SlideShare a Scribd company logo

Web Today, Good Tomorrow? Transactional archiving of web content

•Download as PPTX, PDF•
•
Peter Burnhill
Peter Burnhill

The document discusses the problem of reference rot, where citations and references in scholarly articles break over time as web resources change or disappear. It provides three key points: 1. Scholarly articles are increasingly linking to web resources rather than just other articles, putting those references at risk. Studies show around 20-30% of references are already "missing" or lead to changed content. 2. Content drift is also a problem, where the content behind a cited URL changes significantly over time, undermining the integrity of the citation. Studies found only 25% of references from 2012 articles remained unchanged by 2015. 3. Five options are presented to remedy reference rot: taking snapshots of URLs, helping authors archive references,

1 of 32
Web Today, Gone Tomorrow? => transactional archiving of web content Peter Burnhill University of Edinburgh Professional/Scholarly Publishing (PSP) Division, Association of American Publishers (AAP) Washington DC, 1-4 February 2017
12 ‘Dark Archive Nodes’ in long-lived research institutions in 8 different countries/jurisdictions: North America: Indiana, Rice, Stanford, Virginia, OCLC; Alberta (Ca.) Europe: Edinburgh (UK); Humboldt (Ger); Cattolica dSC (It.) Asia/Pacific: ANU; NII (Japan); UHK Triggered 29 titles so far [1.1 m downloads in 2016] Triggered release at Stanford & EDINA via OpenURL's to local library link-resolvers & CrossRef CLOCKSS Archive Network Library Stewardship: Global & Decentralized not-for-profit joint venture Board: 12 publishers & 12 libraries Cross-sectoral collaboration & innovation Stanford TRAC Certified
① Web-scale not-for-profit archiving agencies: ① National institutions (usually national libraries) … ① Consortia of university libraries & specialist centres … National Science Library, Chinese Academy of Sciences 1. We now have a variety of digital shelving  National Science Library, Chinese Academy of Sciences Good News: a lot of online e-journal content is being kept safe Swiss National Library
… to discover who is looking after whatAn established Global Monitor thekeepers.org 2. We have means to search ‘holdings’ on digital shelves 12 ‘keepers’ (+ Swiss National Library) Funded by: Developed & managed by: on Title or ISSN, using the ISSN Register & ISSN-L as kernel field
3. Use Registry as ‘Observatory’: provide evidence on progress
very many ‘at risk’ e-journals from the “65% of publishers”: the hardest to reach & work with BIG publishers act early but incompletely ** Amber Alert ** a lot of Arts, Humanities, Law & ‘applied’ literature not being archived STEM Journals well archived
Progress as archiving agencies form a Keepers Network  to tackle that Long Tail and ensure completeness => Their recent Statement * endorsed by library community • ARL + CARL + LIBER + RLUK + AUL IARLA : International Alliance of Research Library Associations • Ivy Plus Libraries Collections Group, USA + library groups in Canada, Australasia, South America and Europe * ‘Working Together to Ensure the Future of the Digital Scholarly Record’ http://thekeepers.blogs.edina.ac.uk/keepers-extra/ensuringthefuture => Need support from Publishers & Publisher Associations 1. To read and endorse the Keepers Statement * • be vocal to all publishers in your support of archiving agencies • make it easier for archives to ingest your content & keep it safe 2. To dble-check actual ingest of your content via Keepers Registry
References to Content => Back into Scholarly Publications => Out onto the Web at Large Has ‘fixity’ dynamic , lacks fixity DOI, ISSN CLOCKSS, Portico, CrossRef, etc URLs ‘Web today, gone tomorrow’ Reference RotE-Journal Archiving #keepers #hiberlink Threat to Integrity of scholarly publication => References to Content Now The Bad News: 3 Red Alerts for Publishers
