A General Framework for Discovering Sequential Patterns Based on Fuzzy Concept

一般化的架構來挖掘模糊循序樣式

陳彥良(Yen-Liang Chen) ； 黃正魁(Tony Cheng-Kui Huang) ； 胡蕙玲(Hui-Ling Hu)

《南亞學報》 25期 (2005/12) Pp. 45-58

https://doi.org/10.6989/JN.200512.0045

摘要

隨著資料的大量增加，資料探勘已經被使用在處理資料過剩的問題，並且在既有的資料中，去挖掘有用的、新的和具有潛力的樣式。然而，我們在挖掘量化型的資料時，卻產生傳統不是0就是1的切割問題，而這問題是傳統資料探勘方法無法解決的。為了這個問題，已經有許多學者，運用模糊集合來解決。另一方面，學者U. Fayyad等人，在資料探勘的研究中，有提出一個一般化、以程序為主的架構，來概化知識發現的程序，但很遺憾，他們的架構無法適用在模糊循序樣式探勘中，因此，我們修改了他們的架構，提出另一個架構來一般化這樣的探勘程序。最後，我們提出一個應用來證明這個新架構的可用性。

關鍵字

架構 ； 資料探勘 ； 循序樣式 ； 模糊集合

並列摘要

With the increase of data, data mining has been introducing to solve the overloading problem and to discover valid, novel, potentially useful patterns in existing data. However, to discover quantitative data, we face a sharp boundary problem that the traditional data mining techniques can not overcome. To pinpoint this problem, a lot of researches have been applied fuzzy sets to discover patterns, especially in sequential patterns . As we know, U. Fayyad et al proposed a unifying process-centric framework for Knowledge Discovery in Database (KDD) which can explain the process from top to toe for the traditional data mining problems. Unfortunately, it fails to be workable in part to describe the process for discovering sequential patterns based on fuzzy concept. Therefore, to have more general viewpoint for discovering of that sort patterns, we devote to proposing a general framework modified from the KDD process. An application is proposed to demonstrate that the framework can be a generalization of the realm problem.

並列關鍵字

framework ； data mining ； sequential patterns ； fuzzy sets