本文關(guān)鍵詞： 興趣遷移 興趣收集 用戶聚類 遺忘函數(shù) 時(shí)間窗　出處：《電子科技大學(xué)》2009年碩士論文　論文類型：學(xué)位論文

【摘要】：當(dāng)今社會(huì)是信息爆炸的社會(huì),對(duì)此,人們?cè)缫褯](méi)有爭(zhēng)議。為了能有效地利用呈指數(shù)級(jí)增長(zhǎng)的海量數(shù)據(jù),而不至于淹沒(méi)在信息的海洋中,人們開(kāi)始對(duì)數(shù)據(jù)挖掘技術(shù)進(jìn)行深入研究,并使之成為計(jì)算機(jī)領(lǐng)域的一個(gè)研究熱點(diǎn)。數(shù)據(jù)挖掘研究的是如何在多樣的、海量的數(shù)據(jù)中提取出蘊(yùn)含的、有用的、潛在的,但不容易被發(fā)現(xiàn)的知識(shí)和模式。 由于網(wǎng)絡(luò)的普及與發(fā)展,出現(xiàn)了新問(wèn)題,就是由于信息太多,要么難以及時(shí)地消化吸收,要么難以有效地使用,即所謂“信息過(guò)載”和“信息迷航”的問(wèn)題。人的接受能力和認(rèn)知能力畢竟有限,時(shí)間精力更是有限。所以,一種普遍的現(xiàn)象就是網(wǎng)絡(luò)用戶面對(duì)鋪天蓋地的資訊只能被動(dòng)接受,而在尋找自己感興趣的信息時(shí)又如同大海撈針。自己不感興趣的常常躲不掉,自己感興趣的往往找起來(lái)又很困難。于是新的需求產(chǎn)生了,用戶希望能及時(shí)獲得自己感興趣的信息,能有滿足自己個(gè)性化要求的服務(wù)。新的需求就是新的市場(chǎng),信息服務(wù)提供商也迫切想開(kāi)發(fā)出針對(duì)用戶個(gè)性化需求,滿足用戶興趣愛(ài)好的信息服務(wù)系統(tǒng)。從而增加競(jìng)爭(zhēng)優(yōu)勢(shì),拓展盈利空間。兩方面的愿望,促成了個(gè)性化信息服務(wù)的誕生。個(gè)性化信息服務(wù)的關(guān)鍵和基礎(chǔ)就是收集用戶興趣,建立用戶興趣模型,掌握用戶興趣的遷移變化。知道了用戶的興趣,才知道該如何“投其所好”,有的放矢的開(kāi)展個(gè)性化服務(wù),比如有針對(duì)性的改進(jìn)站點(diǎn)結(jié)構(gòu)和設(shè)計(jì),推薦信息,定制廣告等。 本文完整介紹了基于Web的用戶個(gè)性化興趣挖掘過(guò)程,并對(duì)用戶興趣收集、用戶興趣建模、用戶聚類和用戶興趣遷移做了重點(diǎn)研究。針對(duì)用戶興趣收集提出了基于用戶隱式興趣反饋和顯式興趣反饋相結(jié)合的用戶興趣收集模型。在用戶興趣建模階段,將用戶興趣分為長(zhǎng)期興趣和短期興趣進(jìn)行詳細(xì)分析,從而提出相應(yīng)的算法。隨著時(shí)間的推移,一些原有的興趣會(huì)過(guò)時(shí),一些新的興趣會(huì)產(chǎn)生,需要對(duì)用戶已有興趣進(jìn)行更新。本文又重點(diǎn)研究了用戶興趣遷移,就是研究用戶興趣隨時(shí)間變化后如何及時(shí)更新興趣,如何淘汰舊興趣添入新興趣。以往的研究通常有時(shí)間窗口法,漸近遺忘法等。本文將多種方法相結(jié)合,提出了采用混合模型進(jìn)行建模的算法,這能準(zhǔn)確的反映用戶興趣遷移變化。
[Abstract]:Today's society is a society of information explosion, and there has been no dispute about it. In order to effectively utilize the massive data with exponential growth, and not to be submerged in the ocean of information, people began to study the technology of data mining in depth. Data mining is how to extract the knowledge and patterns contained, useful, potential, but not easy to find in the diverse and massive data. As a result of the popularity and development of the network, new problems have arisen, that is, because of too much information, it is either difficult to digest and absorb in time or to use it effectively. That is, the so-called "information overload" and "information confusion". After all, people's ability to accept and recognize is limited, and their time and energy are even more limited. Therefore, it is a common phenomenon that network users can only passively accept the overwhelming amount of information. And when you look for information you're interested in, it's like looking for a needle in a haystack. You can't hide what you're not interested in. It's often difficult to find what you're interested in. Users hope to get the information they are interested in in time and to have services that meet their own personalized requirements. The new demand is a new market, and the information service providers are eager to develop personalized requirements for users. The information service system to satisfy the interests and interests of the user, thus increasing the competitive advantage and expanding the profit space. The desire of two aspects has contributed to the birth of the individualized information service. The key and foundation of the personalized information service is to collect the interest of the user. Set up user interest model, master the migration change of user interest. Know the user's interest, know how to "give it what you like", carry out individualized service in a targeted way, such as improving the structure and design of the site, recommending information, Custom advertising, etc. In this paper, the process of user personalized interest mining based on Web is introduced, and the collection of user interest, the modeling of user interest, This paper focuses on user clustering and user interest transfer. A user interest collection model based on implicit user interest feedback and explicit interest feedback is proposed for user interest collection. In the stage of user interest modeling, a user interest collection model based on implicit interest feedback and explicit interest feedback is proposed. The user's interest is divided into long-term interest and short-term interest for detailed analysis, so that the corresponding algorithm. Over time, some of the old interest will become obsolete, some new interest will be generated. This paper also focuses on user interest migration, that is, how to update user interest timely after the change of interest over time. How to eliminate old interests and add new ones. The previous researches usually include time window method, asymptotic forgetting method and so on. In this paper, a hybrid model modeling algorithm is proposed by combining many methods. This can accurately reflect user interest migration changes.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2009
【分類號(hào)】：TP311.13

【引證文獻(xiàn)】

相關(guān)期刊論文 前1條

1 李志浩;聶文匯;成鵬;張宇博;陽(yáng)智敏;;基于分頁(yè)緩存模型的用戶興趣跟蹤方法[J];計(jì)算機(jī)工程與科學(xué);2012年10期

相關(guān)碩士學(xué)位論文 前2條

1 林霞;個(gè)性化信息檢索技術(shù)在勘探門(mén)戶中的應(yīng)用研究[D];西安石油大學(xué);2011年

2 鄧亮;基于OpenFlow的視頻流控制原型系統(tǒng)實(shí)現(xiàn)[D];電子科技大學(xué);2013年

，

本文編號(hào)：1507756