JAVA and sentiment analysis

  java   emotion

graduation design done on sentiment analysis , such as: the input sentence , judgment or criticism praised . Experts who have nothing to disclose what information and technology ?
------ Solution - -------------------------------------------
this a bit difficult, first word , then estimated to be combined with contextual semantics. . .
------ Solution ---------------------------------------- ----
this idea I have is for the landlord reference : Get the input string , and then parse the string , and then create a string array , the array included all the words used to describe the mood of people , word , phrase , and then provide an array of strings each string and input is compared to criticize or judge praised ...
------ Solution ---------- ----------------------------------
the whole into a thesaurus related words starting treatment ah
------ Solution ------------------------------------------ -
thesaurus does have to have it, a simple method is to give each word and the word given different positive and negative values ​​( expressed emotion )
------ For reference only ---- -----------------------------------

  The reply was deleted administrator at 2013-03-19 09:55:04

------ For reference only ---------------------------------- -----
really need to do this thesaurus , the Internet and look it ,
------ For reference only ------------------ ---------------------
it, in which to get hold of the database to determine the thesaurus . This should be no problem
------ For reference only ---------------------------------- -----
impossible to be completely accurate , this is the difference between humans and computers. Of course, these can only say at the time of reply . Come on . You can also go directly Baidu input word , and then returned to find Baidu parsing and analysis. This is also a way to Kazakhstan.
------ For reference only -------------------------------------- -
I also do this, we can explore to explore each other . Currently I have achieved based on the emotional part of Naive Bayesian analysis , but the analysis is poor, still thinking of ways to improve .
------ For reference only -------------------------------------- -


accuracy analysis about how much? I am also an issue of positive and negative sentiment analysis determined that the Internet to find some open source framework for Chinese support is not very good . Now want to write a sentiment analysis program that can give me some advice?