word2vec.c

//  Copyright 2013 Google Inc. All Rights Reserved.
//
//  Licensed under the Apache License, Version 2.0 (the "License");
//  you may not use this file except in compliance with the License.
//  You may obtain a copy of the License at
//
//      http://www.apache.org/licenses/LICENSE-2.0
//
//  Unless required by applicable law or agreed to in writing, software
//  distributed under the License is distributed on an "AS IS" BASIS,
//  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
//  See the License for the specific language governing permissions and
//  limitations under the License.

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <math.h>
#include <pthread.h>

#define MAX_STRING 100//string 类型的最大长度
#define EXP_TABLE_SIZE 1000//这里是用来求sigmoid函数,使用的是一种近似的求法，
#define MAX_EXP 6//只要求球区间为６的即可
#define MAX_SENTENCE_LENGTH 1000//句子最大长度,及包含词数
#define MAX_CODE_LENGTH 40//huffman过程中对word进行按词频的huffman code,每个词的最大长度为４０，也可理解为树的高度不会超过２０

const int vocab_hash_size = 30000000;  // Maximum 30 * 0.7 = 21M words in the vocabulary

typedef float real;                    // Precision of float numbers

struct vocab_word {
  long long cn;//词频
  int *point;//huffman编码对应内节点的路劲
  char *word, *code, codelen;//次数组，huffman编码，编码长度
};

char train_file[MAX_STRING], output_file[MAX_STRING];//训练文件和输出文件
char save_vocab_file[MAX_STRING], read_vocab_file[MAX_STRING];//保存词汇表和读取词汇表的文件。格式：词　词频
struct vocab_word *vocab;//词动态数组
//binary进行二进制文件读取写入，cbow为连续BAG OF WORDS结构，
//window为窗口大小 
//min_count为词频下限，小于该下限忽略；
//num_threads为线程数（多线程时每个线程负责部分训练文件，即将整个训练文件均分给多个线程，
//多个线程更新所有的参数（更新参数时的读取冲突可以忽略），其他参数等所有线程共享），
//min_reduce对词语进行约减
int binary = 0, cbow = 1, debug_mode = 2, window = 5, min_count = 5, num_threads = 12, min_reduce = 1;
int *vocab_hash;//词的hash表
long long vocab_max_size = 1000, vocab_size = 0, layer1_size = 100;
// train_words 训练的单词总数（词频累加）
// word_count_actual 已经训练完的word个数
// file_size 训练文件大小，ftell得到
// classes 输出word clusters的类别数
long long train_words = 0, word_count_actual = 0, iter = 5, file_size = 0, classes = 0;
real alpha = 0.025, starting_alpha, sample = 1e-3;

//syn0为所有word的vector，syn1为huffman tree中所有内部节点的vector，
//syn1neg为 negative sampling时负类sampling的vector表示，
//expTable为exp（）函数的离散表示，为了节省时间。
real *syn0, *syn1, *syn1neg, *expTable;

clock_t start;
//hs表示hierarchical softmax，即层次化softmax替代原来的softmax来减少计算时间，加速训练，默认不采用；
//negative为negative sampling，默认采用
int hs = 0, negative = 5;
const int table_size = 1e8;

//negative sampling时 的分布table
int *table;


//为negative sampling而建的word按词频的分布，以后生成随机数，从table中抽取到word作为negative samplings
//如当前有3 words，词频为 word 0:10，word 1: 5， word 2:5，则将table可以分成4分，table[0] = 0, table[1]=0, table[2]=1, table[3]=2，
//此时table中word分布满足其词频分布，后续用于抽样；在下面的实现中采用词频的幂作为其分布
//可见参考此博文http://blog.csdn.net/itplus/article/details/37998797
//
void InitUnigramTable() {
  int a, i;
  long long train_words_pow = 0;
  real d1, power = 0.75;
  table = (int *)malloc(table_size * sizeof(int));
  for (a = 0; a < vocab_size; a++) train_words_pow += pow(vocab[a].cn, power);
  i = 0;
  d1 = pow(vocab[i].cn, power) / (real)train_words_pow;
  for (a = 0; a < table_size; a++) {
    table[a] = i;
    if (a / (real)table_size > d1) {
      i++;
      d1 += pow(vocab[i].cn, power) / (real)train_words_pow;
    }
    if (i >= vocab_size) i = vocab_size - 1;
  }
}

