int API_DOC2VEC::doc2vec(
vector< string > vecQuaryWord, //[In]quary word
map< vector< string >, vector< float > > &mapQuaryDoc ) //[Out]quary Doc
{
if ( vecQuaryWord.size() == 0 )
{
printf( "API_DOC2VEC::doc2vec err!!vecQuaryWord.size:%ld, do not find KeyWord!!\n", vecQuaryWord.size() );
return -1;
}
/*****************************Init:Commen*****************************/
int nRet = 0;
string word;
long long i, j, Count;
/*****************************Init*****************************/
vector<string> outQueryWords;
vector<string> queryCutWords;
map< string, vector< float > > mapQuaryWord; //[In]quary word
map< string, float > mapQuaryTf; //[In]quary TF
map< string,int >::iterator itStopwords;
map< string, float >::iterator itQuaryTf; //[In]dic word
map< string, vector< float > >::iterator itDicWord; //[In]dic word
//doc2vec_cutword
outQueryWords.clear();
for ( j=0;j<vecQuaryWord.size();j++ )
{
queryCutWords.clear();
//Cut Word
app.cut( vecQuaryWord[j], queryCutWords, METHOD_MIX );
//Check && Save GetFeat
for ( i=0;i<queryCutWords.size();i++ )
{
if ( (queryCutWords[i].empty()) || (queryCutWords[i] == " ") || (queryCutWords[i] == " ")
|| (api_commen.IsDigit2(queryCutWords[i])) || (queryCutWords[i].size()<2) )
{
continue;
}
itStopwords = stopwords.find(queryCutWords[i]);
if (itStopwords == stopwords.end()) // not find
{
outQueryWords.push_back( queryCutWords[i] );
}
}
}
/* //check info
printf("Cut Words:");
for(i=0;i<outQueryWords.size();i++)
printf( "%s ", outQueryWords[i].c_str() );
printf( "\n" );*/
mapQuaryWord.clear(); //[In]quary word
mapQuaryTf.clear(); //[In]quary TF
//Count Quary TF by Quary Vector
for (i = 0; i < outQueryWords.size(); i++)
{
word = outQueryWords[i];
//Quary Vector
itDicWord = mapDicWord.find(word);
if (itDicWord != mapDicWord.end()) // find it
mapQuaryWord[word] = itDicWord->second;
else
continue;
//Quary TF
itQuaryTf = mapQuaryTf.find(word);
if (itQuaryTf != mapQuaryTf.end()) // find it
mapQuaryTf[word]++;
else
mapQuaryTf[word] = 1;
}
//Quary TF
Count = mapQuaryTf.size();
if ( Count == 0 )
{
printf( "idQuarySize:%ld, do not find KeyWord!!\n", outQueryWords.size() );
return -1;
}
for(itQuaryTf = mapQuaryTf.begin(); itQuaryTf != mapQuaryTf.end(); itQuaryTf++)
{
mapQuaryTf[itQuaryTf->first] = itQuaryTf->second*1.0/Count;
}
//word2doc
mapQuaryDoc.clear();
nRet = doc2vec_word2doc( mapQuaryWord, mapQuaryTf, mapQuaryDoc );
if ( nRet != 0 )
{
printf("doc2vec_train:word2doc err!!\n");
return nRet;
}
return nRet;
}
/*************************************************************************
> File Name: ./app.cpp
> Author:
> Mail:
> Created Time: Mon 03 Aug 2015 10:13:19 AM CST
************************************************************************/
#include<iostream>
int a = 10;
std::string str("hello");
CppJieba::Application app("./dict/jieba.dict.utf8",
"./dict/hmm_model.utf8",
"./dict/user.dict.utf8",
"./dict/idf.utf8",
"./dict/stop_words.utf8");
using namespace std;
void test()
{
app.cut(std::string, std::vector<std::string>)
{
}
}