[WIP][VIZ] handle single definition across multiple files - SPM example

[remiadon]

@cmaumet here is the graph I get for SPM default, simply running

python bids_prov/visualize.py examples/spm_default/*.json -o spm_defautl.png

It's a large graph, maybe we can shorten names and identifiers for readability.

[cmaumet]

Hi @remiadon! Great! It looks like a few links are missing (which might help with getting the schema a bit more vertical): for example segment activity has no used relationship? Or anat unzipped is not used by any activity?

[remiadon]

@cmaumet here is tthe graph I get now, running

python bids_prov/visualize.py examples/spm_default/*.json --omit-details -o spm_defautl.png

I added an --omit-details flag, to skip optional fields and make the graph clearer

[cmaumet]

Discussed w/ @remiadon on Sept 22:

• [x] segment should use "updated anat nii header"
• [x] check all grey entities (undefined)

[remiadon]

@cmaumet I went on the W3C doc and searched for "qualified relationship" According to these graphs, it seems that those relations won't help with simplifying the overall representation of our graph, as such relationship is modeled as an extra node

[remiadon]

@cmaumet I tried to represent the "edit file" operation using prov:wasInfluencedBy. I get a graph with a single SPM.mat file

[remiadon]

Note that this is the best looking graph I ever had, "raw"-labeled data sources are located at the top while produced outputs lie at the bottom

[remiadon]

@cmaumet can you comment on the graph above please ? If you are OK we'll have to add this in new_features.md

[cmaumet]

Thanks @remiadon! The graph looks great!!

Agreed on qualified relationship adding complexity to the graph...

I'm not sure about wasInfluencedBy, I'll have to double-check.

[cmaumet]

I'd say let's merge this PR and fix the remaining few things in a separate PR (that do not include the multi-file reading).