// Reads a single word from a file, assuming space + tab + EOL to be word boundaries
//读取一个单词，假设每个单词以空格或者tab或者换行符为结尾
void ReadWord(char *word, FILE *fin) {
  int a = 0, ch;
  while (!feof(fin)) {
    ch = fgetc(fin);
    if (ch == 13) continue;
    if ((ch == ' ') || (ch == '\t') || (ch == '\n')) {
      if (a > 0) {
        if (ch == '\n') ungetc(ch, fin);
        break;
      }
      if (ch == '\n') {
        strcpy(word, (char *)"</s>");
        return;
      } else continue;
    }
    word[a] = ch;
    a++;
    if (a >= MAX_STRING - 1) a--;   // Truncate too long words
  }
  word[a] = 0;
}

// Returns hash value of a word
//单词的hash值
int GetWordHash(char *word) {
  unsigned long long a, hash = 0;
  for (a = 0; a < strlen(word); a++) hash = hash * 257 + word[a];
  hash = hash % vocab_hash_size;
  return hash;
}

// Returns position of a word in the vocabulary; if the word is not found, returns -1
//返回word 在词汇hash表中的的位置
int SearchVocab(char *word) {
  unsigned int hash = GetWordHash(word);
  while (1) {
    if (vocab_hash[hash] == -1) return -1;
    if (!strcmp(word, vocab[vocab_hash[hash]].word)) return vocab_hash[hash];
    hash = (hash + 1) % vocab_hash_size;
  }
  return -1;
}

// Reads a word and returns its index in the vocabulary
//从文件中读取一个单词并返回它在词汇hash表中的下标
int ReadWordIndex(FILE *fin) {
  char word[MAX_STRING];
  ReadWord(word, fin);
  if (feof(fin)) return -1;
  return SearchVocab(word);
}

// Adds a word to the vocabulary
//向词汇表中添加一个单词
int AddWordToVocab(char *word) {
  unsigned int hash, length = strlen(word) + 1;
  if (length > MAX_STRING) length = MAX_STRING;
  vocab[vocab_size].word = (char *)calloc(length, sizeof(char));
  strcpy(vocab[vocab_size].word, word);
  vocab[vocab_size].cn = 0;
  vocab_size++;
  // Reallocate memory if needed 
  if (vocab_size + 2 >= vocab_max_size) {
    vocab_max_size += 1000;
    vocab = (struct vocab_word *)realloc(vocab, vocab_max_size * sizeof(struct vocab_word));
  }
  hash = GetWordHash(word);
  while (vocab_hash[hash] != -1) hash = (hash + 1) % vocab_hash_size;
  vocab_hash[hash] = vocab_size - 1;
  return vocab_size - 1;
}

// Used later for sorting by word counts
//比较两个单词的出现频率
int VocabCompare(const void *a, const void *b) {
    return ((struct vocab_word *)b)->cn - ((struct vocab_word *)a)->cn;
}

// Sorts the vocabulary by frequency using word counts
//对词汇表进行排序，利用单词出现的频率
//并去掉低频词
void SortVocab() {
  int a, size;
  unsigned int hash;
  // Sort the vocabulary and keep </s> at the first position
  //</s> 是一个特殊的字符
  qsort(&vocab[1], vocab_size - 1, sizeof(struct vocab_word), VocabCompare);
  //对重排之后的词汇hash表进行更新
  for (a = 0; a < vocab_hash_size; a++) vocab_hash[a] = -1;
  size = vocab_size;
  train_words = 0;
  //去掉低频词
  for (a = 0; a < size; a++) {
    // Words occuring less than min_count times will be discarded from the vocab
    if ((vocab[a].cn < min_count) && (a != 0)) {
      vocab_size--;
      free(vocab[a].word);
    } else {
      // Hash will be re-computed, as after the sorting it is not actual
      hash=GetWordHash(vocab[a].word);
      while (vocab_hash[hash] != -1) hash = (hash + 1) % vocab_hash_size;
      vocab_hash[hash] = a;
      train_words += vocab[a].cn;
    }
  }
  vocab = (struct vocab_word *)realloc(vocab, (vocab_size + 1) * sizeof(struct vocab_word));
  // Allocate memory for the binary tree construction
  for (a = 0; a < vocab_size; a++) {
    vocab[a].code = (char *)calloc(MAX_CODE_LENGTH, sizeof(char));
    vocab[a].point = (int *)calloc(MAX_CODE_LENGTH, sizeof(int));
  }
}

