Benefits of a Graph Modelling Language

[wassfila]

• When do we need a Graph Modelling Language ?
• What is the alternative to using a Graph Modelling Language ?

Short answer

A graph modelling language is a specific data format that can inform about data relations to allow indexing and bidirectional linking by a framework, which can help speed up later on filtering and search algorithms.

Data Model vs Graph Data Model

• Any JSON or xml document can represent a compact form of a graph definition
• JSON files often use references to other nodes in the same document using their keys.
• References often rely on implicit type definition and assumption on where to find the list of referred to items
• a Graph Format is a document that removes all implicit relations assumptions and lists each edge explicitly
• Also in a Graph Format, it's still possible to decide to have attributes that do not represent their own nodes and connect with relationships
• An example is Tagging and Grouping, that could be visualized with either edges or a Box englobing all participants
• Another example is coloring the nodes of the same Type with the same color
• In General, attributes, properties and Tags can have an equal and reversible representation as a graph node

Assumptions

• Using a Graph Modelling language for a small amount of data like a list of users is an overkill
• Using plain JSON/XML document does not scale well as soon as number of nodes and relationship grows and an abstracted analysis is required
• It should be possible to maintain a double representation as a collection of Model documents and a parser that synthesizes the data into a graph database for appropriate representation and analysis
• If the graph database does not bring any value to the users in term of analysis or perception through representation then the effort of the graph database is not worth it
• If the graph database results in easier maintenance and contribution model, it is already a good argument to use it even if analysis or representation are not needed

[wassfila]

A graph database, being a database, lacks the historical versionning capabilities of source code versionning like git. But git history does not scale nor give advantage for a big number of nodes where a binary compact format is required.