search and rescue
DESCRIPTION
Eighth Annual Reference Research Forum, American Library Association Annual Conference, Atlanta, June 16, 2002. Search and Rescue. Repair Strategies of Remote Users Searching the Online Catalog. Nancy B. Turner, Electronic Resources Librarian, Syracuse University - PowerPoint PPT PresentationTRANSCRIPT
Search and Rescue
Repair Strategies of Remote Users Searching the Online Catalog
Nancy B. Turner, Electronic Resources Librarian, Syracuse University Susan E. Beck, Head of Reference & Research Services, New Mexico State University
Eighth Annual Reference Research Forum, American Library Association Annual Conference,
Atlanta, June 16, 2002
Research Questions
Do users in remote locations repair searches?
What types of repair strategies do they employ?
How can this information inform our work as reference librarians?
Transaction Logs
Traditional OPAC LogsWeb Server LogsAppropriate UsesDrawbacks
User Population
Carnegie I—Research Extensive/ Hispanic Serving
15,000 enrollmentDoctoral degrees in
5 colleges: Agriculture, Arts & Sciences, Business, Education, Engineering
Physical Environment
Two library buildings: Branson & Zuhl
Two periodicals areasTwo reference desk areasTwo stack areas
Endeavor catalog
Search Screen
Methodology: Data collection
Apache server log extraction 7 two hour intervals over 3 month period Various times: am, pm, weekends
Created Library IP address map Charted search locations by IP address:
Campus (NMSU domain) Off campus (non-NMSU domain) Staff Zuhl & Branson public service areas
Methodology: Data treatment I
Text files imported to spreadsheetImported three columns only
IP address, Date/Time, Search QueryMapped IP’s to library locationsSelected only unmediated, remote
locations to reviewDivided spreadsheet into 3
worksheets All, Unmediated, Staff
Methodology: Data Treatment II
Search_Code = TALL replaced with Search=TitleSearch_Code = JALL replaced with Search=Journal TitleSearch_Code = FT*& replaced with Search=Keyword Search_Code=NAME replaced with Search=AuthorSearch_Code=SUBJECT replaced with Search=SubjectSearch_Code=CALL replaced with Search=Call Number
%2B = replaced with + %3F = replaced with ?%3A = replaced with : %2C= replaced with , %27 = replaced with ‘ %22 = replaced with “
Cgi search syntax translation
1. GETSearch_Arg=tankers+full+of+trouble&Search_ Code=TALL&PID=19913&SEQ=20020314130448&CNT=25&HIST=1&SEARCH_FROM_TITLES_PAGE
2. GETSearch_Arg=tankers full of trouble &Search=TITLE &PID=19913&SEQ= 20020314130448&CNT=25&HIST=1&SEARCH_FROM_ TITLES_PAGE
User searching for the title, “Tankers full of trouble” from the Titles display listing page
Search query translation
Methodololgy: Data Treatment III
Selected “searches” based on
syntax GETSearch_Arg=
Coded searches by search type JT, T, S, A, CN, B, K, C
Tallied searches by type & area
March 14 Tallies
Bran Per 0 0 0 0 0 0 0 0 0 0Bran Ref 269 88 10 21 3 21 6 27 0 0Bran Stacks 53 28 0 4 6 4 13 0 1 0Campus 421 177 28 31 37 42 25 14 0 0Off Campus 111 30 5 7 2 7 9 0 0 0Zuhl Per 19 9 0 2 1 5 1 0 0 0Zuhl Ref 259 93 6 14 4 25 38 6 0 0Zuhl Stacks 122 41 1 0 0 26 11 3 0 0Total: 1254 466 50 79 53 130 103 50 1 0All Users: 2353Unmediated Inputs % 53
GETSearch_Arg=new mexico litter controll&Search=Keyword
GETv1=1&ti=1,1&Search_Arg=new mexico litter controll&Search=Keyword
GETti=21,0&Search_Arg=new mexico litter controll&Search=Keyword
GETPAGE=sbSearch
GETSearch_Arg=new mexico litter&Search=Keyword
GETti=11,0&Search_Arg=new mexico litter&Search=Keyword
GETti=21,0&Search_Arg=new mexico litter&Search=Keyword
GETti=31,0&Search_Arg=new mexico litter&Search=Keyword
GETSearch_Arg=texas litter&Search=Keyword
GETSearch_Arg=texas litter laws&Search=Keyword
GETSearch_Arg= litter laws&Search=Keyword
Excerpt from search log
Data Analysis
After coding all discrete searches, we then re-coded those searches based on repair types.
