See also: IRC log
<OnnoPaap> The Integrated Operations project is based on ISO 15926
Computas works largely on business process modeling & improvements, largely for the public sector
Approach is to make data-driven decisions quickly and accurately
How can we avoid making the same mistake > 1
And how do you manage knowledge through workforce lifecycle, including generational changes, retirements, etc
<FrankC> socio is good
Focusing on the social aspects of knowledge
Use case: experience transfer in the area of well construction & drilling
<FrankC> socio-technical is better
"Intraproject team transfer"
Tried information retrieval but it wasn't sufficient alone
Started explicitly annotating information, as well as deriving metadata from existing data
<ivan> Slides of Computas
Use an ontology to guide information retrieval and for annotations (as a kind of controlled vocabulary for terms, it seems)
Not so much for inference
Using "semantic distance" notion in their IR
AKSIO drilling ontology created with domain experts, using an interview & observation process
"What can cause bore hole instability?" and then capture the domain knowledge derived
Evolving this bottom-up ontology into a more formal ontology and reusing it on another project
Benefits of this approach: increased quality of knowledge; increased rate of knowledge reuse; leading to better decisions, i.e., better drill plans
Question: is the AKSIO ontology able to be reused by others? Chevron people want it.
<LeeF> answer: yes
leef: thx
Chevron trying to populate ontologies in this area
Side-discussion about something; scribe is confused...Monotonicity, I think...
LeeF: No, the weather changes today were responsible for my slide yesterday :>
<LeeF> it's a strange world, this.
Wants to align AKSIO stuff, and other, with ISO 15926 and other things
Computas focuses more on "long lived business processes" rather than execution-BPEL sorts of things
Difficult to elicit knowledge from the experts
Someone repeat teh question for teh scribe?
and the back of the room
?
<rdecarlo73> "were the abbreviations projects specific or company wide?"
thx
Knowledge elicitation included variance & fault reporting
<FrankC> Chevron is helping to promote SPE's (http://www.spe.org) Open Oilfield Ontology Repository initiative
Question: was the ontology really just a taxonomy or an ontology proper? (axiomatizations)
Answer: It was basically a thesaurus, initially
<FrankC> Supporting long lived work processes -> workflow
Can we align OWL-DL and BPMN
Use case: Daily Production Optimization process
<FrankC> Good question
Turned (something...) into a BPMN ontology
Upshot is that processes and data are described in the same model (ontology)
Lets you consistency check diagrams (cutting edge!)
Extensible via the underlying formalism (OWL-DL)
Commentary: This kind of approach helps explicitly mark information gaps
Fuels the integration-analysis feedback loop
Computas' conclusions: work at a higher level of KR than XML, i.e., RDF and OWL
Work process & KM is a way of linking data to process, i.e., data-drive biz processes
<FrankC> Computas supports the creation of a O&G Interest Group!
And SemWeb tech can be a key enabler of O&G IT
<ram> do not want to do uncertainty stuff with semantic web.. (Ivan quips: Not yet)
<steve> paper on uncertainty reasons for the world wide web : http://www.w3.org/2005/Incubator/urw3/XGR-urw3-20080331/
<FrankC> Semantic Days 2009, May 19-20 in Stravanger
There is some software that does uncertainty in OWL-DL -- http://pellet.owldl.com/pronto -- Scribe's privilege :)
Earth science also a really world class OWL ontology: http://sweet.jpl.nasa.gov/ontology/
What do other operating companies think about this open cooperation required to make standards...
<steve> Ontologies being developed to support integration ot geology, geophysics, geography, etc. ...
<steve> http://www.geongrid.org/index.php/home/search_results/e71139bdc4c23dadfec1d4579ebd1525/
<ivan> slides
Speaker did phd in improving ontology alignment algorithms, including plugin for Protege
<FrankC> Speaker is Jennifer Sampson
Epsis focused on integrated operations in O&G
Working in various ways on O&G standards related to semweb
POSC Caesar, etc
<Robin> PCA RDL is RDS/WIP discussed yesterday
Question: can you say more about extension work to ISO 15926
<Robin> RDL: Reference Data Library
<Robin> RDS: Reference Data System
<Robin> WIP: Work In Progress
<Robin> Part 2: ISO 15926 Data Model
How about we let her give her talk?