// Reduces the vocabulary by removing infrequent tokens
//
void ReduceVocab() {
  int a, b = 0;
  unsigned int hash;
  for (a = 0; a < vocab_size; a++) 
    if (vocab[a].cn > min_reduce) {
      vocab[b].cn = vocab[a].cn;
      vocab[b].word = vocab[a].word;
      b++;
    } else free(vocab[a].word);
  vocab_size = b;
  for (a = 0; a < vocab_hash_size; a++) vocab_hash[a] = -1;
  for (a = 0; a < vocab_size; a++) {
    // Hash will be re-computed, as it is not actual
    hash = GetWordHash(vocab[a].word);
    while (vocab_hash[hash] != -1) hash = (hash + 1) % vocab_hash_size;
    vocab_hash[hash] = a;
  }
  //清空输出缓冲区
  fflush(stdout);
  min_reduce++;
}

// Create binary Huffman tree using the word counts
// Frequent words will have short uniqe binary codes
//创建Huffman二叉树
void CreateBinaryTree() {
  long long a, b, i, min1i, min2i, pos1, pos2, point[MAX_CODE_LENGTH];
  char code[MAX_CODE_LENGTH];
  long long *count = (long long *)calloc(vocab_size * 2 + 1, sizeof(long long));
  long long *binary = (long long *)calloc(vocab_size * 2 + 1, sizeof(long long));
  long long *parent_node = (long long *)calloc(vocab_size * 2 + 1, sizeof(long long));
  for (a = 0; a < vocab_size; a++)
      count[a] = vocab[a].cn;
  for (a = vocab_size; a < vocab_size * 2; a++)
      count[a] = 1e15;
  pos1 = vocab_size - 1;
  pos2 = vocab_size;
  // Following algorithm constructs the Huffman tree by adding one node at a time
  //count 数组是从大到小排序
  for (a = 0; a < vocab_size - 1; a++) {
    // First, find two smallest nodes 'min1, min2'
    if (pos1 >= 0) {
      if (count[pos1] < count[pos2]) {
        min1i = pos1;
        pos1--;
      } else {
        min1i = pos2;
        pos2++;
      }
    } else {
      min1i = pos2;
      pos2++;
    }
    if (pos1 >= 0) {
      if (count[pos1] < count[pos2]) {
        min2i = pos1;
        pos1--;
      } else {
        min2i = pos2;
        pos2++;
      }
    } else {
      min2i = pos2;
      pos2++;
    }
    count[vocab_size + a] = count[min1i] + count[min2i];
    parent_node[min1i] = vocab_size + a;
    parent_node[min2i] = vocab_size + a;
    binary[min2i] = 1;
  }
  // Now assign binary code to each vocabulary word
  for (a = 0; a < vocab_size; a++) {
    b = a;
    i = 0;
    while (1) {
      code[i] = binary[b];
      point[i] = b;
      i++;
      b = parent_node[b];
      if (b == vocab_size * 2 - 2) break;//到达根节点
    }
    vocab[a].codelen = i;
    vocab[a].point[0] = vocab_size - 2;
    for (b = 0; b < i; b++) {
      vocab[a].code[i - b - 1] = code[b];
      vocab[a].point[i - b] = point[b] - vocab_size;//这里没看懂
    }
  }
  free(count);
  free(binary);
  free(parent_node);
}

//从训练文件中得到词汇频率表
void LearnVocabFromTrainFile() {
  char word[MAX_STRING];
  FILE *fin;
  long long a, i;
  for (a = 0; a < vocab_hash_size; a++) vocab_hash[a] = -1;
  fin = fopen(train_file, "rb");
  if (fin == NULL) {
    printf("ERROR: training data file not found!\n");
    exit(1);
  }
  //第一个为特殊词汇
  vocab_size = 0;
  AddWordToVocab((char *)"</s>");
  while (1) {
    ReadWord(word, fin);
    if (feof(fin)) break;
    train_words++;
    if ((debug_mode > 1) && (train_words % 100000 == 0)) {
      printf("%lldK%c", train_words / 1000, 13);
      fflush(stdout);
    }
    i = SearchVocab(word);
    if (i == -1) {
      a = AddWordToVocab(word);
      vocab[a].cn = 1;
    } else vocab[i].cn++;
    //如果词汇量太大，需要对低频词汇进行约减
    if (vocab_size > vocab_hash_size * 0.7) ReduceVocab();
  }
  SortVocab();
  if (debug_mode > 0) {
    printf("Vocab size: %lld\n", vocab_size);
    printf("Words in train file: %lld\n", train_words);
  }
  file_size = ftell(fin);
  fclose(fin);
}

//保存词汇频率表
void SaveVocab() {
  long long i;
  FILE *fo = fopen(save_vocab_file, "wb");
  for (i = 0; i < vocab_size; i++) fprintf(fo, "%s %lld\n", vocab[i].word, vocab[i].cn);
  fclose(fo);
}

