Choosing a terminology to differentiate metapaths versus source-metapath-target instances

[dhimmel]

The following concepts play a crucial role in this study:

1. metapath instance: referring to a type of path such as Compound–downregulates–Gene–participates–Molecular Function (CdGpMF). This concept is defined a sequence of metaedges.

2. source-metapath-target instance: referring to a type of path between two specific nodes, such as Raltitrexed–downregulates–Gene–participates–histone serine kinase activity. This concept is defined a sequence of metaedges with the a specific source and target node.

3. source-path-target instance: referring to a path between two specific nodes, such as Raltitrexed – downregulates – AURKB – participates – histone serine kinase activity. This concept is defined a sequence of edges.

I think we can refer to 3 as a path, and there won't be confusion because paths are generally understood as traversing specific nodes & edges, hence implying specific source and target nodes.

1 & 2 are more prone to confusion. In 1, we're referring to an abstract type of path, before selecting a specific source and target node. 2 is also a metapath, but now refers to the specific connectivity between two nodes (source and target).

[dhimmel]

If we go with "metapath" to refer to 1, here are some possibilities for 2:

• pinned metapath
• applied metapath
• rooted metapath
• imposed metapath
• source-metapath-target (instance)
• metapath between the node pair
• path count (instance): this captures that we care about source-metapath-target instances in order to assess their path counts, so it is more descriptive of our application than the concept generally.

I don't love any of the above options. @kkloste @bdsullivan @mmayers12: are you familiar with any terms of art that could help us differentiate 1 and 2.

[dhimmel]

We never created any terminology that is more succinct than the explicit long forms here.

Suggestions or insights still welcome below.