SPIRES and INSPIRE Travis Brooks SLAC National Accelerator Laboratory INSPIRE Collaboration PPA Computing 1 July 2010.

SPIRES and INSPIRETravis Brooks

SLAC National Accelerator LaboratoryINSPIRE Collaboration

PPA Computing1 July 2010

Infrastructure

• The basic facilities, services and installations needed for the functioning of a community or society wiktionary.org

Community• ~30,000 researchers worldwide

• Questions like:

• What is the universe made of?

• What happened 3µs after the Big Bang?

• Distinction between Theory and Experiment

~15,000 HEP scientists smash stuff at the speed of light to produce new stuff

…and it works!

LHC re-discovering known particles for starters.First needles in the haystack: one in a million.

Another 15,000 HEP researchers scratch their heads to make sense of all that stuff and then

some more

Community

• Experiment

• Large, global collaborations ( > 2000 authors!)

• Big centers of research distributed globally

• SLAC, Fermilab, CERN, DESY, KEK

• Theory

• Small, but global collaborations (avg 3 authors)

• Self-contained papers

1960’s - 1970’s

• HEP Lab Libraries store paper preprints

• Distributed via postal mail to major centers

• “Institute-pays” Open Access

• “SPIRES” catalogs (and distributes) preprints received at SLAC

• Centralized, community-driven model

• Users query SPIRES via terminal login accounts

1990’s

CERN Invents WWWUsers query SPIRES at SLAC via 1st Web Site in the U.S.

1991: arXiv.org

Preprint Culture

• Connections/trust/expertise

• Infrastructure from Labs

• SPIRES, WWW, arXiv

• Researcher desire for rapid communication                 

2000’s

• 2007 survey of 2,000 physicists by CERN, DESY, Fermilab and SLAC. “What is your primary HEP Information Resource?” Gentil-Beccot et al, Information Resources in High-Energy Physics: Surveying the Present Landscape and Charting the Future Course. J.Am.Soc.Inf.Sci.60:150-160,2009 arXiv:0804.2701

•97% of published literature freely available on arXiv

• No Mandates – No Debates

Researchers want speed

Researchers want speed

• SPIRES counts: citations to/from preprints/articles

• Citation peaks at publications

• Scientific discourse proceeds on discipline repository

Citation Advantage

• When an arXiv paper is published, it has already surpassed the citation count a non-arXiv paper will have after 2 years

Read Journals?Gentil-Beccot et al.

arxiv:0906.5418

As many scientists as analyzed here go straight to arXiv so80% arXiv users becomes 90% arXiv users

arXiv 82%

Publisher server 18%

∼30,000 clicks (choice between arXiv and journal)

Benefits to Researchers

• Centralized discipline-based repository with curated metadata/search

• Includes Peer reviewed literature

• Links to every known copy

• dois, urls, arXiv

Numbers

• 834,049 (as of Oct 15)

• 50,077 (During 2008)

• 82,719 (Oct 15 - typical)

• 178 (Last week - typical)

What is SPIRES?• Deep, carefully curated metadata

• Authors, Affiliations, Citations, Keywords

• Carefully, intentionally limited to HEP

• Associated community information

• Conferences, Institutions, People, Jobs

Future of HEP Information

• Conversations on arXiv

• Noting, but not waiting for peer review.

• blog/wiki - like

• Rapid turnaround

• Freely accessible content

• Community driven

• Use technology to tighten this relationship further…with an existing community

2010• Past 40 Years: Information

Infrastructure in response to user needs

• Community Needs in 2010:

• Preserve Quality

• Promote Access

• Archive Research Artifacts

2010• Past 40 Years: Information

Infrastructure in response to user needs

• Community Needs:

• Preserve Quality - SCOAP3

• Promote Access - INSPIRE

• Archive Research Artifacts - INSPIRE/HEPData

Quality via Peer Review

• Peer Review and other journal services currently funded by HEP libraries paying for access...

..to material that is freely available

HEP Open Access• LHC scientists (8000 scientists from 54

countries):

• "We strongly […] support the principles of Open Access Publishing, which includes granting free access of our publications to all. Furthermore, we encourage all our members to publish papers in easily accessible journals, following the principles of the Open Access Paradigm."

SCOAP3 Model• An international consortium to convert

existing (and new) top-quality HEP journals to OA

• Libraries re-direct subscriptions to SCOAP3

• SCOAP3 pays centrally for peer-review service

• Price-per-article established by call for tender

• Articles are (free and libre) Open Access

SCOAP3 Partnerships

SCOAP3 Outlook

• Reach critical mass

• Partnership in Asia and Latin America

• Engage publishers in a call for tender

• Go/No-Go decision

2010• Past 40 Years: Information

Infrastructure in response to user needs

• Community Needs:

• Preserve Quality - SCOAP3

• Promote Access - INSPIRE

• Archive Research Artifacts - INSPIRE/HEPData

Future of HEP Information

• Conversations on arXiv

• Noting, but not waiting for peer review.

