本文選題：DBFS + Lucene�。� 參考：《南京航空航天大學(xué)》2012年碩士論文

【摘要】：傳統(tǒng)的基于目錄和文件的層級(jí)文件系統(tǒng)沿用至今，雖然這樣的樹型文件結(jié)構(gòu)給用戶提供了簡(jiǎn)單易用的文件存放和修改方法，但同時(shí)也帶來(lái)了定位文件和目錄的困難。隨著硬件性能和磁盤存儲(chǔ)容量的不斷提高，計(jì)算機(jī)中的文件數(shù)目不斷增加，而傳統(tǒng)的層級(jí)文件系統(tǒng)的諸多缺點(diǎn)（如對(duì)文件進(jìn)行描述的元數(shù)據(jù)信息匱乏且不易擴(kuò)充）越來(lái)越明顯。尤其是在單個(gè)文件體積小、數(shù)量級(jí)大的文件管理情形下，，用傳統(tǒng)的層級(jí)文件系統(tǒng)對(duì)文件進(jìn)行查找和定位非常困難，而這一情形下的文件管理也成為一個(gè)難題。 本通過(guò)對(duì)用戶態(tài)下的DBFS（Database-based File System，數(shù)據(jù)庫(kù)文件系統(tǒng)）技術(shù)的研究來(lái)解決文件體積小、數(shù)量級(jí)大的文件管理難題，而現(xiàn)有的用戶態(tài)下的DBFS技術(shù)無(wú)法很好地解決這一問(wèn)題，因此本文對(duì)現(xiàn)有DBFS技術(shù)進(jìn)行完善，通過(guò)對(duì)數(shù)據(jù)庫(kù)技術(shù)和全文檢索技術(shù)的研究對(duì)比，提出利用嵌入式數(shù)據(jù)庫(kù)SQLite和全文檢索引擎Lucene對(duì)現(xiàn)有的DBFS模型進(jìn)行改進(jìn)，使其既能提供基于文件元數(shù)據(jù)也能提供基于文本內(nèi)容的快速檢索。通過(guò)對(duì)Lucene的深入學(xué)習(xí)，本文對(duì)全文檢索的相關(guān)技術(shù)進(jìn)行如下研究和改進(jìn)：1、針對(duì)應(yīng)用領(lǐng)域的用戶需求，對(duì)倒排索引進(jìn)行個(gè)性化改進(jìn)；2、結(jié)合TF-IDF加權(quán)算法對(duì)Lucene現(xiàn)有結(jié)果排序算法進(jìn)行改進(jìn)，使其在體現(xiàn)文檔和特征詞關(guān)聯(lián)度的基礎(chǔ)上，更好地體現(xiàn)用戶對(duì)不同信息的不同側(cè)重程度，從而更好地滿足實(shí)際檢索需求。最后結(jié)合遠(yuǎn)洋運(yùn)輸業(yè)務(wù)郵件管理需求，構(gòu)建了基于DBFS的遠(yuǎn)洋運(yùn)輸業(yè)務(wù)郵件管理原型系統(tǒng)，與Uniwell（H.K.）公司的實(shí)際業(yè)務(wù)數(shù)據(jù)相結(jié)合，進(jìn)行應(yīng)用研究分析，驗(yàn)證本文改進(jìn)的DBFS模型在單個(gè)文件體積小、數(shù)量級(jí)大的小文件管理方面的有效性，并且通過(guò)實(shí)驗(yàn)數(shù)據(jù)驗(yàn)證了其對(duì)于倒排索引的個(gè)性化研究和結(jié)果排序算法改進(jìn)的有效性。本課題的研究為類似Uniwell（H.K.）這樣的遠(yuǎn)洋運(yùn)輸公司提供了對(duì)積累的海量遠(yuǎn)洋運(yùn)輸業(yè)務(wù)郵件的快速查找和管理方法，提高了遠(yuǎn)洋運(yùn)輸公司在累積的海量信息中迅速獲得航次決策信息的效率，從而為有效的航次決策提供支持。綜上所述，論文的研究成果具有較高的理論意義和實(shí)用價(jià)值。
[Abstract]:The traditional hierarchical file system based on directories and files has been used up to now. Although this tree file structure provides users with a simple and easy to use file storage and modification methods, it also brings difficulties in locating files and directories. With the improvement of hardware performance and disk storage capacity, the number of files in the computer is increasing, and many disadvantages of traditional hierarchical file system (such as the lack of metadata information described to the file) are becoming more and more obvious. Especially in the case of single file with small size and large order of magnitude, it is very difficult to find and locate files by traditional hierarchical file system, and file management becomes a difficult problem in this case. In this paper, we study DBFS (Database based File system) technology in user state to solve the file management problem of small size and large order of magnitude, but the existing DBFS technology in user state can not solve this problem very well. In this paper, the existing DBFS technology is improved, and the existing DBFS model is improved by using the embedded database SQLite and the full-text retrieval engine Lucene through the research and comparison of database technology and full-text retrieval technology. It can provide both file-based metadata and text-based content-based fast retrieval. Through the in-depth study of Lucene, this paper carries on the following research and improvement to the related technology of full-text retrieval, aiming at the user demand of the application domain, carries on the personalized improvement to the inverted index; 2. Based on the TF-IDF weighted algorithm, Lucene's existing result sorting algorithm is improved to better reflect the different emphasis degree of users on different information on the basis of reflecting the correlation degree of documents and feature words, so as to better meet the actual retrieval requirements. Finally, combining with the mail management requirements of ocean transportation business, the prototype system of ocean shipping mail management based on DBFS is constructed, and Uniwell (H.K.) Combining the actual business data of the company, the application research and analysis are carried out to verify the effectiveness of the improved DBFS model in the management of small files with small size and large order of magnitude. The effectiveness of the personalized research on inverted index and the improvement of the result sorting algorithm are verified by the experimental data. The research of this subject is similar to that of Uniwell (H. K.) Such ocean shipping companies provide a fast search and management method for the accumulated mass of ocean shipping business mail, and improve the efficiency of ocean shipping companies to quickly obtain voyage decision information from the accumulated mass information. It provides support for effective voyage decision making. To sum up, the research results have high theoretical significance and practical value.
【學(xué)位授予單位】：南京航空航天大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2012
【分類號(hào)】：F270.7;F550.6

