Scholarly Communications in an Electronic Age ASIST 2003 Panel Session Bradley Hemminger School of Information and Library Science University of North Carolina at Chapel Hill bmh@ils.unc.edu

Public Storehouse of Knowledge Multiple open digital archives, holding all the world’s knowledge. A single logical universal archive, created by dynamic federation of all public archives. Contains everything: archive holds grey literature (publicly deposited) and gold literature (refereed articles). No barriers to access. Knowledge is freely available to anyone, any time, anywhere. Access to information and knowledge correlates to society’s quality of life.

University of California Electronic Repository UNC Chapel Hill psycprints Harvester (NeoRef) Virginia Tech ETD arXix University of California Electronic Repository

Archive Model (NeoRef) All material and metadata are author contributed to a public OAI archive (author retains ownership). OAI archives have automated or manual moderator to filter out “junk”. Everything--articles, reviews, comments, indexings, etc., are stored as digital content items on archive using the same mechanism. Reviews contain quantitative score, qualitative grade, qualitative comments. All materials universally available via search engines that harvest metadata from OAI archives. Retrieval is through Google like one stop shopping search interface, with dynamic filtering based on metadata and reviews to limit hits to manageable number to review.

Challenges are in Retrieval All material is archived (good and bad) Metrics (some new) are used to differentiate type, content, and quality. Dynamic Searching allows quickly finding materials of most interest. Search on Type article=Review AND date > 1950 Content (schizophrenia AND GeneX) Quality: Peer reviewed {journals}, citation rate > XYZ

How Peer Review might work… Author submits article to her institution’s open archive (DOI uncch:sils/0007548.pdf). Author “submits” to journal EMEDICINE by providing DOI of article. Journal Editor schedules two reviewers. Reviewers review article, and submit their reviews (cornell:0191.pdf, ucb:0084.pdf). Author revises, and places revised article (DOI uncch:sils/0007957.pdf) on archive, and submits this final version to EMEDICINE. Journal submits review (EMED:0023424.pdf) which is final statement from journal (editor), and indicates acceptance of uncch:sils/0007957.pdf as EMED article).

Scholarly Communications Process Today’s Example Present to colleagues V2 Present at conference V3 Idea V1 Submit to journal V4 Revision to include additional new results V8 Revision to update analysis V7 Referees Revision for journal V5 Journal Final Revision V6

Scholarly Communications Process: What’s Captured Today Journal Final Revision V6 Only one version is captured, and the same community then pays to buy back access to article

Scholarly Communications Process:What I’d like to see saved! formulate discussion discussion, revision Present to colleagues V2 Present at conference V3 Idea V1 Submit to journal V4 comments comments comments Author revision Revision to include additional new results V8 Revision to correct analysis V7 Referees Revision for journal V5 Journal Final Revision V6 Copyproofing Criticisms, new thoughts, revision new results, revision Two peer reviews

Change the Process! Think of scholarly communication as continuous process instead of single product (journal publication). Capture significant changes/versions of a work. Include all criticisms and comments about work (all stages). Support normal scholarly discourse, including authors responses as well as others comments. Add reviewer’s quantitative rating of material to allow better filtering based on absolute quality metric during retrieval. Add machine (automated) reviews.

Can we save the Gold and Grey? formulate discussion discussion, revision Present to colleagues V2 Present at conference V3 Idea V1 Submit to journal V4 comments comments comments Author revision Revision to include additional new results V8 Revision to correct analysis V7 Referees Revision for journal V5 Journal Final Revision V6 Copyproofing Criticisms, new thoughts, revision new results, revision Two peer reviews

NeoRef Storage Model Digital Archive Auto-indexing Revision to include additional results and analyses V8 Author Indexing Journal Final Revision V6 Comments on V3 Journal Submission V4 Comments on V6 Material expressing content Conference paper (v3) Local powerpoint Presentation v2 Two peer reviews Machine Review Digital Archive Filter (Moderate) Author Grey Literature Automated Content Item Recognized Expert Top Tier (Keep Forever) Open (anyone)

Selected Technical Challenges Self Contribution Author indexed Author supplied metadata (Dublin Core) as part of authoring process (i.e. not separate after the fact). Automatic extraction of metadata from document. Archive file(s) must be in standard open format NeoRef: PDF/A with DC elements in tags for automatic extraction of metadata. Expect migration to XML as we continue to divorce content from presentation.

Challenges Searching DC metadata to allow coarse discovery. Specialized searching within domain after locating material (based on metadata field indicating appropriate search interface). Interactive searching to allow refinement to most desirable set within a few seconds. Use reviews to help filter search (Facultyof1000). Google searching on full text (covers all materials, but generates large number of hits, lower specificity). Automated agents to bring material of interest to your attention (California digital library). Example: article scores > 7.0, refereed, citation count above 10, type=research article, search terms = schizophrenia, geneX)

Challenges Knowledge Representation Extend DC to include “concepts” and “claims” (ScholOnto) to allow higher level searching compared to simple indexing. Make OAI and DC representation more robust by always supporting DOI to uniquely identify materials. Support unique identification of authors as well. Making all content items submitted permanent Use DC fields to link related items, new version of paper to old version.

