[jira] Created: (LUCENE-1629) contrib intelligent Analyzer for Chinese

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12709880#action_12709880 ]

Robert Muir commented on LUCENE-1629:
-------------------------------------

if you acquire the big5 resources, do you think it would be possible to create a single dictionary that works with both Simplified & Traditional?

(i.e. merge the big5 resources with the gb resources)

The reason I say this, is the existing chinese analyzers, although they tokenize in a less intelligent way, they are agnostic to Simplified/Traditional issues...

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12709885#action_12709885 ]

Robert Muir commented on LUCENE-1629:
-------------------------------------

another potential issue with big5 i want to point out is that many of the big5 character sets such as HKSCS have characters that are mapped into regions of unicode outside of the BMP.

just glancing at the code, some things will need to be modified for this to work correctly with surrogate pairs, various functions that take char will need to take codepoint (int), etc.

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12709974#action_12709974 ]

Mingfai Ma commented on LUCENE-1629:
------------------------------------

could we use CC-CEDICT's dictionary instead? it is using Creative Commons Attribution-Share Alike 3.0 license

http://www.mdbg.net/chindict/chindict.php?page=cc-cedict

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12710035#action_12710035 ]

Koji Sekiguchi commented on LUCENE-1629:
----------------------------------------

Just an FYI. There have been a working for mapping between simplified and traditional chinese characters in Solr 1.4. (but you need to define mapping rules in mapping.txt)
See SOLR-822 and the attached JPG for chinese mapping sample.
I opened LUCENE-1466 for Lucene. :)

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12710070#action_12710070 ]

Robert Muir commented on LUCENE-1629:
-------------------------------------

koji, have you considered using icu transforms for this behavior?
Not only is the rule-based language very nice (you can define variables, use context, etc), but many transformations such as "Traditional-Simplified" are already defined.

http://userguide.icu-project.org/transforms/general

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12710118#action_12710118 ]

Koji Sekiguchi commented on LUCENE-1629:
----------------------------------------

bq. koji, have you considered using icu transforms for this behavior?

Not yet.

bq. Not only is the rule-based language very nice (you can define variables, use context, etc), but many transformations such as "Traditional-Simplified" are already defined.

Right. CharFilter framework that I introduced in SOLR-822 is not only for rule-based mapping, but it can use existing library like ICU to transform/normalize characters.

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12710147#action_12710147 ]

Mingfai Ma commented on LUCENE-1629:
------------------------------------

i'm not sure if the character mapping is a feasible approach. The CC-CEDICT dictionary has terms in both simplified and traditional chinese already and we need not to define any rule. how much effort is needed to define all mapping rules for the simplified and traditional chinese thesaurus?

Besides, simplified and traditional Chinese conversion is not as simple as mapping the code. For the sample meaning, it may use different words in the SC and TC.

if the approach accepted in this issue is ok, I just need to figure out how to convert CC-CEDICT to the dct / mem format, and i suppose it is doable.

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12710712#action_12710712 ]

Xiaoping Gao commented on LUCENE-1629:
--------------------------------------

The dictionary is loaded in to 2 classes:
BigramDictionary.java
WordDictionary.java

so you can read from the loading section to get "dct" format

mem files are just arrays that has been serialized to object files, you can get the format from the code and comments.

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12714285#action_12714285 ]

Otis Gospodnetic commented on LUCENE-1629:
------------------------------------------

I just got to look at this code and I only scanned it quickly.  Is all of the code really Chinese-specific?
Would any of it be applicable to other languages, say Japanese or Korean? (assuming we have dictionaries in suitable format)

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12714293#action_12714293 ]

Xiaoping Gao commented on LUCENE-1629:
--------------------------------------

I think the algorithm of Hidden Markov Model is applicable,
but I doubt that Japanese and Korean may need some language specific
analysis, such as "片假名かたかな.平假名ひらがな" in Japanese, as far as I know, the same
word can have different spelling. This may be a problem in the application?

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12714586#action_12714586 ]

Otis Gospodnetic commented on LUCENE-1629:
------------------------------------------

Hm, my Japanese is a little weak, so I'm not sure what exactly that means and what exactly
different spelling means in this context... :)

Google says "片假名かたかな.平假名ひらがな" means "How do假名piece. Hiragana假名flat"

Which word in the above Japanese text is the same as another word, yet with a different spelling?  Is this a question of synonyms,
such as "auto", "automobile", and "car", and even "sedan" in English?

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12714592#action_12714592 ]

Robert Muir commented on LUCENE-1629:
-------------------------------------

otis if you are interested in japanese/korean you might find this link interesting:

http://bugs.icu-project.org/trac/ticket/2229

similar to the thai approach (in contrib) but with log probabilities.

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12714670#action_12714670 ]

Mingfai Ma commented on LUCENE-1629:
------------------------------------

re. 平假名 and  片假名 in Japanese
http://en.wikipedia.org/wiki/Kana
http://en.wikipedia.org/wiki/Hiragana
http://en.wikipedia.org/wiki/Katakana

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12729373#action_12729373 ]

KuroSaka TeruHiko commented on LUCENE-1629:
-------------------------------------------

WordTokenizer extends Tokenizer, but it's constructor takes a TokenStream rather than a Reader.  
Shouldn't WordTokenizer rather extends TokenFilter, and if so, shouldn't it be named WordTokenFilter?

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...

    [ https://issues.apache.org/jira/browse/LUCENE-1629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12729381#action_12729381 ]

Robert Muir commented on LUCENE-1629:
-------------------------------------

bq. Shouldn't WordTokenizer rather extends TokenFilter, and if so, shouldn't it be named WordTokenFilter?

yes, you are correct. see LUCENE-1728 where we have proposed correcting this.

> contrib intelligent Analyzer for Chinese
> ----------------------------------------
>
>                 Key: LUCENE-1629
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1629
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: contrib/analyzers
>    Affects Versions: 2.4.1
>         Environment: for java 1.5 or higher, lucene 2.4.1
>            Reporter: Xiaoping Gao
>            Assignee: Michael McCandless
>             Fix For: 2.9
>
>         Attachments: analysis-data.zip, bigramdict.mem, build-resources-with-folder.patch, build-resources.patch, build-resources.patch, coredict.mem, LUCENE-1629-encoding-fix.patch, LUCENE-1629-java1.4.patch
>
>
> I wrote a Analyzer for apache lucene for analyzing sentences in Chinese language. it's called "imdict-chinese-analyzer", the project on google code is here: http://code.google.com/p/imdict-chinese-analyzer/
> In Chinese, "我是中国人"(I am Chinese), should be tokenized as "我"(I)   "是"(am)   "中国人"(Chinese), not "我" "是中" "国人". So the analyzer must handle each sentence properly, or there will be mis-understandings everywhere in the index constructed by Lucene, and the accuracy of the search engine will be affected seriously!
> Although there are two analyzer packages in apache repository which can handle Chinese: ChineseAnalyzer and CJKAnalyzer, they take each character or every two adjoining characters as a single word, this is obviously not true in reality, also this strategy will increase the index size and hurt the performance baddly.
> The algorithm of imdict-chinese-analyzer is based on Hidden Markov Model (HMM), so it can tokenize chinese sentence in a really intelligent way. Tokenizaion accuracy of this model is above 90% according to the paper "HHMM-based Chinese Lexical analyzer ICTCLAL" while other analyzer's is about 60%.
> As imdict-chinese-analyzer is a really fast and intelligent. I want to contribute it to the apache lucene repository.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@...
For additional commands, e-mail: java-dev-help@...