[
https://issues.apache.org/jira/browse/LUCENE-7444?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Uwe Schindler updated LUCENE-7444:
----------------------------------
Description:
Yonik said on LUCENE-7318:
{quote}
bq. I think it would make a good default for most Lucene users, and we should
graduate it from the analyzers module into core, and make it the default for
IndexWriter.
This "StandardAnalyzer" is specific to English, as it removes English stopwords.
That seems to be an odd choice now for a few reasons:
- It was argued in the past (rather vehemently) that Solr should not prefer
english in it's default "text" field
- AFAIK, removing stopwords is no longer considered best practice.
Given that removal of english stopwords is the only thing that really makes
this analyzer english-centric (and given the negative impact that can have on
other languages), it seems like the stopword filter should be removed from
StandardAnalyzer.
{quote}
When trying to fix the backwards incompatibility issues in LUCENE-7318, it
looks like most unrelated code moved from analysis module to core (and changing
package names!!!! :( ) was related to word list loading and superclasses of
StopFilter. If we follow Yonik's suggestion, we can revert all those changes. I
agree with hin, an "universal" analyzer should not have any language specific
stop-words.
The other thing is LowercaseFilter, but I'd suggest to simply add a clone of it
to Lucene core and leave the analysis-module self-contained.
was:
Yonik said on LUCENE-7318:
{quote}
bq. I think it would make a good default for most Lucene users, and we should
graduate it from the analyzers module into core, and make it the default for
IndexWriter.
This "StandardAnalyzer" is specific to English, as it removes English stopwords.
That seems to be an odd choice now for a few reasons:
- It was argued in the past (rather vehemently) that Solr should not prefer
english in it's default "text" field
- AFAIK, removing stopwords is no longer considered best practice.
Given that removal of english stopwords is the only thing that really makes
this analyzer english-centric (and given the negative impact that can have on
other languages), it seems like the stopword filter should be removed from
StandardAnalyzer.
{quote}
When trying to fix the backwards incompatibility issues in LUCENE-7318, it
looks like most unrelated code moved from analysis module to core (and changing
package names!!!! :( ) was related to word list loading. If we follow Yonik's
suggestion, we can revert all those changes. I agree with hin, an "universal"
analyzer should not have any language specific stop-words.
The other thing is LowercaseFilter, but I'd suggest to simply add a clone of it
to Lucene core and leave the analysis-module self-contained.
> Remove StopFilter from StandardAnalyzer in Lucene-Core
> ------------------------------------------------------
>
> Key: LUCENE-7444
> URL: https://issues.apache.org/jira/browse/LUCENE-7444
> Project: Lucene - Core
> Issue Type: Task
> Components: core/other, modules/analysis
> Affects Versions: 6.2
> Reporter: Uwe Schindler
>
> Yonik said on LUCENE-7318:
> {quote}
> bq. I think it would make a good default for most Lucene users, and we should
> graduate it from the analyzers module into core, and make it the default for
> IndexWriter.
> This "StandardAnalyzer" is specific to English, as it removes English
> stopwords.
> That seems to be an odd choice now for a few reasons:
> - It was argued in the past (rather vehemently) that Solr should not prefer
> english in it's default "text" field
> - AFAIK, removing stopwords is no longer considered best practice.
> Given that removal of english stopwords is the only thing that really makes
> this analyzer english-centric (and given the negative impact that can have on
> other languages), it seems like the stopword filter should be removed from
> StandardAnalyzer.
> {quote}
> When trying to fix the backwards incompatibility issues in LUCENE-7318, it
> looks like most unrelated code moved from analysis module to core (and
> changing package names!!!! :( ) was related to word list loading and
> superclasses of StopFilter. If we follow Yonik's suggestion, we can revert
> all those changes. I agree with hin, an "universal" analyzer should not have
> any language specific stop-words.
> The other thing is LowercaseFilter, but I'd suggest to simply add a clone of
> it to Lucene core and leave the analysis-module self-contained.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
