NTL > TLR Archive > Index Page


A Tale of Two Thesauri: UMTRI's Move to Adopt the TRT

Bob Sweet, University of Michigan Transportation Research Institute Library, 13 May 2010

Slide 2:  A Tale of Two Thesauri: UMTRI’s
            Move to Adopt the Transportation
            Research Thesaurus

They were the best of terms. They were the worst of terms.

Bob Sweet
Information Resources Manager
bsweet@umich.edu

Slide 3: The Problem

  • UMTRI Thesaurus out of date

  • Short of staff for updating

  • TKN era calls for collaboration, networking, uniformity, and a “common language”

“One of the challenges to finding and managing transportation information is the
limited use of common metadata and indexing terminology within the
transportation community.” --NCHRP Project 20-90 problem statement

Slide 4:  The Solution

  • Adopt the TRT
  • TRT is maintained by a team of dedicated professionals
  • TRT is increasingly accepted as a standard

Slide 5:  Some History

  • UMTRI began as HSRI with auto industry money

  • In 1967, very few, if any, libraries devoted to highway safety

  • Unique, multidisciplinary, “missionbased” collection required “unique classification and indexing technique”

Slide 6: Some History

  • High value placed on the information organization and dissemination mission
  • Who else but an electrical engineer to develop a thesaurus?

Slide 7:  HSR IC Library/Information Center

Image : February 1969

Slide 8:  HIGHWAY SAFETY RESEARCH INFORMATION CENTER  SUBJECT INDEX FACETS

Image : Contents

Slide 9:  The Highway Safety Research Information Center

Image

Slide 10:  Image

Slide 11:  1967

Image

Slide 12:  2004

Image

Slide 13:  An Eye-Opener

  •  Adaptive cruise control rocked our world
  •  Speed Control Devices/ Cruise Control falls under Base Engine Components

Slide 14:  UMTRI Thesaurus

  •  Best of terms
    • When it was first created and for most of our history there was no substitute.
    • Highly granular
  •  Worst of terms
    Cultivated Terrain, Prefabrication, Brake Tramp

Slide 15:  TRT

  • Best of terms
    • Maintained by information professionals
      • More widely accepted
  •  Worst of terms
    • Not granular enough
    • No Rearward Amplification, Restraint System Injuries, Fatality Patterns, Crash Speed

Slide 16:  So How Do We Do It?

  • Do we just stop using UMTRI terms and start using TRT terms?
    • All items cataloged after this date use TRT?
  •  How will we find earlier materials indexed with earlier terms?
    • Airbag restraints (UMTRI)    Air bags (TRT)
  •  How do we educate users?
    • Signs at terminals? Notes on search screen?
  •  Do we even need subject analysis?

Slide  17:  Who Ya’ Gonna Call?

  • Michael Kleiber, Sandy Tucker, Roberto Sarmiento, Paul Burley, Rita Evans
  •  Michael took us into the deep end. Map all terms.
  •  3,000 not 1,500

Slide 18:  Let the Mapping Begin

 Sometimes near-equivalent (close enough) terms, but not within a logically corresponding place in the hierarchy.

Billboards (UMTRI); top level is Roads/ Ways

becomes

Roadside advertising (TRT); top level is Management and organization

 If a term was rarely used (yielded few hits in a search), it was easier to go with a close-enough term

Face Protectors (UMTRI)         Protective clothing (TRT)

Slide 19:  Mapping, con’t.

In many cases, where UMTRI terms were much more specific, we made the decision to go up in the UMTRI hierarchy to find a less specific term for which there was a good TRT match.

Tire Cornering Power (UMTRI)             Tire mechanics (TRT)

NVM                Tire Mechanics
NVMBB           Tire Longitudinal Force
NVMBBB         Braking Force/ Tire Tractive Force
NVMBCB         Tire Cornering Power

Slide 20:  Mapping, con’t.

Sometimes, going to a higher level in the hierarchy would mean duplication of terms.

Phototropic glass (UMTRI)                     Glass (TRT)

BK                Glass
BKG             Glazing             (use Glass)
BKGP           Phototropic glass         (use Glass)

Slide 21:  Mapping, con’t.

  • Sometimes equivalents couldn‟t be found because we didn't understand the UMTRI terms. We were thrilled when term definitions appeared in TRT, but UMTRI terms never were defined.
  • Standing Wave Phenomenon (UMTRI) *

Tire mechanics (TRT)

NVM      Tire Mechanics

NVMI     Tire Dynamics

NVMIS Standing Wave Phenomenon

*Fifteen hits doesn’t earn a trip to Winkler.

Slide 22:  Mapping, cont.

  • Some terms were a little too specific, or vague, as evidenced by there being no hits for those terms in the catalog:

Shift Buttons, Cultivated Terrain, Engine Flooding, Magnetic Cores, Prefabrication

Slide 24:  Mapping, cont.

  • For many UMTRI terms we had to come up with postcoordinate TRT terms to capture the complexity that had been represented by single UMTRI terms or to account for the puzzling tendency in the UMTRI Thesaurus to compound terms and separate them with a slash.

Fibers/ Textiles (UMTRI)

becomes

Fibers (TRT)
Textiles (TRT)

Transmission Fluid (UMTRI)

becomes
Transmissions(TRT)
Fluids(TRT)

Slide 25:  The Programming Challenge

  • In October, our friends in the U-M Library systems office gave us their typical, “Sure, we can do that.”
  •  When we went back to them in March (Feb?), we had a new twist (or two).
  • Paul Burley was adamant that the value of the UMTRI terms ought not be lost and so suggested that they all remain. It made perfectly good sense to leave them in there, whether they remained in the record as subject headings or simply offered keyword access.
  • They systems folks also agreed to our other twist, which was to add MESH terms for those anatomical and biomedical terms that didn't have an equivalent in TRT.

Slide 26:  The Programming Challenge

  • Margaret Kelly, systems librarian, and Excel to the rescue
  • Our spreadsheet included UMTRI terms (with hierarchical codes) and their corresponding TRT terms (with hierarchical codes); it also showed MESH terms and a list of UMTRI terms for which no satisfactory TRT or MESH equivalent could be found.

Slide 27:  Image


Slide 28:  Programming, cont.

  • 2 tab-delimited text files, 1 for TRT terms and one for MESH terms
  • Extracted copies of all 490,010 UMTRI subject terms from the library's Aleph MARC database.
  • Used Perl to compare each „umitrist‟ subject term with the source list, and where there were matches wrote the substitute or substitutes to a new file, constructing them in Aleph MARC format for reloading.
  • Did rigorous „stress testing‟ of the matching protocols, so that „Pain',  for example, did not match „'Paint.'
  • In Excel, combined into one file the untouched copies of the 490,010 UMTRI terms and the new transformed terms produced by Perl program.
  • Using Excel, removed duplicates, keeping one copy of the TRT terms and discarding the „umitrist‟ or TRT duplicates.

Slide 29:  Programming, cont.

  • Repeated steps with MESH terms.
  • Resulted in a file of 835,114 new and preexisting terms, including the new TRT and MESH terms, and the untouched umitrist, Library of Congress, or other subjects, all written in the correct loading format.
  • Copied it back to a text file and used one of the Aleph built-in programs to overlay the block of subject terms in each record with the block of processed terms from the file.

Slide 30:  Dénouement

  • Now nearly twice as many terms providing subject access for items already cataloged. Redundant? Perhaps.
  • Moving ahead with TRT only.
  • Still in development; TRT terms not in production catalog.
  • No more searching with hierarchy codes.
  • Submitted list of 252 UMTRI terms to members of the TRT subcommittee on February 22 for consideration as TRT terms.