points fall within bounding boxes.
• Query Engine Extensions - Allows extending the query engine with specialized graph logic written in C++ or Python.
• Vehicle Routing Systems - Optimizes routes for fleets of vehicles visiting multiple geographical locations.
• Concurrent Read-Write Transactions - Implements concurrent transaction execution to maximize CPU core utilization during database operations.
• Hierarchical Community Clustering - Groups related nodes into communities to organize data for structured summarization.
• Leiden Community Detection - Groups nodes into clusters using the Leiden algorithm to identify connected communities.
• Network Community Partitioning - Partitions nodes into clusters using Leiden, Louvain, or Spectral Clustering algorithms.
• Bulk Data Ingestion - Imports large volumes of nodes and relationships using high-performance parallel processing and batching.
• Query-Based Bulk Ingestion - Transforms external datasets into queries and uses UNWIND to create graph objects in bulk.
• Cross-Database Data Migrations - Transitions existing graph data from other graph database systems using dedicated migration modules.
• Data Action Triggers - Automatically executes queries or modules in response to changes in nodes or relationships.
• CSV Import Managers - Reads data from CSV files and binds rows to variables for use in graph queries.
• In-Memory Storage Modes - Configures storage to operate entirely in memory to maximize processing speed for analytical workloads.
• Graph Subset Selectors - The product creates projected graphs by extracting nodes and relationships from paths.
• Point-in-Time Database Snapshots - Triggers immediate writes of storage to disk to capture a consistent point-in-time state for recovery.
• Model-Level Integrity Enforcement - The product validates that node properties meet requirements for existence, uniqueness, or data type.
• Interactive Migration Managers - Provides an interface to modify proposed table-to-node mappings step-by-step before executing data migration.
• Data Lake Ingestion - Streams tables from data lake storage into the graph database using high-performance protocols.
• Data Projections - The product transforms nodes and relationships into custom map structures by selecting properties.
• Geospatial Data Types - The product creates coordinate points and calculates distances or bounding box containment.
• Query Result Transformations - Controls result set output by ordering data and limiting returned records.
• Memory Optimization - The product reduces resource consumption by using specialized types instead of generic strings.
• Directed Graph Analysis - The product computes reciprocity and triadic census to determine if a graph is a tournament.
• Distance-Based Node Discovery - The product locates nodes at a specific distance from a starting point via relationship filters.
• File-Based Data Import - Loads data from CSV, JSON, and Parquet files from local disks or cloud storage.
• Geospatial Search - Retrieves data and location-based insights based on spatial constraints and coordinates.
• Graph Clustering Metrics - The product measures how nodes cluster together using clustering coefficients or Jaccard coefficients.
• Connectivity Analysis - Determines if connections exist between distant nodes within a specified maximum number of hops.
• Bipartite Matching Algorithms - Identifies the largest set of edges in bipartite graphs that share no endpoints.
• Graph Coloring Algorithms - Implements graph coloring algorithms to assign colors to nodes under adjacency constraints.
• Graph Cycle Detection - The product identifies all paths in an undirected graph where the starting and ending nodes match.
• Bulk Data Importers - Parses JSONL documents from various sources to load data into the graph database.
• Relational-to-Graph Importers - Transfers rows and relationships from a relational database directly into a graph structure.
• Graph Attribute Removal - Deletes specific labels and properties from nodes or relationships to update element metadata.
• Graph Database Migrations - Transfers graph data from Neo4j instances using queries to import nodes and relationships.
• Temporal - The product creates and manages dates and durations to track time-based events.
• Graph Element Upserting - Implements logic to create or update nodes and relationships based on a specified pattern.
• Graph Metadata Inspection - The product returns distinct nodes and relationships to visualize structure and associated properties.
• Graph Path Manipulations - The product creates, slices, or converts paths into ordered lists of nodes and relationships.
• Natural Language Querying - Translates natural language questions into precise graph database queries to retrieve analytical results.
• Graph Similarity Calculations - The product computes similarity scores between node pairs to identify related entities based on neighbors.
• Lineage Traversals - The product identifies all ancestor or descendant nodes connected to a specific starting point.
• Hybrid Storage Modes - Toggles between transactional, analytical, or on-disk storage modes to balance consistency and speed.
• Interactive Graph Visualizers - Provides a user interface for exploring and manipulating graph structures through interactive visual maps.
• GNN Data Serialization - Provides specialized data serialization to export graph structures for Graph Neural Network training and updates.
• Key-Value Data Manipulation - Creates, merges, and flattens collections of key-value pairs including recursive key removal.
• Knowledge Graph Builders - The product transforms structured or unstructured data into a network to provide context for AI.
• Graph Data Projections - The product returns specific nodes, properties, or expressions from a result set with aliasing.
• Map Data Conversions - Transforms lists of pairs or properties into structured maps for use in database queries.
• Memory Deallocation Management - Triggers garbage collection of unused objects manually or via periodic cycles to reclaim system memory.
• Network Topology Analysis - The product locates bridges and articulation points to find minimum removal sets for disconnection.
• Biconnected Component Identification - The product finds subgraphs where every node remains reachable after any single node is removed.
• Node Reachability Analyses - The product finds ancestors and descendants or verifies if a path exists between two nodes.
• Path Cost Calculations - The product aggregates weights along a traversed path to determine the total cost of a journey.
• Path Filters - The product explores and filters paths and relationship sequences between nodes using custom criteria.
• Query Plan Caching - Stores the most efficient execution path for a specific query to avoid repeating cost-estimation.
• Structural Pattern Filtering - Evaluates whether specific structural patterns exist to filter results during execution.
