Tracking Online Document Usage from Your Catalog

Tracking Online Document Usage from Your Catalog

Using statistics to demonstrate the value of government publications

Six State Virtual Government Information Conference August 11-13 and 16-17, 2010

Susan Kendall, Coordinator for Collection Development And Government PublicationsSan José State University, Dr. Martin Luther King, Jr. LibraryEmail: [email protected]

Reasons for tracking usages

1. Usage shows how government publications support students’ & faculty’s research.

2. Statistics can become a part of the assessment & evaluation done during accreditation visits.

3. Data can show how government publications can inform the public in news-worthy events.


1. Usage shows how government publications support students’

& faculty’s research.In April, 2010, we had students in Health Sciences researching breast cancer. Several government publications were accessed. Here are two of our most requested government publications that month:


2. Statistics can become a part of the assessment & evaluation

done during accreditation visits.

At San José State, the library was asked for a 5 year review to present to the University Faculty Senate. Statistics from e-Government Publications were an

important part of the report.


3. Data can show how government publications can inform the

public in news-worthy events.Recent reports on the safety of bottled water and their plastic containers led toa GAO study being one of our top accessed e-government publications in April, 2010.


Accessing eGovPubs 2005-2009


Steps to producing our statistics report

The following slides are from a presentation given at the NASIG

conference May 20, 2005 by Susan Kendall and Celia Bakke


Government Publications Architecture

Front-end

stat_govpub.htm parameters receive: -bibNum-vendor url

Collects data:-bibNum-suDoc #-class-title

Gov_Pub DB

stat_govPub_(month).txt (text file)

Stores data:- bibNum

Stores data:-bibNum-suDoc#-class-title

Redirect to vendor website

insert into

insert into

* Extract data using cfhttp to initiate a one-way request from information from a remote server (the library catalog) http://mill1.sjlibrary.org/search/.bibNum/.bibNum/1,1,1,B/marc~bibNum

extract data using cfhttp *

Back-end

Client

MS SQL DB Server

Programming languages: COLDFUSION, HTML, CSS Database: Miscrosoft SQL Database (MS SQL DB)

Lyna Nguyen

Government Publications Architecture

Front-end

Retrieves/Groups/Counts data:-count-bibNum-suDoc #-class-title

Gov_Pub DB

stat_govPub_(month).txt (text file)

Stores hit data:- bibNum

Data from DB:-bibNum-suDoc#-class-title

Back-end

Admin.

stat_govPub.htm-login/logoff-view by month & year-sort by: a-z, suDocs, highest hits-search by bibNum, suDocs#, title read file

query data

MS SQL DB Server

user submits

display to web browser

extract data

connect to db

Programming languages: COLDFUSION, HTML, CSSDatabase: Miscrosoft SQL Database (MS SQL DB)

Lyna Nguyen


Steps to Modifying the Bibliographic Record

1. Identify the Bibliographic Record2. One of our most accessed records this

Spring Semester:

TITLE: Miss Atom Bomb


Identify the Bibliographic Record Number

B35936691 – Bibliographic Record Number

856 field

Then add the bibliographic record number to the prefix

The prefix is: http://univ-intranet.sjlibrary.org/scripts/database_statistics/stat_govpub.htm?id=35936691


Find the URL for the record

The 856 field will have the URL address

http://purl.access.gpo.gov/GPO/LPS101060 SJSU


Then add tracking information

856 40 |uhttp://www.sjlibrary.org/scripts/ stat_govpub.htm?id=35936691&path=http:// purl.access.gpo.gov/GPO/LPS101060|xSJSU


Overall Database changes:

Search records in the database using Review Files– Initial search strategies:

• 856 has “GPO” (Bibliographic records search)• Since most of these records have a “purl”,

records are grouped into one review of batch changes.


Next step:

• Use a script/macro to copy the bibliographic record number for each record

• Add the prefix to the URL.• Also consider using a “do loop” in the

script to perform batch changes• Majority of records can be batch

processed with the scriptSix State Virtual Government Information Conference August 11-13 and 16-17, 2010

Time required for initial run

• 37,000 bibliographic records / 50,000 entries

• 2 weeks to run initial database change• Many records had non standard URLs

attached


On-going monthly maintenance

• Search for records to be changed after downloading monthly Marcive records

• Use script to do an automatic search• Scan the records to check URLs• Run a script/macro to batch change the

records


On-going monthly maintenance and time

consideration

• Total staff time: less than 10 minutes

• Total Machine time: less than 2 hours


Program can sort by the number of “hits”


Program can sort by Title:


Program can sort by SuDocs Number

What I’ve learned from our usage statistics project:

• Our e-govpubs continued to be accessed via the catalog.

• The catalog is still a viable point for accessing information.