<Robin> Part 4: ISO 15926 Reference Data
Three generations if IO development; gen 0 focused on infrastructure level; gen 1 (currently) work process management, collaboration, viz
gen 2: 'smart stuff' -- agents, prediction, integrated scheduling
Question: Most of the value creation will be in gen 2 & 3, right? Answer: yes
Question: Gen 0 was negative value creation? Yes, it's mostly just capital investment.
Data is outstripping processing capabilities
Combine top-down and bottom-up ontology development
with, roughly, top-down aligned to "knowledge" and bottom-up aligned to "experience"
Environment Web Project -- official system of record for emissions & discharges in Norway related to O&G operations
Scribe gives up
Participants: Danny Ducharme, BP, Roger Cutler, Chevron,
Richard Sears, Shell, Raphaële Henri-Bally, Total, Bertrand du Castel,
Shclumberger;
Moderator: Ivan Herman, W3C
ivan: panel discussion. what are the areas we can work together?
ivan: where is there room for cooperation and work?
richard: not so interested in philosophic
ontological development discussions
... there is an equal jargon disconnect with the industry folks AND with the
technical folks
... need to be convinced that tools are going to help business
raphaële: common problem of attracting new generation - they need a way to quickly get information they want, which is what this workshop is all about
roger: clear to me that this is a technology with
potential use in the industry
... in chevron we've taken baby steps and have had a lot of troubles and
solved them ourselves
... would like to be able to share experiences & best practices with
others in the industry
... wouldn't lead to competitive advantage - should be able to cooperate
there
... another area - if major capital projects expect vendors to use 15926 (or
other standard), then we ought to agree on that
ivan: what issues here are specific to the O&G industry?
roger: more specifically, the vendors that supply
us with interpretation system are specific to this industry.
... (G&G interpretation systems) - if we expect the next gen
interpretation systems to have some semantic content, we could expect our
vendors to supply us with standards-based technologies
danny: misery loves company
... as an industry, skipped middleware approaches, only taken small steps to
SOA, this is another wave for approaching integration
... from IT it is appealing--how will engineering view this?
... community can help us with the challenge of what the real tangible things
are that can be brought to the business
... where should we be on the adoption curve?
... what are the benefits of participating early?
bertrand: upstream - can separate exploration & production - we see a lot of new ventures in the deep sea, ventures where cost of deploying personnel is high
<scribe> ... new techniques that go beyond human capabilities
UNKNOWN_SPEAKER: this points to long-term
tendency towards automation
... one of our challenges is to define the articulations - the points at which
things come together
raphaële: present & future projects have
exploration & production working hand in hand - this is an incentive for
G&G to get into a semantically organized world
... business & IT systems that we share are another force pointing in this
direction
roger: not just expl & prod - just in prod
domain, if you're producing a field - have scientists speaking a different
language from the folks working the equipment & different from the
maintenance people
... so just producing a field depends on integration
... initial semantic wins are infrastructure, under the surface
... hard to make a sexy case for SW. sexy case for HCLS is inference through
complicated pathways to find new answers
richard: production world is data-driven e.g.
optimizing reservoir models, fluid flow, history matching
... inference there is about connecting data/numbers differently & finding
significant aspects of models
... and finding that sooner and deeper than currently
... exploration is not numbers based -- more storytelling and strong/subtle
inference about geologic models/history & petroleum systems
... these are very complex inferences that semantic web can help with
... the way we do this now is by moving people & their knowledge around
the world
roger: is there room to collaborate in taking this forward?
richard: Yes.
... we tend to be partners in lots of ventures anyway
roger: are you suggesting some sort of inquiry
into the semantics involved with exploration decision making / knowledge
gathering?
