摘 要
近年來，數(shù)據(jù)挖掘(Data Mining，簡稱DM)，受到國際人工智能與數(shù)據(jù)庫界的廣泛重視。但是隨著網(wǎng)絡(luò)時(shí)代的到來，傳統(tǒng)的數(shù)據(jù)挖掘的對象發(fā)生了改變，這對于數(shù)據(jù)挖掘和知識發(fā)現(xiàn)提出了新的挑戰(zhàn)，Web挖掘正是這樣的背景下提出的。Web挖掘就是從Web世界的各種數(shù)據(jù)中識別出有效的、新穎的、潛在有用的，以及最終可理解的模式的過程。Web挖掘已經(jīng)成為Web信息決策的重要手段，而Web使用挖掘因?yàn)槠浍@得挖掘數(shù)據(jù)的便利性及準(zhǔn)確性，更是成為Web挖掘中的重要研究方向之一。
目前我國的互聯(lián)網(wǎng)已經(jīng)十分普及，成為人們獲取各種信息的主要手段之一?；ヂ?lián)網(wǎng)與實(shí)體經(jīng)濟(jì)不斷融合，利用互聯(lián)網(wǎng)改造和提升傳統(tǒng)產(chǎn)業(yè)，帶動了傳統(tǒng)產(chǎn)業(yè)結(jié)構(gòu)調(diào)整和經(jīng)濟(jì)增長方式的轉(zhuǎn)變，互聯(lián)網(wǎng)已經(jīng)成為我國發(fā)展低碳經(jīng)濟(jì)的新型戰(zhàn)略性產(chǎn)業(yè)。工信部發(fā)布的互聯(lián)網(wǎng)產(chǎn)業(yè)數(shù)據(jù)顯示，截至2009年底，國內(nèi)網(wǎng)站數(shù)量達(dá)到323萬個(gè)，年增長率12.3%，網(wǎng)民人數(shù)達(dá)到4．04億，信息產(chǎn)業(yè)占國內(nèi)生產(chǎn)總值的比重達(dá)到10％左右。隨著互聯(lián)網(wǎng)產(chǎn)業(yè)的不斷發(fā)展，網(wǎng)站之間的競爭達(dá)到了白熱化程度，如何在日益激烈的網(wǎng)站競爭中脫穎而出是網(wǎng)站決策者面臨的主要問題。“以用戶為核心”的網(wǎng)站構(gòu)建思想已經(jīng)成為趨勢。這就需要網(wǎng)站經(jīng)營者了解用戶對于網(wǎng)站訪問的感受，同時(shí)根據(jù)用戶的需要及時(shí)對于網(wǎng)站進(jìn)行合理的改進(jìn)，從而贏得用戶的青睞。日志文件是網(wǎng)站能夠直接獲得的最為全面的用戶訪問記錄，日志文件中記錄了用戶訪問過程的全部信息。Web使用挖掘正是從Web日志文件中發(fā)現(xiàn)用戶的訪問習(xí)慣和訪問模式，從而對于網(wǎng)站的運(yùn)行布局和結(jié)構(gòu)進(jìn)行優(yōu)化，進(jìn)而提升網(wǎng)站的用戶滿意度。
本文結(jié)合“江蘇招生考試網(wǎng)”的真實(shí)運(yùn)行數(shù)據(jù)，通過Web使用挖掘技術(shù)對于網(wǎng)站的運(yùn)行日志文件進(jìn)行全面的挖掘分析，從中發(fā)現(xiàn)用戶的訪問習(xí)慣和訪問模式，進(jìn)而發(fā)現(xiàn)網(wǎng)站的運(yùn)行現(xiàn)狀以及頁面之間的關(guān)聯(lián)性、時(shí)序性，最終根據(jù)挖掘結(jié)果幫助網(wǎng)站決策者制定優(yōu)化策略，這對于網(wǎng)站適應(yīng)未來發(fā)展趨勢、加快自身發(fā)展、應(yīng)對競爭和挑戰(zhàn)有著極具價(jià)值的現(xiàn)實(shí)意義。
論文創(chuàng)新之處主要體現(xiàn)于：全面梳理了Web使用挖掘的相關(guān)理論知識；針對Web使用挖掘的整個(gè)過程進(jìn)行了深入探討，特別針對數(shù)據(jù)預(yù)處理中的主要問題提出相應(yīng)的解決辦法；在理論研究的基礎(chǔ)上，綜合運(yùn)用計(jì)算機(jī)技術(shù)、數(shù)據(jù)庫技術(shù)、數(shù)據(jù)挖掘等手段，建立了“基于Web使用挖掘的網(wǎng)站優(yōu)化系統(tǒng)”，為Web使用挖掘的實(shí)際應(yīng)用做出了有益的嘗試。

關(guān)鍵詞：數(shù)據(jù)挖掘，Web使用挖掘，數(shù)據(jù)預(yù)處理，關(guān)聯(lián)規(guī)則
Abstract
In recent years, Data Mining has being paid fairly attention by international artificial intelligence and data base field. With web age’s coming, objects of traditional data mining change, which brings the new challenge to data mining as well as knowledge discovery. And Web mining, introduced from such a background, that is a course of recognizing effective, new, potencially useful, comprehensible mode. It has become a significant means for web information decision-making, meanwhile, become an essential academic interest of web mining for mining data’s convenience and accuracy. Our country internetwork’s preva lence promotes itself to become one of the main manners for people achineving kinds of information. It brings along traditional industry’s structural readjustment and economic growth manner’s tranforming through gradual convergence of internet and the real economy, or utilization of transforming, advancing traditional industry. Internetwork has become our country’s new type strategic industry of low-carbon economy development. Internet industrial data announced by Ministry of Industry and Information Technology shows that until the end of 2009, domestic web sites reach 3,230,000; annual rate of growth is 12.3%; netizen reach 4.04 hundred million; information industry holds about 10% in GDP. With the gradual development of internet industry, competitions between web sites is to the fierce degree. How to occupy the top point in this fierce competition is a main problem confronted by web decision-makers. “User-centering”, the trend of web buliding, demands web operators understanding users’ visiting recept, then according to it, transforming relative improvement for users’ satisfication. Log files are the most direct complete records of user visit and contain the whole information about user visiting process. Hence,Web mining finds out users’ visiting habit and visting mode from log files in order to realize web running placement and structural optimization, and then rising users’ satisfication degree. This paper does an entire mining analysis of web running log files, basing on the real data from “Jiang Su enrollment examination web site”. It assits web site dicision-maker to make optimizational strategy finally by mining consequnce which is formed by discovering relevance & timing sequnence between web site running actuality and page layout. Therefore, data mining remains actural valuable meaning to the respect of web site adapting to futural trend, self-development fastering, competition&chanllenge confronting. This paper’s creation: comprehesively combs data mining relatie theoratical knowledge; explores the overall process of web minning, especially on the resolving methods to data pretreatment; builds “web mining optimizaition system”, basing on the theoratical study and means of applying computer technology, data base technology and data mining; does an profitable attempt of web mining practice and application..