發(fā)布時(shí)間：2019-01-25 19:37

【摘要】：目前,信息與計(jì)算機(jī)技術(shù)的迅速發(fā)展,Internet中的信息資源急劇增加,包含了—些簡(jiǎn)單的文本數(shù)據(jù),還包括了大量的圖像、視頻等多媒體信息。因此,如何高效、準(zhǔn)確地從海量Web圖像資源中檢索到用戶興趣信息成為當(dāng)前的一個(gè)研究熱點(diǎn)。同時(shí),對(duì)于Web圖像檢索,目前的大多搜索引擎提供的圖像檢索服務(wù)沒(méi)有考慮到用戶需求的差異性,因此,隨著Web中圖像數(shù)量的迅速增加,便會(huì)消耗大量的檢索時(shí)間,降低圖像檢索的效率。于是人們希望能夠及時(shí)獲得所感興趣的信息資源,針對(duì)不同的需求為自己提供個(gè)性化的服務(wù)。 針對(duì)以上問(wèn)題,并結(jié)合Web圖像檢索的自身特點(diǎn),本文提出了個(gè)性化Web圖像檢索技術(shù)研究。首先,根據(jù)目前的信息檢索服務(wù)沒(méi)有考慮用戶的差異,而導(dǎo)致檢索效率比較低的問(wèn)題,本文提出了一種基于用戶興趣模型的個(gè)性化Web圖像檢索算法。首先給出了用戶興趣模型的形式化定義；同時(shí)考慮到隨著時(shí)間的變化,用戶興趣會(huì)發(fā)生改變這個(gè)問(wèn)題,本文又引入了新奇因子,有效地結(jié)合了短期興趣和長(zhǎng)期興趣；然后采用顯式跟蹤和隱式跟蹤相結(jié)合的方法對(duì)用戶的興趣進(jìn)行學(xué)習(xí),以不斷完善用戶的興趣信息。通過(guò)用戶興趣模型可以根據(jù)不同用戶的不同興趣愛(ài)好為用戶提供個(gè)性化的Web圖像檢索服務(wù),極大地提高了圖像檢索的效率。 個(gè)性化Web圖像檢索中還存在另外一個(gè)關(guān)鍵的技術(shù)問(wèn)題,那便是用戶興趣的遷移問(wèn)題,即具有相似興趣愛(ài)好用戶之間的興趣信息轉(zhuǎn)移問(wèn)題。目前,這種技術(shù)也稱(chēng)為個(gè)性化推薦。針對(duì)這一問(wèn)題,本文提出了基于用戶興趣模型的個(gè)性化用戶興趣推薦算法,此算法采用了SVD技術(shù)和K-means聚類(lèi)相融合的協(xié)同過(guò)濾方法,有效克服了評(píng)分矩陣數(shù)據(jù)的稀疏問(wèn)題,同時(shí)有效解決了個(gè)性化Web圖像檢索中用戶興趣轉(zhuǎn)移問(wèn)題,為新老用戶提供了個(gè)性化的推薦服務(wù),大大提高了用戶對(duì)感興趣信息的檢索速度和效率。 最后,本文完成了一個(gè)支持多模態(tài)查詢的個(gè)性化Web圖像檢索系統(tǒng),對(duì)全文的工作進(jìn)行了總結(jié),并給出了未來(lái)本課題需要進(jìn)一步研究和提高的問(wèn)題。
[Abstract]:At present, with the rapid development of information and computer technology, the information resources in Internet increase rapidly, including some simple text data, and a lot of multimedia information, such as images, video and so on. Therefore, how to efficiently and accurately retrieve user interest information from massive Web image resources has become a hot research topic. At the same time, for Web image retrieval, most of the current search engines provide image retrieval services that do not take into account the difference of user needs. Therefore, with the rapid increase of the number of images in Web, a large amount of retrieval time will be consumed. Reduce the efficiency of image retrieval. So people hope to be able to get interested information resources in time, to provide personalized services for different needs. Aiming at the above problems and combining the characteristics of Web image retrieval, this paper puts forward the research of personalized Web image retrieval technology. Firstly, a personalized Web image retrieval algorithm based on user interest model is proposed in this paper, because the current information retrieval service does not consider the difference of users, which leads to low retrieval efficiency. Firstly, the formal definition of user interest model is given, and considering the problem that user interest will change with time, the novelty factor is introduced in this paper, which effectively combines short-term interest with long-term interest. Then we use the combination of explicit tracking and implicit tracking to study the interest of users in order to improve the interest information of users. The user interest model can provide users with personalized Web image retrieval service according to different users' different interests, which greatly improves the efficiency of image retrieval. Another key technical problem in personalized Web image retrieval is the migration of user interest, that is, the problem of information transfer between users with similar interests. At present, this technology is also called personalized recommendation. In order to solve this problem, this paper proposes a personalized user interest recommendation algorithm based on user interest model. This algorithm uses SVD technology and K-means clustering fusion to effectively overcome the sparse problem of scoring matrix data. At the same time, it effectively solves the problem of user interest transfer in personalized Web image retrieval, provides personalized recommendation services for new and old users, and greatly improves the speed and efficiency of users' retrieval of information of interest. Finally, a personalized Web image retrieval system supporting multi-modal query is completed, and the work of the thesis is summarized, and the problems that need to be further studied and improved in the future are given.
【學(xué)位授予單位】：東北林業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2012
【分類(lèi)號(hào)】：TP391.3

