基于能量?jī)?yōu)化的微博用戶轉(zhuǎn)發(fā)行為預(yù)測(cè)
發(fā)布時(shí)間:2018-11-08 18:00
【摘要】:微博用戶轉(zhuǎn)發(fā)行為預(yù)測(cè)是微博社交網(wǎng)絡(luò)消息擴(kuò)散模型構(gòu)建的基礎(chǔ),在輿情監(jiān)控、市場(chǎng)營(yíng)銷(xiāo)與政治選舉等領(lǐng)域有著廣泛的應(yīng)用.為了提高用戶轉(zhuǎn)發(fā)行為預(yù)測(cè)的精度,本文在MRF(Markov Random Field)能量?jī)?yōu)化框架下綜合分析了用戶屬性與微博內(nèi)容特征、用戶轉(zhuǎn)發(fā)行為約束、群體轉(zhuǎn)發(fā)先驗(yàn)等因素對(duì)用戶轉(zhuǎn)發(fā)行為的影響,并在邏輯回歸模型的基礎(chǔ)上構(gòu)造了相應(yīng)的能量函數(shù)對(duì)用戶轉(zhuǎn)發(fā)行為進(jìn)行了全局性的預(yù)測(cè).實(shí)驗(yàn)結(jié)果表明,微博用戶轉(zhuǎn)發(fā)行為不僅取決于用戶屬性、微博內(nèi)容等特征,而且也受到用戶轉(zhuǎn)發(fā)行為約束、群體轉(zhuǎn)發(fā)先驗(yàn)等因素不同程度的影響.相對(duì)于傳統(tǒng)算法,本文算法可以更準(zhǔn)確地對(duì)用戶轉(zhuǎn)發(fā)行為進(jìn)行建模,因而可獲得更好的預(yù)測(cè)結(jié)果.
[Abstract]:Weibo's user forwarding behavior prediction is the basis for building a social network message diffusion model, which is widely used in the fields of public opinion monitoring, marketing and political election. In order to improve the accuracy of user forwarding behavior prediction, the user attributes and Weibo content characteristics, user forwarding behavior constraints are comprehensively analyzed in this paper under the framework of MRF (Markov Random Field) energy optimization. Based on the logical regression model, a corresponding energy function is constructed to predict the user forwarding behavior globally. The experimental results show that the user forwarding behavior of Weibo is not only dependent on the characteristics of user attributes, Weibo content, but also affected by user forwarding behavior constraints and group forwarding prior factors. Compared with the traditional algorithm, the proposed algorithm can model the user forwarding behavior more accurately, so that better prediction results can be obtained.
【作者單位】: 周口師范學(xué)院網(wǎng)絡(luò)工程學(xué)院;
【基金】:國(guó)家自然科學(xué)基金(No.U1404620,No.U1404622) 河南省自然科學(xué)基金(No.162300410347) 河南省科技攻關(guān)項(xiàng)目(No.172102310727,No.162102310589,No.162102210396,No.162102310590) 河南省高校重點(diǎn)科研項(xiàng)目(No.17A520018,No.17A520019,No.15A520116,No.16B520034,No.16A520105) 周口師范學(xué)院高層次人才科研啟動(dòng)基金(No.zknuc2015103)
【分類(lèi)號(hào)】:TP393.092
本文編號(hào):2319256
[Abstract]:Weibo's user forwarding behavior prediction is the basis for building a social network message diffusion model, which is widely used in the fields of public opinion monitoring, marketing and political election. In order to improve the accuracy of user forwarding behavior prediction, the user attributes and Weibo content characteristics, user forwarding behavior constraints are comprehensively analyzed in this paper under the framework of MRF (Markov Random Field) energy optimization. Based on the logical regression model, a corresponding energy function is constructed to predict the user forwarding behavior globally. The experimental results show that the user forwarding behavior of Weibo is not only dependent on the characteristics of user attributes, Weibo content, but also affected by user forwarding behavior constraints and group forwarding prior factors. Compared with the traditional algorithm, the proposed algorithm can model the user forwarding behavior more accurately, so that better prediction results can be obtained.
【作者單位】: 周口師范學(xué)院網(wǎng)絡(luò)工程學(xué)院;
【基金】:國(guó)家自然科學(xué)基金(No.U1404620,No.U1404622) 河南省自然科學(xué)基金(No.162300410347) 河南省科技攻關(guān)項(xiàng)目(No.172102310727,No.162102310589,No.162102210396,No.162102310590) 河南省高校重點(diǎn)科研項(xiàng)目(No.17A520018,No.17A520019,No.15A520116,No.16B520034,No.16A520105) 周口師范學(xué)院高層次人才科研啟動(dòng)基金(No.zknuc2015103)
【分類(lèi)號(hào)】:TP393.092
【相似文獻(xiàn)】
相關(guān)期刊論文 前4條
1 施步青;李春林;鐘景秀;;基于時(shí)間約束的Ad Hoc網(wǎng)格能量?jī)?yōu)化算法[J];武漢理工大學(xué)學(xué)報(bào)(交通科學(xué)與工程版);2009年02期
2 呂芹;;行為預(yù)測(cè)即將降臨社交網(wǎng)絡(luò)[J];互聯(lián)網(wǎng)周刊;2014年02期
3 謝婧;劉功申;蘇波;孟魁;;社交網(wǎng)絡(luò)中的用戶轉(zhuǎn)發(fā)行為預(yù)測(cè)[J];上海交通大學(xué)學(xué)報(bào);2013年04期
4 劉臣;田占偉;于晶;單偉;;在線社會(huì)網(wǎng)絡(luò)用戶的信息分享行為預(yù)測(cè)研究[J];計(jì)算機(jī)應(yīng)用研究;2013年04期
相關(guān)碩士學(xué)位論文 前4條
1 馬瑩瑩;微博用戶轉(zhuǎn)發(fā)行為及情感預(yù)測(cè)研究[D];哈爾濱工業(yè)大學(xué);2015年
2 劉敏;微博傳播效果的分析與預(yù)測(cè)[D];太原理工大學(xué);2016年
3 童智高;基于用戶行為和位置感知的邊際服務(wù)加載優(yōu)化研究[D];浙江大學(xué);2017年
4 粟武林;一種基于邏輯回歸的微博用戶可信度評(píng)估方法[D];河北大學(xué);2015年
,本文編號(hào):2319256
本文鏈接:http://sikaile.net/guanlilunwen/ydhl/2319256.html
最近更新
教材專(zhuān)著