[jira] [Commented] (LUCENE-7824) Multi-word synonyms rule with common terms at the same position are buggy

2017-05-14 Thread ASF subversion and git services (JIRA)

[ 
https://issues.apache.org/jira/browse/LUCENE-7824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16009823#comment-16009823
 ] 

ASF subversion and git services commented on LUCENE-7824:
-

Commit 55bad6fec3c984d4ef56f94f0f50b9f1b2e6dba3 in lucene-solr's branch 
refs/heads/branch_6_6 from [~jimczi]
[ https://git-wip-us.apache.org/repos/asf?p=lucene-solr.git;h=55bad6f ]

LUCENE-7824: Fix graph query analysis for multi-word synonym rules with common 
terms (eg. new york, new york city).


> Multi-word synonyms rule with common terms at the same position are buggy
> -
>
> Key: LUCENE-7824
> URL: https://issues.apache.org/jira/browse/LUCENE-7824
> Project: Lucene - Core
>  Issue Type: Bug
>Affects Versions: master (7.0), 6.5.1
>Reporter: Jim Ferenczi
> Fix For: master (7.0), 6.6
>
> Attachments: LUCENE-7824.patch
>
>
> The automaton built from the graph token stream tries to pack common terms in 
> multi word synonyms that appear at the same position. This means that some 
> states inside a multi word synonym can have multiple transitions.
> As a result the intersection point of the graph are not computed correctly.
> For example the synonym rule: "ny, new york city, new york" is not applied 
> correctly to the query "ny police".
> In this case "police" is detected as part of the multi synonyms path and we 
> create the disjunction between:
>  "ny police", "new york police", ...
> I pushed a patch that removes this optim (and creates a single transition 
> from each state) in order to ensure that the intersection points of the graph 
> always showed up at the end of the multi synonym paths.
> [~mattweber] can you take a look ?



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



[jira] [Commented] (LUCENE-7824) Multi-word synonyms rule with common terms at the same position are buggy

2017-05-14 Thread ASF subversion and git services (JIRA)

[ 
https://issues.apache.org/jira/browse/LUCENE-7824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16009822#comment-16009822
 ] 

ASF subversion and git services commented on LUCENE-7824:
-

Commit 84b8b5a1d895ba2fa2d7fbad8cd4ea50321e0dd3 in lucene-solr's branch 
refs/heads/branch_6x from [~jimczi]
[ https://git-wip-us.apache.org/repos/asf?p=lucene-solr.git;h=84b8b5a ]

LUCENE-7824: Fix graph query analysis for multi-word synonym rules with common 
terms (eg. new york, new york city).


> Multi-word synonyms rule with common terms at the same position are buggy
> -
>
> Key: LUCENE-7824
> URL: https://issues.apache.org/jira/browse/LUCENE-7824
> Project: Lucene - Core
>  Issue Type: Bug
>Affects Versions: master (7.0), 6.5.1
>Reporter: Jim Ferenczi
> Fix For: master (7.0), 6.6
>
> Attachments: LUCENE-7824.patch
>
>
> The automaton built from the graph token stream tries to pack common terms in 
> multi word synonyms that appear at the same position. This means that some 
> states inside a multi word synonym can have multiple transitions.
> As a result the intersection point of the graph are not computed correctly.
> For example the synonym rule: "ny, new york city, new york" is not applied 
> correctly to the query "ny police".
> In this case "police" is detected as part of the multi synonyms path and we 
> create the disjunction between:
>  "ny police", "new york police", ...
> I pushed a patch that removes this optim (and creates a single transition 
> from each state) in order to ensure that the intersection points of the graph 
> always showed up at the end of the multi synonym paths.
> [~mattweber] can you take a look ?



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



[jira] [Commented] (LUCENE-7824) Multi-word synonyms rule with common terms at the same position are buggy

2017-05-14 Thread ASF subversion and git services (JIRA)

[ 
https://issues.apache.org/jira/browse/LUCENE-7824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16009821#comment-16009821
 ] 

ASF subversion and git services commented on LUCENE-7824:
-

Commit 21362a3ba4c1e936416635667f257b36235b00ab in lucene-solr's branch 
refs/heads/master from [~jimczi]
[ https://git-wip-us.apache.org/repos/asf?p=lucene-solr.git;h=21362a3 ]

LUCENE-7824: Fix graph query analysis for multi-word synonym rules with common 
terms (eg. new york, new york city).