//读取词汇表
void ReadVocab() {
  long long a, i = 0;
  char c;
  char word[MAX_STRING];
  FILE *fin = fopen(read_vocab_file, "rb");
  if (fin == NULL) {
    printf("Vocabulary file not found\n");
    exit(1);
  }
  //初始化词汇hash表
  for (a = 0; a < vocab_hash_size; a++) vocab_hash[a] = -1;

  vocab_size = 0;
  while (1) {
    ReadWord(word, fin);
    if (feof(fin)) break;
    a = AddWordToVocab(word);
    fscanf(fin, "%lld%c", &vocab[a].cn, &c);
    i++;
  }
  SortVocab();
  if (debug_mode > 0) {
    printf("Vocab size: %lld\n", vocab_size);
    printf("Words in train file: %lld\n", train_words);
  }
  fin = fopen(train_file, "rb");
  if (fin == NULL) {
    printf("ERROR: training data file not found!\n");
    exit(1);
  }
  fseek(fin, 0, SEEK_END);
  file_size = ftell(fin);
  fclose(fin);
}

//
void InitNet() {
  long long a, b;
  unsigned long long next_random = 1;
  //posix_memalign是用来对齐函数
  a = posix_memalign((void **)&syn0, 128, (long long)vocab_size * layer1_size * sizeof(real));
  if (syn0 == NULL) {printf("Memory allocation failed\n"); exit(1);}
  //Hierarchical Softmax 模型
  if (hs) {
    a = posix_memalign((void **)&syn1, 128, (long long)vocab_size * layer1_size * sizeof(real));
    if (syn1 == NULL) {printf("Memory allocation failed\n"); exit(1);}
    for (a = 0; a < vocab_size; a++)
        for (b = 0; b < layer1_size; b++)
            syn1[a * layer1_size + b] = 0;
  }
  //Negative Sampling 模型
  if (negative > 0) {
    a = posix_memalign((void **)&syn1neg, 128, (long long)vocab_size * layer1_size * sizeof(real));
    if (syn1neg == NULL) {printf("Memory allocation failed\n"); exit(1);}
    for (a = 0; a < vocab_size; a++)
        for (b = 0; b < layer1_size; b++)
            syn1neg[a * layer1_size + b] = 0;
  }
  //随机初始化，没怎么看懂
  for (a = 0; a < vocab_size; a++)
      for (b = 0; b < layer1_size; b++) {
        next_random = next_random * (unsigned long long)25214903917 + 11;
        syn0[a * layer1_size + b] = (((next_random & 0xFFFF) / (real)65536) - 0.5) / layer1_size;
    }
  CreateBinaryTree();
}

void *TrainModelThread(void *id) {
   // word 向sen中添加单词用，句子完成后表示句子中的当前单词
  // last_word 上一个单词，辅助扫描窗口
  // sentence_length 当前句子的长度（单词数）
  // sentence_position 当前单词在当前句子中的index
  long long a, b, d, cw, word, last_word, sentence_length = 0, sentence_position = 0;
  // word_count 已训练语料总长度
  // last_word_count 保存值，以便在新训练语料长度超过某个值时输出信息
  // sen 单词数组，表示句子
  long long word_count = 0, last_word_count = 0, sen[MAX_SENTENCE_LENGTH + 1];
  // l1 ns中表示word在concatenated word vectors中的起始位置，之后layer1_size是对应的word vector，因为把矩阵拉成长向量了
  // l2 cbow或ns中权重向量的起始位置，之后layer1_size是对应的syn1或syn1neg，因为把矩阵拉成长向量了
  // c 循环中的计数作用
  // target ns中当前的sample
  // label ns中当前sample的label
  long long l1, l2, c, target, label, local_iter = iter;
  // id 线程创建的时候传入，辅助随机数生成
  unsigned long long next_random = (long long)id;
   // f e^x / (1/e^x)，fs中指当前编码为是0（父亲的左子节点为0，右为1）的概率，ns中指label是1的概率
  // g 误差(f与真实值的偏离)与学习速率的乘积
  real f, g;
  // 当前时间，和start比较计算算法效率
  clock_t now;

  real *neu1 = (real *)calloc(layer1_size, sizeof(real)); // 隐层节点
  real *neu1e = (real *)calloc(layer1_size, sizeof(real)); // 误差累计项，其实对应的是Gneu1
  FILE *fi = fopen(train_file, "rb");
  //基于字节来切分训练语料
  fseek(fi, file_size / (long long)num_threads * (long long)id, SEEK_SET);

