本文選題：網(wǎng)絡廣告 + 數(shù)據(jù)挖掘��； 參考：《山東科技大學》2007年碩士論文

【摘要】： 可定向投放是網(wǎng)絡廣告的一大特點。數(shù)字時代的新技術使網(wǎng)絡廣告的定向投放成為可能：網(wǎng)絡廣告可以投放給某些特定的目標人群，甚至可以做到一對一的定向投放。數(shù)據(jù)挖掘作為一種先進的數(shù)據(jù)分析方法，，是實現(xiàn)對客戶數(shù)據(jù)進行深入分析的有效工具，其中聚類分析是其中重要的一種作用，可實現(xiàn)在沒有訓練的條件下把樣本分為若干類。網(wǎng)絡廣告精確營銷系統(tǒng)是聚類算法在網(wǎng)絡廣告投放中的應用體現(xiàn)，幫助用戶鎖定目標受眾進行合適的網(wǎng)絡廣告投放。 本文在研究了網(wǎng)絡廣告的特點和發(fā)展歷史，以及數(shù)據(jù)挖掘的基本知識的知識背景下，提出了細分受眾的精確網(wǎng)絡廣告營銷的思路，并在此基礎上提出了基于并行K-均值算法的網(wǎng)絡廣告精確營銷系統(tǒng)模型，并詳細說明了數(shù)據(jù)庫管理，前端處理系統(tǒng)，數(shù)據(jù)采集系統(tǒng)，后臺數(shù)據(jù)分析系統(tǒng)和效果計費系統(tǒng)的設計原理，并重點描述了基于并行K-均值算法的受眾聚類系統(tǒng)的設計。根據(jù)受眾的點擊率、點擊數(shù)、轉(zhuǎn)化率，近期訪問量以及曝光時間總和五個屬性確定受眾評價函數(shù)，利用PVM搭建并行虛擬機，選用Master／Slave模式實現(xiàn)了細分受眾模塊的程序設計，并最后通過實際效果證明了本系統(tǒng)的優(yōu)越性。在本文最后還對系統(tǒng)的進一步改進作了總結(jié)，為下一步工作指明了方向。 通過并行算法細分受眾，快速的鎖定目標受眾群，實施合適的網(wǎng)絡廣告營銷策略，這是網(wǎng)絡廣告投放市場的一次嘗試，具有重要的現(xiàn)實意義和理論意義。
[Abstract]:Targeted delivery is a major feature of online advertising. The new technology of the digital age makes it possible to direct the network advertisement: the network advertisement can be delivered to some specific target groups, and even can be directed to one to one. As an advanced data analysis method, data mining is an effective tool for in-depth analysis of customer data, among which cluster analysis is an important role, which can be used to classify samples into several classes without training. The network advertisement precise marketing system is the application embodiment of clustering algorithm in the network advertisement putting in, helps the user to lock in the target audience to carry on the suitable network advertisement to put in. Under the background of studying the characteristics and development history of network advertisement and the basic knowledge of data mining, this paper puts forward the idea of accurate network advertising marketing of subdividing audience. On this basis, the model of network advertisement precise marketing system based on parallel K-means algorithm is put forward, and the database management, front-end processing system and data acquisition system are described in detail. The design principle of background data analysis system and effect accounting system is described, and the design of audience clustering system based on parallel K-means algorithm is described. According to the five attributes of audience click rate, click number, conversion rate, recent visit amount and exposure time summation, the audience evaluation function is determined, and the parallel virtual machine is built by using PVM, and the program design of subdivision audience module is realized by using Master/Slave mode. Finally, the superiority of the system is proved by the actual effect. At the end of this paper, the further improvement of the system is summarized, and the direction of the next work is pointed out. Subdividing the audience through parallel algorithm, locking the target audience quickly, and implementing the appropriate network advertising marketing strategy is an attempt to put online advertising into the market, which has important practical and theoretical significance.
【學位授予單位】：山東科技大學
【學位級別】：碩士
【學位授予年份】：2007
【分類號】：TP311.52

【引證文獻】

相關碩士學位論文 前1條

1 張小飛;模糊綜合評判方法在煙草績效考核中的應用研究[D];東北大學;2009年

本文編號：1786937