一种基于DA_FASTTEXT的文档分类研究  

Study on Document Classification Based on DA_FASTTEXT

作　　者：王栋平 穆宁 王峥[2] 张晓燕 WANG Dong-ping;MU Ning;WANG Zheng;ZHANG Xiao-yan(Nanjing Fiberhome Starrysky Co.,Ltd.,Nanjing 210019,China;Fiber Home Telecommunication Technologies Co.,Ltd.,Wuhan 430205,China)

机构地区：[1]南京烽火星空通信发展有限公司,南京210019 [2]烽火通信科技股份有限公司,武汉430205

出　　处：《价值工程》2025年第6期145-147,共3页Value Engineering

基　　金：国家重点研发计划(2023YFB3106900);江苏省科技厅产业前瞻与关键核心技术重点项目(BE2023004)。

摘　　要：传统文档分类系统都是基于文档的词属性,利用庞大的词典支持和复杂的切词处理实现文档分类,导致很难兼顾分类准确性和分类速度。本文研究基于FASTTEXT算法的中文文档分类,使中文文档分类在保证分类准确性的同时还降低了时间开销,同时利用蜻蜓优化算法(DA)实现FASTTEXT参数优化,解决FASTTEXT算法参数过多、模型效果依赖于参数值的设定优化问题。本文提出了DA_FASTTEXT分类方法,实现了一个基于DA_FASTTEXT分类方法的中文文档分类系统。测试结果表明其具有更优的分类准确性和分类速度综合性能。Traditional document classification systems rely on the lexical attributes of documents,utilizing extensive dictionaries and intricate word segmentation processes.This often makes it challenging to balance both classification accuracy and speed.This paper focuses on Chinese document classification based on the FASTTEXT algorithm,which not only ensures classification accuracy but also reduces time costs.Additionally,the Dragonfly Algorithm(DA)is employed to optimize the parameters of FASTTEXT,solving the problem of excessive parameters in the FASTTEXT algorithm and the dependence of model performance on parameter value settings.This article proposes the DA_FASTTEXT classification method and implements a Chinese document classification system based on the DA_FASTTEXT classification method.Testing results demonstrate that this document classification system exhibits superior overall performance in terms of classification accuracy and speed.

关 键 词：文档分类 蜻蜓优化算法(DA) 参数优化 FASTTEXT算法 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]