發(fā)布時間：2019-04-21 17:38

【摘要】：面對大量的信息和信息的多元化,人們可獲得的信息越來越多,但是快速有效而又準(zhǔn)確地獲取自己所需的信息卻是相當(dāng)困難的。搜索引擎的出現(xiàn)解決了用戶的檢索需求,但卻不能適時地為用戶主動提供其需要的信息。當(dāng)用戶不確定所需信息的關(guān)鍵詞描述,無法進(jìn)行搜索；又或者用戶自己不愿進(jìn)行搜索,只是希望獲得一些信息的推薦時,就需要一個比搜索引擎更主動的系統(tǒng)來滿足這些需求,智能推薦服務(wù)系統(tǒng)因此誕生,推薦技術(shù)也日益成為一個重要的研究課題。 本文針對用戶的信息推薦需求,設(shè)計并實現(xiàn)了一種網(wǎng)頁推薦系統(tǒng)。本系統(tǒng)以基于奇異值分解的協(xié)同過濾算法為關(guān)鍵技術(shù),用文本分類中的特征選擇方法來預(yù)處理網(wǎng)頁信息,并運(yùn)用特征選擇方法對收集到的網(wǎng)頁以及用戶潛在興趣頁面進(jìn)行處理,同時引入聚類分析算法解決服務(wù)器負(fù)荷大和推薦結(jié)果可能不準(zhǔn)確的問題,最后通過用戶反饋和用戶使用時間段對推薦結(jié)果列表進(jìn)行優(yōu)化,使推薦結(jié)果的準(zhǔn)確度提高,使得推薦結(jié)果更加符合用戶的需求,更加人性化。本文設(shè)計的網(wǎng)頁推薦系統(tǒng)是而向用戶的,結(jié)合用戶興趣并采用基于奇異值分解的協(xié)同過濾推薦算法實現(xiàn)的混合推薦系統(tǒng)。本推薦系統(tǒng)綜合考慮用戶正在瀏覽的頁面、用戶的興趣和用戶歷史訪問記錄,結(jié)合用戶使用推薦系統(tǒng)的時間段來為用戶提供網(wǎng)頁推薦服務(wù)。最終實現(xiàn)的推薦系統(tǒng)可以發(fā)現(xiàn)用戶的興趣和愛好,并向目標(biāo)用戶推薦符合其興趣愛好的信息或物品,參考用戶反饋和用戶使用時段來進(jìn)行智能推薦。 論文首先對網(wǎng)頁推薦系統(tǒng)的研究背景、研究現(xiàn)狀和研究內(nèi)容以及在實際中的應(yīng)用進(jìn)行了綜述。在對網(wǎng)頁推薦系統(tǒng)進(jìn)行了需求分析的基礎(chǔ)上,論文提出了網(wǎng)頁推薦系統(tǒng)的總體架構(gòu)和概要設(shè)計,對模塊進(jìn)行了詳細(xì)設(shè)計和編碼實現(xiàn),并針對采用的關(guān)鍵技術(shù)和方法進(jìn)行了詳細(xì)的介紹和說明。論文的最后對網(wǎng)頁推薦系統(tǒng)進(jìn)行了實驗測試,并指出了網(wǎng)頁推薦系統(tǒng)需要進(jìn)一步改進(jìn)和完善的方面。
[Abstract]:In the face of a large number of information and the diversity of information, people can obtain more and more information, but it is very difficult to get the information they need quickly, effectively and accurately. The emergence of search engine solves the user's need of retrieval, but it can not provide users with the information they need at the right time. When the user is uncertain about the keyword description of the required information, the search can not be carried out; Or if users do not want to search themselves, just want to obtain some information recommendations, they need a more proactive system than the search engine to meet these needs, intelligent recommendation service system was born. Recommendation technology has increasingly become an important research topic. In this paper, a web recommendation system is designed and implemented according to the information recommendation requirements of users. This system takes the collaborative filtering algorithm based on singular value decomposition as the key technology, uses the feature selection method in text classification to pre-process the web page information, and uses the feature selection method to process the collected web page and the user's potential interest page. At the same time, cluster analysis algorithm is introduced to solve the problem that the server load is heavy and the recommendation result may not be accurate. Finally, the recommended results list is optimized by user feedback and user usage time, so that the accuracy of recommendation results can be improved. So that the recommendation results more in line with the needs of users, more user-friendly. The web recommendation system designed in this paper is a hybrid recommendation system which combines user's interest and adopts collaborative filtering recommendation algorithm based on singular value decomposition (SVD). This recommendation system takes into account the pages that the user is browsing, the user's interest and the user's historical visit record, and combines the time period of the user's use of the recommendation system to provide the web page recommendation service for the users. The final implementation of the recommendation system can discover the interests and interests of users and recommend to the target users information or items in accordance with their interests, reference to user feedback and user use time to carry out intelligent recommendations. Firstly, this paper summarizes the research background, research status and application in practice of web recommendation system. On the basis of analyzing the requirement of web recommendation system, this paper puts forward the overall structure and outline design of web page recommendation system, and designs the module in detail and implements the coding. The key technologies and methods adopted are introduced and explained in detail. At the end of the paper, we test the web recommendation system, and point out the aspects that need to be further improved and perfected.
【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】：2013
【分類號】：TP391.3

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 朱敏;蘇博;;基于奇異值分解的協(xié)同過濾推薦算法研究[J];計算機(jī)安全;2010年07期

2 方俊;;電子商務(wù)系統(tǒng)商品推薦方法淺析[J];大眾科技;2010年08期

3 王燕紅;;基于聚類分析的新書推薦[J];計算機(jī)光盤軟件與應(yīng)用;2012年03期

4 董祥和;齊莉麗;董榮和;;優(yōu)化的協(xié)作過濾推薦算法[J];計算機(jī)工程與應(yīng)用;2009年08期

5 李燕;馮博琴;魯曉鋒;;Web日志挖掘中的數(shù)據(jù)預(yù)處理技術(shù)[J];計算機(jī)工程;2009年22期

6 吳兵;葉春明;;基于效用的個性化推薦方法[J];計算機(jī)工程;2012年04期

7 曾小波;魏祖寬;金在弘;;協(xié)同過濾系統(tǒng)的矩陣稀疏性問題的研究[J];計算機(jī)應(yīng)用;2010年04期

8 鮑雷;楊天奇;;基于用戶瀏覽行為的個性化網(wǎng)頁推薦[J];微計算機(jī)信息;2010年06期

9 余小高;;基于分布式數(shù)據(jù)挖掘的電子商務(wù)推薦系統(tǒng)[J];計算機(jī)系統(tǒng)應(yīng)用;2009年11期

10 彭飛;鄧浩江;劉磊;;加入用戶評分偏置的推薦系統(tǒng)排名模型[J];西安交通大學(xué)學(xué)報;2012年06期

相關(guān)碩士學(xué)位論文 前3條

1 陳默;基于神經(jīng)網(wǎng)絡(luò)的元搜索引擎[D];浙江大學(xué);2006年

2 程偉想;網(wǎng)格聚類算法的研究[D];華北電力大學(xué)（河北）;2008年

3 李柯;基于用戶訪問矩陣的網(wǎng)頁推薦模型研究[D];江蘇大學(xué);2010年

，

本文編號：2462414