EDIT WP5

advertisement
EDIT WP5.2 3RD MODELERS’ MEETING
Budapest, May 7th 2007 (Day 1)
In attendance: IK Imre Kilian (HNHM), AG Andras Gubanyi (NHNM), DT Dave Taylor
(RBGK), VK Veronika Kiss (HNHM), MCE Malte Ebach (BGBM), DL Dilan Laetif (SNMS),
MD Markus Döring (BGBM), MJ Mark Jackson (RBGK) & LS Lutz Suhrbier (FUB).
1. Revisionary Model update






Not interviewed so far: Belgium, France, Spain and Russia. Problems with
communication with Russia and Spain (but former now resolved). Trips arranged to
Russia and Belgium. Paris will be done shortly.
Final Step: Discussion about interviews, lists of software and resources and putting
together a simple example of the revisionary model.
Requires feedback from the draft report. So far DT has been getting very little
feedback: MCE has suggested that he use the WP5 blog and WP6 website to
“advertise” the report for feedback. Given that there is little feedback it is difficult to
find out what WP5 & 6.
DT summarizes Prelim. Rev. Taxo. report:
o There is a common revision model in the coarse scale.
o Difficult to accurately quantify how long individual processes in the model
have taken (a-elapsed time from start to end, b-amount of time spend actually
working) since some processes have taken over 18 months and interviewees
have not kept records. Required to define bottlenecks objectively
o DT & MJ suggested soliciting further info by sending out questionnaires that
state the rev. taxo. results, i.e. the unified model (e.g what the average
taxonomists does), and then ask for a response on the rev. model including a
list of software that people use and bottlenecks they see. Maybe ask for
priorities in component developments too (gather feature requests).
There is a mismatch between the bottlenecks provided by WP6 taxonomists and those
uncovered in the Rev. Taxo. Model (i.e. WP6 taxos user requirements refer to post
revision stage bottlenecks with a new technology, not with existing bottlenecks in the
taxo. process).
The final Rev. Taxo. report and model will be submitted at the end of July (see
below).
2. ATBI Bottleneck Analysis







Klaus Riede and the Budapest team have not been able to meet due to time constraints.
IK and VK share responsibilities (both half-time positions).
See VK talk (see WP5 wiki: http://wp5.e-taxonomy.eu/wiki/AtbiModels).
The factor of time has been incorporated into the ATBI model.
Results are presented as detailed data (e.g. MSExcel spreadsheets etc.).
VS presents her results of the interviewing process. Three types of interviews took
place (including questionnaires) involving 8 biologists:
o Rev. Taxo. Process.
o ATBI+M Process.
o Timesheet: monitoring the time used during the ATBI+M Process.
IK and VS seem confident that the timesheets are a reliable reflection of the time used


in the ATBI modelling process.
It was agreed at the last impromptu modeller’s meeting in Stuttgart that the same
timesheet model for both ATBI and Rev. Taxo. will be helpful for the platform in
several ways:
o Targeting certain existing practises in the taxo. workflow.
o Introducing possible solutions (e.g. recommending a time saving component).
Relevant bottlenecks need to be identified namely those that can be directly addressed
by the platform.
3. ATBI Workflow Modelling









See VK talk (see WP5 wiki: http://wp5.e-taxonomy.eu/wiki/AtbiModels).
IK differs from DT models in the:
o Level of detail (DT captures detail in text document, IK in the model itself)
o Include simple data and control flow
o Indicate activities that could be computerized with colors
Rev. Tax. Model and ATBI model will not result in a single application.
Our intentions need to be made clear (i.e. what is the platform etc.).
WP5.2 need to promote what the platform is to taxonomists (i.e. through mock-ups
etc.) in order to explain what we mean and in order to get feedback. Once we have
obtained this feedback we may be able to decide whether the two models can be
combined.
If it will only take a week for IK and DT to meet and work together in order to create a
unified or merged model, then not much time has been wasted (i.e. unified model is
not a deliverable or milestone) and the data for each model and be cross-checked and
further common bottlenecks can be identified.
A merged, unified model has been proposed.
The Unified Model will contain “subpackages” for specific activities carried out only
by some taxonomists that are skipped by others.
Doubts whether time recording can be done for rev. taxo. (e.g. absolute figures would
relate to differing work units, so may have to capture percentages).
4. ATBI Demand Driven Design