Project 2 years: March 2013 to June 2015 Funder Andrew W. Mellon Foundation Partners University of Edinburgh EDINA & Language Technology Group, School of Informatics Los Alamos National Laboratory ambition 1. Define and measure the extent of ‘Reference Rot’ 2. Scope possible intervention opportunities to stop the rot we did that and went further to 3. Devise sustainable solutions capable of maximal reach The aim today is to 4. Prompt action by those who can make a difference …
arXiv Elsevier corpus PMC Dark solid lines represents URIs to Web-at-large, from 1997/2011 Red Alert 1 Scholarly Articles increasingly link to Web Resources, not just back to other Articles Klein M, Van de Sompel H, Sanderson R, Shankar H, Balakireva L, et al. (2014) Scholarly Context Not Found: One in Five Articles Suffers from Reference Rot. PLoS ONE 9(12): e115253. doi:10.1371/journal.pone.0115253 http://journals.plos.org/plosone/article?id=info:doi/10.1371/journal.pone.0115253 Data: 1.2m articles with URI references, of which 393,000 to ‘Wild Web’ => 1million URIs
Reference Rot = Link Rot + Content Drift When what was referenced & cited ceases to say the same thing, or ‘has ceased to be’ http://www.snorgtees.com/this-parrot-has-ceased-to-be 1. Link Rot: Link stops working => two questions about the 1 million URLs to Web-at-large 1. Do those links (URLs) still work? - on the ‘Live Web’’? 2. Is there a ‘Memento’ of that reference in the ‘Archived Web’?
Klein M, Van de Sompel H, Sanderson R, Shankar H, Balakireva L, et al. (2014) Scholarly Context Not Found: One in Five Articles Suffers from Reference Rot. PLoS ONE 9(12): e115253. doi:10.1371/journal.pone.0115253 http://127.0.0.1:8081/plosone/article?id=info:doi/10.1371/journal.pone.0115253 within 14 days of publication date … PMC Elsevier ‘Not Archived’ 74.5% 75.2% Of those ‘Not Archived’ % % still ‘Live’ on the Web 80 67.3 ‘No longer Live’ on the Web 20% 32.7% Many ‘missing, presumed lost’ Most referenced URIs at risk of loss Team at Harvard Law School established similar evidence • 70% of the URLs within [law] journals & 50% of the URLs within U.S. Supreme Court opinions … “do not produce the information originally cited.” Jonathan Zittrain, Kendra Albert and Lawrence Lessig (2014). Perma: Scoping and Addressing the Problem of Link and Reference Rot in Legal Citations. Legal Information Management 14. doi:10.1017/S1472669614000255. Red Alert 2 Reference Rot is already significant
Content Drift is even scarier! Red Alert 3 when what is at end of cited URL has changed, or gone!! http://dl00.org 2000 http://dl00.org 2004 http://dl00.org 2005 http://dl00.org 2008 (a) Dynamic content as values on webpage changes over time (b) Static content but very different (often unrelated) web pages
‘Similarity’ of Representative Mementos & Live Web Content as at August 2015 by Year of Publication 655,000 Elsevier articles, 1997 to 2012 Jones SM, Van de Sompel H, Shankar H, Klein M, Tobin R, et al. (2016) Scholarly Context Adrift: Three out of Four URI References Lead to Changed Content. PLOS ONE 11(12): e0167475. doi:10.1371/journal.pone.0167475 http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0167475 ‘Similarity’ decreases over time After 3 years, only ¼ of URIs lead to unchanged content + increase in Link Rot 25% * fresh evidence on ‘Content Drift’ *
only about 25% of referenced resources In articles published in 2012 remain unchanged by August of 2015 25% 25% 25% Confirmed in all 3 datasets
=> Content of Citations Rot over Time!! … leading to rotten references for the reader Get Smell Out Copyright © 2017
Rot in References means a Defective Article! undermines the integrity of the scholarly record  http://www.fao.org/wairdocs/tan/x5883e/x5883e01.htm
So what should to expect of the Publisher? Beyond the assurance that the fish / references / articles sold are not rotten Kind permission from Manchester Evening News
5 Options to Remedy Reference Rot Hint: Remedy for fish is ‘Quick Freeze & Store with Date Stamp’ Kind permission from Asia Quality Control Always end on the +ve … !!