  while (1) {
    if (word_count - last_word_count > 10000) {
      word_count_actual += word_count - last_word_count;
      last_word_count = word_count;
      if ((debug_mode > 1)) {
        now=clock();
        printf("%cAlpha: %f  Progress: %.2f%%  Words/thread/sec: %.2fk  ", 13, alpha,
         word_count_actual / (real)(iter * train_words + 1) * 100,
         word_count_actual / ((real)(now - start + 1) / (real)CLOCKS_PER_SEC * 1000));
        fflush(stdout);
      }
      alpha = starting_alpha * (1 - word_count_actual / (real)(iter * train_words + 1)); // 自动调整学习速率
      if (alpha < starting_alpha * 0.0001) alpha = starting_alpha * 0.0001; //学习速率下限
    }
    //当前句子长度为0
    if (sentence_length == 0) {
      while (1) {
        word = ReadWordIndex(fi);
        if (feof(fi)) break;  //文件结尾
        if (word == -1) continue; //单词不存在
        word_count++;
        if (word == 0) break; //是特殊词</s>
        // The subsampling randomly discards frequent words while keeping the ranking same
        // 这里的亚采样是指 Sub-Sampling，Mikolov 在论文指出这种亚采样能够带来 2 到 10 倍的性能提升，并能够提升低频词的表示精度。
        // 低频词被丢弃概率低，高频词被丢弃概率高
        //具体参考博文http://blog.csdn.net/itplus/article/details/37998797
        if (sample > 0) {
          real ran = (sqrt(vocab[word].cn / (sample * train_words)) + 1) * (sample * train_words) / vocab[word].cn;
          next_random = next_random * (unsigned long long)25214903917 + 11;
          if (ran < (next_random & 0xFFFF) / (real)65536) continue;
        }
        sen[sentence_length] = word;
        sentence_length++;
        if (sentence_length >= MAX_SENTENCE_LENGTH) break;
      }
      sentence_position = 0;  // 当前单词在当前句中的index，起始值为0
    }

    //读到文件末尾
    if (feof(fi) || (word_count > train_words / num_threads)) {
      word_count_actual += word_count - last_word_count;
      local_iter--;
      if (local_iter == 0) break;
      word_count = 0;
      last_word_count = 0;
      sentence_length = 0;
      fseek(fi, file_size / (long long)num_threads * (long long)id, SEEK_SET);
      continue;
    }

    word = sen[sentence_position];
    if (word == -1) continue;
    for (c = 0; c < layer1_size; c++) neu1[c] = 0;
    for (c = 0; c < layer1_size; c++) neu1e[c] = 0;
    next_random = next_random * (unsigned long long)25214903917 + 11;
    b = next_random % window; //随机取窗口

    if (cbow) {  //train the cbow architecture
      // in -> hidden
      //输入层到隐藏层
      cw = 0;
      for (a = b; a < window * 2 + 1 - b; a++) if (a != window) {
        c = sentence_position - window + a;
        if (c < 0) continue;
        if (c >= sentence_length) continue;
        last_word = sen[c];
        if (last_word == -1) continue;
        //cbow 将上下文词的vector 相加
        for (c = 0; c < layer1_size; c++) neu1[c] += syn0[c + last_word * layer1_size];
        cw++;
      }

