Page MenuHomePhabricator

Multilingual, Monolingual or Cross language Search
Closed, DuplicatePublic

Description

Discuss T92929#1254424


In Cross language search, a user mainly searches only in English and user story:

  • As a Chinese translator I search for home in English, and choose Chinese from languages filter to view translated messages in Chinese.
  • I can get Spanish translations by selecting Spanish from languages filter.


In Multilingual search, a user can search in any language and user story:

  • As a multilingual translator I search for casa (i.e German), and choose Chinese from languages filter to view translated messages in Chinese.
  • I can get Spanish translations of casa by selecting Spanish from languages filter.

The implementation of retrieval engine may be same for the both modes of search (as mentioned in the diagrams) but we might have to use language analyzers for stemming, etc. if we decide to implement multilingual search.
Depending on the above diagrams, the User Interface will be the same.

@Nikerabbit Am I right? or Did I misunderstood?

Event Timeline

Phoenix303 updated the task description. (Show Details)
Phoenix303 raised the priority of this task from to Normal.
Phoenix303 added a project: Translation-search.
Phoenix303 added subscribers: gerritbot, Amire80, siebrand and 6 others.
Restricted Application added a subscriber: Aklapper. · View Herald TranscriptMay 9 2015, 5:59 AM
Nikerabbit updated the task description. (Show Details)May 25 2015, 11:08 AM
Nikerabbit set Security to None.

we might have to use language analyzers for stemming, etc. if we decide to implement multilingual search

As far as I know, ElasticSearch knows in what language each document is and uses the appropriate analyser in consequence.

Per Nikerabbit's comment, the "retrieval engine" is ElasticSearch; the "fetch" arrow is the process of retrieving messages in the chosen language for the message keys returned by ES. The monolingual search (current only option) doesn't have this second step. So, yes, the initial search (in ES) should be allowed in any language.