• Rapid turnaround of freely accessible content

• Community driven

• Literature growing more complex

• Objects that aren’t papers, but are “information”

• “Datasets”, figures, tables, Computer code

• Use technology to tighten this relationship further…with an existing community

Guts...

SPIRES System

• PL360 Emulated in C!

• SPIRES (non-SQL DBMS + internal scripting language)

• And the clearest, least obfuscated, best documented part of the code base is...

• ...Perl!

INSPIRE• Joint Project of CERN, DESY, Fermilab

and SLAC

• Unify SPIRES content with Invenio platform

• Invenio = Open source digital library

• http://invenio-software.org

• http://inspirebeta.net

INSPIRE Philosophy• Leverage Users

• Clean, maintainable, sharable codebase

• Open Source/Open Standards

• Continue manual curation...

• ...but utilize automation feeds where possible

• Utilize person-power to

• drive user participation

• exercise judgement (author ID, classification)

Invenio: Modern System

• Stable, modern, extensible software stack (LAMP)

• Fast, even with large repository

• Focused on search

• Open Source (GPL) community

• Substantial HEP use (CERN, ILC, …)

• Over 20 production instances worldwide

• Modular architecture

• Based on open standards

• MARCXML, OAI-PMH, etc

Opportunities• Enhanced Search and Discovery

• Automated classification using taxonomy

• User tagging

• Organize your personal papers etc.

• Run a Journal Club

• Author identification

• Claim your papers

User tagging

•Hidden 20 FTE - Can be utilized via interactive techniques

• 2007 survey of 2,000 physicists by CERN, DESY, Fermilab and SLAC Gentil-Beccot et al, Information Resources in High-Energy Physics: Surveying the Present Landscape and Charting the Future Course. J.Am.Soc.Inf.Sci.60:150-160,2009 arXiv:0804.2701

Who do we know?

• HEPNames: 80K entries

• Affiliation history for 20K researchers

• Emails for 25K

• 800K papers with authors and (standardized) affiliations

• 5M ‘signatures’ on papers

• 350K unique name strings

Who• Automatic Disambiguation

• Henning Weiler - PhD student@CERN

• On 963 documents, 21 real authors could be identified for the query "Chen, G".

• 22 orphans remain

• 98% identified

User Accounts

• Tied to academic affiliation

• ...and ORCID....

• Ability to correct information and claim papers

• Corrections still vetted by staff

Sources

• Source of 2008 additions

• Many papers have information from multiple sources

• Many arXiv papers will be published later

arXiv• OAI-PMH Feed

• Rough Metadata (author/title/id)

• LaTeX and/or PDF parsing

• Citations, Authors, Affiliations, Keywords

• Parsed by Perl/Python

• Checked (or redone) by Humans

Journals

Publishers

• APS (Phys.Rev.D, Phys.Rev.Lett.)

• Elsevier (Phys.Lett.B, Nucl.Phys.B)

• Springer (Eur.Phys.C, JHEP(>2010))

• IOP (J.Phys.G, JHEP (<2010))

Feeds• APS in OAI-PMH

• Full Metadata + References

• Elsevier, Springer, JHEP

• In-house XML via FTP

• Rich Metadata, Most with References

• Fall back to screen-scraping HTML

Users• In 2008:

• 173 Added papers directly from users

• 3,800 Papers with user updates/corrections to reference lists

• 4,000 User updated profiles (institutional history, etc)

Export• DOIs, publication information to arXiv,

ADS

• bidirectional exchange of XML

• Currently: Rough “API” with in-house XML formats for Physicists building apps

• INSPIRE:OAI-PMH interface, rich API

• NLM DTD

• MARCXML

2010• Past 40 Years: Information

Infrastructure in response to user needs

• Community Needs:

• Preserve Quality - SCOAP3

• Promote Access - INSPIRE

• Archive Research Artifacts - INSPIRE/HEPData

~15,000 HEP scientists smash stuff at the speed of light to produce new stuff

…and it works!

LHC re-discovering known particles for starters.First needles in the haystack: one in a million.

Data• HEPData - Durham U.

• Stores Data “behind” figures/tables

• Submitted from Experiments

• INSPIRE partners with HEPData

• Provides access, linking and deposition in central community location

• Serve “long-tail” of theorists and others with “misc.” materials

• Enables access citation etc..

Existing Infrastructure

Data

• Trusted Community Infrastructure

• Future?

• DPHEP Study Group

• Continuing conversation with researchers to develop data preservation strategy

Conclusion• Access, Quality, and Artifacts

• Emerging from community of researchers

• Aligned with community needs

• Target what scientists need

• Quality - Speed - Completeness

• Building on existing, trusted infrastructures

Infrastructure

• The basic facilities, services and installations needed for the functioning of a community or society wiktionary.org

Questions?

• For more information on INSPIRE see

http://www.projecthepinspire.net http://inspirebeta.net