• With our statistics, we have made a case for the Chancellor’s Office to consider funding the Marcive program “Documents without Shelves” for all 23 CSU libraries.


I would like to acknowledge the work done by:

Shirley Hwang, Lyna Nguyen, Celia Bakke and April Gilbert

fromSJSU University Library


Tracking Online Document Usage From Your Catalog

Christopher C. BrownUniversity of Denver, Penrose Library

[email protected]


Documents’ Electronic Environment

• 97% of current GPO catalog records contain links to online content

• Library ILS systems provide circulation stats for books, but have no ability to provide stats for online clickthroughs

• Library directors increasingly believe depository collections are not used, because we cannot provide them the most relevant stats


Statistics we now know

• Documents Received• Circulation Statistics (from our ILS reports)

• GPO PURL Referral Statistics (see http://www.fdlp.gov/component/docman/cat_view/178-collection-management/249-purl-referrals for individual library statistics; see also http://fdlp.gov/collections/building-collections/618-purl-referrals-reporting for discussion of recent issues)


http://www.fdlp.gov/component/docman/cat_view/178-collection-management/249-purl-referrals

http://www.fdlp.gov/component/docman/cat_view/178-collection-management/249-purl-referrals

http://fdlp.gov/collections/building-collections/618-purl-referrals-reporting

Statistics we don’t know

• Visits to online docs URLs by our users – we are clueless!

• How many times URLs are visited by our users

• What titles are visited by our users• What agencies are most popular with our

users• We don’t know the whole picture


Objective

• To track online government document clickthroughs when accessed via the online catalog

oNot possible to capture every use of government info by our usersoBut is possible to capture all clickthroughs via the OPAC


Why we need URL statistics

• Justify our depository status to administrators

• Assist with item selections• GPO cannot provide them• URL maintenance• “Knowing where they’re going” is always

helpful


Why Statistics are Difficult to Gather

• Not all government URLs are PURLed• In 2004 I counted over 1,400 servers

hosting government documents to which our catalog pointed. We can’t expect 1,400 sites to provide us statistics.


URLs at University of Denver

• Nearly 600,000 documents URLs in our catalog

• Many of these not in CGP• “Aggressively” added URLs


URL Addition Projects at DU


How the Clickthrough Works

• URL click sends user to a database• Database records URL, date, time• Database instantly refers user to URL


The URL Prefix is Appended before the URL/PURL


Stats are Logged, and User is Redirected to Desired URL


An Access Database is Used to Manage the Project Stats


Click-through to Catalog

URL Testing


Benefits of Clickthrough Project1. We can provide meaningful stats to

the library director2. We can see high-use and low-use

areas3. We can tell if users benefit from our

special projects4. We can do reactive URL

maintenance5. We can see turnaways and other

problem


1. Providing Meaningful Stats



• Older Docs Content Gets Visits

FY04 FY05 FY06 FY07 FY08 FY09 FY10

Total Clicks 3809 4504 4686 5217 6342 7660 7921

Up to 10 years 3542 4155 4170 4369 4996 5600 5871

percent 93.0% 92.3% 89.0% 83.7% 78.8% 73.1% 74.1%

Over 10 years 267 349 516 848 1346 2060 2050

percent 7.0% 7.7% 11.0% 16.3% 21.2% 26.9% 25.9%



Comparison of Online Access with Physical Circulation of Documents


2. High-Use Areas by Agency


2. High-Use Areas by SuDocs


Clickthroughs in Relation to Number of Records

Fiscal Year Total Docs Bib Recs Bib Recs with URLs Clickthroughs to Docs

FY2004 358,215 43,307 3,809

FY2005 373,200 55,508 4,504

FY2006 388,610 62,374 4,686

FY2007 401,454 103,021 5,217

FY2008 429,122 159,543 6,342

FY2009 711,315 463,121 7,660

FY2010 860,346 594,431 7,921


3. Special Project Usage

Project URL Count Coverage Dates

Tracking Time Span URL Clicks

Unique URL Clicks

% Unique

AccessedTopographic Maps 456 1991 – 2001 Sept. 2003 – June 2009 101 76 16.6%NASA Technical Reports

24,825 1976 – 2001 April 2007 – June 2009 310 263 1.06%

GAO Reports (older) 9,559 1976 – 1999 Aug. 2007 – June 2009 184 161 1.68%LexisNexis Digital Hearings/Committee Prints

57,200 1850 – 1995 July 2007 – June 2009 1027 851 1.49%

Readex Digital Serial Set

248,134 1817 – 1948 Sept. 2008 – June 2009 239 205 0.08%

OSTI Reports 19,901 2002 – 2006 July 2008 – June 2009 476 375 1.88%


4. URL Maintenance

• Proactive approach– Works well with fewer URLs– Is a nice luxury, if time allows

• Reactive approach– Fix broken links, since you know when they

are used– Don’t worry about broken links that nobody

clicks


4. Reactive URL Maintenance

• Two approaches: Proactive approach• My approach: Reactive approach – with

nearly half-a-million docs URLs in our OPAC, we can’t afford to be proactive.

