基于图书销售数据的出版选题预测应用研究

发布时间：2018-06-05 13:52

本文选题：选题预测 + 数据挖掘　；参考：《北方工业大学》2017年硕士论文

【摘要】：近年来,数据挖掘技术的广泛研究与应用引起了各界的极大关注,各行各业都迫切需要将企业数据转换成有用的信息和知识。图书出版行业在挑战与机遇中发展,互联网正在冲击着包括出版产业在内的各行各业,与此同时也为出版产业带来新的契机,出版产业必将不断融合新型计算机技术,并最终诞生全新的产业模式和产业链条,图书出版业采用先进数据挖掘技术势在必行。针对出版选题策划依赖主观经验的问题,对用户需求与当前社会热点调研不准确的潜在问题,通过对图书销售市场的研究分析,根据图书销售市场短期的波动性与中长期的周期性特点,提出了综合时间序列预测算法与神经网络算法分别对中长期与短期选题进行预测。中长期预测方面,采用Holt-Winters时间序列预测模型按图书类别预测图书销量,为出版单位做出合理的选题类别策划提供依据,辅助以选题耗费时间预测,避免出版单位因延误销售旺季造成巨大损失,减少图书出版过程中大量人、财、物力的不必要消耗。短期预测方面,采用神经网络模型对各地域的指定图书选题提供印刷量的预测,其中通过对作者热度加权的改进模型提高对热门事件的预测精准度。具体采用JSOUP框架爬取新浪微博热搜作者与热搜词语,构建热搜信息库,对作者信息与内容信息通过热点判断后,进行热门作者与热门内容加权。更加精准的把握类似"诺奖效应"带来的巨大收益,辅助图书选题工作人员在印刷量安排上做出正确的判断。本课题设计了基于图书销售数据的选题预测微信公众账号,为出版单位工作人员提供可靠的选题类别预测与印刷量预测,通过选题预测可有效把控市场规律,迎合用户消费倾向,有效减少错过最佳销售时机与印刷量分配不均的情况,避免造成库存积压导致大量人力、物力与财力的消耗,有效的提高出版单位经济效益。
[Abstract]:In recent years, the extensive research and application of data mining technology has aroused great concern from all walks of life. All kinds of industries urgently need to convert enterprise data into useful information and knowledge. The book publishing industry is developing in challenges and opportunities. The Internet is impacting all kinds of industries, including the publishing industry. At the same time, it also brings new opportunities for the publishing industry, and the publishing industry will continue to integrate new computer technology. Finally, a brand-new industrial model and industrial chain were born. It is imperative for the book publishing industry to adopt advanced data mining technology. In view of the problem that the topic planning of publishing depends on subjective experience and the potential problem that the user's demand and the current social hot spot investigation are not accurate, through the research and analysis of the book sales market, According to the short-term volatility of book sales market and the periodicity of medium and long term, a comprehensive time series prediction algorithm and a neural network algorithm are proposed to predict the short term and long term topics, respectively. In the aspect of medium and long term prediction, the Holt-Winters time series model is used to predict the book sales according to the book category, which provides the basis for the publishing unit to make a reasonable topic selection planning, and assists in the time-consuming prediction with the topic selection. In order to avoid the huge loss caused by the late sales season, the publishing unit will reduce the unnecessary consumption of a large number of people, money and material resources in the course of book publishing. In the aspect of short-term prediction, the neural network model is used to predict the number of selected titles of selected books in various regions, in which the accuracy of prediction of hot events is improved by the improved model of the author's heat weighting. The author and the hot search words of Sina Weibo are crawled by JSOUP frame, and the hot search information database is constructed. After the author's information and content information are judged by hot spot, the hot author and hot content are weighted. More accurate grasp similar to the "Nobel Prize effect" bring huge benefits, assist book selection staff in the print volume arrangements to make a correct judgment. In this paper, the author designs a public account of subject selection prediction based on book sales data, which can provide reliable prediction of subject selection category and print quantity for the staff of publishing units, and can effectively control the market rules through the topic selection prediction. In order to meet the consumption tendency of users, it can effectively reduce the situation of missing the best sales opportunity and the uneven distribution of printing quantity, avoid causing a large amount of manpower, material and financial resources to be consumed as a result of overstocking of stock, and effectively improve the economic benefits of publishing units.
【学位授予单位】：北方工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F274;TP311.52;TP183

【参考文献】