SPEC Kit 341: Digital Collections Assessment and Outreach · 125
INDIANA UNIVERSITY BLOOMINGTON
IN Harmony Query Logs Analysis Study: Final Report
http://webapp1.dlib.indiana.edu/inharmony/projectinfo/userStudies/finalReport_
INHarmonyLogsAnalysis.pdf
-3 -
I. Executive Summary
The IN Harmony Sheet Music Query Logs Analysis study was designed to understand how users search
for sheet music in order to inform the:
selection of a metadata schema and associated vocabularies for sheet music description
design of a sheet music cataloging tool
design of end-user browse and search interfaces
This analysis is the first of a two-phase usability study. Remaining and emerging questions will be
addressed in a second study, which will be conducted with representative users during the months of June
and July 2005.
Six months of data, June-November 2004, was collected from the IU Sheet Music and UCLA Sheet
Music Consortium websites. A ten percent random sample of 2,542 log entries (IU, n=1,116 and UCLA,
n=1,426) were analyzed to determine the number of:
Browse, search and advanced searches
User-specified keyword, subject, names/composer, etc. queries
o Mappings of keyword searches to specific fields
o Mappings of subject searches to subject sub-categories (topical, form, genre, style,
temporal, geographic, etc.)
o Known-item versus unknown-item queries
o Year search filters used
Search inputs were also analyzed in terms of kinds of content (e.g. lyrics), syntax/operators used, and use
of advanced search fields.
The data was processed, charted and graphed using Mircrosoft’s Excel spreadsheet program. Keyword
and user-specified subject searches were further analyzed with Jenn Riley’s, IU Metadata Librarian and
music domain expert, help.
Name and title keyword searches for sheet music were the most popular 37% keyword name and 29%
keyword title for IU Sheet Music and 27% keyword name and 52% keyword title for UCLA Sheet Music.
The most common user-specified search fields are also name and title 23% name and 19% title for IU
and 28% name and 12% title for UCLA. Subject-specified searches are few (IU and UCLA, 3%)
however, a significant number of subject keyword searches were conducted (19%, n=1,695). The most
common year filter used on both websites was for pre-1923 records. A more detailed discussion of the
findings can be found in section VI and VII of this report.
Key findings and recommendations are:
Cataloging Tool
Tendency towards name searching should require name authority control and integration of the
cross-referencing structures in order for users to more reliably find name variants
Preference toward keyword searching suggests the need for rich, descriptive records including the
recording of lyrics, chorus lines, and other metadata that may enrich the record to increase recall
when free-text searching
Common subject searches such as genre/form/style, instrumentation and topic should require the
use of controlled vocabularies suitable for describing these specific subject areas
Previous Page Next Page