Analysis of Several Influencing Factors on Word2Vec Text Classification Effect
The Word2Vec vector model has numerous parameters,and its classification effect varies in different scenarios.It is necessary to analyze its influencing factors.Starting from the basic principles of the Word2Vec model,this paper analyzes and discusses the impact of three major factors of pre trained corpus,pre trained parameters of word vectors,and classification model parameters on the model's classification effect.The results indicate that the effect of limited domain prediction is better than that of wide domain prediction.And the larger the vector dimension in the pre trained parameters,the better the effect.There is an optimal value in window size,and the classification algorithm has little impact.The learning rate,activation function and batch size of the classification model parameters have a greater impact on the classification effect of the model,and the training round is relatively small.
国家科技期刊平台
NSTL
万方数据
维普