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 向友君;謝勝利;;圖像檢索技術(shù)綜述[J];重慶郵電學(xué)院學(xué)報(bào)(自然科學(xué)版);2006年03期

2 陳冬玲;王大玲;于戈;于芳;;基于PLSA方法的用戶興趣聚類(lèi)[J];東北大學(xué)學(xué)報(bào)(自然科學(xué)版);2008年01期

3 邱兆文;張?zhí)镂?;基于用戶多媒體數(shù)據(jù)管理模型的個(gè)性化圖像檢索[J];電子學(xué)報(bào);2008年09期

4 李春妍;王勇;;個(gè)性化服務(wù)中用戶興趣聚類(lèi)算法研究[J];信息技術(shù);2007年10期

5 邱兆文,張?zhí)镂?一種新的圖像顏色特征提取方法[J];哈爾濱工業(yè)大學(xué)學(xué)報(bào);2004年12期

6 邱兆文;龐俊彪;張?zhí)镂?梁可;;圖像檢索中基于二次距離的相關(guān)反饋[J];哈爾濱工業(yè)大學(xué)學(xué)報(bào);2006年09期

7 王惠鋒 ,孫正興 ,王箭;語(yǔ)義圖像檢索研究進(jìn)展[J];計(jì)算機(jī)研究與發(fā)展;2002年05期

8 張鋒;常會(huì)友;;使用BP神經(jīng)網(wǎng)絡(luò)緩解協(xié)同過(guò)濾推薦算法的稀疏性問(wèn)題[J];計(jì)算機(jī)研究與發(fā)展;2006年04期

9 葉志偉;夏彬;周欣;張彥超;;一種改進(jìn)的基于顏色直方圖的圖像檢索算法[J];吉首大學(xué)學(xué)報(bào)(自然科學(xué)版);2009年05期

10 楊小平,丁浩,黃都培;基于向量空間模型的中文信息檢索技術(shù)研究[J];計(jì)算機(jī)工程與應(yīng)用;2003年15期

相關(guān)博士學(xué)位論文 前2條

1 魯珂;流形學(xué)習(xí)方法在Web圖像檢索中的應(yīng)用研究[D];電子科技大學(xué);2006年

2 邱兆文;面向用戶的Web圖像檢索關(guān)鍵技術(shù)研究[D];哈爾濱工業(yè)大學(xué);2009年

相關(guān)碩士學(xué)位論文 前1條

1 滕躍;基于用戶興趣的個(gè)性化WEB檢索[D];清華大學(xué);2004年

，

本文編號(hào)：2415123