①Take Snapshot of what is at end of URL & put in safe place until needed by reader • Various web archives support on-demand creation of snapshots of URLs: – archive.is / Internet Archive / perma.cc / webcitation.org Archive-It @archiveitorg perma.cc @permacc
Decide where to intervene for best effect? Activity Actor Snapshot Quality 1. Preparation Author/reference tool best 2. Submission /Issue Editor/manuscript system good 3. Access (post-publication) Aggregator/ publisher platform better late than not 4. Shelving Librarian/IR, journal archive better than nothing Need to put the means of re-creating fixity within the software being used in each workflow
‘Best’ would be to help authors do right thing - at earliest moment of capture! http://the-animals-biography.blogspot.co.uk/2014/04/kingfisher.html
.. when the Authors are trawling for content
• Preparation -> Study -> Compose -> Submission => Good News: something already exists … • Hiberlink Project: EDINA developed code for Zotero [open source] Note University of Edinburgh now investigating how to assist doctoral students with their references to web resources in e-theses ② Help the Author record their dependencies? • ‘transactional archiving’ of referenced web content • do it when noted & citation created • OK, but how to effect change in note-taking software? eg EndNote, Mendeley, Reference Manager, RefMe, Zotero
Need to create a time-based record of what an Author regards as significant …
… or needs to provide as evidence! Alexander Lexén https://www.flydreamers.com/en/photo/alexander-lexen-s-fly-fishing-catch-of-a-european-brown-trout-fly-dreamers-pic291999
More Good News: Metadata for the citation of that Snapshot Three key elements should be recorded in the citation: 1. Original URL 2. Snapshot URL where the web content was archived 3. Date/Time when the snapshot was taken (& archived) A proposed standard ‘Robust Links’ syntax is set out at http://robustlinks.mementoweb.org/
③ Adapt the publisher process to ‘stop the rot’ • Submission -> Editing -> (Revision) -> Acceptance -> Issue a) Publishers should create Snapshots in web archives • Editors to use citations with the 3 Robust Link elements b) Submission systems should accept citations submitted with Robust Link syntax! • Engage / amend / use ‘Robust Links’ syntax => Yet More Good News: something already exists … Hiberlink Project: algorithm created for OJS [open source] ; code in GitHub
④Value in having ‘Hibernator’ Infrastructure Publishing platform ‘Hibernator’ External archival service e.g. Internet Archive, Perma cc • Asynchronous - returns Hiberlink in Robust Link format • Distributed - archived in different locations • Lightweight - leveraging HTTP & what already exists as middleware which simplifies interaction between publisher systems & web archives Note University of Edinburgh is building the Hibernator for its doctoral students to support references used as evidence in e-theses
Activity Responsibility Snapshot Quality 3. Access Platform better late than not ⑤Act to help the Reader, given rot Access/Post-Publication -> Reader Access -> Use • Install ‘Link Decoration’: enable readers to employ Memento for search web archives for content ‘around time of submission’ Finish on this Good News: Herbert Van de Sompel et al. (2015) Robust Links - Link Decorations http://robustlinks.mementoweb.org/spec/
Thank You: Questions Welcome p.burnhill@ed.ac.uk With kind permission from 'Feather Saturnfly' on flickr, All Rights Reserved
Useful links – that still work  Hiberlink.org Scholarly Context Adrift: Three out of Four URI References Lead to Changed Content. PLOS ONE 11(12) doi:10.1371/journal.pone.0167475 http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0167475 The Cobweb: Can the Internet be archived? New Yorker, Annals of Technology, January 2015 http://www.newyorker.com/magazine/2015/01/26/cobweb The growing problem of Internet “link rot” and best practices for media and online publishers https://journalistsresource.org/studies/society/internet/website-linking-best-practices-media-online-publishers Law Library of Congress Implements Solution for Link and Reference Rot https://www.digitalgov.gov/2016/04/13/law-library-of-congress-implements-solution-for-link-and-reference-rot/

More Related Content

Web Today, Good Tomorrow? Transactional archiving of web content

