Conclusions

1
Conclusions
2
LSIDs suck (sadly)
3
suckis a technical term
4
DOIs suck ( )
5
Handles suck less
6
Metadata matters
7
RDF rocks
8
XML schema suck
9
What we need
10
Unique identifiers
11
Resolvable
12
Have metadata
13
Taxonomic names arent enough
14
Names have too much information
Cherie Booth
15
Names can change when circumstances change
Cherie Booth
Cherie Blair
16
Names carry meaning
Jonathon Roughgarden
Joan Roughgarden
17
Semantically opaque
18
identifier has no meaning
19
trouble with meaning
20
Zbtb7
21
POK erythroid myeloid ontogenic factor
22
POK erythroid myeloid ontogenic factor
23
Pokemon gene
24
(No Transcript)
25
Pokemon causes cancer
26
Funny!
27
(No Transcript)
28
Not funny
29
Zbtb7
30
Zbtb7
31
LSID parts
32
Opaque is a myth
33
Credit card
34
LSIDs are nice
35
Explict metadata and data access
36
(No Transcript)
37
(No Transcript)
38
LSIDs suck
39
Have to fuss with DNS
40
Reliant on Internet address
41
What about DOIs ?
42
Resolve this
43
doi10.1080/10635150490264996
44
What do you get?
45
Have subscription?
46
(No Transcript)
47
What, no subscription?
48
Metadata?
49
Human-readable documents
50
Cant predict what you get
51
Handles might be useful
52
hdl2254/20971
53
Handle to HTML
54
HTML is XML
55
GUID resolving to metadata
56
Metadata matters
57
RDF
58
Resource Description Framework
59
Simple format (e.g., XML)
60
Everything is a resource
61
or a literal
62
supports inference
63
underpins Semantic Web
64
property
subject
object
65
triple
66
http//www.w3.org
World Wide Web consortium
dcpublisher
67
(No Transcript)
68
(No Transcript)
69
RDF is everywhere
70
RDF is everywhere
71
RDF is everywhere
72
Existing vocabularies
73
Basic metadata (Dublin Core)
74
Geography (WGS 84)
75
Publications (PRISM)
76
People (FOAF)
77
Rights (Creative Commons)
78
Requires you to have URIs for objects
79
URIs include
80
URL
81
URN
82
DOI
83
LSID
84
RDF documents can be independent
85
Can be as small as one triple
86
Aggregate triples from different sources
87
Store in a triple store
88
Make new inferences
89
There are known knowns, things we know that we
know
90
There are known unknowns, things we now know we
dont know
91
But there are also unknown unknowns, things we do
not know we don't know
92
unknown knowns
93
things we dont know we know
94
latent knowledge
95
International Plant Names Index (IPNI)
Poissonia heterantha
Tephrosia heterantha
basionym
Coursetia heterantha
Tephrosia heterantha
basionym
96
International Plant Names Index (IPNI)
Poissonia heterantha
basionym
Tephrosia heterantha
Coursetia heterantha
basionym
Poissonia heterantha
Coursetia heterantha
?
97
IPNI knows these names are synonyms
98
But doesnt know it knows it
99
Melissotarsus insularis
100
Melissotarsus insularis
no hit
101
label
HNS30687
Melissotarsus insularis
subject
CASENT0107663-D01
HNS30687
source
CASENT0107663-D01
DQ176312
subject
TaxId342313
label
TaxId342313
Melissotarsus sp. BLF m1
102
GUIDs are trivial
103
Its about metadata
104
Its about inference
105
Lessons from DOIs
106
Crossref
107
reference backbone
108
add value to electronic publications
109

110
(No Transcript)
111
(No Transcript)
112
(No Transcript)
113
Crossref assigns DOI prefix
114
Crossref stores metadata
115
Crossref can be searched
116
Sound familiar?
117
(No Transcript)
118
Oh, the vision thing George Bush (Snr), 1987
119
GBIF assigns handles
120
Data sources provide identifiers
121
Data sources provide metadata and data
122
GBIF has metadata standards
123
Data sources supply metadata to GBIF
124
Low barrier to entry
125
Data sources install Java Handle System (or
whatever)
126
If use RDF then GBIF can support inference
127
We learn something we didnt know
128
Are we ready to do this?
129
Quality of service
130
24
131
7
132
365
133
downtime
134
(No Transcript)
135
Memorandum
136
ASMX
137
404
138
GUIDs are easy
139
Metadata is what counts
140
and persistence
141
(No Transcript)