Cyclic relation discovered in "sample_data/vertebrate_anatomy.obo"

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Load "sample_data/vertebrate_anatomy.obo" into OntoWiz
2. Start the pre-processing.
3. Use the debug-graph-generator (still at the development stage).

What is the expected output? What do you see instead?

Had not expected getting a cyle, i.e. assumed it was a DAG. In the graph we get 
the cycle:
{{
   VAO:0000015 ---develops_from---> VAO:0000012
   VAO:0000012 -------produces----> VAO:0000013
   VAO:0000013 --------is_a-------> VAO:0000015
}}

Any suggestions handling situations like these?
-- These cycles are detected; an easy way doing it is to print a warning when 
it happens. 
-- A better solution would be to add rules to handle stations like this; any 
ideas?

Original issue reported on code.google.com by oeks...@gmail.com on 4 Mar 2013 at 4:09

[GoogleCodeExporter]

Hi Ole,

i'm afraid I was wrong yesterday about the set of paths we need.
On a second thought I realised we do need paths composed of different types
of relations (TypeDef's). This is required for constructing property
chains.

Vladimir

Original comment by vladimir.n.mironov@gmail.com on 6 Mar 2013 at 6:26

• Added labels: ****
• Removed labels: ****

[GoogleCodeExporter]

Indeed; came to same conclusion after introspection about ONTO-PERLs 
functionality:
(1) the set of distinct vertices (get_all_ancestors/descendents) and
(2) the set of paths
requires this non-interuptive construction. 
-- In this context, would you w.r.t. the final document, be be able to write a 
discussion* of why you think we are not getting the correct answer when 
building the paths discreetly? 
<-- With the term "discreetly" I imply building separate sets of paths for 
"is_a", "part_of", "produces", "develops_from", etc.
* If you would, I might be able to find a loophole giving alterantive room for 
efficiency gains ;)

Original comment by oeks...@gmail.com on 6 Mar 2013 at 8:50

• Added labels: ****
• Removed labels: ****