  • 1. Web Today, Gone Tomorrow? => transactional archiving of web content Peter Burnhill University of Edinburgh Professional/Scholarly Publishing (PSP) Division, Association of American Publishers (AAP) Washington DC, 1-4 February 2017
  • 2. 12 ‘Dark Archive Nodes’ in long-lived research institutions in 8 different countries/jurisdictions: North America: Indiana, Rice, Stanford, Virginia, OCLC; Alberta (Ca.) Europe: Edinburgh (UK); Humboldt (Ger); Cattolica dSC (It.) Asia/Pacific: ANU; NII (Japan); UHK Triggered 29 titles so far [1.1 m downloads in 2016] Triggered release at Stanford & EDINA via OpenURL's to local library link-resolvers & CrossRef CLOCKSS Archive Network Library Stewardship: Global & Decentralized not-for-profit joint venture Board: 12 publishers & 12 libraries Cross-sectoral collaboration & innovation Stanford TRAC Certified
  • 3. ① Web-scale not-for-profit archiving agencies: ① National institutions (usually national libraries) … ① Consortia of university libraries & specialist centres … National Science Library, Chinese Academy of Sciences 1. We now have a variety of digital shelving  National Science Library, Chinese Academy of Sciences Good News: a lot of online e-journal content is being kept safe Swiss National Library
  • 4. … to discover who is looking after whatAn established Global Monitor thekeepers.org 2. We have means to search ‘holdings’ on digital shelves 12 ‘keepers’ (+ Swiss National Library) Funded by: Developed & managed by: on Title or ISSN, using the ISSN Register & ISSN-L as kernel field
  • 5. 3. Use Registry as ‘Observatory’: provide evidence on progress
  • 6. very many ‘at risk’ e-journals from the “65% of publishers”: the hardest to reach & work with BIG publishers act early but incompletely ** Amber Alert ** a lot of Arts, Humanities, Law & ‘applied’ literature not being archived STEM Journals well archived
  • 7. Progress as archiving agencies form a Keepers Network  to tackle that Long Tail and ensure completeness => Their recent Statement * endorsed by library community • ARL + CARL + LIBER + RLUK + AUL IARLA : International Alliance of Research Library Associations • Ivy Plus Libraries Collections Group, USA + library groups in Canada, Australasia, South America and Europe * ‘Working Together to Ensure the Future of the Digital Scholarly Record’ http://thekeepers.blogs.edina.ac.uk/keepers-extra/ensuringthefuture => Need support from Publishers & Publisher Associations 1. To read and endorse the Keepers Statement * • be vocal to all publishers in your support of archiving agencies • make it easier for archives to ingest your content & keep it safe 2. To dble-check actual ingest of your content via Keepers Registry
  • 8. References to Content => Back into Scholarly Publications => Out onto the Web at Large Has ‘fixity’ dynamic , lacks fixity DOI, ISSN CLOCKSS, Portico, CrossRef, etc URLs ‘Web today, gone tomorrow’ Reference RotE-Journal Archiving #keepers #hiberlink Threat to Integrity of scholarly publication => References to Content Now The Bad News: 3 Red Alerts for Publishers
  • 9. Project 2 years: March 2013 to June 2015 Funder Andrew W. Mellon Foundation Partners University of Edinburgh EDINA & Language Technology Group, School of Informatics Los Alamos National Laboratory ambition 1. Define and measure the extent of ‘Reference Rot’ 2. Scope possible intervention opportunities to stop the rot we did that and went further to 3. Devise sustainable solutions capable of maximal reach The aim today is to 4. Prompt action by those who can make a difference …
  • 10. arXiv Elsevier corpus PMC Dark solid lines represents URIs to Web-at-large, from 1997/2011 Red Alert 1 Scholarly Articles increasingly link to Web Resources, not just back to other Articles Klein M, Van de Sompel H, Sanderson R, Shankar H, Balakireva L, et al. (2014) Scholarly Context Not Found: One in Five Articles Suffers from Reference Rot. PLoS ONE 9(12): e115253. doi:10.1371/journal.pone.0115253 http://journals.plos.org/plosone/article?id=info:doi/10.1371/journal.pone.0115253 Data: 1.2m articles with URI references, of which 393,000 to ‘Wild Web’ => 1million URIs
  • 11. Reference Rot = Link Rot + Content Drift When what was referenced & cited ceases to say the same thing, or ‘has ceased to be’ http://www.snorgtees.com/this-parrot-has-ceased-to-be 1. Link Rot: Link stops working => two questions about the 1 million URLs to Web-at-large 1. Do those links (URLs) still work? - on the ‘Live Web’’? 2. Is there a ‘Memento’ of that reference in the ‘Archived Web’?