> Multi-word synonyms rule with common terms at the same position are buggy
> -
>
> Key: LUCENE-7824
> URL: https://issues.apache.org/jira/browse/LUCENE-7824
> Project: Lucene - Core
>  Issue Type: Bug
>Affects Versions: master (7.0), 6.5.1
>Reporter: Jim Ferenczi
> Fix For: master (7.0), 6.6
>
> Attachments: LUCENE-7824.patch
>
>
> The automaton built from the graph token stream tries to pack common terms in 
> multi word synonyms that appear at the same position. This means that some 
> states inside a multi word synonym can have multiple transitions.
> As a result the intersection point of the graph are not computed correctly.
> For example the synonym rule: "ny, new york city, new york" is not applied 
> correctly to the query "ny police".
> In this case "police" is detected as part of the multi synonyms path and we 
> create the disjunction between:
>  "ny police", "new york police", ...
> I pushed a patch that removes this optim (and creates a single transition 
> from each state) in order to ensure that the intersection points of the graph 
> always showed up at the end of the multi synonym paths.
> [~mattweber] can you take a look ?



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



[jira] [Commented] (LUCENE-7824) Multi-word synonyms rule with common terms at the same position are buggy

2017-05-12 Thread Matt Weber (JIRA)

[ 
https://issues.apache.org/jira/browse/LUCENE-7824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16008311#comment-16008311
 ] 

Matt Weber commented on LUCENE-7824:


Sure, looks good then!

> Multi-word synonyms rule with common terms at the same position are buggy
> -
>
> Key: LUCENE-7824
> URL: https://issues.apache.org/jira/browse/LUCENE-7824
> Project: Lucene - Core
>  Issue Type: Bug
>Reporter: Jim Ferenczi
> Attachments: LUCENE-7824.patch
>
>
> The automaton built from the graph token stream tries to pack common terms in 
> multi word synonyms that appear at the same position. This means that some 
> states inside a multi word synonym can have multiple transitions.
> As a result the intersection point of the graph are not computed correctly.
> For example the synonym rule: "ny, new york city, new york" is not applied 
> correctly to the query "ny police".
> In this case "police" is detected as part of the multi synonyms path and we 
> create the disjunction between:
>  "ny police", "new york police", ...
> I pushed a patch that removes this optim (and creates a single transition 
> from each state) in order to ensure that the intersection points of the graph 
> always showed up at the end of the multi synonym paths.
> [~mattweber] can you take a look ?



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



[jira] [Commented] (LUCENE-7824) Multi-word synonyms rule with common terms at the same position are buggy

2017-05-12 Thread Jim Ferenczi (JIRA)

[ 
https://issues.apache.org/jira/browse/LUCENE-7824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16008305#comment-16008305
 ] 

Jim Ferenczi commented on LUCENE-7824:
--

I don't think we should try to optimize here. The number of terms should be 
small in a query so I would prefer to keep it simple and just create a new 
entry for each token like the cached token stream does.

> Multi-word synonyms rule with common terms at the same position are buggy
> -
>
> Key: LUCENE-7824
> URL: https://issues.apache.org/jira/browse/LUCENE-7824
> Project: Lucene - Core
>  Issue Type: Bug
>Reporter: Jim Ferenczi
> Attachments: LUCENE-7824.patch
>
>
> The automaton built from the graph token stream tries to pack common terms in 
> multi word synonyms that appear at the same position. This means that some 
> states inside a multi word synonym can have multiple transitions.
> As a result the intersection point of the graph are not computed correctly.
> For example the synonym rule: "ny, new york city, new york" is not applied 
> correctly to the query "ny police".
> In this case "police" is detected as part of the multi synonyms path and we 
> create the disjunction between:
>  "ny police", "new york police", ...
> I pushed a patch that removes this optim (and creates a single transition 
> from each state) in order to ensure that the intersection points of the graph 
> always showed up at the end of the multi synonym paths.
> [~mattweber] can you take a look ?



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



[jira] [Commented] (LUCENE-7824) Multi-word synonyms rule with common terms at the same position are buggy

2017-05-12 Thread Matt Weber (JIRA)

[ 
https://issues.apache.org/jira/browse/LUCENE-7824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16008209#comment-16008209
 ] 

Matt Weber commented on LUCENE-7824:


[~jim.ferenczi] Maybe use a {{BytesRefHash}} and maintain a id-to-hash map so 
we still only have single copy of common term in memory and still have a unique 
id?

> Multi-word synonyms rule with common terms at the same position are buggy
> -
>
> Key: LUCENE-7824
> URL: https://issues.apache.org/jira/browse/LUCENE-7824
> Project: Lucene - Core
>  Issue Type: Bug
>Reporter: Jim Ferenczi
> Attachments: LUCENE-7824.patch
>
>
> The automaton built from the graph token stream tries to pack common terms in 
> multi word synonyms that appear at the same position. This means that some 
> states inside a multi word synonym can have multiple transitions.
> As a result the intersection point of the graph are not computed correctly.
> For example the synonym rule: "ny, new york city, new york" is not applied 
> correctly to the query "ny police".
> In this case "police" is detected as part of the multi synonyms path and we 
> create the disjunction between:
>  "ny police", "new york police", ...
> I pushed a patch that removes this optim (and creates a single transition 
> from each state) in order to ensure that the intersection points of the graph 
> always showed up at the end of the multi synonym paths.
> [~mattweber] can you take a look ?



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org