Create bipartite graph item from files

[yaslena]

Description

Bipartite graphs are special case of a two-mode graph. They have two node sets and edges that only connect nodes from opposite sets. Those types of graphs are ideal for investigating connections between two different entities, for example authors and their publications, politicians and public debates, employees and their work places. This type of graph is often used in affiliation networks but can be also useful in other scenarios where the relationship between two entities is examined. If the network should contain more than two types of nodes (authors - publications - text genres, or politicians - debates - political spectrum) then the bipartite graph type will not be suitable.

Bipartite network steps

1. onboard edges file

Copy a file containing tabular edges data (most likely csv, but possibly also json, yaml, ...) from the local file system (or possibly remote/url) into the kiara data store.

• Inputs:
  file_path (type: string)

• Outputs:
  file_item (type: file, qualities: tabular, at least 2 columns: 'source'/'target', optionally edge attributes like weight)

2. onboard two nodes files

Copy two files containing tabular node data (most likely csv, but possibly also json, yaml, ...) from the local file system (or possibly remote/url) into the kiara data store.

• Inputs:
  file_pathA (type: string)
  file_pathB (type:string)

• Outputs:
  file_itemA (type: file, qualties: tabular, one column referring to the node id, other columns describing attributes for each node)
  file_itemB (type: file, qualties: tabular, one column referring to the node id, other columns describing attributes for each node)

3. create network graph item

Read the onboarded files from step 1 and 2 and parse their content into a database item.

• Inputs:
  edges file_item: (type: file, from step 1)
  nodes file_itemA: (type: file, from step 2); in addition to unipartite indicate partition: B.add_nodes_from(nodes1, bipartite=0)
  nodes file_itemB: (type: file, from step 2); indicate partition: B.add_nodes_from(nodes2, bipartite=1)
  graph type (type: string) [directed, undirected, multi-directed, multi-undirected]

• Outputs:

network_graph (type: network_graph, stored as sqlite-db-file, qualities: three tables 'edges'/'nodes1'/'nodes2')