  • 12. Klein M, Van de Sompel H, Sanderson R, Shankar H, Balakireva L, et al. (2014) Scholarly Context Not Found: One in Five Articles Suffers from Reference Rot. PLoS ONE 9(12): e115253. doi:10.1371/journal.pone.0115253 http://127.0.0.1:8081/plosone/article?id=info:doi/10.1371/journal.pone.0115253 within 14 days of publication date … PMC Elsevier ‘Not Archived’ 74.5% 75.2% Of those ‘Not Archived’ % % still ‘Live’ on the Web 80 67.3 ‘No longer Live’ on the Web 20% 32.7% Many ‘missing, presumed lost’ Most referenced URIs at risk of loss Team at Harvard Law School established similar evidence • 70% of the URLs within [law] journals & 50% of the URLs within U.S. Supreme Court opinions … “do not produce the information originally cited.” Jonathan Zittrain, Kendra Albert and Lawrence Lessig (2014). Perma: Scoping and Addressing the Problem of Link and Reference Rot in Legal Citations. Legal Information Management 14. doi:10.1017/S1472669614000255. Red Alert 2 Reference Rot is already significant
  • 13. Content Drift is even scarier! Red Alert 3 when what is at end of cited URL has changed, or gone!! http://dl00.org 2000 http://dl00.org 2004 http://dl00.org 2005 http://dl00.org 2008 (a) Dynamic content as values on webpage changes over time (b) Static content but very different (often unrelated) web pages
  • 14. ‘Similarity’ of Representative Mementos & Live Web Content as at August 2015 by Year of Publication 655,000 Elsevier articles, 1997 to 2012 Jones SM, Van de Sompel H, Shankar H, Klein M, Tobin R, et al. (2016) Scholarly Context Adrift: Three out of Four URI References Lead to Changed Content. PLOS ONE 11(12): e0167475. doi:10.1371/journal.pone.0167475 http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0167475 ‘Similarity’ decreases over time After 3 years, only ¼ of URIs lead to unchanged content + increase in Link Rot 25% * fresh evidence on ‘Content Drift’ *
  • 15. only about 25% of referenced resources In articles published in 2012 remain unchanged by August of 2015 25% 25% 25% Confirmed in all 3 datasets
  • 16. => Content of Citations Rot over Time!! … leading to rotten references for the reader Get Smell Out Copyright © 2017
  • 17. Rot in References means a Defective Article! undermines the integrity of the scholarly record  http://www.fao.org/wairdocs/tan/x5883e/x5883e01.htm
  • 18. So what should to expect of the Publisher? Beyond the assurance that the fish / references / articles sold are not rotten Kind permission from Manchester Evening News
  • 19. 5 Options to Remedy Reference Rot Hint: Remedy for fish is ‘Quick Freeze & Store with Date Stamp’ Kind permission from Asia Quality Control Always end on the +ve … !!