      if (cw) {
        //归一化，上下文词的个数
        for (c = 0; c < layer1_size; c++) neu1[c] /= cw;
        //hierarchical softmax
        //参考博文http://blog.csdn.net/itplus/article/details/37969519
        if (hs){
          for (d = 0; d < vocab[word].codelen; d++) {
            f = 0;
            l2 = vocab[word].point[d] * layer1_size;
            // Propagate hidden -> output
            for (c = 0; c < layer1_size; c++) 
              f += neu1[c] * syn1[c + l2];
            if (f <= -MAX_EXP) continue;
            else if (f >= MAX_EXP) continue;
            else f = expTable[(int)((f + MAX_EXP) * (EXP_TABLE_SIZE / MAX_EXP / 2))];
            // 'g' is the gradient multiplied by the learning rate
            // g 是梯度乘以学习速率
            g = (1 - vocab[word].code[d] - f) * alpha;
            // Propagate errors output -> hidden
            // 累计误差率
            for (c = 0; c < layer1_size; c++) neu1e[c] += g * syn1[c + l2];
            // Learn weights hidden -> output
            // 更新参数权重值
            for (c = 0; c < layer1_size; c++) syn1[c + l2] += g * neu1[c];
          }
        }
        // NEGATIVE SAMPLING
        // 负采样 算法
        if (negative > 0) 
        {
          for (d = 0; d < negative + 1; d++) {
            if (d == 0) {
              target = word;
              label = 1;
            } else {
              //进行随机采样
              next_random = next_random * (unsigned long long)25214903917 + 11;
              target = table[(next_random >> 16) % table_size];
              if (target == 0) target = next_random % (vocab_size - 1) + 1;
              if (target == word) continue;
              label = 0;
            }
            l2 = target * layer1_size;
            f = 0;
            for (c = 0; c < layer1_size; c++) f += neu1[c] * syn1neg[c + l2];
            if (f > MAX_EXP) g = (label - 1) * alpha;
            else if (f < -MAX_EXP) g = (label - 0) * alpha;
            // 梯度 乘以 学习速率
            else g = (label - expTable[(int)((f + MAX_EXP) * (EXP_TABLE_SIZE / MAX_EXP / 2))]) * alpha;
            // 累计误差
            for (c = 0; c < layer1_size; c++) neu1e[c] += g * syn1neg[c + l2];
            // 更新权值
            for (c = 0; c < layer1_size; c++) syn1neg[c + l2] += g * neu1[c];
          }
        }
          // hidden -> in
        for (a = b; a < window * 2 + 1 - b; a++) if (a != window) {
          c = sentence_position - window + a;
          if (c < 0) continue;
          if (c >= sentence_length) continue;
          last_word = sen[c];
          if (last_word == -1) continue;
          // 更新词向量
          for (c = 0; c < layer1_size; c++) syn0[c + last_word * layer1_size] += neu1e[c];
        }
      }
    } else {  //train skip-gram
      for (a = b; a < window * 2 + 1 - b; a++) if (a != window) {
        c = sentence_position - window + a;
        if (c < 0) continue;
        if (c >= sentence_length) continue;
        last_word = sen[c];
        if (last_word == -1) continue;
        l1 = last_word * layer1_size;
        for (c = 0; c < layer1_size; c++) neu1e[c] = 0;
        // HIERARCHICAL SOFTMAX
        if (hs) for (d = 0; d < vocab[word].codelen; d++) {
          f = 0;
          l2 = vocab[word].point[d] * layer1_size;
          // Propagate hidden -> output
          for (c = 0; c < layer1_size; c++) f += syn0[c + l1] * syn1[c + l2];
          // 不在expTable内的舍弃掉，有网友发邮件问过作者，作者回答说计算精度有限，怕有不好印象
          // 可以改成太小的都是0，太大的都是1，运行结果还是有差别的
          //参考注释 http://songchengru.eicp.net/code/word2vec.html
          if (f <= -MAX_EXP) continue;
          else if (f >= MAX_EXP) continue;
          else f = expTable[(int)((f + MAX_EXP) * (EXP_TABLE_SIZE / MAX_EXP / 2))];
          // 'g' is the gradient multiplied by the learning rate
          g = (1 - vocab[word].code[d] - f) * alpha;
          // Propagate errors output -> hidden
          //记录累计误差
          for (c = 0; c < layer1_size; c++) neu1e[c] += g * syn1[c + l2];
          // Learn weights hidden -> output
          // 
          for (c = 0; c < layer1_size; c++) syn1[c + l2] += g * syn0[c + l1];
        }
        // NEGATIVE SAMPLING
        if (negative > 0) for (d = 0; d < negative + 1; d++) {
          if (d == 0) {
            target = word;
            label = 1;
          } else {
            next_random = next_random * (unsigned long long)25214903917 + 11;
            target = table[(next_random >> 16) % table_size];
            if (target == 0) target = next_random % (vocab_size - 1) + 1;
            if (target == word) continue;
            label = 0;
          }
          l2 = target * layer1_size;
          f = 0;
          for (c = 0; c < layer1_size; c++) f += syn0[c + l1] * syn1neg[c + l2];
          if (f > MAX_EXP) g = (label - 1) * alpha;
          else if (f < -MAX_EXP) g = (label - 0) * alpha;
          else g = (label - expTable[(int)((f + MAX_EXP) * (EXP_TABLE_SIZE / MAX_EXP / 2))]) * alpha;
          for (c = 0; c < layer1_size; c++) neu1e[c] += g * syn1neg[c + l2];
          for (c = 0; c < layer1_size; c++) syn1neg[c + l2] += g * syn0[c + l1];
        }
        // Learn weights input -> hidden
        for (c = 0; c < layer1_size; c++) syn0[c + l1] += neu1e[c];
      }
    }
    sentence_position++;
    if (sentence_position >= sentence_length) {
      sentence_length = 0;
      continue;
    }
  }
  fclose(fi);
  free(neu1);
  free(neu1e);
  pthread_exit(NULL);
}

