Neo4j

Neo4j is a native property-graph database, first released in 2007 and implemented in Java. It stores data as a labeled property graph: nodes carry one or more labels and key/value properties, and directed, typed relationships connect them and can themselves hold properties. Rather than computing joins at query time, Neo4j uses index-free adjacency, so each node holds direct pointers to its relationships. It is an ACID-compliant, transactional store and is accessed over the binary Bolt protocol or an HTTP endpoint.

Queries are written in Cypher, Neo4j's declarative graph language, whose ASCII-art pattern syntax such as (a)-[:KNOWS]->(b) expresses traversals far more concisely than SQL joins. Cypher was opened as openCypher and is the main basis for the ISO GQL standard, so related engines like Memgraph share much of its dialect.

Data typically enters via CSV import, the bulk importer, or streaming from Kafka and Spark. The Neo4j APOC procedure library adds import and export for JSON, CSV, GraphML, and Cypher scripts; GraphML round-trips to visualization tools such as Gephi and Cytoscape, and Neo4j also emits its own Neo4j JSON and PG-JSON shapes.

Strengths are deep traversals, relationship-heavy queries, and a mature ecosystem. Limitations include memory-bound performance on very large graphs and the operational cost of horizontal scaling, since native graphs are hard to partition cleanly.