社交網(wǎng)絡(luò)專業(yè)領(lǐng)域社區(qū)關(guān)鍵技術(shù)研究與應(yīng)用
發(fā)布時間：2018-02-07 11:52
本文關(guān)鍵詞： 社交網(wǎng)絡(luò) 社區(qū)發(fā)現(xiàn) 話題模型 分布式計算 領(lǐng)域?qū)＜?/strong> 出處：《北京郵電大學(xué)》2015年碩士論文 論文類型：學(xué)位論文
【摘要】：近年來,社交網(wǎng)絡(luò)服務(wù)迅猛發(fā)展,用戶人數(shù)呈爆炸式增長。通過社交網(wǎng)絡(luò)服務(wù),人們除了進行日常的社交行為,則更多的是將其當作公共媒體平臺。調(diào)查發(fā)現(xiàn),除了跟好友保持聯(lián)系之外,人們使用社交網(wǎng)絡(luò)大多是用來獲取專業(yè)的知識分享以及跟蹤自己感興趣的事件或話題。社交網(wǎng)絡(luò)中,人們的交互有明顯的社區(qū)性,相同社區(qū)內(nèi)的用戶多具有相同興趣或關(guān)注點并交流密切,不同社區(qū)通過關(guān)聯(lián)節(jié)點進行連接。同時,由于社交網(wǎng)絡(luò)的用戶眾多,每天都會產(chǎn)生成千上萬的信息,對于個人來說,很難有效的從海量的數(shù)據(jù)中找到自己所關(guān)注的內(nèi)容,因而我們有必要研究合適的方法,來幫助用戶更加高效的使用社交網(wǎng)絡(luò)。 針對上述背景,本文主要研究了社交網(wǎng)絡(luò)專業(yè)領(lǐng)域社區(qū)發(fā)現(xiàn)問題和專業(yè)領(lǐng)域用戶社區(qū)話題監(jiān)測問題。文章首先建立了社交網(wǎng)絡(luò)專業(yè)領(lǐng)域社區(qū)發(fā)現(xiàn)模型,該模型針對用戶在社交網(wǎng)絡(luò)上對專業(yè)領(lǐng)域知識的需求,在充分利用社交網(wǎng)絡(luò)數(shù)據(jù)信息的基礎(chǔ)上,提出了能夠準確識別專業(yè)領(lǐng)域?qū)＜矣脩舻膶I(yè)領(lǐng)域?qū)＜矣脩艚缍ㄋ惴�。在識別出的專家用戶群基礎(chǔ)上,完成了專家用戶社交網(wǎng)絡(luò)的構(gòu)建及連接強度的評估,并提出了基于用戶連接強度的社區(qū)劃分算法。然后,本文構(gòu)建了專業(yè)領(lǐng)域用戶社區(qū)話題監(jiān)測模型,該模型針對用戶面對專業(yè)領(lǐng)域?qū)＜疑鐓^(qū)中產(chǎn)生的海量數(shù)據(jù)無法有效的獲知其所討論話題的問題,在充分分析社交網(wǎng)絡(luò)數(shù)據(jù)特征及話題分布特征的基礎(chǔ)上,提出了有監(jiān)督的層次狄利克雷分配算法,并給出了分布式解決方案,從而可以高效的監(jiān)測專業(yè)領(lǐng)域用戶社區(qū)中的熱門話題。經(jīng)過在真實數(shù)據(jù)的驗證表明,上述兩個模型相比現(xiàn)有的解決方案,都具有更好的性能優(yōu)勢。 最后,基于本文研究的社交網(wǎng)絡(luò)專業(yè)領(lǐng)域社區(qū)發(fā)現(xiàn)模型和專業(yè)領(lǐng)域用戶社區(qū)話題監(jiān)測模型,構(gòu)建了社交網(wǎng)絡(luò)專業(yè)領(lǐng)域社區(qū)話題監(jiān)測系統(tǒng)。文章對該系統(tǒng)的整體架構(gòu)、各模塊設(shè)計、開發(fā)環(huán)境與運行平臺、系統(tǒng)的運行結(jié)果以及性能分析做了詳細的介紹。
[Abstract]:In recent years, social networking services have grown rapidly and the number of users has exploded. Through social networking services, people use them more as a public media platform than in their daily social activities. In addition to keeping in touch with friends, most people use social networks to gain professional knowledge sharing and track events or topics of interest to them. Most users in the same community share the same interests or concerns and communicate closely. Different communities connect through connected nodes. At the same time, because of the large number of users on social networks, thousands of messages are generated every day. It is difficult to find out what we are concerned about from the massive data, so it is necessary to study appropriate methods to help users to use social network more efficiently. In view of the above background, this paper mainly studies the social network domain community discovery problem and the specialized domain user community topic monitoring question. Firstly, the paper establishes the social network specialized domain community discovery model. This model aims at the needs of users' knowledge of professional domain on social network, and makes full use of social network data and information. Based on the expert user group identified, the construction of the social network and the evaluation of the connection intensity of the expert user are completed. A community partition algorithm based on user connection strength is proposed. Then, a model of user community topic monitoring in professional field is constructed in this paper. This model aims at the problem that users can not effectively know the topic they are discussing in the face of the massive data generated in the professional domain expert community, and based on the analysis of the social network data characteristics and topic distribution features, A supervised hierarchical Delikley assignment algorithm is proposed, and a distributed solution is given, which can efficiently monitor the hot topics in the user community in the professional field. The above two models have better performance advantages than the existing solutions. Finally, based on the social network domain community discovery model and the professional domain user community topic monitoring model, a social network professional domain community topic monitoring system is constructed. Each module design, development environment and running platform, system running results and performance analysis are introduced in detail.
【學(xué)位授予單位】：北京郵電大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：TP393.09
【參考文獻】
相關(guān)期刊論文 前1條
1 張志飛;苗奪謙;高燦;;基于LDA主題模型的短文本分類方法[J];計算機應(yīng)用;2013年06期
，本文編號：1494314
本文鏈接：http://sikaile.net/guanlilunwen/ydhl/1494314.html
最近更新
教材專著