... what would be the first step?
richard: It's about inference in how we are
bringing lots of data together from lots of different sources
... not sure how to take that forward
... one way is to sit with explorers and learn about how they go about the
process now
Onno: from an engineering perspective
... problem we started with is losing lots of money through lack of
interoperability
... needed to make data model that would last throughout engineering,
maintenance, 40-50 years and beyond many changes of software
... e.g. give a manufacturing item - a piece of pipe - a URI - as well as
other pipe, wells, coatings, etc. all get unique IDs that remain the same
throughout manufacturing, logistics, construction, operations, maintenance
... clear advantage for engineering & related vendors. also have
operations & maintenance -- benefits accumulate over 40-50 years of
operations
raphaële: think there is also ROI in G&G
... as soon as we share interest in a lease, decide to develop together, then
there is money to gain from increased collaboration
... makes sense to share more information so that more people are looking at
data to get the most sensible understanding of subsurface
... == less risk if implementing a well
... if we don't share information, the operator will ask for more details that
will take a lot of time out
ivan: ontologies being shared & reused is a
big part of the Semantic Web culture.
... this wasn't easy for HCLS partners to realize / accept
... though they had existing examples (e.g. Gene Ontology (GO))
... what is the business culture like here?
danny: i see great support for sharing ontologies
/ methodologies
... we wouldn't share data or business processes
... but approaches are fine
Alan: sharing continuum
... definitely share what an ontology is, how they are built - no hesitation
to share there
... further out to the point of defining ontologies - still great incentive
for sharing & learning from each other
... further out to the way we use ontologies & link with business
processes - concern - less sharing
... at the other end, data - no sharing at all
bertrand: exactly the reverse
... we share many details of our operations (expl & prod)
... this would be a case where we're sharing what we're sharing already but
would be morepractical for automation
richard: what we don't want to share is the
inference of how we connect the dots to create the potential for a new field
... beyond that, there's a lot that we can share
... semantic web will help me give geoscientists a bit more information at
their fingertips to make their own geologic judgments
roger: very granular access control is a huge
problem in our industry
... this is a problem in semantic web
... i don't understand why HCLS hasn't encountered this as a huge problem
ivan: HCLS has had this problem - but they also have huge public databases
roger: we have some [public data], but the real
core of what we work with is under highly granular access control
... since we have this problem, we can either collaborate to solve it, or
demand that a solution be found
richard: in some cases, there is undue importance
attached to the bits, rather than to the inference atop the bits
... if something is clearly visible in the data, then it's already been done -
because an earlier geologist would have inferred it 5 years before
Jean-Francois: when you want to access data -
this access needs to be very precise
... you can use for this access a SPARQL query
... and this is where you can put security access
roger: perhaps not that simple
kendall: wihle there is a culture of openness/sharing, there's nothing technical about the Semantic Web technologies that prevent you from doing very fine-grained restrictions
richard: further challenge - in many countries, the company doesn't own anything (and hence doesn't decide what can be shared)
roger: regardless of the dispute, this is an
important issue in our industry
... we can cooperate on understanding & working this issue
May: as an O&G outsider - from defense
perspective - as with G&G, analysts spend a ton of time finding/preparing
data, rather than doing their job
... so maybe IT is not doing their job
... semantic technology can help with this sort of data integration
... might be a good area for collaboration
<kendall> For the record, though I'm not an Oracle shill, there's *nothing* about Oracle 11g's OWL inference capabilities that in any way vitiates Oracle's access control technologies.
bertrand: W3C is a government organization--
ivan: no it is not
bertrand: --this is my perspective
<kendall> he may mean "governance"
bertrand: actual acting members are academic
types
... in that mindset, you find that sharing data is perfectly fine
<general murmur of disagreement>
roger: in our 8 year experience in W3C, we find
that not to be true
... most active members are companies like MS, Oracle, HP, ...
... tends to be very industry focused
<Robin> tech industry focused?
