Using Web Mining Techniques to Identify A Company's External Environment on the Web

綱頁探勘技術於辨識公司外部環境之應用

隨著全球資訊網的蓬勃發展，許多公司會由他們的商業網站來散佈資訊。從策略規劃的角度來說，辨識一個公司的外部環境可以幫助創造其企業價值。因此，藉由網站來辨識公司的外部環境變得相當重要。 傳統方法（例如，網站日誌檔或註冊資訊的分析）常囿於資料蒐集的有限性或不正確性。相反地，網站內容分類可以被使用來辨識一個公司的外部環境，而且，由於網頁間的關係近似於真實世界的社會互動關係，因此可用於輔助分類外部環境。我們因此提出一個分類器叫CNB-HI，它使用綱頁內容及超連結結構來辨識公司外部環境的角色。兩個實驗被用來檢視所提方法的績效。在第一個實驗中，我們比較了CNB及天真貝氏分類器的其它變形，並得到CNB可獲得較佳績效的結論。第二個實驗並進一步顯示CNB-HI的績效相對於CNB有顯著地改善。本研究所提方法的可行性因此獲得證明。

As the World Wide Web prevails nowadays, many companies tend to disseminate information through their commercial Web sites. From a strategic planning viewpoint, identifying a company's external environment helps to create its business values. Therefore, it becomes essential for companies to identify its external environment through the Web. Traditional approaches such as analysis of access log or registration suffer from limited or incorrect data collection. In contrast, Web content classification can be used for a company's external environment identification. Furthermore, relationships among Web pages resemble social interactions in the real world and contribute to classifying the external environment. We, therefore, propose a classifier, CNB-HI, that utilizes Web contents and hyperlink structure to identify the roles of a company's external environment. Two experiments are conducted to examine the performance. In the first experiment, we compare CNB with variants of Naive Bayes classifiers, and conclude that CNB achieves a better performance. The second experiment further shows that the performance of CNB-HI improves markedly compared to CNB. The feasibility of our proposed approach is thus justified.