The document discusses using nanopublications to address issues with how scientific findings are currently published and communicated. Nanopublications involve subdividing findings into small atomic assertions, attaching provenance and metadata to each assertion, and packaging them together as linked data packages called nanopublications. This allows for automated understanding and integration of scientific claims directly from publications. The approach enables reliable, decentralized publishing of scientific data on the web through the use of trustworthy identifiers.
1. Linked Data Publishing with Nanopublications
Tobias Kuhn
http://www.tkuhn.org
@txkuhn
Department of Computer Science, VU University Amsterdam
IOS Press 30 Year Anniversary
Amsterdam, Netherlands
4 April 2017
2. Problem: We Communicate through Papers
that Software Can’t Understand
scientific paper
scientist
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 2 / 16
3. Problem: We Communicate through Papers
that Software Can’t Understand
millions of new
papers every year
scientific paper
?!
scientist
Which genes are
related to
mental diseases?
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 2 / 16
4. Problem: We Communicate through Papers
that Software Can’t Understand
millions of new
papers every year
scientific
databases
software
scientific paper
?!
scientist
Which genes are
related to
mental diseases?
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 2 / 16
5. Automatic Text Mining is
Not Good Enough
World-leading text mining on
chemical–disease relations:
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 3 / 16
6. Automatic Text Mining is
Not Good Enough
World-leading text mining on
chemical–disease relations:
Manual Text Mining is
Slow and Expensive
Around 50 biocurators employed to
feed European protein databases:
read papers &
feed databases
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 3 / 16
7. New Paradigms of Scientific Publishing?
scientist other
scientists
scientific
papers
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 4 / 16
8. Where are we Now? Where is the Data?
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 5 / 16
9. Where is the Data?
In the Supplementary Material
...
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 6 / 16
10. New Paradigms of Scientific Publishing?
scientist other
scientists
scientific
papers
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 7 / 16
11. A New Paradigm of Scientific Publishing
scientist
bits of formally
structured
knowledge
scientific
database
causes(
GeneX,
DiseaseY
)
other
scientists
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 8 / 16
12. Nanopublications: Linked Data Containers for
Provenance-Aware Semantic Publishing
assertion
provenance
publication info
nanopublication
http://nanopub.org
@nanopub org
• Subdivide scientific findings into the
smallest possible atomic pieces
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 9 / 16
13. Nanopublications: Linked Data Containers for
Provenance-Aware Semantic Publishing
assertion
provenance
publication info
nanopublication
http://nanopub.org
@nanopub org
• Subdivide scientific findings into the
smallest possible atomic pieces
• Attach provenance and metadata on
that atomic level
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 9 / 16
14. Nanopublications: Linked Data Containers for
Provenance-Aware Semantic Publishing
assertion
provenance
publication info
nanopublication
http://nanopub.org
@nanopub org
• Subdivide scientific findings into the
smallest possible atomic pieces
• Attach provenance and metadata on
that atomic level
• Represent everything as Linked Data
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 9 / 16
15. Nanopublications: Linked Data Containers for
Provenance-Aware Semantic Publishing
assertion
provenance
publication info
nanopublication
http://nanopub.org
@nanopub org
• Subdivide scientific findings into the
smallest possible atomic pieces
• Attach provenance and metadata on
that atomic level
• Represent everything as Linked Data
• Make a small package out of these
three parts: assertion, provenance,
publication info
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 9 / 16
16. Nanopublications: Linked Data Containers for
Provenance-Aware Semantic Publishing
assertion
provenance
publication info
nanopublication
http://nanopub.org
@nanopub org
• Subdivide scientific findings into the
smallest possible atomic pieces
• Attach provenance and metadata on
that atomic level
• Represent everything as Linked Data
• Make a small package out of these
three parts: assertion, provenance,
publication info
• Then we treat each of these small
packages as an independent
publication, and we call them
nanopublications
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 9 / 16
17. Nanopublication Example
:assertion {
:p occursIn: mesh:D004730 .
:p geneProductOf: hgnc:3763 .
}
:provenance {
:assertion prov:hadPrimarySource
pubmed:12891700 .
}
:pubinfo { :np
dct:created 2014-07-03 ;
pav:createdBy
orcid:0000-0001-6818-334X . }
Complete example: https://goo.gl/f7iPKK
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 10 / 16
19. Reliable Identifiers
(with Cryptographic Hashes)
Make nanpublications ...
Verifiable
+
Immutable
+
Permanent
.trighttp://example.org/r1. RA 5AbXdpz5DcaYXCh9l3eI9ruBosiL5XDU3rxBbBaUO70
http://trustyuri.net/
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 12 / 16
20. Decentralized and Reliable Publishing with a
Nanopublication Server Network
Nanopublications
with Trusty URIs
Publication
Retrieval
Propagation /
Archiving
http://purl.org/nanopub/monitor
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 13 / 16
22. Highly Reliable Data Publishing and Retrieval
Reliable even when done automatically by software.
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 15 / 16
23. Highly Reliable Data Publishing and Retrieval
Reliable even when done automatically by software.
So, be prepared for the raise of the Science Bots!
S C I E N C E B O T S
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 15 / 16
24. Thank you for your attention!
Further information:
• Nanopublications: http://nanopub.org
• Trusty URIs: http://trustyuri.net
• More: http://www.tkuhn.org
Tobias Kuhn, VU University Amsterdam Linked Data Publishing with Nanopublications 16 / 16