novel-pinyin 基于自然语言处理的拼音输入法 - 第3页

**chaisave** · 07-10-16, 13:07

两位过谦了。如果 epico 有需要的话，我可以帮忙打打下手。8-)

**epico** · 07-10-16, 13:36

引用:

作者: chaisave

两位过谦了。如果 epico 有需要的话，我可以帮忙打打下手。8-)

版主过谦了。欢迎加入。呵呵。

**liuspider** · 07-10-16, 17:41

glad you guys can work together

when you want more publicity, let me know and I will add this to http://www.scim-im.org/projects/imengines

**findsun** · 07-10-17, 00:33

引用:

作者: liuspider

glad you guys can work together

when you want more publicity, let me know and I will add this to http://www.scim-im.org/projects/imengines

Liuspider, yes, please help to add scim-sunpinyin on http:// scim-im.org/projects/imengines!

If you need any information about scim-sunpinyin or sunpinyin, you may send query to TCHAIKOV at GMAIL dot COM or YONG.SUN at SUN dot COM.

Here is the project page for input-method project on OS.o, www.opensolaris.org/os/project/input-method.

Thanks so much!

**dgod** · 07-10-17, 09:05

搞中文输入法的人自己竟然用英文聊天，受不了。

**epico** · 07-10-17, 10:43

引用:

作者: liuspider

glad you guys can work together

when you want more publicity, let me know and I will add this to http://www.scim-im.org/projects/imengines

Thanks in advance! Currently novel-pinyin is under heavily development, please be patient.When it is ready, I will tell you in the first time.

**epico** · 07-11-16, 11:53

Novel Pinyin Version 1 TODO List
http://alex-epico.blogspot.com/2007/...todo-list.html

**epico** · 07-11-16, 11:53

Novel Pinyin Version 1 TODO List
http://alex-epico.blogspot.com/2007/...todo-list.html

很抱歉，帖子发重复了。

**epico** · 07-11-16, 11:55

引用:

作者: epico

Novel Pinyin Version 1 TODO List
http://alex-epico.blogspot.com/2007/...todo-list.html

Storage:

* N-gram file

Training:

* Training using parameters in prototype system.

Lookup:

* Original Lookup with Candidate Selection.

Self-Learning:

* Learn User Sentence when Commit String.

novel-imengine:

* Special Table support rewrite from scim-pinyin in c.
* Scim UI Config Module.

**colinzhengj** · 07-11-16, 15:24

I'll definitely use it if you don't have to rely on SCIM

**epico** · 07-12-20, 14:05

引用:

作者: colinzhengj

I'll definitely use it if you don't have to rely on SCIM

I will consider it. But the first version will be on scim.

**epico** · 07-12-20, 14:06

Finished n-gram storage code, and update code on http://novel-pinyin.cvs.sourceforge..../novel-pinyin/.

On currently implementation, I modified the P(P|W) from ratio which is computed from scim-pinyin, to k/n, k indicates k matched pinyins, n indicates total n pinyins for word W.
I don't know how it will influence the result of HMM correct rate, hopes it will not be bad.

Tested this approach on research prototype, the correct rate is lower. And test with manual input, it don't work too bad.

And counting the computing complexity, it seems that the speed is sufficient.

**epico** · 07-12-20, 14:07

Upload new novel-pinyin code to sourceforge, currently finished segment & training part.

In this place, I use a modified interpolation method to ease implementation.
The parameter optimization part is done in research prototype.
So the code in novel-pinyin is relatively simple, just use parameters computed from prototype.

The word segment use shortest path algorithm to segment words, and prepare the data to training part.

**bearson** · 08-05-26, 03:26

不推荐加入不完整双拼，
双拼，就是要两个字幕一个汉字，这个功能也只是对频繁使用的词语起到很有限的加速效果，
既然这个输入法要做到整句输入，就不要这个功能。

**epico** · 08-05-28, 15:58

引用:

作者: bearson

不推荐加入不完整双拼，
双拼，就是要两个字幕一个汉字，这个功能也只是对频繁使用的词语起到很有限的加速效果，
既然这个输入法要做到整句输入，就不要这个功能。

我也觉得不完整双拼，和整句输入法不太好结合在一起。
当然如果有人能够提出好的想法的话，也可以参考实现一下。
呵呵。

07-10-16, 13:07	第 31 帖
chaisave ★版主☆ 注册日期: Mar 2003 帖子: 2,402 精华: 5	两位过谦了。如果 epico 有需要的话，我可以帮忙打打下手。8-) __________________ Have you Debianned today?

07-10-16, 17:41	第 33 帖
liuspider 注册会员注册日期: Jan 2003 帖子: 1,280 精华: 4	glad you guys can work together when you want more publicity, let me know and I will add this to http://www.scim-im.org/projects/imengines __________________ 生于忧患，死于安乐！ www.liucougar.net

07-11-16, 11:53	第 38 帖
epico 注册会员注册日期: May 2006 帖子: 45 精华: 0	Novel Pinyin Version 1 TODO List http://alex-epico.blogspot.com/2007/...todo-list.html 很抱歉，帖子发重复了。此帖于 07-11-16 12:06 被 epico 编辑. 原因: 发重复了

欢迎来到LinuxSir.Org!
您还未登录，请登录后查看论坛，或者点击论坛上方的注册链接注册新账号。

07-10-17, 09:05	第 35 帖
dgod 注册会员注册日期: Sep 2004 帖子: 634 精华: 3	搞中文输入法的人自己竟然用英文聊天，受不了。

07-11-16, 11:53	第 37 帖
epico 注册会员注册日期: May 2006 帖子: 45 精华: 0	Novel Pinyin Version 1 TODO List http://alex-epico.blogspot.com/2007/...todo-list.html

07-11-16, 15:24	第 40 帖
colinzhengj 注册会员注册日期: Nov 2006 我的住址: Austin 帖子: 26 精华: 0	I'll definitely use it if you don't have to rely on SCIM

07-12-20, 14:06	第 42 帖
epico 注册会员注册日期: May 2006 帖子: 45 精华: 0	Finished n-gram storage code, and update code on http://novel-pinyin.cvs.sourceforge..../novel-pinyin/. On currently implementation, I modified the P(P\|W) from ratio which is computed from scim-pinyin, to k/n, k indicates k matched pinyins, n indicates total n pinyins for word W. I don't know how it will influence the result of HMM correct rate, hopes it will not be bad. Tested this approach on research prototype, the correct rate is lower. And test with manual input, it don't work too bad. And counting the computing complexity, it seems that the speed is sufficient.

07-12-20, 14:07	第 43 帖
epico 注册会员注册日期: May 2006 帖子: 45 精华: 0	Upload new novel-pinyin code to sourceforge, currently finished segment & training part. In this place, I use a modified interpolation method to ease implementation. The parameter optimization part is done in research prototype. So the code in novel-pinyin is relatively simple, just use parameters computed from prototype. The word segment use shortest path algorithm to segment words, and prepare the data to training part.

08-05-26, 03:26	第 44 帖
bearson 注册会员注册日期: Oct 2007 帖子: 168 精华: 0	不推荐加入不完整双拼，双拼，就是要两个字幕一个汉字，这个功能也只是对频繁使用的词语起到很有限的加速效果，既然这个输入法要做到整句输入，就不要这个功能。