help make my code more efficient for long speeches

2. identify unique word-pairs and calculate their frequencies.

bool wordpairMapping( vector& sentences, map< pair, int> &wordpairFreq_map); 

Given a list of sentences stored in the first argument sentences, this function identifies all the all the unique word-pairs and each word-pair's frequency. The identified (word-pair, frequency)'s will be stored into wordpariFreq_map, which is a map of (key, value) pairs. The key of this map a word-pair and the value is the frequency of this word-pair. This function will return true if the mapping is successful; false otherwise.

Note that

Tokens are case insensitive. We will consider lower case in this project. Whitespaces will be the token delimiter.

The two words in a word-pair are different. For example, event though the first sentence above contains two the, you are not going to construct a word pair

Order does not matter between two words in a word-pair. For example, the word-pair is the same as . You are recommended to arrange the two words in lexicographical order before inserting the pair into the map.

Suggestions:

Use istringstream to tokenize a sentence.

Use set to store all the unique tokens identified in a sentence.

Assume sentences consists of the following 3 sentences:

The first story is about connecting the dots. The first story is about connecting the dots. The first story is about connecting the dots. 

This function is going to identify a total of 21 word-pairs as follows:

: 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3 : 3

My code:

bool wordpairMapping( vector<string>& sentences, map< pair<string,string>, int> &wordpairFreq_map){

bool found = false;

int count =0;

vector<string> vec;

vector<string> vect;

vector<string> tokens;

for(unsigned int i = 0; i < sentences.size(); i ++){

string str = sentences[i];

char *cstr = new char[str.length() + 1];

strcpy(cstr, str.c_str());

char * pch;

pch = strtok (cstr,".");

while (pch != NULL)

{

tokens.push_back(pch);

pch = strtok (NULL, ".");

}

}

map < pair<string, string>, int > :: iterator it;

string tempo;

string line ="";

for(unsigned int i =0; i < tokens.size(); i++){

line = tokens[i];

for (int j =0; j < line.length(); j++){

if(line[j] == '-'){

line.erase(line.begin()+j);

}

}

vect.push_back(line);

}

int temp =0;

for(unsigned int i =0; i < vect.size(); i++){

vector<string> str;

string line = vect[i];

for(int i =0; i < line.length(); i ++){

if(!isspace(line[i])){

temp =i;

while(!isspace(line[i]))

i++;

string word = line.substr(temp, i-temp);

transform(word.begin(), word.end(), word.begin(), ::tolower);

str.push_back(word);

}

}

for(unsigned int i =0; i < str.size(); i++){

for(int j =i+1; j < str.size(); j++){

if((str[j] < str[i]) && (str[i] != str[j])){

tempo = str[i];

str[i] = str[j];

str[j] = tempo;

}

}

}

for(unsigned int i =0; i < str.size(); i++){

for(unsigned int j =i+1; j < str.size(); j++){

if(str[i] == str[j]){

str.erase(str.begin()+j);

}

}

}

for( unsigned int k =0; k < str.size(); k ++){

for(unsigned int l =k+1; l < str.size(); l++){

if(str[k] != str[l]){

bool check = true;

pair<string, string> p;

p.first = str[k];

p.second = str[l];

if(!wordpairFreq_map.empty()){

for(it = wordpairFreq_map.begin(); it != wordpairFreq_map.end(); it++){

if((it->first.first == p.first && it->first.second == p.second )|| (it->first.second == p.first &&it->first.first == p.second) ){

it->second++;

check = false;

count ++;

}

}

if(check) {

wordpairFreq_map.insert(make_pair(make_pair(p.first, p.second), 1));

count++;

}

}

if(wordpairFreq_map.empty()) {

wordpairFreq_map.insert(make_pair(make_pair(p.first, p.second), 1));

count++;

}

}

}

}

found = true;

}

return found;

}