TDWG2010 Annotations Session 1: Existing Systems

• Micheal Giddens, Silver Biology :

State Louisiana imaging specimen sheet project: digital annotation are properties of morphological characters, determination (annotation of image) share annotation among community: generate key ... not reside in the image

A more general sense of annotation: tag, comments. information other than the original information. Different system may have different interpretation of the meaning of annotation and what is an annotation

Should we start with absolute abstract level? How to expose them

Observation

• Jason Best, BRIT

extracting label from specimen sheet with human interaction define region of interest of texual part of ocr

• Amanda Neill, BRIT

view annotation histroy group level, specimen level determination,confirmation, other, context, notes generate annotation label delete etc maintainence functions

• Rod Spears, Specify

Through workbench, send request to mexcio, Gbif, get information back to auguament their record with duplicate and become annotation Gbif is a little messy

• Silver lighting project

subcripber increase . Scale up pub-sub-hub

• BiSciCol:

collection tracker collection are distributed, how to share infformation among instituations who share the same collection related publication

a web of information with relation (duplicate) maitained by a cenrreal serveice ,(harverst every thing have guid

subscriber need automatically pull information in interaction

• BioCase:

ABCD schema 1000 elements annotation should be able deal with that amount of possible elemnt base on raw xml and diff, online of data source.

• Morph banck:

Image annotation: Image is annotation of clollection , image need annotated to which it is related meta data can link to an ontology term

• FilteredPush:

Three use cases about annotating distributed data Find duplicates: avoid retyping, establish connection make annotation: distributed based connection quality control: annotation injection - dispacting annotation based on system knowledge(without explicit addressing) message volume, security

annotation for quality control annotation ontology: common properties of annotation subject, content, motivation, evidence, annotator,guid,expectaiton, time Content is domain specific

• Walter:

need a more general discussion annotator may not know they are annotating for others holder may not have interest of getting annotation back to database

RAW NOTES TAKEN BY Pete Ruhl, USGS

Michael Giddins

---

digitizing herbarium sheets

- want to allow users to add information to the "digital sheets" - i.e. to annotate them.

herbarium conceptual model. 1. you have the physical specimen 2. you mount it 3. you annotate it (anything other than the physical specimen itself is "an annotation".

Pete Ruhl's speculation: 1. the physical specimen is given GUID. 2. the location and ownership properties for the specimen identified by GUID are *NOT* annotations. 3. *Any* information that is attached to the GUID by the object owner is an "annotation".

Jason Best (Apiary project).

1. digitized specimens 2. extracting information from the herbaria sheets (that stuff are all "annotations".

 - primary label
 - handwritten notes on mounting sheet

Pete's Comment: "Annotation" is a domain-specific term for "data about a physical specimen".

Atrium Biodiversity Information System - can annotate at collections level, or for individual sheets.

note: looks like annotations can be structured, scalar or unstructured (e.g. block of text notes written by investigator).

Specify Workbench

investigators add value and "correct" or fill out, "complete" a record. The modified data are an "annotation", not part of the "original" record.

pmruhl: I see value-added addendum to the original records.

Morphbank Aaron Steele

Morpbank publishes annotations. Others subscribe. When you have lot's of subscribers it gets to be a problem.

PubSubHub protocol

BiSciCol Biological Science Collections Tracker

Trying to build architecture for linking and transmitting information about "same things" stored in many different places in distributed system.

(seems like a messaging architecture with a central messaging hub to me).

BioCase project: EDIT Specimen Portal

Annotation service ABCD schema used. Complicated to add annotation.

- Their approach: add annotation to ABCD schema. Use edits, xml stored in subversion, then show diffs.

(note: again, annotation is addition to, or modification of, original or "master" record.)

Greg Riccardi, Morphbank/Morphster/Specify

Cypriniforme Tree of Life Project

FISH !!!

Morphbank image bank, images annotated to i,llustrate characteristics of morphology.

annotations (arrows and such) were embedded in the image itself. Should be separated from the image and discoverable and describable in it's own right.

NEW PROJECT

Specify as spec catalog, Morphbank for images, Morphster and Ontobrowser

Filtered Push Use Cases

1. Find Duplicates 2. Make Annotation 3. Quality Control Data

Herbaria: many sheets created from collection of "one" specimen. Collect - mount on several sheets. Send duplicates to friends. Wouldn't it be nice if basic information was distributed to all holders of those sheets? Wouldn't it be great if all annotations made (where ever) are distributed to all holders of those sheets? Wouldn't it be nice if QA procedures and outcomes were distributed to all holders of the sheets? -- sheets = digital representations that are out there in cyberspace.

Filtered Push Network.

So - what is the concern of an annotation