Skip to content

Lucune Group Meeting 2022 10 12

Jump to bottom
Dulip Withanage edited this page Feb 10, 2023 · 5 revisions

2022-10-12 Lucene

Present: Dulip, Antti-Jussi, Martin Braendle, Maja Ronne, Leonhard Maylein, Benedikt Kroll, Nils Weiher, Richard Higgins, Marcos (?), Marc Bria

Introductions

  • Solr/Lucene in OJS, but also external search tools in PKP software
  • Dulip
    • TIB (Hannover) but also PKP. Mostly interested in multilingual search in OJS, OMP, OPS.
  • Antti-Jussi
    • Antti-Jussi Nyborg, federation of Finnish societies. Maintains journal.fi. Also works with PKP part-time.
    • Finnish installation is too large for built-in tools.
    • Interested
  • Alec
    • Lucene plugin hasn’t received a lot of attention; features haven’t been maintained well and aren’t well understood.
    • Containers – may see more uptake
  • Benedikt
    • Munich
    • Trying to integrate OJS into other search tools/infrastructure that already uses Lucene/Solr
    • Also interested in multilingual support
  • Leo
    • Heidelberg
    • At the moment, use an external Solr index to search all OA platforms (OJS, OMP, also others: digitization workflow)
    • Not sure if we’ll use the plugin, but there may be some journals that will.
    • Could also contribute experience from our Solr index
  • Marc Bria
    • Autonomous University of Barcelona
    • Interested in making search work in multiple single-tenant installations (of OJS) – sort of aggregation by search.
  • Marcos
    • Used to work for X Univeristy Press; now freelancer in Spain for various institutions using OJS.
    • Some institutions with big installations / multiple installations; performance and aggregating content
  • Martin
    • Working for 2 institutions: University of Zurich (responsible for publication repository and OJS, along with team); also Swiss chemical society (several OJS instances there). More content coming online soon.
    • Used Lucene/Solr in OJS 2.4, mostly b/c of multilingual content. One journal uses 7 languages. Did some testing for 3.1 or 3.2, when Lucene plugin was first redeveloped (?).
    • Still waiting to properly provision a server with enough RAM.
    • Also have a lot of experience with ElasticSearch.
  • Nils Weiher
    • University of Heidelberg on OJS instances. Colleague working on search isn’t available, so covering for them.
    • Schema mapping from OJS to Solr is a problem; when sharing with several applications you need a common schema. ▪ Richard: Possibly approached (schema mapping) by Drupal plugin
  • Richard
    • 2 installations of OJS, one large, one small; would like to bring content into a search engine.
    • Trouble getting OJS 3.2 working with plugin.
    • Have been working with OJS since ~2015.
    • Wanting to integrate search.
  • Maja
    • Working for Royal Danish Library. Installation for ~200 journals.
    • Is a programmer.
    • Just upgraded to version 3.3.
    • Running on PostgreSQL database.

Presentation: Lucene plugin

  • Mattermost channel access seems interesting to the group

  • Call for the group to participate in testing

Question: Priority. What to prioritize?

  • Richard
    • Can’t get it to connect with remote Solr core.
    • Need to upgrade to 3.3.
    • Directions/documentation were about Solr; plugin itself needs to be documented.
  • Martin
    • In our case, ElasticSearch is provided as an institutional service.
  • Antti-Jussi
    • Had to install Lucene plugin during a crash situation (upgrading to 3.2 led to search overwhelming the feature).
    • Had to comment out some Lucene plugin features that were causing problems. (Expedient.)
    • Just want “good results” matching data – trim down extra features.
    • In the process of upgrading to 3.3; will report back.

Areas of work

  • Removing extra features (based on polls)
  • Fix potential ordering issue of results (Antti-Jussi – based on sorting)
  • Just release the plugin for 3.3.
  • Increasing transparency (logging; interface clarity) and configurability (matching expectations e.g. compared to Drupal integration)
  • Clarify documentation
    • Move into docs hub as a plugin guide
    • Deployments
      • Some people will have an institutional Lucene, will not need to learn it, will have expertise
      • Some people will know OJS and will not know anything about Lucene.

Alec’s Notes

  • Richard and Maja use PostgreSQL and report struggles. Include them in community testing.
  • Alec to confirm that 3.3 plugin works better than 3.2 to Dulip
  • Lucene plugin should be simplified.
  • Dulip to get a Mattermost channel going and send invites (probably via Michael)

PKP PJS Lucene Plugin

Clone this wiki locally