Forums

Articles
Create
cancel
Showing results for 
Search instead for 
Did you mean: 

Similar word forms (lexem) are not searched for non english words

Arthur Foss January 8, 2019

In some languages, the word can be written in different ways depending on the sentence. For example, this is word "instruction" in Russian - "инструкция".

But it can also be written as "инструкциИ/инструкциЕЙ/инструкциЮ/инструкциЯ/инструкциЯХ/инструкциЙ/инструкциЯМ/инструкциЯМИ/инструкциЕЮ.

If i searched the world "log" in confluence it finds words in articles like "log/logs/logged/logging". But if I look for one of the Russian words, I see articles that have only completely the same word in results.

Is it possible to do something with this problem?

1 answer

0 votes
Diego
Atlassian Team
Atlassian Team members are employees working across the company in a wide variety of roles.
January 9, 2019

Hello there Arthur!

This behaviour seems to be the expected one. If we take a look, the word log is present partially in logs, logged, logging. The results come due to this fact, log is present in the string presented.

I would like to understand if more results are provided if the search is based on just part of the original word, инструкция. Looking at it, it seems that инст is a common part of all the words presented.

With this in mind, could you clarify what happens when you search for инст instead of the entire word?

Arthur Foss January 10, 2019

Hello @Diego

 

No, if I write the word 'инст', nothing will be found. Only completely identical words are searched for russian words.

I wanted to say that it works differently for English.

For English,

  • if I write the word 'correctly' in results will be articles with words 'correct/correctly/corrected/corrects/correcting'
  • if I write the word 'corrected' in results will be articles with words  'correct/correctly/corrected/corrects/correcting'
  • if I write the word 'correct' in results will be articles with words  'correct/correctly/corrected/corrects/correcting'
  • if I write the word 'corrects' in results will be articles with words  'correct/correctly/corrected/corrects/correcting'
  • if I write the word 'correcting' in results will be articles with words  'correct/correctly/corrected/corrects/correcting'.

 

For Russian:

  • if I write the word 'инст' nothing will be found.
  • if I write the word 'инструкци' nothing will be found (it is not an existing word).
  • if I write the word 'инструкция' in results will be articles with word 'инструкция'
  • if I write the word 'инструкции' in results will be articles with word 'инструкции'
  • if I write the word 'инструкциям' in results will be articles with word 'инструкциям'.

 

As far as I understand the search knows some grammatical rules for English language and can search for different forms of the word. But search does not know the grammar of the Russian language and is looking for only 100% matches of all character of words.

Diego
Atlassian Team
Atlassian Team members are employees working across the company in a wide variety of roles.
January 10, 2019

Hello there again @Arthur Foss!

I was able to replicate the behaviour brought. With this in mind, I have raised a requested here:

Confluence Search - Similar Words in Russian

Also, Confluence still does not implement partial word search, which is why инст returns no results:

Confluence Search Result Should Include Partial Word Match

Currently, the best course of action would be to vote and watch the reports. Voting increases the report visibility and watching will keep you up to date!

Suggest an answer

Log in or Sign up to answer
TAGS
AUG Leaders

Atlassian Community Events