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 王濤;劉紀(jì)平;毋河海;;基于排序預(yù)處理的等高線提取算法[J];測(cè)繪學(xué)報(bào);2006年04期

2 何芳原;;淺談海量數(shù)據(jù)處理技術(shù)研究[J];硅谷;2009年08期

3 程濤;施水才;王霞;呂學(xué)強(qiáng);;基于同義詞詞林的中文文本主題詞提取[J];廣西師范大學(xué)學(xué)報(bào)(自然科學(xué)版);2007年02期

4 林潔;李丹寧;吳曉;;基于用戶的個(gè)性化綜合倒排索引[J];杭州師范大學(xué)學(xué)報(bào)(自然科學(xué)版);2008年03期

5 胡正華;航次決策支持系統(tǒng)分析與設(shè)計(jì)[J];世界海運(yùn);2002年02期

6 王遠(yuǎn)定;梁久禎;;利用關(guān)鍵詞倒排表實(shí)時(shí)檢索中文網(wǎng)頁(yè)[J];計(jì)算機(jī)工程與應(yīng)用;2010年28期

7 趙珂;逯鵬;李永強(qiáng);;基于Lucene的搜索引擎設(shè)計(jì)與實(shí)現(xiàn)[J];計(jì)算機(jī)工程;2011年16期

8 馮勇;方欣;徐紅艷;;帶有高效索引的語(yǔ)義Web服務(wù)I/O匹配優(yōu)化方法[J];計(jì)算機(jī)應(yīng)用;2011年03期

9 周漢平;;Levenshtein距離在編程題自動(dòng)評(píng)閱中的應(yīng)用研究[J];計(jì)算機(jī)應(yīng)用與軟件;2011年05期

10 周秀霞;隋會(huì)民;;TRS信息資源整合的模式及其局限研究[J];情報(bào)科學(xué);2005年11期

相關(guān)會(huì)議論文 前1條

1 魏環(huán)宇;陽(yáng)國(guó)貴;;一個(gè)基于數(shù)據(jù)庫(kù)的文件系統(tǒng)(XFS)的設(shè)計(jì)與實(shí)現(xiàn)[A];2008通信理論與技術(shù)新進(jìn)展——第十三屆全國(guó)青年通信學(xué)術(shù)會(huì)議論文集（上）[C];2008年

相關(guān)碩士學(xué)位論文 前5條

1 陳仙桃;面向遠(yuǎn)洋運(yùn)輸業(yè)的船貨匹配方法研究及應(yīng)用[D];南京航空航天大學(xué);2010年

2 李清;基于數(shù)據(jù)庫(kù)技術(shù)的文件系統(tǒng)XDBFS的設(shè)計(jì)與實(shí)現(xiàn)[D];國(guó)防科學(xué)技術(shù)大學(xué);2006年

3 楊光宇;全文檢索系統(tǒng)Lucene的分析與擴(kuò)展[D];吉林大學(xué);2009年

4 魏環(huán)宇;一個(gè)集成桌面搜索的數(shù)據(jù)庫(kù)文件系統(tǒng)的研究與實(shí)現(xiàn)[D];國(guó)防科學(xué)技術(shù)大學(xué);2008年

5 高欣;基于Lucene的全文檢索系統(tǒng)的研究與實(shí)現(xiàn)[D];天津師范大學(xué);2010年

本文編號(hào)：2101505