通用词向量模型在文本多目标分类中的性能比较  

PERFORMANCE COMPARISON OF GENERAL WORD VECTOR MODEL IN TEXT MULTI-OBJECTIVE CLASSIFICATION

作　　者：王德志 梁俊艳[2] Wang Dezhi;Liang Junyan(Department of Computer Engineering,North China Institute of Science and Technology,Langfang 065201,Hebei,China;Library,North China Institute of Science and Technology,Langfang 065201,Hebei,China)

机构地区：[1]华北科技学院计算机学院,河北廊坊065201 [2]华北科技学院图书馆,河北廊坊065201

出　　处：《计算机应用与软件》2022年第7期188-194,共7页Computer Applications and Software

基　　金：国家重点研发计划项目(2018YFC0808306);河北省物联网监控工程技术研究中心项目(3142018055)。

摘　　要：针对通用词向量模型在文本多目标分类中的不同性能评价比较问题,基于微博灾害数据集,设计四种多目标分类神经网络模型。通过实验,对比分析同一词向量模型在不同分类模型中的性能差异;分析不同词向量模型在分类模型中的性能特点;对模型训练时间和测试准确性进行分析。实验结果表明,Word2vec模型在CNN和LSTM网络模型中针对微博灾害数据准确率最高。Aiming at the problem of different performance evaluation and comparison of general word vector model in text multi-objective classification, we design four neural network models with multi-objective classification based on Weibo disaster data set. Through experiments, we compared and analyzed the performance differences of the same word vector model in different classification models. We analyzed the performance characteristics of different word vector models in classification models. And then the training time and testing accuracy of the model were analyzed. The experimental results show that Word2 vec model has the highest accuracy for microblog disaster data in CNN and LSTM network models.

关 键 词：词向量模型 情感分类 灾害数据集 自然语言处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]