FY Clicks Errors Error Rate

FY04 3809 202 5.30%

FY05 4504 231 5.13%

FY06 4686 299 6.38%

FY07 5217 217 4.16%

FY08 6342 179 2.82%

FY09 7660 177 2.31%

FY10 7921 181 2.29%


It is important to report broken PURLs to GPO. They are repaired very quickly.


5. Turnaway Problems: The Dark Days of 2009

From August 24 through September 7, 2009 the PURL server was down. We were able to track each of the turnaways from government information and amend our records as needed.

6. Search Engine Attacks• CUIL (http://www.cuil.com/) CUIL attacked many OPACs – at least Millennium OPACs. We were

attacked two times. Our project uncovered the attacks!• August, 2007 and February, 2008• The CUIL clickthroughs were subsequently omitted from the project stats


http://www.cuil.com/

Clickthrough Methods Compared

Library Prepend Pattern

Auburn U. http://lib.auburn.edu/AuCataloging/en/clickCount/countAndRedirect.jsf?bibId=1536469&url=http://purl.access.gpo.gov/GPO/LPS29932 [try it]

San Jose State U.

http://www.sjlibrary.org/scripts/stat_govpub.htm?id=17434373&path=http://purl.access.gpo.gov/GPO/LPS29932[try it]

LUIS (Louisiana) http://utils.louislibraries.org/cgi-bin/lz0050.x?sitecode=LALU?http://purl.access.gpo.gov/GPO/LPS29932[try it]

U. of Denver (old ColdFusion)

http://lib-anubis.cair.du.edu/findit/peak/redirect.cfm?LinkURL=http://purl.access.gpo.gov/GPO/LPS29932[try it]

U. of Denver(present PHP)

http://library.du.edu/clickthrough/index.php/clicks/?type=gov&url=http://purl.access.gpo.gov/GPO/LPS29932[try it]

Mercer U. http://tarver.mercer.edu/inc/opachits/getcha.php?type_stat=gov&URL=http://purl.access.gpo.gov/GPO/LPS29932[try it]

Any scripting language can be used to accomplish this project!


http://lib.auburn.edu/AuCataloging/en/clickCount/countAndRedirect.jsf?bibId=1536469&url=http://purl.access.gpo.gov/GPO/LPS29932

Writing Your Specs

• Be nice to your systems person.• Let the systems person decide which

scripting code to use.• Here are the specs I am using for version

2.0 of our specs.


Specs for the New DU Clickthrough System

• Project hosted on stable server (such as library Web server).• Should be able to handle long URLs – up to 700 characters.• Prepended URL sends request to library server.• Included in prepended URL is cataloger-supplied 3-letter code

of URL type (ex: gov, cou, ran – any 3-letter combination that may be needed in future).

• Server records date/time, IP address of requestor, 3-letter code of URL type, and URL requested.

• Server redirects user to desired URL.• Reporting mechanism available to gather clickthroughs.• Archiving function available to archive stats.• Ability to view archived records.• Secure login for authorized users.


For More Information:Brown, Christopher C. “Statistics for Online Document Use.” Presentation given at the Fall 2003 Depository Library Conference, 22 October 2003, Arlington, VA. Published in the Proceedings of the 12th Annual Depository Library Conference, Oct. 19-22, 2003.

Brown, Christopher C. 2004. “Knowing Where They're Going: Statistics for Online Government Document Access through the OPAC”. Online Information Review 28 (6), 396-409. DOI: 10.1108/14684520410570526

Brown, Christopher C. and Susan Xue. “Local Access Statistics for Federal Documents: Tracking Web Page and Online Catalog Usage.” Presentation given at the Fall 2004 Depository Library Conference, 20 October 2004, Washington, DC. Published in the Proceedings of the 13th Annual Depository Library Conference, Oct. 17-20, 2004. [view]

Brown, Christopher C., Stephanie Braunstein, Susan Kendall, Liza Weisbrod, Jennifer Gerke, and Shane Cole. “Tracking Online Document Usage from the Catalog: Experiences from the Field.” Presentation given at the Fall 2009 Depository Library Conference, 19 October 2009, Arlington, VA [view].

Brown, Christopher C. 2011. “Knowing Where They Went: Six Years of Online Statistics via the OPAC for Federal Government Information.” College & Research Libraries 72 (1), forthcoming 2011. Preprint:http://www.ala.org/ala/mgrps/divs/acrl/publications/crljournal/preprints/crl-068.pdf.

http://sites.google.com/site/librariancorner/url-clickthrough-project