See VK talk (see WP5 wiki: http://wp5.e-taxonomy.eu/wiki/AtbiModels).
IK showed the data from the timesheet questionnaire (imported as an MSExcel file in
to MSProject) as a gannt diagram.
There seems to be very little communication between Riede and the Budapest team
(Riede having no time to dedicate to studying the model and providing feedback).
5. Actions





IK and DT will meet in order to merge the models (in Budapest) ASAP and agree
future work practices to ensure divergence does not occur.
The unified model will be stored in a common server (using EA with subversion at
the BGBM).
Once merged, IK and DT will further develop the unified model as they gather further
information, working independently but liaising with each other as necessary.
Delay the decision on publishing the results of the models (i.e. journals etc.).
All modelling deliverables will be posted on the WP5 blog.



DT to develop spreadsheet to capture timings (sensu ATBI timesheets) based on the
unified model, and organise interviews by self with taxonomists in London, and by
key EDIT partners (e.g. RMCA, MIZPAN, HNHM, IBSAS, BGBM, SMNS, CSIC,
MNHN, UKBH) in order to extract this information .
D5.13 and 5.14 (due Month 29) will be combined as a single deliverable in the next
JPA as only one platform will exist.
D5.8 will be delayed by one month (new due date Month 17) in order to add in results
of additional timing assessments.
EDIT WP5.2 3RD MODELERS’ MEETING
Budapest, May 8th 2007 (Day 2)
In attendance: IK Imre Kilian (HNHM), AG Andras Gubanyi (NHNM), DT Dave Taylor
(RBGK), VK Veronika Kiss (HNHM), MCE Malte Ebach (BGBM), DL Dilan Laetif (SNMS),
MD Markus Döring (BGBM), MJ Mark Jackson (RBGK) & LS Lutz Suhrbier (FUB).
1. The Internet Platform, the user interface and the ATBI+M and Rev. Taxo. Model





Definitions:
o Platform = interface definitions to allow interoperability of components.
Infrastructure/Middleware is closest term we can think of now. The platform is
not “visible” as a website or alike.
o User Interfaces exist for the different application components. There is no
single interface for the platform.
With a loose set of tools a user (taxonomist) needs guidance how to find them and
how/when to use them. The software tracker allows to find applications, but doesn’t
show people how to do a revision with the platform. The unified workflow model can
guide users to the appropiate tools for their work. The workflow can be entered at any
point. A user having data can start with the publishing activity. Behind all activities
there should be a page describing the activity, link to tools, explain differences and
maybe explain alternatives.
Descriptive data is core part of taxonomic revision and is hard to separate from
taxonomic data. Ideally all characters of all examined specimens would be measured,
but manually this is clearly too much. So current taxonomic practice is actually
derived from a “bottleneck” situation, but not seen as such.
Need to define the data coming out of a taxonomic revision to build Interfaces.
Character states in a spreadsheet? Protolog/Diagnosis as unstructured text?
Agreements summarised in plenum:
o couple descriptions with taxon THIS IS THE TAXONOMIC CORE
o have a common data model
o platform=infrastructure
o no single workflow mandated, but standard workflow recommended. enter
workflow at any step
o facts, i.e. TDWGs TaxonDataModel has to be dealt with. separate from tax
core, doesn’t alter a taxon. That includes distribution!
o platform can be used by individual researchers or collaboratively in a group.
o need technical committee meeting to harmonize all of WP5 activities
o hosting services need to be adressed in ISTC, esp. what to do with TBytes of
Images? Does one institute provide this service as “integration” measure or
do we outsource this?
2. Data modelling in UML



See IK talk (see WP5 wiki).
UML modelling is the preferred method for data modelling.
Klaus Riede’s database was discussed and several queries were made:
o We need to assess what the purpose of the ATBI database is and why WP7
does not use existing software such as Recorder or BioOffice.
o The Budapest team and DL need to know what the user requirements are from
WP7.
o Klaus Riede, DL and IK need to discuss what data model they need and
whether existing software is worthwhile using.
o Consider looking into accessing better ways of using taxonomic data from
existing databases (e.g. Fauna Europea). Sounds like a typical platform
component.
3. Further WP5 IT Meetings


This meeting is the last Modeller’s meeting. Meetings from now on will involve all
WP5, 6 & 7 developers.
The tentative meeting date for a two-day “EDIT Developers Meeting” is 6-7
September, 2007 in Berlin (BGBM).
Download