  • 20. ①Take Snapshot of what is at end of URL & put in safe place until needed by reader • Various web archives support on-demand creation of snapshots of URLs: – archive.is / Internet Archive / perma.cc / webcitation.org Archive-It @archiveitorg perma.cc @permacc
  • 21. Decide where to intervene for best effect? Activity Actor Snapshot Quality 1. Preparation Author/reference tool best 2. Submission /Issue Editor/manuscript system good 3. Access (post-publication) Aggregator/ publisher platform better late than not 4. Shelving Librarian/IR, journal archive better than nothing Need to put the means of re-creating fixity within the software being used in each workflow
  • 22. ‘Best’ would be to help authors do right thing - at earliest moment of capture! http://the-animals-biography.blogspot.co.uk/2014/04/kingfisher.html
  • 23. .. when the Authors are trawling for content
  • 24. • Preparation -> Study -> Compose -> Submission => Good News: something already exists … • Hiberlink Project: EDINA developed code for Zotero [open source] Note University of Edinburgh now investigating how to assist doctoral students with their references to web resources in e-theses ② Help the Author record their dependencies? • ‘transactional archiving’ of referenced web content • do it when noted & citation created • OK, but how to effect change in note-taking software? eg EndNote, Mendeley, Reference Manager, RefMe, Zotero
  • 25. Need to create a time-based record of what an Author regards as significant …
  • 26. … or needs to provide as evidence! Alexander Lexén https://www.flydreamers.com/en/photo/alexander-lexen-s-fly-fishing-catch-of-a-european-brown-trout-fly-dreamers-pic291999
  • 27. More Good News: Metadata for the citation of that Snapshot Three key elements should be recorded in the citation: 1. Original URL 2. Snapshot URL where the web content was archived 3. Date/Time when the snapshot was taken (& archived) A proposed standard ‘Robust Links’ syntax is set out at http://robustlinks.mementoweb.org/
  • 28. ③ Adapt the publisher process to ‘stop the rot’ • Submission -> Editing -> (Revision) -> Acceptance -> Issue a) Publishers should create Snapshots in web archives • Editors to use citations with the 3 Robust Link elements b) Submission systems should accept citations submitted with Robust Link syntax! • Engage / amend / use ‘Robust Links’ syntax => Yet More Good News: something already exists … Hiberlink Project: algorithm created for OJS [open source] ; code in GitHub
  • 29. ④Value in having ‘Hibernator’ Infrastructure Publishing platform ‘Hibernator’ External archival service e.g. Internet Archive, Perma cc • Asynchronous - returns Hiberlink in Robust Link format • Distributed - archived in different locations • Lightweight - leveraging HTTP & what already exists as middleware which simplifies interaction between publisher systems & web archives Note University of Edinburgh is building the Hibernator for its doctoral students to support references used as evidence in e-theses
  • 30. Activity Responsibility Snapshot Quality 3. Access Platform better late than not ⑤Act to help the Reader, given rot Access/Post-Publication -> Reader Access -> Use • Install ‘Link Decoration’: enable readers to employ Memento for search web archives for content ‘around time of submission’ Finish on this Good News: Herbert Van de Sompel et al. (2015) Robust Links - Link Decorations http://robustlinks.mementoweb.org/spec/
  • 31. Thank You: Questions Welcome p.burnhill@ed.ac.uk With kind permission from 'Feather Saturnfly' on flickr, All Rights Reserved
  • 32. Useful links – that still work  Hiberlink.org Scholarly Context Adrift: Three out of Four URI References Lead to Changed Content. PLOS ONE 11(12) doi:10.1371/journal.pone.0167475 http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0167475 The Cobweb: Can the Internet be archived? New Yorker, Annals of Technology, January 2015 http://www.newyorker.com/magazine/2015/01/26/cobweb The growing problem of Internet “link rot” and best practices for media and online publishers https://journalistsresource.org/studies/society/internet/website-linking-best-practices-media-online-publishers Law Library of Congress Implements Solution for Link and Reference Rot https://www.digitalgov.gov/2016/04/13/law-library-of-congress-implements-solution-for-link-and-reference-rot/