基于大数据南昌天气微博排名神经网络模型建构研究
Research on Neural Network Model Construction of Nanchang Weather Microblog Ranking Based on Big Data
陈丽红 1甘广绩 2吴菲菲1
作者信息
- 1. 南昌市气象局,330038,南昌
- 2. 成均馆大学计算机学院,03063,首尔
- 折叠
摘要
@南昌天气微博 2011 年 9 月开通以来,为推送天气预报产品、讲好南昌气象故事做了大量工作,赢得气象政务微博 15 名左右、江西前 10 的佳绩.但随着名次前移,排名增速放缓趋势日益明显.新浪微博排名指标权值非透明,努力方向迷茫.为了解决指标权值不透明导致排名递进难的问题,构建了神经网络模型,并通过逐次训练,得到指标调整前后的稳定神经网络排名预测模型.其中,评论量、点赞量、用户访问量、播放量、转发量为影响排名的重要指标,更新后排名指标权重向峰值指标倾斜明显.经过检验,模型预测准确率符合预期,值得从业者关注.
Abstract
Since the opening of@Nanchang weather microblog in September 2011,it had been done a lot of work to push weather forecast products and tell good Nanchang weather stories,and won a-bout 15 meteorological government microblogs and the top 10 in Jiangxi.However,as the ranking moved forward,the slowdown in ranking growth was becoming increasingly obvious.The weight of Sina Weibo ranking indicators was not transparent,and the direction of efforts was unclear.In order to solve the problem of opaque indicator weights causing difficulty in ranking progression,a neural network model was constructed and trained step by step to obtain a stable neural network ranking prediction model before and after indicator adjustment.Among them,comments,likes,user visits,playback,and forwarding were important influencing indicators that affected ranking.After upda-ting,the weight of ranking indicators tilted significantly towards the peak indicator.After testing,the accuracy of the model prediction met expectations and was worth the attention of practitioners.
关键词
神经网络/微博排名/影响指标/大数据分析Key words
neural network/microblog ranking/impact indicator/big data analysis引用本文复制引用
基金项目
江西省气象局软科学研究计划(赣气软科验字[2022]第3号)
南昌市气象局科研项目(洪气发[2023]14号)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据