PANDORA Australia s Web Archive Library Science Talks SNL CERN

PANDORA Australia’s Web Archive Library Science Talks SNL/CERN, September 2004 Paul Koerbin Digital Archiving Branch National Library of Australia pkoerbin@nla. gov. au

PANDORA Australia’s Web Archive 1. Background approach to web archiving 2. The management system (PANDAS) 3. Workflows and procedures 4. Issues and future directions

PANDORA Australia’s Web Archive 1. Background approach to web archiving in Australia - PANDORA

PANDORA Australia’s Web Archive Beginnings • Name originally an acronym for: ‘Preserving and Accessing Networked Documentary Resources of Australia’ • Now: ‘Australia’s Web Archive’ • Began in mid-1996 (selecting) • Began archiving in late 1996 -early 1997

PANDORA Australia’s Web Archive Approach • • Practical and pragmatic Began as: Proof-of-concept project Now: Routine National Library activity Achieving outcomes while continuing to develop and extend processes and systems • Best use of available resources and infrastructure

PANDORA Australia’s Web Archive Resources • Existing technical services staff - librarians • Digital Archiving Branch has the business responsibility • Information technology staff from within the Library for development and support • PANDORA partner institutions (10 including the NLA)

PANDORA Australia’s Web Archive Mandate and responsibilities • National Library of Australia’s statutory responsibilities • National Library Act, 1960 • Maintain and develop a national collection of ‘library material’ • Comprehensive collection relating to Australia and the Australian people

PANDORA Australia’s Web Archive Mandate and responsibilities • National Library has a leadership role for the Australian library community Legal deposit • Legal deposit in the federal jurisdiction in Australia does not cover electronic resources

PANDORA Australia’s Web Archive Some key characteristics • Selective approach to archiving online resources • Scalable to available resources and do-able • Negotiate permission to archive • Apply manual quality assurance processes to harvested resources • Provide access to the archived resources

PANDORA Australia’s Web Archive Shortcomings of selective approach • Can’t collect everything that future researchers may want • Labour intensive tasks • Does not retain the full complexity of the linking structure of the Internet

PANDORA Australia’s Web Archive Indicative statistics as at August 2004 • • 6, 500+ titles 13, 000+ archived instances 21 million files* 680 gigabytes* *These figures are for the display copy only. Two more preservation copies plus preservation metadata are maintained.

PANDORA Australia’s Web Archive 2. The management system: PANDAS

PANDORA Australia’s Web Archive • PANDAS – PANDORA Digital Archiving System • Integrated web based system • Workflow management system • Developed specifically to manage the web archiving processes at the National Library of Australia • Used by PANDORA’s partners located throughout Australia

PANDORA Australia’s Web Archive • Developed in-house at the NLA • Replaced multiple non-integrated systems used between 1996 and 2001 • Written in Java on Apple Web. Objects application development platform • First version released in June 2001 • Second version released August 2002 • Ongoing enhancement and development program

PANDORA Australia’s Web Archive

PANDORA Australia’s Web Archive PANDAS system architecture consists of 4 layers • 1) Presentation layer – client applications for visual presentation to the end user • 2) Application layer – the core application functionality such as PANDAS and PANDORA

PANDORA Australia’s Web Archive PANDAS system architecture consists of 4 layers • 3) Business layer – application access to the data storage and communication infrastructure • 4) Data layer – third party infrastructure products, e. g. Oracle database and Web. DAV accessible files servers

PANDORA Australia’s Web Archive Nomenclature • PANDORA – the whole enterprise • PANDAS – the whole management system • PANDAS – the system component providing a web-based user application to manage workflows • PANDORA – the system component that creates the public interface

PANDORA Australia’s Web Archive

PANDORA Australia’s Web Archive PANDAS is used to: • Record administrative metadata about titles selected (or rejected or monitored) for archiving • Schedule and initiate harvesting • Manage quality assurance checking and problem fixing

PANDORA Australia’s Web Archive PANDAS is used to: • Prepare items for public display through the PANDORA home page • Manage access restrictions • Generate management reports

PANDORA Australia’s Web Archive PANDAS is a workflow system that: • Connects with and utilises other software and protocols for specific functions • Provides an interface to the harvesting software – currently this is HTTrack (http: //www. httrack. com)

PANDORA Australia’s Web Archive PANDAS is a workflow system that: • Uses Web. DAV protocol to provide content managers with remote access to the harvested files • Uses Z 39. 50 protocol to access the National Bibliographic Database to extract metadata from the MARC record

PANDORA Australia’s Web Archive PANDORA public interface component • Title and subject listings and title entry pages are generated ‘on-the-fly’ from PANDAS metadata • Some static web pages (documents, information) • Search engine

PANDORA Australia’s Web Archive Persistent identifiers and URLs • Running number generated by PANDAS • Persistent URL applied to title entry page http: //nla. gov. au/nla. arc-21220 • Logically extended to any resource in the Archive http: //nla. gov. au/nla. arc-21220 -20030822 www. ipjp. org/september 2002/schweitzer-ed. html • Citation generator on public interface

PANDORA Australia’s Web Archive 3. Workflows and procedures

PANDORA Australia’s Web Archive • • Identifying and selecting Recording administrative metadata Harvesting Quality assurance processing Archiving Preparing for public display Creating resource discovery metadata Reporting

PANDORA Australia’s Web Archive Identifying and selecting • Selection guidelines – each partner has their own guidelines • Just guidelines … not rules nor ideology • Selection priorities in guidelines (NLA) • Notification networks – indexing agencies, staff, publishers, public • NLA selection guidelines available at: http: //pandora. nla. gov. au/selectionguidelines. html