<kendall> ... Most academic participants are Invited Experts, and there are some university members, but a small minority
roger: accessing things outside the firewall is something that's not in our corporate culture
ivan: there is a version of wikipedia (called
dbpedia) accessible on the Semantic Web
... a number of applications today are developed that make use of this data
... my question is whether this industry would be comfortable with that sort
of setup
roger: very uncomfortable
steve: i'm not from SemWeb culture - there's
nothing intrinsic about SemWeb that requires openness
... standards in general require some notion of cooperation
... and there are benefits to sharing
... intelligence community is very interested in Semantic Web -- though they
of course have similar security concerns
... but there is an opportunity here to pull in more public data, if you
want
... protecting data sources in the semantic web is just like protecting data
sources in the Web
... if anything, Semantic Web may give you the opportunity to have more
modular access control than you have now
... but that's a research question
<kendall> it's not a research question, actually... it's an R&D question, i.e., it's further along
(scribe agrees with kendall, esp. since scribe was one source of the topic during dinner last night :-) )
frank: at chevron we're gaining advantage partnering with academia (e.g. USC) - it's a strategic decision that does well for us
May: MIT has RDF model for security
roger: access control aspects of the Web are totally unsatisfactory for us
bertrand: specific example
... we need to be able to say that data needs to reside in certai public
places and can only be accessed from public places
... when we talk about requirements to W3C, we can pinpoint to very specific
examples
ivan: clear that this [access control] would be an issue that needs to be discussed in anything going forward
David: you do share a lot of data with your suppliers, right?
raphaelle: there are strong feelings about the
data being acquired by service providers but owned by operator. if anything
goes wrong, the operators face the consequences
... e.g. corrupted well, lost device...
... there's a tricky situation right now where it's hard to separate the raw
data from the technology used to get that data
roger: yes we share data with suppliers, but we don't have it under control.
danny: draw a distinction between public sharing and sharing with suppliers in the presence of contracts
bertrand: i love the euphemism behind the word 'sharing'
Neil: when we talk sharing, i'm thinking
discourse - licenseweb - that's already there
... i think the picture being painted here is bleaker than reality
... that makes it tough to put a real value proposition together to start
addressing real issues
... I once asked how Semantic Web handles units of measure - general kerfuffle
ensued
... i wondered how an organization can underpin future data sharing without
having solved units of measure
... now 4 years later I re-asked the question - where is the semantic web's
answer to this problem?
... meanwhile other organizations (e.g. Energistics) have solved it
themselves, yet Semantic Web has not incorporated this
frank: IEEE has excellent upper ontology on units
of measure
... we're looking at that in refinery
kendall: but that's one approach to units of
measure - there is a completely different approach with different pros and cons
- this is a legitimate question and would make a legitimate task force/XG
... this is a hard but small enough problem that it could be tackled
<general observation that this is not O&G specific>
bertrand: i think it is industry specific and it
is worthless with respect to semantic web
... industry specific because we have practices around the world with concepts
not found in other unit/measure systems
... either semantic web brings new possibilities in our industry or it's not
worth going into it
... we're already at the 5th generation of units systems
raphaelle: <example of sharing data but not
sharing details of how the data was arrived at [scribe didn't catch the
details]>
... once we get to the point where everyone is using the same process, then
that can be in common
steve: what do you see as where you can get the
biggest ROI out of semantic web technologies?
... reducing cost of development? of maintenance? of interoperability?
... how much is spent on converting data & maintaining conversions?
<neilmcn> URL for my editorial - http://www.oilit.com/2journal/2article/0403_3.htm Neil McNaughton
steve: and what's the potential increase in renvenue by discovering things that you otherwise would not have discovered (or as easily) without
roger: biggest bang that I see is Semantic Web as
secret sauce for putting things together from domain specific areas in a way
that scales and does not become fragile/non-maintainable
... investment is lower in scaling
... our experience is that using SW technologies has made the cost of scaling
out into other areas lower
... and that gives opportunities for spreading to more domains
danny: cost savings are clear - more scalable and more leverageable
<david> for those interested: there are work on units of measures and the semantic web e.g. http://idi.fundacionctic.org/muo/swig-20081021/
danny: concerned about challenge of collecting semantics from those with the expertise
bertrand: only measure of return is if we help
our clients find oil better or produce it more efficiently
... otherwise we can't justify investment in a technology
richard: big value creation step in O&G is
discovery
... in my mind, huge value opportunity is to enable explorers to make
connections that wouldn't have happened otherwise
... may be improvements on production side: operability, better models,
understanding systems
... but that's an order of magnitude less compared with exploration
raphaële: depends who you talk to
... if you talk to IT: time & money savings
... if you talk to production business: free time to tackle new
opportunities
Alan: over time, we've improved formalisms and by
doing that improved ability to automate things
... semantic web is another step in this direction
... even within a single area (no integration) - the ability to clarify subtle
semantics & differences in our businesses using semantic web technologies
will open up many opportunities for automation
... seems to be proven by WITSML experience
<kendall> Link to a different approach to units & quantities in OWL, using datatype reasoning: http://www.webont.org/owled/2008/papers/owled2008eu_submission_44.pdf
<scribe> Scribe: on the units of measure thread, there is a recent (as in ongoing) thread on semantic-web@w3.org touching on this. http://lists.w3.org/Archives/Public/semantic-web/2008Dec/0081.html
Amol: there is a lot of power in the
relationships - the simple things that can be made more tangible in your
datasets
... can add a lot of power for the user via something that is trivial from an
IT perspective
... lets people see relationships that otherwise would have taken a lot of
time or might have been impossible
<steve> I would love to see W3C (working with some others) develop "core" ontoloties -- those that cut across just about any field of knowledge: people, geolocation, date/time, units of measure, etc.
