1	An Arizona Model for Web Preservation and Access Richard Pearce-Moses Director of Digital Government Information Arizona State Library, Archives and Public Records Phoenix
2	The World Has Changed The web realized and expanded Vannevar Bush’s vision of memex The web catalyzed the information explosion The web challenges the traditional definition of publication
3	Libraries Must Respond Reassess criteria for collecting agency publications Scale workflows to enormous numbers Adapt curatorial practices Retool core skill sets Recognize profound impact on preservation programs
4	Curation Activities (Means) Identification and selection Acquisition Description Reference Preservation Principles and traditions (Objectives)
5	Curatorial Responses Bibliocentric Model Item-level control, highly selective No significant difference between traditional print and digital materials Technocentric Model Captures everything, patrons select Relies on full-text searching If it’s not online, it doesn’t exist
6	An Arizona Model An archival approach Designed to manage large numbers of documents with limited resources Based on observation that websites are similar to archival collections Documents share the same provenance Creators organize documents into groups (series / directories) Aggregates (provenance and series) serve as the basic units of curation
7	Identification Domain is a clue to provenance adobe.com adot.state.az.us (Transportation) azgovernor.gov (Governor) phoenixvis.net (Environmental Quality) Scalable 1,500 distinct domains on 50,000 pages 500 not obvious, needed to be checked 160 domains have in-scope content
8	Selection/Appraisal Based on aggregates, not items Directories contain similar materials Reflects web masters’ sense of order Analysis of URLs reveals hierarchy _derived (scripts) Browse (many directories) ContactUs (ephemera) AnnualReports (blank forms) Publications (nearly empty)
9	Selection/Appraisal – 2 Browse/ DRTF (Drought Task Force) AMA (Area management plans) Surface (Surface water agreements) WQARF (Water Quality Assurance Fund) Website had ~5,000 documents ~50 directories and subdirectories ~25 were in-scope
10	Description Interpolation not transcription Collections, series have no title page Many web documents have little or no publication information Many web documents have poor quality publication information Mapping the forest Humans describe the abstract Computers describe the concrete
11	Description – 2 Series titles Series/directory names transformed from URLs to English Scope and contents note Why are these documents valuable? What are they about? Who’s responsible for them? Access points Associates controlled vocabulary with series
12	Acquisition Largely automated Can create a pick list Many repositories will accept default Harvesting software restricted to In-scope domains Selected series (directories) Specified file types (.doc, .html) Packages documents With supplied and internal metadata As complete units
13	Reference Google defines patrons’ expectations Classification can improve retrieval Classes eliminate noise by grouping similar documents Headings refine queries by suggesting related concepts
14	Reference – 2 Classification based on controlled vocabulary assigned to series
15	Preservation: Two senses Conservation Long-term accessibility Much work needs to be done Pear-Preserves To capture what will be lost until the documents can be conserved
16	Web Archives Workbench Library of Congress National Digital Information Infrastructure Preservation Program (NDIIPP) University of Illinois at Urbana-Champaign OCLC Five state libraries (AZ, CT, IL, NC, WI) Other content providers Practical tools and a workbench
17	Tools – Being Tested Domain tool Returns a list of distinct domains on seed sites Tracks decisions: in-scope, out-of-scope, new Entity tool Links domains to agencies and subordinates Organizes agencies, subordinates into a taxonomy Records information about agencies, subordinates Version 0.1 being tested now
18	Tools – Coming Soon Analysis tool Shows organization of website series Records selection decisions about series Associates supplied metadata with series Title, Scope note, Dates, Access points Harvester Downloads documents Packages document with supplied and internal metadata Delivers METS packages to be loaded on another system Due for testing January 2006
19	Stay tuned! Comments and criticisms welcome Richard Pearce-Moses rpm@lib.az.us Paper and slides available at www.lib.az.us/diggovt/azmodel/GODORT/