Elasticsearch Support Based on PeopleSoft Application Version

In order to utilize Elasticsearch for your PeopleSoft application, you will need to apply updates and fixes from your application’s most current update image.

See PeopleSoft Applications: Elasticsearch Updates and Fixes on My Oracle Support (Doc ID: 2181988.1) for more information on which update images have the required fixes for your application.

This table lists the key functionality of Elasticsearch from the perspective of the use and implementation within a PeopleSoft application system.

Item

Elasticsearch Functionality

High availability

The recommended approach to high availability is by deploying an Elasticsearch cluster with at least 3 nodes on machines that are not on the same physical server and by following the configuration guidelines.

Fail-over mechanism

Elasticsearch search engine is distributed by nature, that is, it knows how to manage multiple nodes. A cluster can contain one or more nodes and one of the nodes act as a master node. Elasticsearch automatically distributes the indexed data to the nodes based on its constructs of primary shards and replica shards, so indexed data is available even when a node fails.

Phonetic

Elasticsearch supports phonetic search using a plug-in that is installed when you install Elasticsearch in a PeopleSoft implementation.

Note: For custom search pages, PeopleSoft Applications can use PeopleSoft APIs for phonetic queries in Elasticsearch.

Wild card search

Elasticsearch supports using a wild card as a leading character in a search text. For example, *racle.

Stop Word

Stop Words are words that are commonly used (such as articles, prepositions) and do not contain important significance to be used in a search text.

If a word is in the Elasticsearch’s stop word list, it would be ignored.

Attachment processing

In Elasticsearch, PeopleSoft Search Framework downloads any attachment specified in a search definition, and directly pushes the attachment to the Elasticsearch search engine using the libcurl library bypassing Integration Gateway. Elasticsearch uses the ingest- attachment plug-in to extract the attachment contents and indexes the attachment data. This avoids multiple log-in attempts during crawling.

Stemming

Stemming is the process of reducing inflected word to its root or stem. For example, talking, talked, talks will reduce to the root — talk.

Elasticsearch supports limited stemming support, for example, talk/talked/talks. Synonyms are supported by using a file to store the required synonyms as per your business needs.

Elasticsearch supports stemming search for the following languages:

  • Arabic

  • Czech

  • Danish

  • Dutch

  • English

  • Finnish

  • French

  • German

  • Greek

  • Hungarian

  • Italian

  • Japanese

  • Korean

  • Norwegian

  • Portuguese

  • Romanian

  • Russian

  • Simplified Chinese

  • Spanish

  • Swedish

  • Thai

  • Turkish

  • Traditional Chinese

Relevancy Scoring

Relevancy scoring is performed by Elasticsearch, which uses an internal scoring algorithm. For more information, refer to Elasticsearch Reference [7.10].

Custom Scoring

In Elasticsearch, custom scoring is supported through Search Framework APIs.

Synonyms

Elasticsearch uses a file (synonyms_grammar.txt) to store synonyms or any abbreviations that you want to include. If synonyms provided are not adequate for your installation, you can modify this file. The file is located at config\analysis\synonym_grammar.txt.

You can enter synonyms in two ways:

"i-pod, i pod => ipod"

OR

"universe, cosmos"