本文選題：微博轉(zhuǎn)發(fā)　切入點：情感分析　出處：《北京郵電大學》2015年碩士論文　論文類型：學位論文

【摘要】：隨著互聯(lián)網(wǎng)和Web2.0技術(shù)的迅速發(fā)展和普及,互聯(lián)網(wǎng)上出現(xiàn)了大量的在線社交網(wǎng)絡(luò)。這些社交網(wǎng)絡(luò)成為人們分享信息,傳播信息,獲取信息的主要平臺。微博是一個基于用戶關(guān)注粉絲關(guān)系的信息分享、傳播以及獲取平臺。用戶可以通過瀏覽器,智能手機客戶端發(fā)布140個字以內(nèi)的文字信息,圖片和視頻,實現(xiàn)即時分享。作為一個媒體平臺,最重要的作用是信息的傳播。在微博網(wǎng)絡(luò)中,信息的傳播主要是通過微博轉(zhuǎn)發(fā)實現(xiàn)的。因此,微博的轉(zhuǎn)發(fā)數(shù)量可以作為衡量微博傳播效果的重要指標。研究信息在微博網(wǎng)絡(luò)中的轉(zhuǎn)發(fā)行為,對微博轉(zhuǎn)發(fā)規(guī)模、數(shù)量的預測,在產(chǎn)品營銷,熱點提取和控制敏感信息的傳播方面有重要作用和現(xiàn)實意義。本文從兩個方面研究微博轉(zhuǎn)發(fā)行為：1.微博能否被轉(zhuǎn)發(fā)；2.微博的轉(zhuǎn)發(fā)規(guī)模和數(shù)量。 在微博能否被轉(zhuǎn)發(fā)方面,本文從微博文本情感和用戶角色兩個方面入手,分析這兩方面對微博轉(zhuǎn)發(fā)行為的影響。在微博文本情感方面,本文構(gòu)造了一個文本情感分析引擎來分析文本情感對微博轉(zhuǎn)發(fā)行為的影響。在用戶角色方面,通過對用戶的聚類,驗證了不同的用戶角色的微博在轉(zhuǎn)發(fā)行為方面的巨大差異。 在研究微博的轉(zhuǎn)發(fā)規(guī)模和數(shù)量方面,本文建立了兩個微博轉(zhuǎn)發(fā)量預測模型,分別是兩階段微博轉(zhuǎn)發(fā)量預測模型和基于粉絲轉(zhuǎn)發(fā)意愿和轉(zhuǎn)發(fā)影響力的預測模型。在兩階段模型中,通過分類模型與回歸模型的組合,有效的降低了數(shù)據(jù)不平衡的影響,得到了比較好的預測效果。通過分析微博轉(zhuǎn)發(fā)量的產(chǎn)生機制,建立了基于粉絲轉(zhuǎn)發(fā)意愿和粉絲轉(zhuǎn)發(fā)影響力的微博轉(zhuǎn)發(fā)預測模型,該模型有效率學習了微博轉(zhuǎn)發(fā)量的增長過程,得到了比兩階段模型更好的預測效果。
[Abstract]:With the rapid development and popularization of the Internet and Web2.0 technology, a large number of online social networks have emerged on the Internet. Weibo is a platform for sharing, disseminating and acquiring information based on the user's focus on fan relationships. Users can publish 140 characters of text information, pictures and videos through browsers and smartphone clients. Realize instant sharing. As a media platform, the most important role is the dissemination of information. In Weibo's network, the dissemination of information is mainly realized through the transmission of Weibo. Therefore, Weibo's quantity of retweets can be used as an important indicator to measure the effect of Weibo's dissemination. Research on the behavior of forwarding information in the Weibo network, the prediction of the scale and quantity of the forwarding of Weibo, and the marketing of the products. Hot spot extraction and control of the dissemination of sensitive information play an important role and practical significance. This paper studies the forwarding behavior of Weibo:: 1. Whether Weibo can be forwarded. 2. The scale and quantity of the retweeting of Weibo. In terms of whether Weibo can be forwarded, this paper analyzes the influence of the two aspects on Weibo's retweeting behavior from two aspects: the emotion and the role of the user. In this paper, a text emotion analysis engine is constructed to analyze the influence of text emotion on Weibo's forwarding behavior. In the aspect of studying Weibo's forwarding scale and quantity, this paper sets up two forecasting models of Weibo's forwarding quantity. The two stage Weibo forwarder prediction model and the prediction model based on fan forwarding will and forwarding influence respectively. In the two-stage model, through the combination of classification model and regression model, the effect of data imbalance is effectively reduced. By analyzing the generation mechanism of Weibo's retweeting quantity, we set up a Weibo forwarding prediction model based on fans' forwarding will and fan's forwarding influence. This model can effectively learn the process of the growth of Weibo's retweeting quantity. The prediction results are better than that of the two-stage model.
【學位授予單位】：北京郵電大學
【學位級別】：碩士
【學位授予年份】：2015
【分類號】：TP393.092

【參考文獻】

相關(guān)期刊論文 前4條

1 梁冰;張玉忠;金鐸;;協(xié)同過濾中的矩陣分解算法研究[J];廣東通信技術(shù);2013年08期

2 謝麗星;周明;孫茂松;;基于層次結(jié)構(gòu)的多策略中文微博情感分析和特征抽取[J];中文信息學報;2012年01期

3 張e，

本文編號：1575054