kenlm

run.py

@@ -5,43 +5,51 @@ from nltk.tokenize import RegexpTokenizer
 from nltk import trigrams
 import regex as re
 import lzma
+import kenlm
 
 
 class WordPred:
 
     def __init__(self):
         self.tokenizer = RegexpTokenizer(r"\w+")
-        self.model = defaultdict(lambda: defaultdict(lambda: 0))       
-        self.vocab = set()
-        self.alpha = 0.001
+        # self.model = defaultdict(lambda: defaultdict(lambda: 0))
+        self.model = kenlm.Model("model.binary")
+        self.words = set()
 
     def read_file(self, file):
         for line in file:
             text = line.split("\t")
-            yield re.sub(r"[^\w\d'\s]+", '', re.sub(' +', ' ', ' '.join([text[6], text[7]]).replace("\\n"," ").replace("\n","").lower()))
+            yield re.sub(r"[^\w\d'\s]+", '',
+                         re.sub(' +', ' ', ' '.join([text[6], text[7]]).replace("\\n", " ").replace("\n", "").lower()))
 
     def read_file_7(self, file):
         for line in file:
             text = line.split("\t")
             yield re.sub(r"[^\w\d'\s]+", '', re.sub(' +', ' ', text[7].replace("\\n", " ").replace("\n", "").lower()))
 
-    def read_train_data(self, file_path):
+    def fill_words(self, file_path, output_file):
+        with open(output_file, 'w') as out:
+            with lzma.open(file_path, mode='rt') as file:
+                for text in self.read_file(file):
+                    for word in text.split(" "):
+                        if word not in self.words:
+                            out.write(word + "\n")
+                        self.words.add(word)
+
+    def read_words(self, file_path):
+        with open(file_path, 'r') as fin:
+            for word in fin.readline():
+                self.words.add(word.replace("\n",""))
+
+
+    def create_train_file(self, file_path, output_path, rows=10000):
+        with open(output_path, 'w') as outputfile:
             with lzma.open(file_path, mode='rt') as file:
                 for index, text in enumerate(self.read_file(file)):
-                tokens = self.tokenizer.tokenize(text)
-                for w1, w2, w3 in trigrams(tokens, pad_right=True, pad_left=True):
-                    if w1 and w2 and w3:
-                        self.model[(w2, w3)][w1] += 1
-                        self.vocab.add(w1)
-                        self.vocab.add(w2)
-                        self.vocab.add(w3)
-                if index == 300000:
+                    outputfile.write(text)
+                    if index == rows:
                         break
-            
-            for word_pair in self.model:
-                num_n_grams = float(sum(self.model[word_pair].values()))
-                for word in self.model[word_pair]:
-                    self.model[word_pair][word] = (self.model[word_pair][word] + self.alpha) / (num_n_grams + self.alpha*len(self.vocab))
+                outputfile.close()
 
     def generate_outputs(self, input_file, output_file):
         with open(output_file, 'w') as outputf:
@@ -55,8 +63,7 @@ class WordPred:
                     outputf.write(prediction + '\n')
 
     def predict_probs(self, word1, word2):
-        predictions = dict(self.model[word1, word2])
-        most_common = dict(Counter(predictions).most_common(6))
+
 
         total_prob = 0.0
         str_prediction = ''
@@ -75,7 +82,7 @@ class WordPred:
 
         return str_prediction
 
+if __name__ == "__main__":
     wp = WordPred()
-wp.read_train_data('train/in.tsv.xz')
-wp.generate_outputs('dev-0/in.tsv.xz', 'dev-0/out.tsv')
-wp.generate_outputs('test-A/in.tsv.xz', 'test-A/out.tsv')
+    # wp.create_train_file("train/in.tsv.xz", "train/in.txt")
+    # wp.fill_words("train/in.tsv.xz", "words.txt")