首页|期刊导航|四川轻化工大学学报(自然科学版)|基于乡村自媒体的流量建模与数据挖掘

基于乡村自媒体的流量建模与数据挖掘OA

Traffic Modeling and Data Mining Based on Rural Self-media

中文摘要英文摘要

乡村自媒体凭借其成本低、入行门槛低、用户群体广泛的特点,在推动乡村经济多元化发展、促进乡村文化的传承与创新等方面展现出巨大的潜力与价值.然而,乡村自媒体存在内容同质化、缺乏创新、创作方向不明确等问题.针对这些问题,本研究通过主成分分析(PCA)实现流量建模,通过潜在狄利克雷分配(LDA)结合BERT构建多源文本融合主题分类模型.结合流量建模和主题分类模型对乡村自媒体数据进行深度挖掘;通过挖掘乡村自媒体主题流量的发展趋势和变化规律,推动乡村自媒体更好地发展.实验结果表明:LDA结合BERT能有效提高作品主题分类模型的准确率,最终准确率为92.15%,较其他模型提高1.3~3.9个百分点.乡村动物相关的自媒体作品关注度较高,该主题中的猫和狗的主题在不同影响力的创作者作品中均表现出最高的关注度.研究为乡村创作者的内容创作提供了数据驱动的优化策略,对乡村数字经济的发展具有重要意义.

Rural self-media,with the characteristics of low cost,minimal entry barriers,and broad user base,demonstrates great potential in promoting both rural economic diversification and cultural preservation as well as innovation.However,it suffers from issues such as content homogeneity,lack of innovation,and unclear creative direction.To address these problems,this study employs Principal Component Analysis(PCA)to construct a traffic model,and combines Latent Dirichlet Allocation(LDA)with BERT to build a multi-source text fusion topic classification model.By integrating the traffic model and the topic classification model,this study conducts in-depth mining of rural self-media data.Through exploring the development trends and variation patterns of topic traffic in rural self-media,the study aims to foster better growth of rural self-media.Experimental results show that LDA combined with BERT effectively improves the accuracy of the content topic classification model,achieving a final accuracy of 92.15%,which is 1.3 to 3.9 percentage points higher than that of other models.Rural self-media content related to animals attracts higher attention,with topics about cats and dogs showing the highest engagement across works created by influencers of varying impact levels.This study provides data-driven optimization strategies for rural content creators and holds significant importance for the development of the rural digital economy.

陈鸿;李伊陶;熊兴中

四川轻化工大学 自动化与信息工程学院,四川 宜宾 644000四川轻化工大学 自动化与信息工程学院,四川 宜宾 644000||智能感知与控制四川省重点实验室,四川 宜宾 644000四川轻化工大学 自动化与信息工程学院,四川 宜宾 644000||智能感知与控制四川省重点实验室,四川 宜宾 644000

信息技术与安全科学

数据挖掘主题建模乡村自媒体深度学习

data miningtopic modelingrural self-mediadeep learning

《四川轻化工大学学报(自然科学版)》 2026 (4)

117-128,12

科技部重大专项项目(2022YFE03050004).

10.11863/j.suse.2026.04.10

评论