DataTOL meeting notes, 2010-06-16
Participants: Bertram, Anand, Biva, Paolo, Manish

Hi this is Bertram. Please enter your name in the upper right corner (I think), as you join the call.
Also, please log into MarraTech and goto the Silver Room (and Skype for backup).

Anand's report: http://groups.google.com/group/datatol/web/pc1-mockup-split-and-capture-opm
Biva's report: 
Are we off by one week, or OK? 
(here: http://groups.google.com/group/datatol/web/weekly-reports?_done=%2Fgroup%2Fdatatol%3F) 

Anand:
split FPC-RWS wf in two parts, worked ok
captured as OPM (with Dan's help)
captured native provenence
connection string problem re-solved
trying to get to the actual data; not just the refs!?
-> conceptual

Biva:
use case 
converting OPM.RDF to OPM.XML 
trying to see whether Kepler and Taverna OPM.XML are compatible
working on conceptual model

Manish: when looking at different formats, check whether anything is missing
OPM has 'was-derived-from', etc. ==> check whether somethings have been dropped

Bertram: contact mentors and DToL group right away with question

Manish: OPM specification, share Dan's (Kepler) schema with others

Paolo: OPM spec is "rich", don't get stuck there; baseline 3-4 types of nodes and similarly edges:
(don't loose sleep on inference rules ;-)

Need to tackle the conceptual model (type of information we want to capture)
different from architecture/implementation 

Let's focus on D3: graph given
in Taverna, rendering of processors and execution

Graphviz/dotfiles will result in different renderings of graphs 
==> Paolo to provide slides that shows /explains dot-graph
OPM to dot
get the OPM toolbox (again!) from http://www.openprovenance.org
the the binary
unzip
go to bin/
you will find a number of utils including
opmconvert (rdf <-> xml)
opm2dot  <-- this is what will give you a .dot as well as a .pdf file from an OPM graph

opm toolbox to convert opm to dot file or equivalently a pdf file to produce a graph. create the graph for both the  splitted workflow and the whole workflow. convert all the 3 opm files to dot files.  create an artifact conceptually the schema , ER diagram?  conceptual schema for the shared provenance, processes and artifacts(links to data)   

PC3, Paolo's graph:
http://twiki.ipaw.info/bin/view/Challenge/UoM report page from PC3 / myGrid team (Paolo)
and here is a link to the presentation I gave at the PC3 meeting
http://groups.google.co.uk/group/datatol/web/PC3-report.pdf

UC Davis "import" into Taverna :
algorithm that looks at OPM graph; 
translated to internal Taverna graph;
need to "guess" a Taverna wf that may have produced the graph
(why does one need this?)

native OPM <=> other OPM 
failed (different naming schemes)
Problems:
1. Conceptual model mapping
2. identifiers mapping

Difference to PC3:
Then (PC3): Kepler <==using OPM===> Taverna
Now: Kepler <==OPM==> Shared Provence Space (SPS)  <==OPM==> Taverna

Overall story: 
A user U1 runs a Taverna wf W1 and deposits her data D1 in SDS (Shared Data Space)
and puts the trace T1 in SPS

A user U2 runs a Kepler wf W2  that happens to use some parts of D1,
creating a new data product D2 in SDS, and a new trace T2 in SPS.

Now when browsing or querying the provenance (T2) of D2, we need to be able to "unravel" the processing history of not only U2's run, but also go back "through" D1/T1 all the way to the inputs of U1's run of W1.

Biva: SPS for Kepler, Taverna? 
Should one be able to discern whose wf-run this was? (Kepler, Taverna, etc)

Paolo: look at two examples  (OPM graphs from Taverna and Kepler, dotfiles, PDF files)
OPM++  ("Profile" late) 

Processes (Invocations), Artifacts (Refs-to-Data)
mapping of internal identifiers to 




OPM 1.1: 
Definition 1 (Artifact). Immutable piece of state, which may have a physi- 
cal embodiment in a physical object, or a digital representation in a computer 
system. 

Definition 2 (Process). Action or series of actions performed on or caused 
by artifacts, and resulting in new artifacts. 

Definition 3 (Agent). Contextual entity acting as a catalyst of a process, en- 
abling, facilitating, control ling, or affecting its execution. 


we do make distiction b/w data and  traces in the general store, but sometimes the distinction is not very strong because when the data is a constant, we can directly refer to the data itself rather than accessing it via a reference.

http://taverna.opm.org/converter?it=2-0
http://taverna.opm.org/t2:ref//3123868a-40ad-42c8-9f3c-7fb3112651e7?b97d1bae-67d4-45d6-94cb-20983b913ef3

Glossary 
lists the DToL terms and their (informal) definitons

NEXT GOAL/Deliverable
- develop, finish conceptual model for SPS 

NEXT ACTIONS
==> Anand, Biva (and Saumen when available)  to create and link graphical versions of the three wf runs/traces
(OPM graphs)
==> summarize what you see / what's different in the 2 (or 3 with COMAD) versions
These findings will inform the SPS-CM

==> Anand, Biva: create google  spreadsheet with "Dictionary" of maybe this form:
DToL| Kepler |  Taverna | OPM |   Is part of SPS-CM??   
 proc-actor   | actor |  processor | process | .yes
 data-item|   data-token |  ?data | ~artifact |  yes 
 run
 trace
 workflow
 user
 date-time started
 data-time finished..
^^^ above goes the beginning of the SPS-CM  
Please include all OPM entities (A, P, AG) and all OPM relations (5 or so of them)
in the OPM column!

==> create Glossary with (informal) definitions key terms from the DToL column of the above table
==> also check provenance vocabulary by Jun Zhao (Paolo's ref)
==> also check glossary of Tim at UC Davis