Originally posted by **candlecao** October 2, 2024
Personally, I don't think it's necessarily worthwhile to reconcile it.
The more universal or general a node is, the less valuable it becomes for reconciliation.
To put it in another way:
LinkedMusic is essentially a knowledge network, a knowledge graph, or a graph database, When it comes to efficient metadata retrieval, it's wise to avoid creating a "super node"—a node that connects to many or most other nodes in the graph, and a node due to excessive reconciliation.
> This is because :
- (1) Super nodes are so well-known that we don't have to create a link to illustrate it, e.g., "D major" "songs".
>> So it requires certain (music) expertise to determine whether a value of property is potentially "super".
- (2) A reconciled super node will otherwise hinder the efficiency of metadata retrieval and "community detection" on the network and the recommendation based on the graph structure.
That is why perhaps we won't get rid of the awareness of distinguishing between an objectProperty and a dataProperty, which is inspired by OWL.
- (3) However, in a certain case, the super nodes (property value) do needs reconciliation, particularly when asserting the type or class of the entity.
Discussed in https://github.com/DDMAL/linkedmusic-datalake/discussions/205
The more universal or general a node is, the less valuable it becomes for reconciliation. To put it in another way:
LinkedMusic is essentially a knowledge network, a knowledge graph, or a graph database, When it comes to efficient metadata retrieval, it's wise to avoid creating a "super node"—a node that connects to many or most other nodes in the graph, and a node due to excessive reconciliation. > This is because : - (1) Super nodes are so well-known that we don't have to create a link to illustrate it, e.g., "D major" "songs". >> So it requires certain (music) expertise to determine whether a value of property is potentially "super". - (2) A reconciled super node will otherwise hinder the efficiency of metadata retrieval and "community detection" on the network and the recommendation based on the graph structure. That is why perhaps we won't get rid of the awareness of distinguishing between an objectProperty and a dataProperty, which is inspired by OWL. - (3) However, in a certain case, the super nodes (property value) do needs reconciliation, particularly when asserting the type or class of the entity.