ivan: closing remarks?
roger: 3 aspects: expressivity, flexibility, and inference - there is potential value in all 3
<kendall> And where else can you get all 3 from one family of technologies? :>
<steve> I'd add "interoperability"
bertrand: we are a long way from understanding what we need to share
richard: agree with roger's 3 points
... answers are often not apparent in the data
... care more about competing with the earth than competing with other
companies
... the earth is hard to understand
... lots of indirect measurements and huge investments to check
gueses/inferences
raphaelle: similar mindsets - what makes us difference is what we do with the information
<kendall> W3C will need to pull in the machine learning folks to really make O&G exploration inference practical and useful.
raphaelle: challenege is to find information that we do not know exists
<kendall> s/machine learning & statistical inference/
<ivan> slides
<ivan> Frank: what are your experiences with upper ontologies like time?
<ivan> JF: we reuse completely those ontologies and we correlate it with others using, eg, rules in Corese
<ivan> ... for using open ontology: we are very pleased to have that
<ivan> ... we made some mistakes, eg, i made a quick presentation of the ontology structure, at the beginning we did not have that in mind
<ivan> ... and we took the ontology structure of others
<ivan> ... in one year with experts in partners: we take one year to be able to talk to one another
<ivan> ... it was difficult
<ivan> Frank: what do you use for conceptual maps
<ivan> JF: it is a free tool
<ivan> Bertrand: what is corese?
<ivan> JF: it is a graph oriented resolution system developed by inria
<ivan> ... it is the tool we use for inferences for the project
<ivan> slides
<ivan> Frank: do you use estimatic declarations in description logic or is it rule base?
<ivan> emilio: the rules are all in the axioms
<ivan> frank: why ontologies for decisions? we do lot of decision analysis and we have templates of spreadsheets for this. Why ontologies?
<ivan> emilio: i think what I am trying to do is to automate the process as much as possible; try to use the ontologies to capture the knowledge
<ivan> ... the spreadsheets are not well documented, hard to understand
<ivan> ... we get lot of new people who are young and that helps to capture the knowledge of older people
<ivan> frank: are these ontologies public?
<ivan> emilio: yes
Participants: Ivan Herman, W3C, Frank Chum, Chevron, Alan
Doniger, Energistics, Onno Paap, Fluor, Anil Rode, Shell, Robin Benjamins,
Bechtel
Moderator: Jim Crompton, Chevron
frank: try to capture the outcome and next steps
jim: draw a line here to say that we had a successful meeting, then: what's next?
<mario> Jim: the meeting has been a success
jim: are there real opportuniteis in the oil and
gas ind: yes
... are there opportunities for cooperation: yes
... is there a demonstrated interest: yes
... to ivan: other interest groups has been started, what have other
industries done, scope, drive?
ivan: in w3c we don't really have industry
specs
... however has been working on XBRL
... separate group on eGovernment, not only semantic web
... informally with lawyers
... HCLS is the only one
... all groups plan to produce best practice kind of documents, for their
industry
... all also work on outreach
... and they often come back to the rest of the W3C with technical issues.
