<<   <   >   >>

2019-05-31 | Apps | Generating Gane and Sarson DFDs

Graphviz can do Gane and Sarson pretty much OOTB. I decided on some conventions:

• The process is a subject, and is always first.

• An entity, process, or data store can all be objectsx.

• Track both directions of data flow individually, but roll them up into a single line with appropriate arrows. Note that this can be tallied by putting subject+object in a hash. Test to see if there is an opposite flow to the current test as well as the same flow with the opposite direction (object->subject). Since only processes are set as subject, this works to create a hash that is complete as far as flow.

• Use short names, numbers, and alphanumeric (D13) for entities, processes, and data stores. Put the longer names in a tab delimited field for each entry.

• Put all processes in the objects file so you can easily add to the triples.

Here is an example Graphviz file:

digraph {
    charset="utf-8";overlap="false";
    splines="true";
    sep="+20";
    node [shape=record];
    Entity [label="Entity \nDescription"];
    D1 [label=“<f0>D1|<f1> Data Store \nDescription"];
   1 [label=“{<f0> 1|<f1> Process \nDescription  

   }" shape=Mrecord];
    1 -> D1 [dir="both"];
    1 -> Entity [dir="back"];
}

Run:

sfdp -Tsvg < i.gv > o.svg

To render this:

This is how I imagine that the existing ontologies map to the Gane and Sarson triples. I realize that it might require more detail, but this is all I have right now.

Process:
planned process
http://purl.obolibrary.org/obo/OBI_0000011

Data at rest:
data storage
http://purl.obolibrary.org/obo/IAO_0000030

Entity
role.
http://purl.obolibrary.org/obo/BFO_0000023

Subprocesses:
contains process http://purl.obolibrary.org/obo/BFO_0000067

Predicates for data flow:
has specified input
http://purl.obolibrary.org/obo/OBI_0000293

has specified output
http://purl.obolibrary.org/obo/OBI_0000299

dfd diagramming git graphviz