This is very cool stuff, if they ever manage to make it all work.
Christine Chichester - Knewco Inc.
"Community peer review in Wiki environment"
goal: distill down every unique scientific concept to a unique identifier (the "knowlet")
Many challenges in current biomedical research
* volume of data
* distributed systems and databases
* incompatible data formats
[Brazilian Portugese sp etc]
*** ambiguity of terminology
* inability to share knowledge
"Too much to read" indicates major trends
* from reading to consulting
* from reading to meta analysis
* from texts to facts
... to central and community annotation
difficult homonym disambiguation issue: use context
- first order symantic enrichment
a knowlet is a triple
- wiki annotation
- concept profile match
- sequence similarity
build an association matrix for large data sources
- disambiguation of author names
[Dr. somebody has algorithms]
1 million disambiguated authors
- from MEDLINE
1 million for genes, drugs and ?proteins
Assignment of protein function and discovery of new nucleolar proteins based on automtic analysis of medline
Martijn Schuemie, Christine Chichester, Frederique Lisaceck, Yahoo Coute, Peter-Jan Roes, Jean Charles Sanches, Barend Mons
Special issue Systems Biology in Protemics, 2008 (in press)
put discovered hypotheses into WikiProf and then if approved into e.g. swissprot
- GO gene ontology
- NLM UNMS??? UNLS?
runs on OmegaWiki which uses MediaWiki
[knowledge space knowlet thing]
Wikiproteins Peer Review: ??? automated selector/requestor for peer review of annotations ???