Getting this feedback is important for W3C
... HCLS not today asking w3c to store an ontology or so, but may come up
jim: survey of panel on "now so what"
robin: we are relatively new to it, but implementing. We need to colaborate. Next steps: forming broader communication, what is the right entity to work in?
ivan: obvisouly we need the feedback. Traditional
members are large companies, for the HCLS-people they got together with the
IBMs and the Oracles
... half of the interesting tools are made by smaller companies
... we can provide a meeting place
anil: oil industry not trad. been part of w3cx
... don't expect a large participation from oil industry
... however useful things we can do with w3c
... hard to pitch back to shell. What will I have to put in, and what will I
get out
... bandwidth may restrict this to one domain, which one?
... help to set the expectations right
... we need the value of going to w3c
alan: value in the education area. Value in archiving reachable targets
three different target areas: concepts, representations, software
onno: ... oil and gas is a narrow field for us
... to ivan- > what is the idea of "phase 1"
ivan: we need to see if there is enough energy if there is purpose to set up new workshops
onno: posc caesar has an interop group, we propose an action item for this to join w3c
frank: objectives of HCLS IG is similar to us,
see http://www.w3.org/2001/sw/hcls/
... we need to talk to our management to tell them what the benefits are
roger: I was member of HCLS
ivan: when the group started it was narrower, within a year we had a whole range of companies and areas
roger: what did it look like in the beginning
ivan: in practice we need one or two persons 6-7
months to build up the community and to build up a charter
... first action item is to find a person (fellow) to be sent to us to start
work on this
... without this I think it will not happen, we need a champion
jim: who sponsor this? If community cannot come up with this person. Or we dont see a growth in w3c membership. Is that a message to w3c that we are not enough interessted?
ivan: we cannot afford without interest and
finances
... we are accountable to our members
... we need that type of champion coming from the industry
jim: value proposition is important. But we have
ideas that o&g is too small and too large.
... discussion on scope vs. sponsorship
robin: we dont want to get to get in another
group that needs to be funded, why not use an existing? but leverage under
w3c
... under the umbrella of w3c
... there are fiatech, posc caesar, energetics..
steve: going back to HCLS, there are also here
plenty of other org
... w3c was the neutral part
... they came to us where vendors was driving the agenda, legal framework
etc.
... if this is not worth $ 60k compared to the the benefits, we should not be
doing this
... membership fee + people
bertrand: beeing chairman of other org. 1
succeded, another succeded well, two failed
... the ones that "we get the candy" works.
... here we will not have a common enemy, noone is ahead of anyone... so what
will bring us together
anil: how do we found this person? unique skill. Few and hard to give up.
richard: what will drive us? the value
proposition might be around technology and knwoldege and there is real value in
capturing that and feeding back to the business
... about scope: who are we going to attract? look at where we are making
money. Energy is more glamorous but we make our money in oil and gas.
jim: what will bring us together - the great
knowledge exchange and resources are probably scarce.
... who drives? Chevron for the industry? A few leading light companies? What
is the critical minimum that drives the early stages of such a
collaboration?
anil: a majors. a few service providers
ivan: several issues. Issue of the person and the
issue of number of real participants
... the champion has to be seen as a independent person
... not good if the person comes from chevron
... two or three of the major major, two or three of the service providors
... only chevron .... it is crazy. Not seen as serious attempt
alan: it's going to happen in the companies
anyway. Organizations as posc caesar, energetics etc. is going to follow this
anyway
... we can expect energetics to join W3C independent of W3C forming a group
... limited bandwith if w3c don't form a group. If there is a group, then
broadband, public and shared. I see all sorts of dynamics
... i see w3c having an important role in this industry in any case
roger: who drive - I would like to see this
driven by w3c. Too much history in the organizations. Too much happening
outside the industry
... the involvement by organizations is very very important, but I would like
to see it driven by w3c
robin... I just want to see it driven
jim: last words?
ivan: need to continue to see if we can find a
champion, if so, we will try to have this champion to work with us to seed this
organization
... if we do not ahve this champion within 2-3 months
... contacts by energetics will be more narrow
... if anyone wants to be a champion, contact me
jim: wrapping up
... there is a search for a champion
... while we have not reach closure, we have generated exitement.
... I think there is momentum building. Thanks everyone
steve: thanks to the programming committee and chevron for hosting