I have an application using MySQL as the DB and hibernate and struts. I’m working on converting this to use Neo4J instead of Hibernate and MySQL. The structure is a little different, a new way of thinking of relationships – they are more life-like. Anyway, I have a person node that has a relationship to a company node and the relationship is employed. So a person is employed at a company. A person also has skills and these skills could be obtained while being employed at a company. so I need to somehow go from an employed relationship to a skill.
I know you can’t go from a relationship to a relationship as they are simply ties from a node to a node (an edge). So I’m sure I just need to think of a way to relate these objects better. I was thinking of somehow splitting up the employed relationship into a node as experience perhaps. So a person has experiences. They got this experience by being employed at a company. They got these skills from this experience. Which makes sense but I think of nodes as “concrete” information where I could start from and work my way back to other “concrete” information, and having experiences in a node doesn’t seem right. It seems to vague, it seems like something that should be stored within a relationship, which is how I have it now, so I’m back to square one.
Another thing I’m struggling with, which splitting out the experience into a node might help with this, is that a person can gain multiple types of experiences from the same company – working their way up the food chain so to speak. So I would need to have a person with multiple relationships to different experiences which point back to the same company. And that is fine – it makes sense to have multiple relationships to have multiple experiences. It seems to be a long way to go for a company to see who they have employed.
Maybe the experience could be more like a position. A company has a position and that position is known to have these skills associated with them. The position is filled by a person and the person could add skills to the what is already associated to that position. But those skills they add do not need to be added as default, so there would need to e some flag in the relationship from position to skill.
Hmm that might work. Anyone else struggle with these kinds of domain building issues with Neo4J or any graph database really. It’s definitely a new way of thinking of data structures. It’s almost best to forget all the rdbms knowledge and “rules” you learned.
Yeah, you’ve encountered the hyperedge. Take a look at http://docs.neo4j.org/chunked/stable/cypher-cookbook-hyperedges.html
One of the equivalents in your case could be a Person related to an Experience/Position node. The Experience node has two types of relations- one is EmployedAt and one is AquiredSkill.
Example:
(Person)-[:has]->(Experience)
(Experience)-[:EmployedAt]->(Company)
(Experience)-[:AquiredSkill]->(Skill)
And you have multiple experiences for a person and multiple skills for an experience/position.
That way you can trace a persons skills/lifetime at a company through his experiences, and if you also share the skill, then you can trace skill sets across companies.
Think that makes sense?