PANDORA Australia’s Web Archive Selection – what sort of publications? • • • Titles – the entities to be archived Defined during the selection process Document-like publications, e. g. PDF Whole web sites Parts of web sites

PANDORA Australia’s Web Archive Selection – what sort of publications? • Focus on content – substantial, unique • Special events or issues • Format or potential technical problems are not, in principle, a selection consideration • One-off archiving • Scheduled archiving – whole entity, not an update

PANDORA Australia’s Web Archive Recording administrative metadata • Four types of records – Title – Publisher – Indexer – Collection • Selection status • Additional details associated with status (standing)

PANDORA Australia’s Web Archive Administrative metadata • • • Publisher details Archiving permission status Access restrictions Notes Assigning ownership of titles Transfer titles between agencies

PANDORA Australia’s Web Archive Harvesting • Mostly harvesting from the Web • Also able to upload from local drives (Web. DAV protocol) • Third party software – HTTrack • PANDAS interface to set up harvesting rules

PANDORA Australia’s Web Archive Harvesting • Define extent of selected resource to be archived • Set gather filters and gather settings • Set gather schedule • Initiate harvesting

PANDORA Australia’s Web Archive Scheduling harvesting • Significant function of PANDAS • Regular schedules, e. g. weekly, monthly, annual • Specific dates • Harvest now • Combination of scheduling options

PANDORA Australia’s Web Archive Harvesting - filters and settings • Default settings • Ignore robot. txt rules because permission to archive has been obtained from publisher • Gather sub-directories • Gather ‘near files’, e. g. linked images • Limit on depth – sufficient for any web site but to prevent abuse of host server

PANDORA Australia’s Web Archive Harvesting - filters and settings • • Gather filters are critical Selection based on specific content Archiving permission for specific content Efficient use of resources (bandwidth, storage)

PANDORA Australia’s Web Archive Quality assurance • Important process for PANDORA • Owner of title notified when harvest is complete • Visual, manual checking process • Check for completeness and functionality • Check that content is new (if previously archived) • Check that there is no extraneous material

PANDORA Australia’s Web Archive Quality assurance • Harvested files in a working area – not ‘archived’ at this stage • Web. DAV (protocol) access to the working area • Problem analysis and fixing • Missing files, broken links • Complex problems referred to IT support through PANDAS error reporting module

PANDORA Australia’s Web Archive Quality assurance • Problems due to limitations of harvesting software • Excessive use of Java. Script • Deep web resources • Traps such as metafiles, absolute links • Other methods of acquisition (CD, FTP) • Business decision whether or not to accept the harvested instance

PANDORA Australia’s Web Archive Archiving • Harvested instance is accepted • One-click process for PANDAS user • Transfers instance from working area to Digital Object Storage System • Creates preservation and display copies • Perl scripts – e. g. re-write external links

PANDORA Australia’s Web Archive Archiving – preservation master copies • Preservation master – incl. harvest log files • Display master – includes changes made to the harvested instance (manual and scripts) • Metadata master – http header responses • Gzip compressed TARball (Tape Archive format) on Digital Object Storage System (DOSS) • Access (display) copy on web server

PANDORA Australia’s Web Archive Preparing for public access – title entry pages • Generated ‘on-the-fly’ from content of PANDAS database • Partner branding • Link to publisher’s site • Links to dated archived instances • Manual additions – notes, links to serial issues, copyright statement

PANDORA Australia’s Web Archive Preparing for public access – listings and collections • • Subject listings Title listings Partner views Collections – events, sampling over specific time period

PANDORA Australia’s Web Archive Public access – restrictions • • Period Date Authentication IP addresses/subnet mask (i. e. physical locations such as a single PC in the NLA main reading room) • PANDAS manages automatically – can be manually enabled/disabled

PANDORA Australia’s Web Archive Creating resource discovery metadata • • MARC record for each title National Library of Australia OPAC National Bibliographic Database Metadata derived from the catalogue record is embedded in the title entry pages • Indexing/abstracting services’ citations

PANDORA Australia’s Web Archive Reporting • Pre-defined reports from PANDAS UI • Statistical and data reports • SQL query on Oracle database (not through PANDAS interface) • Pro. Clarity for user defined data cube reporting and analysis • Link. Scan for broken publisher URL links

PANDORA Australia’s Web Archive 4. Issues and future directions

PANDORA Australia’s Web Archive Current issues • Commitment to selective, quality assessed, accessible web archiving • Efficient identification – automated selection • Legal deposit (when? ) • Blanket permission – government agencies

PANDORA Australia’s Web Archive Current issues • Ongoing development and enhancement of PANDAS • Improve robustness of system • Re-engineer PANDAS software • Need to achieve greater efficiencies and increase scale of web archiving activity

PANDORA Australia’s Web Archive Future directions • Automatically ingest and process larger volume of online publications and associated metadata – batches • Comply with international standards and adopt standard tools – IIPC • Incorporate other collection methods – domain harvesting, deep web, deposit

PANDORA Australia’s Web Archive Future directions • Automate collection of more preservation metadata and develop metadata management interface • Improve access and discovery paths to the Archive’s resources as it continues to grow

PANDORA Australia’s Web Archive More information • PANDORA home page http: //pandora. nla. gov. au/ • Key documents (background, technical, PIs) http: //pandora. nla. gov. au/documents. html • PANDAS manual http: //pandora. nla. gov. au/manual/pandas • Papers and presentations http: //pandora. nla. gov. au/papers. html

PANDORA Australia’s Web Archive Questions?

PANDORA Australia’s Web Archive http: //pandora. nla. gov. au