本文選題：選題預測 + 數(shù)據(jù)挖掘�。� 參考：《北方工業(yè)大學》2017年碩士論文

【摘要】：近年來,數(shù)據(jù)挖掘技術(shù)的廣泛研究與應(yīng)用引起了各界的極大關(guān)注,各行各業(yè)都迫切需要將企業(yè)數(shù)據(jù)轉(zhuǎn)換成有用的信息和知識。圖書出版行業(yè)在挑戰(zhàn)與機遇中發(fā)展,互聯(lián)網(wǎng)正在沖擊著包括出版產(chǎn)業(yè)在內(nèi)的各行各業(yè),與此同時也為出版產(chǎn)業(yè)帶來新的契機,出版產(chǎn)業(yè)必將不斷融合新型計算機技術(shù),并最終誕生全新的產(chǎn)業(yè)模式和產(chǎn)業(yè)鏈條,圖書出版業(yè)采用先進數(shù)據(jù)挖掘技術(shù)勢在必行。針對出版選題策劃依賴主觀經(jīng)驗的問題,對用戶需求與當前社會熱點調(diào)研不準確的潛在問題,通過對圖書銷售市場的研究分析,根據(jù)圖書銷售市場短期的波動性與中長期的周期性特點,提出了綜合時間序列預測算法與神經(jīng)網(wǎng)絡(luò)算法分別對中長期與短期選題進行預測。中長期預測方面,采用Holt-Winters時間序列預測模型按圖書類別預測圖書銷量,為出版單位做出合理的選題類別策劃提供依據(jù),輔助以選題耗費時間預測,避免出版單位因延誤銷售旺季造成巨大損失,減少圖書出版過程中大量人、財、物力的不必要消耗。短期預測方面,采用神經(jīng)網(wǎng)絡(luò)模型對各地域的指定圖書選題提供印刷量的預測,其中通過對作者熱度加權(quán)的改進模型提高對熱門事件的預測精準度。具體采用JSOUP框架爬取新浪微博熱搜作者與熱搜詞語,構(gòu)建熱搜信息庫,對作者信息與內(nèi)容信息通過熱點判斷后,進行熱門作者與熱門內(nèi)容加權(quán)。更加精準的把握類似"諾獎效應(yīng)"帶來的巨大收益,輔助圖書選題工作人員在印刷量安排上做出正確的判斷。本課題設(shè)計了基于圖書銷售數(shù)據(jù)的選題預測微信公眾賬號,為出版單位工作人員提供可靠的選題類別預測與印刷量預測,通過選題預測可有效把控市場規(guī)律,迎合用戶消費傾向,有效減少錯過最佳銷售時機與印刷量分配不均的情況,避免造成庫存積壓導致大量人力、物力與財力的消耗,有效的提高出版單位經(jīng)濟效益。
[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.
【學位授予單位】：北方工業(yè)大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：F274;TP311.52;TP183

【參考文獻】

相關(guān)期刊論文 前10條

1 林海;康寶中;;基于Holt-Winters時間序列的圖書選題預測模型[J];數(shù)字技術(shù)與應(yīng)用;2017年03期

2 傅祚華;;選題策劃與出版流程[J];中國編輯;2016年02期

3 倫淑嫻;林健;姚顯雙;;基于小世界回聲狀態(tài)網(wǎng)的時間序列預測[J];自動化學報;2015年09期

4 洪鵬;余世明;;基于時間序列分析的自動售貨機銷量預測[J];計算機科學;2015年S1期

5 韓敏;許美玲;穆大蕓;;無核相關(guān)向量機在時間序列預測中的應(yīng)用[J];計算機學報;2014年12期

6 彭帥英;李廣杰;彭文;馬建全;王雪冬;秦勝伍;;基于改進遺傳算法的Holt-Winters模型在采空沉陷預測中的應(yīng)用[J];吉林大學學報(地球科學版);2013年02期

7 李長青;;如何透視圖書銷售數(shù)據(jù)背后的奧秘——圖書銷售數(shù)據(jù)定量分析方法淺探[J];科技與出版;2008年03期

8 蔡鑫;;電信數(shù)據(jù)挖掘數(shù)據(jù)準備過程的規(guī)范化設(shè)計[J];計算機工程;2007年24期

9 劉雨露;;數(shù)據(jù)挖掘在高校學生管理決策中的應(yīng)用模式分析[J];成都信息工程學院學報;2006年03期

10 張穎,楊蘭英;時間數(shù)列分析中的加法模型與乘法模型[J];統(tǒng)計與信息論壇;2005年04期

相關(guān)碩士學位論文 前4條

1 陳東河;基于圖書市場分析的選題方案研究及應(yīng)用[D];北方工業(yè)大學;2016年

2 莫海燕;隨機神經(jīng)網(wǎng)絡(luò)預測模型及金融市場波動和相關(guān)性的研究[D];北京交通大學;2015年

3 沙川;Holt-Winters時間序列模型參數(shù)估計和預測[D];南京大學;2011年

4 吳瑕;數(shù)據(jù)挖掘技術(shù)在教學管理中的應(yīng)用研究[D];哈爾濱工程大學;2007年

，

本文編號：1982206