SimRecall/predict.py

#coding utf-8
import sys
from gensim import models
import numpy as np

if __name__=="__main__":
    #model = models.word2vec.Word2Vec.load('word2vec.txt')
    #print(model.wx)
    f1 = open('word2vec.txt')
    word_dict = {}
    while True:
        line = f1.readline()
        if not line:
            break
        items = line.strip().split(" ")
        if len(items)<64:
            continue
        arr = []
        for w in items[1:]:
            arr.append(float(w))
        word_dict[items[0]] = arr
    #print(word_dict)
    f  = open(sys.argv[1])
    num = 0
    while True:
        line = f.readline()
        if not line:
            break
        num = num+1
        if num == 1:
            continue
        items = line.split("\t")
        if len(items)<2:
            continue
        vid = items[0]
        title_arr = items[1].split(" ")
        title_info = np.zeros(64)
        word_len = 0
        for word in title_arr:
            if word in word_dict:
                 #print(title_info)
                 #print(word)
                 word_vec = word_dict[word]
                 #print(word_vec)
                 title_info = np.add(title_info, word_vec)
                 word_len +=1
        #print(title_info)
        title_info_list = []
        if word_len<=0:
            continue
        for j in title_info:
            title_info_list.append(j/word_len)
        #print("title_info_list:", title_info_list)
        print(vid,"\t",title_info_list)