Challenges Rights Administration Support mechanisms to allow authors to set permissions as they desire, and enforce this. NeoRef supports Creative Commons through DC rights element. OAI recent supported rights administration using Creative Commons (and looking at how to handle collections etc where DC rights element may be insufficient).

What do users want? The ALPSP survey was intended to discover the views of academics, both as authors and as readers. Some 14,000 scholars were contacted across all disciplines and all parts of the world, and with almost 9% responding. Alma Swan and Sheridan Brown. Authors and Electronic Publishing: The ALPSP Research Study on Authors' and Readers’ Views of Electronic Research Communication. (West Sussex, UK: The Association of Learned and Professional Society Publishers, 2002). http://www.alpsp.org/pub5.htm

Importance of journal features

Importance of the peer review process

Importance of publishers’ roles Factor Responses as authors Responses as readers Peer review 81 80 Gathering articles together to enable browsing of content 64 49 Selection of relevant and quality-controlled content 71 54 Content editing and improvement of articles 60 39 Language or copy editing 50 34 Checking of citations/adding links 46 28 Marketing (maximising visibility of journal) 44 20

Survey (Project Romeo) Authors want quick and convenient dissemination of their work Free access to others papers Not overly concerned (or aware) of copyright issues unless it stops them from freely distributing their work or accessing others.

Survey (Zhang 1999) Important to authors are Permanence and Quality of electronically survey archived resources Better (faster, more accurate) searching capabilities, i.e. using metadata instead of just search engines.

Report version: http://www.jisc.ac.uk/uploaded_documents/rowland.pdf Survey (Rowland) 16% said that the referees would no longer be anonymous 27% said that traditional peer review would be supplemented by post-publication commentary 45% expected to see some changes in the peer-review system within the next five years Fytton Rowland, “The Peer-Review Process,” Learned Publishing 15 no. 4 (October 2002): 247-258. Report version: http://www.jisc.ac.uk/uploaded_documents/rowland.pdf

Extra Material….

Additional Challenges Archive Hosting Off the shelf computer system with lots of disk space and public domain archiving application (DSpace, Eprints). Who maintains the material? {Library (MIT DSpace), Grad School, University (California), Publisher (PLoS, BioMedCentral), Society (arXiv)} Where are comments and reviews held (after the fact content items that reference original)?

Challenges Make content universally available Export OAI items so they can be harvested Have public domain quality harvesters that support quick and simple searching (i.e. Google for metadata).

Challenges Peer Review Make more public. Make available comments on articles. Add quantitative scoring as well as qualitative.

Overview of Peer Review Qualitative Quantitative Grade Score (1-10) Quantitative Filter Published Article Review Peer, Open, Machine Accept, reject, revise with respect to XYZ standards Article submitted Send elsewhere Comments to Author Reject

General Review Model Parallels In general, you have sample (material) which is judged/scored quantitatively and qualitatively by an identified observer with respect to some standard.

NeoRef for Movies, Products,… The same process used by NeoRef to support Scholarly Communication could be used for most any communication of informaiton purpose. All that is required is storage of Digital Content Items, and linking of reviews, comments, etc to them. DocSouth: self cataloged and indexed items are Grey; librarian/archivist cataloged and indexed items are Gold. Movies: Grey is everyone’s reviews; Gold is Siskel and Ebert reviews. Consumer Products: product reviews by Consumer Reports (gold), user reviews (grey).

Current Peer Review Options Quantitative √ Qualitative √ (relative) √ (absolute) √ Human Judgement Expert peer review (status quo) Certified expert peer review Open Peer Review BMJ, BioMed Open comment review psycprints Computer Judgement Computer peer review Human Usage Citation-based (CiteSeer) Usage counts (CiteSeer) Example Quantity of discussion Coarse Categorization Two Tier (grey/gold) Moderator (current arXiv) No review (old arXiv)

Importance of future dissemination channels Dissemination method Very important plus important categories Ranking Traditional print + electronic journal 91 1 Discipline-based electronic reprint archive 78 2 Traditional print journal 77 3 Traditional electronic-only journal 66 4 Institution-based electronic reprint archive 60 5 New forms of electronic-only journal 49 6 Discipline-based electronic preprint archive 44 7 Institution-based electronic preprint archive 33 8

Provider Service Change What is worth paying for? Quality review (Faculty of 1000) Proofing, citation linking, professional presentation (CiteSeer, Cite-base) Marketing Archival (JStor) Who hosts material: Society (arXiv) Commerical Publishers (Elesiever,BioMedCentral) University Library (MIT Dspace)

CRADLE Center for Research and Development of Digital Libraries Electronic Theses and Disserations Minds of Carolina NeoRef