MongoDB offers a different style of persistent storage by storing a flat set of JSON "documents" that each consist of named fields and attachments. MongoDB is concurrent and clusterable, and may offer a faster and more scalable mechanism for storing graphs than traditional databases. It is accessible through a Java driver.

Consider that a graph node may be represented as a document, with node properties stored as named fields. Since documents are flat, so the unique document ID would be used for the node's identification property. See http://www.mongodb.org/display/DOCS/Trees+in+MongoDB for ways of encoding tree-structures in MongoDB. Like JCR, MongoDB allows referential integrity, but doesn't enforce it, couchDB is type-free
MongoDB has adhoc-querying : http://www.mongodb.org/display/DOCS/Querying