• Query Performance Analyzers - Analyzes query plans to identify performance bottlenecks and optimize data retrieval pipelines.
• Query Performance Tuning - Organizes internal data structures to provide rapid access and reduce the time required to retrieve information.
• Query Grouping - Executes a sequence of queries as a single unit of work to maintain data consistency.
• Real-Time Analytics - The product calculates analytics on evolving datasets in real-time without restarting the process.
• Real-Time Data Streaming - Ingests real-time data from streaming platforms like Kafka, RedPanda, or Pulsar using built-in consumers.
• Schema Migration Validation - Performs pre- and post-migration checks to ensure the accuracy of transformed data.
• Search and Indexing - Combines indexing with search functionality to locate nodes and edges via exact, fuzzy, or regex matches.
• Automatic Index Generation - Generates label and edge-type indices asynchronously when new types are encountered.
• Search Result Filtering - Refines query results using boolean logic, property comparisons, and pattern expressions.
• Simple Path Discovery - The product returns every path between two nodes that does not visit the same node twice.
• Spatial Indexing - The product accelerates spatial queries based on distance or bounding box coordinates.
• Graph-Embedded Procedures - The product runs specialized query modules to perform complex analytics and transformations.
• Subgraph Extractions - The product traverses a graph from a start node to return reachable nodes based on constraints.
• Path Pattern Existence Checks - Checks for the existence of specific structural patterns within filtering clauses.
• Text Pattern Matching - Identifies substring indices and captures matched expressions from text using regular expressions.
• Text Search - Provides keyword and pattern-based text matching to locate specific entities or relationships.
• Embedding Generation - Maps graph entities into low-dimensional vector spaces to support supervised learning tasks.
• Vertex Property Compression - Reduces the memory footprint of the property store by applying compression to vertex properties.
• Virtual Graph Projections - The product creates virtual graphs by aggregating paths and relationships into a consolidated map.
• Voronoi Partitioning - The product groups nodes into cells based on the closest center node using shortest-path distance.
• User-Defined Query Modules - Implements custom logic via query modules to extend available functionality within queries.
• Text String Manipulation Utilities - Joins, formats, and replaces substrings within text using delimiters and regular expressions.
• Topological Sorting - The product produces a linear ordering of nodes where every directed relationship flows forward.
• Algorithm Wrappers - Executes complex network analysis by wrapping external graph algorithms into query modules.
• Flow Hierarchy Analysis - The product determines flow hierarchy by calculating the fraction of relationships not in cycles.
• Maximum Flow Calculation - Finds the maximum flow through a weighted graph from source to sink.
• Maximum Flow Path Identification - Returns the specific paths that constitute the maximum flow within a network.
• Graph Connectivity Analysis - The product determines if two nodes belong to the same connected component using disjoint-sets.
• Set Cover Problems - Identifies the minimum number of sets required to cover all elements in a dataset.
• Analytical Data Streaming - Streams large datasets using Arrow Flight or Spark to ensure efficient data transfer.
• GPU Acceleration - Uses hardware acceleration and specialized drivers to increase the processing speed of complex graph computations.
• Database Engine Memory Limits - Sets maximum memory allocation thresholds for instances, queries, or procedures to prevent system exhaustion.
• Multi-threaded Execution - Parallelizes data scans and aggregations across multiple CPU cores to accelerate large dataset processing.
• Coordinate Distance Calculation - Computes distances between geographic coordinate sets along the earth's surface.
• Virtual Graph Analysis - The product synthesizes temporary relationships from multi-hop paths to run algorithms without modifying storage.
• Topological Sortings - The product arranges nodes in a directed graph so every node appears before those it points to.
• Graph Combinatorial Optimization - Computes optimal solutions for bipartite matching, graph coloring, and set cover problems.
• Minimum Spanning Tree Algorithms - The product identifies a subset of relationships that connects all nodes without creating cycles.
• A-Star Search - The product runs the A* search algorithm using distance and heuristic properties for geospatial data.
• K-Shortest Path Discovery - The product returns a specified number of shortest paths ordered by increasing length.
• Multiple Path Identification - The product retrieves all paths between two nodes that share the same minimum weight sum.
• Traveling Salesman Problem Solvers - Calculates the shortest route that visits every geographical node once and returns.
• Granular Element Permissions - Enforces role-based and label-based permissions directly at the node and edge level.
• Identity-Based Database Routing - Uses OIDC or JWT claims to map sessions to specific databases for tenant isolation.
• Role-Based Access Controls - Enforces security policies by mapping user identities to specific roles at the node and relationship level.
• Disconnected Component Identification - The product groups nodes into separate components where no edges exist between different groups.
• RAG Context Expansion - Retrieves relevant information by traversing hops from a pivot node to ground language models.
• Query Extensions - Allows developers to implement specialized graph analysis logic by extending the query engine with external scripts.
• Transaction Batching - Processes large datasets by breaking heavy queries into smaller, periodic transaction batches to manage memory.
• External API Ingestion Pipelines - Retrieves and ingests organizational data by connecting to external REST APIs.
• Performance Analysis - Returns execution details and timing metrics for each logical operator to identify performance bottlenecks.
• Strongly Connected Component Algorithms - The product identifies biconnected components, strongly connected components, k-components, or maximal cliques.
• Geospatial Distance Calculations - The product calculates distances between points and checks if coordinates fall within bounding boxes.
• Node Control Embeddings - Updates vector representations of nodes in real-time streams to track similarity over time.
• Databases & Data - In-memory graph database for real-time analytics
• Graph Databases - In-memory graph database.
• Backend and Infrastructure - In-memory graph database for real-time analysis.

Star history