Search Repair Levels
WordRepairs occur at word level
ConceptRepairs involve the concept(s)
searched. A concept is typically multi-word or a phrase. It can be multi-concept
SearchRepairs show knowledge of search types
or knowledge of formatting within search types
Word Level Repairs
Spelling (WS) multi-meter multimeter
18th eighteenth
Plural/Singular (WP) duns dun groups group
Capitalization (WC) Doll doll
Word Level Example
GETSearch_Arg=bertrand russel& Search=Keyword
GETSearch_Arg=bertrand russell& Search=Keyword
GET Search_Arg=raceism in unimployment &Search=Keyword &
GET Search_Arg=racism in imployment &Search=Keyword
Concept Level RepairsPunctuation: Add/drop punctuation (CP)
city of ladies “city of ladies”
christine pisan pisan, christine
Broaden: Drop concept/word (CB)
How to use a digital multimeters digital multimeters
Pregnancy—week by week Pregnancy
Change to another concept (CC)
English Men & Manners in the Eighteenth Century 18th Century British Women
Decline and Fall of Rome Edward Gibbon
Narrow: Add concept/word (CN)
“puerto rico” “puerto rico” “English”
pizan pizan city of ladies
Rephrase (CR) English Society in the 18th Century Daily Life in Eighteenth Century England
Ireland travel Ireland guidebook
Concept Level Example
GETSearch_Arg=Patton&Search=Keyword GETSearch_Arg=General Georgo Patton &Search
=Keyword (CN)GETSearch_Arg=Leadership&Search=Subject (CC)GETSearch_Arg=Military Leadership& Search =Subject
(CN, STGETSearch_Arg=General Patton leadership&
Search=Subject (CN)GETSearch_Arg=George Patton leadership&
Search=Subject (CR)GETSearch_Arg=Leadership style of George Patton&
Search=Subject (CR, CN)
Search Level Repairs
Invert words/names (SI)
Edward Gibbon Gibbon, Edward
Add/drop article (SA) A Doll House Doll House
Switch search type (ST)
A Doll House (Subject) A Doll House (Title)
tf 140 s73 (Keyword) tf 140 s73 (Call Number)
Limit search (SL) Home care home care (limited to specific library location or format or language, etc.)
Search Level Example I
GET Search_Arg=child abuse laws&Search=Keyword GET Search_Arg=child abuse laws& Search=Subject
GET Search_Arg=Troy Boone&Search=AuthorGET Search_Arg=Boone, Troy&Search=Author
GET Search_Arg=A Doll House&Search=TitleGET Search_Arg=Doll House&Search=Title
Search Level Example II
GET Search_Arg=Babylonian Mathematics and the Plimpton 322&Search= Keyword
GET Search_Arg=Babylonian Mathematics and the Plimpton 322&Search= Subject
GET Search_Arg=Babylonian Mathematics and the Plimpton 322&Search= Title
…2 minutes pass…
GET Search_Arg=History of Babylonian Mathematics&Search=Title GET Search_Arg=History of Babylonian Mathematics&Search=Journal Title GET Search_Arg=History of Babylonian Mathematics&Search=Subject GET Search_Arg=History of Babylonian Mathematics& Search= Keyword
Preliminary findings: Search types
Although no specific search type emerges as an obvious favorite
Keyword 28% Jrnl Title 15%Subject 19% Author 10%Title 16% Boolean 9%
Call # 3%
44% of all searches are for known items (title, journal title, author & call number)
Preliminary findings: Word level repairs
Repair Code Repair Type TOTAL %
WS Spelling 81 8%WP Plural/Singular 13 1%WC Capitalization 21 2%Total Word Level Repairs 115 12%
Preliminary findings: Concept level repairs
Repair Code Repair Type TOTAL %
CPPunctuation: Add/drop punctuation 30 3%
CB Change to broader concept 116 12%CC Change to another concept 89 9%CN Narrow: Add concept/word 132 14%CR Rephrase 136 14%Total Concept Level Repairs 503 52%
Preliminary findings: Search level repairs
Repair Code Repair Type TOTAL %
SI Invert words/names 22 2%SA Add/drop article 7 1%ST Switch search type 270 28%SL Limit search 17 2%SB Boolean Syntax Change 26 3%Total Search Level Repairs 342 36%
Preliminary findings: Overall repairs
Repair Code Repair Type TOTAL
Total Repairs 960 87%R Repeat Searches 145 13%
Total Redone/Repaired 1105
Total Searches 2132
% Searches Repaired 45%% Searches Redone 52%
Preliminary findings
Over 50% catalog uses are non-staff Unassisted user searches.…………….56% Staff/Mediated user searches………..44%
Higher incidence of concept level repairs than word and search level
Preferred repair strategy is to switch search types
High incidence of retyping exact search
Further study
Extend data collection time periods Increase # of data collection
sessionsInclude different user population Extend analysis to include
Length of time spent repairing search Search types leading to more repairs Number of repairs made per search
Questions?
For more information, contact:
Nancy B. Turner at [email protected] E. Beck at [email protected]