Title: 一個領域獨立並具有自動化學習模式的軟體再利用架構
A Domain-Independent Software Reuse Framework with Automatic Learning Model
Authors: 陳明豐
Ming-Feng Chen
王豐堅
Feng-Jian Wang
資訊科學與工程研究所
Keywords: 軟體再利用;晶面法則;學習模式;以相似程度為基的擷取;Software Reuse;Facet Scheme;Learning Model;Similarity-based Retrieval
Issue Date: 1994
Abstract: 「軟體再利用」被視為提昇軟體生產力及品質的一個有效的方法,當我們
要對於一個元件庫進行再利用的動作時,首先必須先將元件庫中的元件進
行分類的動作,分類後的結果才可以因應程式發展者的搜尋,以便於找出
所需的元件。這樣的過程在以往的作法中,需要每個不同的元件庫進行相
同的過程,形成一種「勞力密集」的人力損耗。這篇論文提出一個「軟體
再利用架構」(SRF), 透過這個領域獨立的再利用架構,允許對於各種不
同的元件庫以一個半自動化的方式來進行分類的動作。除此之外, SRF也
提供不同的搜尋等級以滿足對於元件庫不同熟悉度之發展者的需求。並且
利用一個學習的機制來調整分類的結果,以改善原有不恰當的分類,使
得 SRF 能隨時滿足發展者的需求,找出最適合的元件給使用者。
Software reuse is an effective method to improve the software
productivity and quality. Since the object-oriented
programming is getting popular and OO libraries are getting
bigger, the libraries need be classified to speed up the
search. In previous classification methods, each library goes
on the same classi- fication process. However, different
domains have different libraries. Such kind of
classifications alone cannot provide an effective search
environment. In this thesis, we propose a software reuse
framework (SRF) to overcome the drawback. Based on the
domain-independent features of SRF, the classification
process becomes semi-automatic. On the other hand, SRF
provides four searching levels to satisfy the demands of
developers of different familarity with the library. SRF
also provides a learning mechanism to let developers adjust
the improper classi- fication results. Consequently, SRF is
expected to find out the desired component faster.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT830392020
http://hdl.handle.net/11536/58940
Appears in Collections:Thesis