Map your tabular data into graph entities

This feature is experimental and not ready for use in production. It is only available as part of an Early Access Program, and can go under breaking changes until general availability.

Once the data is prepared, we need to create a mapping model that maps our tables and columns into graph entities. The model is defined as a JSON file.

Model for our sample data

{
  "name": "example", (1)
  "db": "neo4j", (2)
  "nodes": [ (3)
    {
      "source": "USERS", (4)
      "label": "User", (5)
      "key_field": "node_id", (6)
      "properties": { (7)
        "id": "id",
        "age": "age",
        "up_votes": "up_votes"
      }
    },
    {
      "source": "QUESTIONS",
      "label": "Question",
      "key_field": "node_id",
      "properties": {
        "id": "id",
        "view_count": "view_count"
      }
    },
    {
      "source": "ANSWERS",
      "label": "Answer",
      "key_field": "node_id",
      "properties": {
        "id": "id"
      }
    }
  ],
  "edges": [ (8)
    {
      "source": "QUESTIONS", (9)
      "type": "ASKED", (10)
      "source_field": "user_node_id", (11)
      "target_field": "node_id", (12)
      "properties": {} (13)
    },
    {
      "source": "ANSWERS",
      "type": "ANSWERED",
      "source_field": "user_node_id",
      "target_field": "node_id",
      "properties": {}
    },
    {
      "source": "ANSWERS",
      "type": "HAS_ANSWER",
      "source_field": "question_node_id",
      "target_field": "node_id",
      "properties": {}
    }
  ]
}

1	Name of graph projection, will be overridden by `graph_name` argument to the stored procedure.
2	Name of database that the graph projection will be created in, will be overridden by `neo4j_db_name` argument gathered from secrets.
3	List of node mappings.
4	Name of the source table. Could be specified as a regular expression.
5	Static label to use for the projected node records. Could also be replaced by `label_field` property which describes a column of type `STRING` or `ARRAY<STRING>` which retrieves the label(s) from the data.
6	Defines column to be used as the unique identifier.
7	List of properties to attach to the projected node, mapped from source column to node property.
8	List of relationship mappings.
9	Name of the source table. Could be specified as a regular expression.
10	Static relationship type to use for the projected relationship records. Could also be replaced by `type_field` property which describes a column of type `STRING` which retrieves the relationship type from the data.
11	Defines the column to be used to identify the source node.
12	Defines the column to be used to identify the target node.
13	List of properties to attach to the projected relationship, mapped from source column to node property.

Note that Neo4j GDS/AuraDS only supports LONG, DOUBLE, ARRAY<LONG>, ARRAY<FLOAT> and ARRAY<DOUBLE> data types and columns that are not compatible with these types are not supported.

Refer to Supported types for a full list of supported types in Neo4j GDS/AuraDS.

Once you are happy with the mapping, save it as a file into a Google Cloud Storage bucket, e.g. gs://<your-bucket-name>/my-graph-model.json