Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Page Properties

Status

Status
colourYellow
titleIN progress

Description

Alma as currently configured does not map equivalent characters in CJK languages. A special setting and re-index of the database may address this.

Decision summary

Owning group

RM-CMOS

Approver

Operations ?Team

Consulted

Campuses, CJK KGCampus cataloging staff, East Asian Studies CKG, OT, Discovery OST

Informed

sils-cohort

Decision-making process

Priority

High

Target decision date

Date decided

...

Stakeholder group

Impact

Reasoning

Background

By default, Alma does not have the ability to automatically recognize when two versions of a word mean the same thing. For example, 群島 (traditional Chinese) and 群岛 (simplified Chinese) should retrieve the same record. However, they don't.

See https://uc-sils.atlassian.net/l/cp/0PP83b1Z for additional background on the original issue and tickets.

Updates:

Ex Libris enabled the option to Sandbox Test

After consultation with Ex Libris, Ex Libris suggested we enable the option to Search in Hangul (Korean) and return results in both Hangul and Hanjain the sandbox, which they indicated will . They indicated this would also help with the Japanese Kanji and Chinese traditional/simplified issues as well. However, it will not fix similar problems with other languages.

The change was made in UCLA’s and UCSD’s premium sandbox (PSB), as well as the NZ PSB in September 2022. Indexing the sandbox took an extraordinarily long time. NZ indexing started on 8/30/2022 and finished on 9/18/2022. During this time and also broke some other functionality. ExL has indicated that this would not happen in production. period, we were unable to retrieve the majority of records within the system (bib, user, order, etc). After the initial reindexing finished, users and orders still were not searchable and required additional reindexing jobs.

Ex Libris is asserting that all our issues with reindexing in the sandbox are because full reindexing jobs should not be done in the sandbox (only smaller sets). They also note that semiannual indexing has already happened twice in production and we did not encounter any issues either time. However, they have not yet explained why users and orders needed additional indexing after the first indexing had finished. In our 10/17/222 support meeting, Ex Libris support said they’d try to connect us with a developer for more information.

Test Results

Campuses have tested the indexing for Chinese and Japanese and are pleased with the results. Korean testing is still pending. Testing spreadsheet available in GDrive.This does not generally affect searching in Primo. That is a separate issue. Read-only copy of testing spreadsheet

Primo VE Note

Similar issues with Primo CJK indexing are outside the direct scope of this decision.

Options Considered

Option 1

Option 2

Questions to consider

  1. Why did user/order searching not work after the initial sandbox reindexing had finished? This doesn’t seem like it would be related to an under allocation of resources.

Action Log

Action/Point Person

Expected Completion Date

Notes

Status