|
This is version 39.
It is not the current version, and thus it cannot be edited.
[Back to current version]
[Restore this version]
- Updates: Overview and Progress
- General Discussion
- Current issues
- New ideas
- Priorities
- Goals (Short-term, 6-month, 1 year)
- Bertram (overview)
- requirements:
- semantic (in addition to structural) annotation of:
- services (actor) and
- datasets
- for ... discovery, WF design, ...
- prerequisites: what does this annotation langauge have?
- structural types
- array of float; list of record of (int, float, string); etc.
- e.g., relational schema type, ptolemy type
- applies both to data and "functions" (actors)
- essentially, useful for plumbing -- the level of the programmer, or data provider
- semantic type
- formal expression referencing concepts in an ontology
- to provide a high-level, conceptual of data and services
- data(set)
- informal descriptions: eml style metadata, keywords, controlled vocabularly
- formal description
- dataset ~ concept
- attribute (column) ~ concept
- fine-grained:
- query(X1, X2, ..., Xn) ~ concept expression(X1, ..., Xn)
- Xi are "tags" or variables point to substructure
- actors (services)
- overall function
- input/output signature (-> reuse framework from dataset annotation)
- Questions
- How do you require that certain things must be registered, e.g., that a lat lon point has an x and y coord
- How do you handle null values in a table ... e.g., a row doesn't have an x value
- What is the difference between an annotation langauge and a description logic
- Shawn (details of semantic registration language)
- simplified Antweb example
- abundance datasets
- d1,d2 have similar structure (geo-referenced data)
- d3 is differently structured (co-occurence based)
- d4 lists host/parasite relationships (inquilinism)
- Measurement, parasitism ontology (from SWDB'04 paper)
- Presentation
- observation, TaxonID, Parasite, Host
- Example: annotating d1 ...
d1(Ge,Sp,Co,Lt,Ln) ~
Observation(O), value(O,Co), context(O,S), location(S,P), LatLonPoint(P),
latDeg(P,Lt), lonDeg(P,Ln), item(O,A), Abundance(A), property(A,N), SciName(N),
genus(N,Ge), species(N,Se).
- Who is creating these mappings?
- Shawn: use of a tool
- Discussion about interface ...
- Can we use growl browser to do the semantic annotation mappings graphically
- similar in style to xml mapping tools
- what is missing:
- multiple selection of nodes
- difficult to get relevant stuff when you expand nodes ... many nodes you don't care about up the inheritence tree
- just want subclass and derived properties
- graph query language ... to filter ... decouple what you show and select
- e.g., i am only interested in isa links, so you might want to have a power user language, to subset what you are currently showing
- would generalize; think it would be a useful extension to these visual editors, ... "+ isa; - any"
- expression language for power use to turn on and off certain things; incremental search that start from results
- graphic frame logic
- a fundamental problem with visualization formulas as opposed to logical _consequent_ of the formulas
- Jessie: ProFormaVis -- based on Angiosperm ontology
- structures, attributes-values
- reads in ontology; divides the gui into structure hierarchy, selected structure description details
- the ontology doesn't try to define what a plant is, but gives the set of things you might want to say about plants
- trying to describe a specimen, e.g., leaf, with attribute veination pattern, and anastomosing and arcuate...
- the veination patther is anastomosing and arcuate
- the end to end application: the scientist refines the set of things and properties (attribute values) that are useful for a collection; a form is generated / configured based on those filters; for others to classify items to
- Bob Morris, et al., allows general description of things -- Structure of Descriptive Data -- TADWIG
- Commercial program called Lucid 3
- Nico on Taxon / SMS
- partial overlaps
- give me the ants of south america ... give all concepts of ants (may be thousands) ...
- Deana: in biodiversity use case; want to group by function, which is similar to form
- Several international ones that are comprehensive in traits and functions
- Weren't sure the relevance to north america plants
- in KNB now, can do exactly the same search (via ITIS) ... through markup via EML
- rich: given two identifiers how are they similar
-
- follow-up list
- growl extension selective viewing / filtered views based on some keywords / or in general queries / or just let user select nodes in the neighborhoods in the tree
-
Back to the Annual meeting agenda
|
