基于Mahout框架的協(xié)同過濾推薦引擎的研究與實(shí)現(xiàn)
發(fā)布時間：2018-11-26 09:03
【摘要】：當(dāng)今互聯(lián)網(wǎng)應(yīng)用的快速擴(kuò)張給廣大用戶帶來了大量的數(shù)據(jù)信息，很好地滿足了用戶在信息時代對信息的需求，特別是Web2.0的迅速發(fā)展以及移動互聯(lián)網(wǎng)的崛起，用戶自創(chuàng)和分享內(nèi)容變得越來越容易，用戶生成內(nèi)容隨即大量產(chǎn)生�，F(xiàn)有的信息檢索技術(shù)（如搜索引擎）一定程度上解決了激化的矛盾，但是并不能完全滿足社會的需求，當(dāng)用戶自身無法給出有效的關(guān)鍵詞，又或者用戶沒有明確需求時，，搜索引擎被動驅(qū)動的缺點(diǎn)暴露無遺。這時推薦系統(tǒng)作為信息檢索領(lǐng)域新興技術(shù)應(yīng)運(yùn)而生。它依靠其智能挖掘用戶需求、主動推送精準(zhǔn)信息，很快得到了研究者和市場的關(guān)注。 本文的項(xiàng)目目標(biāo)在于探索構(gòu)建一個基于Hadoop平臺的協(xié)同過濾推薦引擎，利用開源框架Mahout實(shí)現(xiàn)傳統(tǒng)協(xié)同過濾算法到MapReduce編程模型的移植。本文首先介紹了推薦引擎的研究背景、選題意義、國內(nèi)外研究現(xiàn)狀，闡述了推薦引擎的理論知識和協(xié)同過濾算法；其次，詳細(xì)敘述了本推薦引擎的總體架構(gòu)和推薦引擎的算法設(shè)計，隨后，重點(diǎn)闡述了本推薦引擎的具體實(shí)現(xiàn)過程，最后給出了推薦引擎的實(shí)驗(yàn)結(jié)果及在現(xiàn)實(shí)中的初步應(yīng)用。 本論文的主要貢獻(xiàn)包括： 1）設(shè)計并實(shí)現(xiàn)了基于Hadoop的協(xié)同過濾推薦引擎，實(shí)現(xiàn)了傳統(tǒng)協(xié)同過濾算法從單機(jī)到MapReduce框架下的移植。 2）設(shè)計并實(shí)現(xiàn)了Web管理系統(tǒng)管理Hadoop平臺上的推薦引擎，支持多策略多任務(wù)執(zhí)行推薦作業(yè)。
[Abstract]:Nowadays, the rapid expansion of Internet applications has brought a lot of data information to the vast number of users, which has well met the information needs of users in the information age, especially the rapid development of Web2.0 and the rise of mobile Internet. It becomes easier and easier for users to create and share content, and user-generated content is generated in large quantities. The existing information retrieval technology (such as search engine) has solved the intensified contradiction to some extent, but can not completely meet the needs of the society, when the user can not give effective keywords, or when the user does not have a clear demand. The shortcomings of passive drive of search engine are exposed. At this time, recommendation system as a new technology in the field of information retrieval came into being. It relies on its intelligent mining of user needs and proactively pushing accurate information, which has attracted the attention of researchers and markets. The goal of this paper is to explore the construction of a collaborative filtering recommendation engine based on Hadoop platform and implement the migration of traditional collaborative filtering algorithm to MapReduce programming model using open source framework Mahout. This paper first introduces the research background of recommendation engine, the significance of choosing the topic, the research status at home and abroad, and expounds the theoretical knowledge of recommendation engine and collaborative filtering algorithm. Secondly, the overall architecture of the recommendation engine and the algorithm design of the recommendation engine are described in detail. Then, the implementation process of the recommendation engine is described in detail. Finally, the experimental results of the recommendation engine and its preliminary application in reality are given. The main contributions of this thesis are as follows: 1) the collaborative filtering recommendation engine based on Hadoop is designed and implemented, and the traditional collaborative filtering algorithm is transplanted from single machine to MapReduce framework. 2) the recommendation engine on Web management Hadoop platform is designed and implemented, which supports multi-strategy and multi-task recommendation.
【學(xué)位授予單位】：華南理工大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2013
【分類號】：TP391.3
本文編號：2358076
[Abstract]:Nowadays, the rapid expansion of Internet applications has brought a lot of data information to the vast number of users, which has well met the information needs of users in the information age, especially the rapid development of Web2.0 and the rise of mobile Internet. It becomes easier and easier for users to create and share content, and user-generated content is generated in large quantities. The existing information retrieval technology (such as search engine) has solved the intensified contradiction to some extent, but can not completely meet the needs of the society, when the user can not give effective keywords, or when the user does not have a clear demand. The shortcomings of passive drive of search engine are exposed. At this time, recommendation system as a new technology in the field of information retrieval came into being. It relies on its intelligent mining of user needs and proactively pushing accurate information, which has attracted the attention of researchers and markets. The goal of this paper is to explore the construction of a collaborative filtering recommendation engine based on Hadoop platform and implement the migration of traditional collaborative filtering algorithm to MapReduce programming model using open source framework Mahout. This paper first introduces the research background of recommendation engine, the significance of choosing the topic, the research status at home and abroad, and expounds the theoretical knowledge of recommendation engine and collaborative filtering algorithm. Secondly, the overall architecture of the recommendation engine and the algorithm design of the recommendation engine are described in detail. Then, the implementation process of the recommendation engine is described in detail. Finally, the experimental results of the recommendation engine and its preliminary application in reality are given. The main contributions of this thesis are as follows: 1) the collaborative filtering recommendation engine based on Hadoop is designed and implemented, and the traditional collaborative filtering algorithm is transplanted from single machine to MapReduce framework. 2) the recommendation engine on Web management Hadoop platform is designed and implemented, which supports multi-strategy and multi-task recommendation.
【學(xué)位授予單位】：華南理工大學(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1條
1 許海玲;吳瀟;李曉東;閻保平;;互聯(lián)網(wǎng)推薦系統(tǒng)比較研究[J];軟件學(xué)報;2009年02期
本文編號：2358076
本文鏈接：http://sikaile.net/kejilunwen/sousuoyinqinglunwen/2358076.html
最近更新
教材專著