//
void TrainModel() {
  long a, b, c, d;
  FILE *fo;
  pthread_t *pt = (pthread_t *)malloc(num_threads * sizeof(pthread_t));
  printf("Starting training using file %s\n", train_file);
  //开始的学习速率
  starting_alpha = alpha;
  //词向量初始化
  if (read_vocab_file[0] != 0) ReadVocab(); else LearnVocabFromTrainFile();
  if (save_vocab_file[0] != 0) SaveVocab();

  if (output_file[0] == 0) return;
  //
  InitNet();
  //对于负采样，带全采样初始化表
  if (negative > 0) InitUnigramTable();
  //
  start = clock();
  for (a = 0; a < num_threads; a++) pthread_create(&pt[a], NULL, TrainModelThread, (void *)a);
  for (a = 0; a < num_threads; a++) pthread_join(pt[a], NULL);
  fo = fopen(output_file, "wb");
  if (classes == 0) {
    // Save the word vectors
    fprintf(fo, "%lld %lld\n", vocab_size, layer1_size);
    for (a = 0; a < vocab_size; a++) {
      fprintf(fo, "%s ", vocab[a].word);
      if (binary) for (b = 0; b < layer1_size; b++)
        fwrite(&syn0[a * layer1_size + b], sizeof(real), 1, fo);
      else for (b = 0; b < layer1_size; b++)
        fprintf(fo, "%lf ", syn0[a * layer1_size + b]);
      fprintf(fo, "\n");
    }
  } else {
    // Run K-means on the word vectors
    int clcn = classes, iter = 10, closeid;
    int *centcn = (int *)malloc(classes * sizeof(int));
    int *cl = (int *)calloc(vocab_size, sizeof(int));
    real closev, x;
    real *cent = (real *)calloc(classes * layer1_size, sizeof(real));

    for (a = 0; a < vocab_size; a++) cl[a] = a % clcn;

    for (a = 0; a < iter; a++) {
      for (b = 0; b < clcn * layer1_size; b++) cent[b] = 0;
      for (b = 0; b < clcn; b++) centcn[b] = 1;
      for (c = 0; c < vocab_size; c++) {
        for (d = 0; d < layer1_size; d++) cent[layer1_size * cl[c] + d] += syn0[c * layer1_size + d];
        centcn[cl[c]]++;
      }
      for (b = 0; b < clcn; b++) {
        closev = 0;
        for (c = 0; c < layer1_size; c++) {
          cent[layer1_size * b + c] /= centcn[b];
          closev += cent[layer1_size * b + c] * cent[layer1_size * b + c];
        }
        closev = sqrt(closev);
        for (c = 0; c < layer1_size; c++) cent[layer1_size * b + c] /= closev;
      }
      for (c = 0; c < vocab_size; c++) {
        closev = -10;
        closeid = 0;
        for (d = 0; d < clcn; d++) {
          x = 0;
          for (b = 0; b < layer1_size; b++) x += cent[layer1_size * d + b] * syn0[c * layer1_size + b];
          if (x > closev) {
            closev = x;
            closeid = d;
          }
        }
        cl[c] = closeid;
      }
    }
    // Save the K-means classes
    for (a = 0; a < vocab_size; a++) fprintf(fo, "%s %d\n", vocab[a].word, cl[a]);
    free(centcn);
    free(cent);
    free(cl);
  }
  fclose(fo);
}

//寻找参数，通过对比str和argv[a]，没有返回-1
int ArgPos(char *str, int argc, char **argv) {
  int a;
  for (a = 1; a < argc; a++) if (!strcmp(str, argv[a])) {
    if (a == argc - 1) {
      printf("Argument missing for %s\n", str);
      exit(1);
    }
    return a;
  }
  return -1;
}

