根據Frequent的變化來分析網路使用者需求趨勢──以104家教網為例

Analyzing the Trends of Web Users Behaviors Based on the Change of Frequent Itemsets

對一個成功的網站管理者來說，了解網路使用者的使用趨勢是一個重要的工作。藉由了解使用者的趨勢變化，可以讓網站管理者制定出更有效率的策略，並提供更好的服務。資料探勘在現今網路時代，已經變成一項重要技術來挖掘這些資訊。在本文中，我們從兩段不同的時間軸中，收集網路使用者對此網站輸入的資料，並利用FP-Growth演算法分別找出其Frequent itemsets，再根據觀察不同時期Frequent itemsets的變化來定義emerging pattern、perished pattern和persistent pattern，並用這三種patterns來分析網路使用者的使用趨勢。在本文中，我們使用104家教網──一個幫學生和教師找家教配對的網站做為資料來源，並利用此方法進行分析，來找出其需求的變化。

Understanding the trends of Web users behavior is an important factor for running a successful website. Owners or adminstrators of websites need to make efficient marketing strategies and provide better services according to the change of users hehaviors. Data mining has become a significant tool to explore such kinds of information in the Internet age. In this paper, FP-Growth algorithm was applied to discover frequent itermsets on cllected data at different periods. Three types of changes for frequent itemsets (i.e., emerging pattern, perished pattern and persistent pattern) were defined to observe the behaviors of Web users. Among the results that data were collected from 104 tutoring Web site, we showed the charactertics of changes of Web users behaviors by analyzing these derived patterns.