【摘要】：微博作為近些年新興的社交網(wǎng)站,對整個互聯(lián)網(wǎng)產(chǎn)生了巨大影響,微博服務(wù)極大的拉近了人與人之間的距離,讓人們擁有了新的合作與交流方式。對于微博網(wǎng)絡(luò)中的用戶,每一個用戶都是信息傳播的入口,這種新的方式可以爆發(fā)式的產(chǎn)生大量的實時消息。因此,用戶之間的關(guān)注關(guān)系與轉(zhuǎn)發(fā)關(guān)系構(gòu)成了一個龐大且復雜的微博網(wǎng)絡(luò)。在微博網(wǎng)絡(luò)逐漸形成的過程中,微博社區(qū)也逐漸形成并且達到一定規(guī)模。這是一種普遍的社會網(wǎng)絡(luò)現(xiàn)象,一般這種社區(qū)的內(nèi)部具有很強的相關(guān)性與聯(lián)系性。如果能夠準確的劃分出各個社區(qū),這對于進一步改善微博服務(wù)質(zhì)量是一個前提條件。通過社區(qū)發(fā)現(xiàn)算法的比較,本文研究在大規(guī)模網(wǎng)絡(luò)中標簽傳播算法(Label Propagation Algorithm,簡稱LPA)是一個快速有效的社區(qū)發(fā)現(xiàn)方法,也是目前應(yīng)用范圍比較廣的一個算法。但是,標簽傳播中的隨機節(jié)點訪問降低了算法的魯棒性,對于識別社區(qū)結(jié)構(gòu)的穩(wěn)定性也造成了一定的影響。因此本文提出一種具有節(jié)點優(yōu)先度的標簽傳播算法(Precedential Label Propagation Algorithm,簡稱PLPA),通過引入節(jié)點訪問優(yōu)先度的機制來改善隨機節(jié)點訪問帶來的魯棒性下降的缺陷。并通過對比實驗驗證算法的有效性與穩(wěn)定性。另一方面,本文針對微博網(wǎng)絡(luò)中的用戶排名進行了系統(tǒng)地研究,而大部分的學者在進行該方面研究時只考慮單一的用戶關(guān)注關(guān)系,或者用戶發(fā)布微博的數(shù)目,而在微博中有著較高影響力的用戶通常會發(fā)布一些價值含量較高的信息,并且有著大量的跟隨用戶。本文提出一種基于時效性以及隨機游走策略的用戶排名算法(本文稱為User-Weibo Rank,簡稱UWRank),在該算法所強調(diào)的用戶排名策略中,同時考慮用戶的關(guān)注關(guān)系與微博發(fā)布影響力這兩大因素。針對微博信息的高實時性特點,融合微博內(nèi)容的時效性作為衡量影響力的因素。并且通過實驗,驗證算法再保持較好結(jié)果的前提下,算法能夠達到較好的收斂效果。
[Abstract]:As a new social network site in recent years, the micro-blog has a great impact on the whole Internet, and the micro-blog service has greatly shortened the distance between people and people, so that people have a new way of cooperation and communication. For users in the microblogging network, each user is an entry for information dissemination, and this new way can generate a large number of real-time messages. Therefore, the relationship between the users and the forwarding relationship form a large and complex micro-blog network. In the process of the gradual formation of the micro-blog network, the micro-blog community is gradually formed and reached a certain scale. This is a common social network phenomenon, and generally the interior of such a community has a strong correlation and contact. This is a prerequisite for further improving the quality of the microblogging service if the communities can be accurately divided. In this paper, the label propagation algorithm (LPA) in a large-scale network is a fast and effective method for community discovery and a wide range of algorithms. However, random node access in the label propagation reduces the robustness of the algorithm and has a certain effect on the identification of the stability of the community structure. This paper proposes a label propagation algorithm (PLPA) with node priority, which can improve the robustness of random node access by introducing the mechanism of node access priority. And the validity and the stability of the algorithm are compared by comparing the experimental verification algorithm. On the other hand, this paper makes a systematic study of the user's ranking in the micro-blog network, and most of the scholars consider only a single user-interest relationship when doing the research, or the number of users to publish the micro-blog, Users with high influence in the system often release high-value information and have a large number of follow-up users. This paper presents a user-ranking algorithm based on time-effectiveness and random walk strategy (referred to herein as" User-Weibo Rank "), in which the user's attention relationship and the influence of microblogging are considered in the user's ranking strategy emphasized by the algorithm. Aiming at the high real-time character of the micro-blog information, the time-effectiveness of the content of the micro-blog is used as a factor to measure the influence. And the algorithm can achieve better convergence effect on the premise of the experiment and the verification algorithm to keep the better result.
【學位授予單位】：哈爾濱工程大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP393.092

【參考文獻】

相關(guān)期刊論文 前7條

1 陳克寒;韓盼盼;吳健;;基于用戶聚類的異構(gòu)社交網(wǎng)絡(luò)推薦算法[J];計算機學報;2013年02期

2 王晟;王子琪;張銘;;個性化微博推薦算法[J];計算機科學與探索;2012年10期

3 郭浩;陸余良;王宇;張亮;;基于信息傳播的微博用戶影響力度量[J];山東大學學報(理學版);2012年05期

4 王曉光;袁毅;滕思琦;;微博社區(qū)交流網(wǎng)絡(luò)結(jié)構(gòu)的實證分析[J];情報雜志;2011年02期

5 陳東明;徐曉偉;;一種基于廣度優(yōu)先搜索的社區(qū)發(fā)現(xiàn)方法[J];東北大學學報(自然科學版);2010年03期

6 孫雷;孫慶蘇;;數(shù)據(jù)挖掘在高校圖書館智能分析中的應(yīng)用[J];現(xiàn)代情報;2009年08期

7 姚小濤,席酉民;社會網(wǎng)絡(luò)理論及其在企業(yè)研究中的應(yīng)用[J];西安交通大學學報(社會科學版);2003年03期

相關(guān)碩士學位論文 前1條

1 周濤;Wiki社群的社會網(wǎng)絡(luò)分析[D];華東師范大學;2005年

，

本文編號：2506156