| import textdistance | |
| import pickle | |
| correction_dict = {} | |
| with open('correct.txt', 'r') as correct_file, open('incorrect.txt', 'r') as incorrect_file: | |
| correct_lines = correct_file.readlines() | |
| incorrect_lines = incorrect_file.readlines() | |
| for correct_line, incorrect_line in zip(correct_lines, incorrect_lines): | |
| correct_words = correct_line.strip().split() | |
| incorrect_words = incorrect_line.strip().split() | |
| for incorrect_word, correct_word in zip(incorrect_words, correct_words): | |
| correction_dict[incorrect_word] = correct_word | |
| def predict_correction(incorrect_word, correction_dict): | |
| if incorrect_word in correction_dict: | |
| return correction_dict[incorrect_word], 0 | |
| else: | |
| best_match = None | |
| best_distance = float('inf') | |
| for correct_word in correction_dict.keys(): | |
| distance = textdistance.levenshtein(incorrect_word, correct_word) | |
| if distance < best_distance: | |
| best_distance = distance | |
| best_match = correct_word | |
| return best_match, best_distance | |
| def correct_text(input_text, correction_dict): | |
| corrected_text = [] | |
| words = input_text.split() | |
| for word in words: | |
| best_correction, distance = predict_correction(word, correction_dict) | |
| if best_correction: | |
| corrected_text.append(best_correction) | |
| else: | |
| corrected_text.append(word) | |
| return ' '.join(corrected_text) | |
| with open('correction_model.pkl', 'wb') as model_file: | |
| pickle.dump(correction_dict, model_file) | |