/*主程序*/
int main(int argc, char **argv) {
  int i;
  if (argc == 1) {
    printf("WORD VECTOR estimation toolkit v 0.1c\n\n");
    printf("Options:\n");
    printf("Parameters for training:\n");
    printf("\t-train <file>\n");
    printf("\t\tUse text data from <file> to train the model\n");
    printf("\t-output <file>\n");
    printf("\t\tUse <file> to save the resulting word vectors / word clusters\n");
    printf("\t-size <int>\n");
    printf("\t\tSet size of word vectors; default is 100\n");
    printf("\t-window <int>\n");
    printf("\t\tSet max skip length between words; default is 5\n");
    printf("\t-sample <float>\n");
    printf("\t\tSet threshold for occurrence of words. Those that appear with higher frequency in the training data\n");
    printf("\t\twill be randomly down-sampled; default is 1e-3, useful range is (0, 1e-5)\n");
    printf("\t-hs <int>\n");
    printf("\t\tUse Hierarchical Softmax; default is 0 (not used)\n");
    printf("\t-negative <int>\n");
    printf("\t\tNumber of negative examples; default is 5, common values are 3 - 10 (0 = not used)\n");
    printf("\t-threads <int>\n");
    printf("\t\tUse <int> threads (default 12)\n");
    printf("\t-iter <int>\n");
    printf("\t\tRun more training iterations (default 5)\n");
    printf("\t-min-count <int>\n");
    printf("\t\tThis will discard words that appear less than <int> times; default is 5\n");
    printf("\t-alpha <float>\n");
    printf("\t\tSet the starting learning rate; default is 0.025 for skip-gram and 0.05 for CBOW\n");
    printf("\t-classes <int>\n");
    printf("\t\tOutput word classes rather than word vectors; default number of classes is 0 (vectors are written)\n");
    printf("\t-debug <int>\n");
    printf("\t\tSet the debug mode (default = 2 = more info during training)\n");
    printf("\t-binary <int>\n");
    printf("\t\tSave the resulting vectors in binary moded; default is 0 (off)\n");
    printf("\t-save-vocab <file>\n");
    printf("\t\tThe vocabulary will be saved to <file>\n");
    printf("\t-read-vocab <file>\n");
    printf("\t\tThe vocabulary will be read from <file>, not constructed from the training data\n");
    printf("\t-cbow <int>\n");
    printf("\t\tUse the continuous bag of words model; default is 1 (use 0 for skip-gram model)\n");
    printf("\nExamples:\n");
    printf("./word2vec -train data.txt -output vec.txt -size 200 -window 5 -sample 1e-4 -negative 5 -hs 0 -binary 0 -cbow 1 -iter 3\n\n");
    return 0;
  }
  output_file[0] = 0;
  save_vocab_file[0] = 0;
  read_vocab_file[0] = 0;
  if ((i = ArgPos((char *)"-size", argc, argv)) > 0) layer1_size = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-train", argc, argv)) > 0) strcpy(train_file, argv[i + 1]);
  if ((i = ArgPos((char *)"-save-vocab", argc, argv)) > 0) strcpy(save_vocab_file, argv[i + 1]);
  if ((i = ArgPos((char *)"-read-vocab", argc, argv)) > 0) strcpy(read_vocab_file, argv[i + 1]);
  if ((i = ArgPos((char *)"-debug", argc, argv)) > 0) debug_mode = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-binary", argc, argv)) > 0) binary = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-cbow", argc, argv)) > 0) cbow = atoi(argv[i + 1]);
  if (cbow) alpha = 0.05;
  if ((i = ArgPos((char *)"-alpha", argc, argv)) > 0) alpha = atof(argv[i + 1]);
  if ((i = ArgPos((char *)"-output", argc, argv)) > 0) strcpy(output_file, argv[i + 1]);
  if ((i = ArgPos((char *)"-window", argc, argv)) > 0) window = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-sample", argc, argv)) > 0) sample = atof(argv[i + 1]);
  if ((i = ArgPos((char *)"-hs", argc, argv)) > 0) hs = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-negative", argc, argv)) > 0) negative = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-threads", argc, argv)) > 0) num_threads = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-iter", argc, argv)) > 0) iter = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-min-count", argc, argv)) > 0) min_count = atoi(argv[i + 1]);
  if ((i = ArgPos((char *)"-classes", argc, argv)) > 0) classes = atoi(argv[i + 1]);
  vocab = (struct vocab_word *)calloc(vocab_max_size, sizeof(struct vocab_word));
  vocab_hash = (int *)calloc(vocab_hash_size, sizeof(int));
  expTable = (real *)malloc((EXP_TABLE_SIZE + 1) * sizeof(real));
  //EXP_TSBLE_SIZE = 1000，MAX_EXP = 6
  //将[-６，６]均分成1000等份,
  for (i = 0; i < EXP_TABLE_SIZE; i++) {
    expTable[i] = exp((i / (real)EXP_TABLE_SIZE * 2 - 1) * MAX_EXP); // Precompute the exp() table
    expTable[i] = expTable[i] / (expTable[i] + 1);                   // Precompute f(x) = x / (x + 1)
  }
  TrainModel();
  return 0;
}