Word2vec的工作原理及应用探究

Exploration of the Working Principle and Application of Word2vec

研究了Word2vec的工作原理及应用,明确了统计语言模型的关键问题,分析了词向量的特点,并对神经网络语言模型、Log_Linear模型和Log_Bilinear模型的基本原理进行了探讨,对Word2vec词向量训练框架的工作原理进行了详细分析,推导出了训练模型的目标函数,介绍了Word2vec工程的主要文件和训练参数,并将Word2vec应用于中文词向量的训练。

This paper studies the working principle and application of Word2vec,defines the key problems of statistical language model, analyzes the characteristics of word vector, probes into the basic principles of neural network language model,Log_Linear model and Log_Bilinear model,makes a detailed analysi on the working principle of word vector’s training framework of word2vec, and derives the objective functions of the training models,and introduces the main files in Word2vec project and training parameters, and applies Word2vec into the training of Chinese word vector.