Combine networks and provide multi-relation networks

[bockthom]

As already mentioned in #11, it would be helpful to have a function which combines networks and keeps edge types, node types and (possibly) attributes.

There are already some existing implementations to combine networks, but they either do not keep the edge types or node types (https://github.com/se-passau/dev-network-growth/blob/master/util.R#L64) or work only for disjoint networks (https://github.com/se-passau/codeface-extraction-r/blob/master/util-networks.R#L877). What we need is a universal function which can combine all kinds of networks and keeps all types (and possibly attributes).

As far as I know, @ecklbarb needs multi-edge-type networks, i.e., networks which consist of networks of different types. For example, think about author networks which contain cochange-based edges and mail-based edges together in one network. As one idea is to combine author networks with mail relation and author networks with cochange relation to achieve that goal, this is related to combining networks in general.

@ecklbarb Please keep us informed with your progress here.

---

As a side note, this issue is also somehow related to #15 (which could be based on the combine-networks function resulting from this issue).

[clhunsen]

I talked to @ecklbarb about the construction of multi-edge-type networks, e.g., supporting several author.relation entries in the network configuration. We came up with a solution that, as we hope, will add solutions for the following problems in the end: multi-edge-type networks, combination of networks, and, therefore, multi-artifact networks (see #15).

Multi-edge-type networks

To support a network-configuration entry such as author.relation = c("mail", "cochange"), we need to carry out the following steps:

1. Change the allowed number of arguments for the author.relation key to Inf here:
   https://github.com/se-passau/codeface-extraction-r/blob/0e90d6152c854766ec0f40496745508cf0e228fa/util-conf.R#L651-L656

2. Split the function construct.network.from.list into two parts here:

   1. The first part of the function is the construction of an edgelist and a set of vertices. From the new function, we need to return a list of the following shape which is compatible with igraph::as.data.frame():

      network.df = list(
          edges = edge.list,
          vertices = data.frame(name = nodes.processed)
      )

      This way, we can transform the list to a network easily, using the function igraph::graph.data.frame().

   2. The second part of the function is the reliable construction of a network from the returned list of edges and vertices. We need to be sure to include the corner-cases we handle right now.

3. In the function get.author.network, we need to build the network by looping over all given author.relation entries given through the network configuration.
   https://github.com/se-passau/codeface-extraction-r/blob/0e90d6152c854766ec0f40496745508cf0e228fa/util-networks.R#L493-L536

   There, we need to perform following steps to ensure the construction of the right network:

   1. Get the edgelist and the set of vertices for all configured relations.
   2. Merge the edge lists and the sets of vertices. We should rely here on the function merge, which should provide all functionality we need for the join operation. We need a separate function for that.
   3. Call the new function from point 2.ii to construct a network from the merged edgelists and vertices.

Combination of networks

When we implement point 3.ii in a public function somewhere in the networks module (e.g., right next to add.edges.for.bipartite.relation or so), we have easy play for the combination of networks here: We just need to add a function with a similar signature that just takes two (or more) networks as parameters and then performs the following steps:

1. Call igraph::as.data.frame() to obtain lists as described in point 2.i above.
2. Merge the result using the function from 3.ii.
3. Call the function from point 2.ii above.

Multi-artifact networks

As the combination of networks also works for differing artifacts, we are safe here.

Transfer for other types of networks

These steps need to be performed also for artifact networks (solving the main idea from #15 this way), bipartite networks, and multi networks. We have to be careful with the bipartite relation especially.

[ecklbarb]

We have to remember to adapt the simplify function to the multi-edge-type networks. Each edge type should be left after the simplification.

[clhunsen]

I suggest that we add the further improvements in another issue (i.e., merging of edge lists instead of networks here and here; and simplification on edge lists instead of networks here) and close this issue after merging of PR #115 for better bookkeeping.