// TextAI class for parsing text and // predicting words using knowledge // about all learned data #include #include #include #include #include #include using namespace std; // some additional functions for work // replacing non-alnum chars from text with spaces void erase_no_words (string& text) { auto pred = [] (const char& c) { return !(isalnum (c) || iswspace (c)); }; replace_if (text.begin(), text.end(), pred, ' '); } // splitting text into words by spaces vector split_words (string text) { vector result; string temp; for (auto c: text) if (!iswspace (c)) temp += c; else if (!temp.empty ()) { result.push_back (temp); temp.clear (); } return result; } class TextAI { struct word_t { // word_data string word; int weight; word_t () : word (""), weight (-1) {} word_t (string s) : word (s), weight (0) {} bool operator== (const word_t& w) const { return word == w.word; } bool operator< (const word_t& w) const { return weight < w.weight; } word_t& operator++ () { weight++; return *this; } }; typedef set w_data; map database; public: TextAI () {} TextAI (string text) { learn (text); } bool word_unknown (string word) const { return database.count (word) == 0; } word_t find_weightest (const w_data& data) const { word_t result; for (auto word: data) if (result < word) result = word; return result; } void learn_pair (string word, string next) { *(database[word].insert (word_t (next))).first++; } void learn (string text) { erase_no_words (text); vector words = split_words (text); for (int i = 0; i+1 < words.size (); i++) learn_pair (words[i], words[i+1]); } string predict_next_word (string word) const { if (word_unknown (word)) return ""; else return find_weightest (database.at (word)).word; } }; int main () { TextAI ai; // try to add more text here! string text = "this is good text, so I think..."; string word = "is"; ai.learn (text); cout << "source text:\n" << text << "\npredicting after '" << word << "': '" << ai.predict_next_word (word) << "'"; return 0; }

Text AI 📚 [ v0.1 ]