本研究主要在應用人工智慧中專家系統知識庫建構的技術理念於平版印刷上的網點變異現象控制。系統中主要可分為三部分：１. 知識庫 (knowledge Base)２. 推論機制 (Inference Engiine)３. 人機介面 (User Interface)。其中知識庫記了一些平版印刷品質管制相關的基本資料；而推論機制中含有一些故障現象及判斷的經驗法則 (HeuristicRules)；另在人機介面中，將透過視窗 (Windows) 軟體下的一套專家系統建構工具Nexpert Object，使得相關的品管知識、資訊得以運作與執行。
在知識工程 (Knowledge Engineering) 中，網點變異的知識擷取(Knowledge Acquisition) 與知識表現 (Knowledge Representation)，將透過Nexpert Object的一些編輯器如法則編輯器 (Rule Editor) 或物件編輯器 (Object Editor) 等將網點擴大、縮小、蠕印、雙影、拖網、填塞等網點變異判斷變成一些物件 (Objects) 與法則 (Rules) 的模式，並經由Donald A. Waterman所提出之知識擷取五個階段性步驟完成雛型系統的建構。
本研究在現階段中獲致的主要結論為：１.以網點變異現象控制作為平版印刷品質管制的問題，適予發展專家系統。２.可作為新進的平版印刷機操作人員的訓練教材及品管圈相關人員的決策支援。３.保存平版印刷網點變異知識及故障點排除經驗。

Printing Quality Control Expert System (PQCES) mainly describes the use of an expert system, a branch of Artificial Intelligence (Al), in controlling offset printing dot variations. The system can be largely divided into three parts: t base, which data and fault judgements can be acquired from some quality control devices; the inference engine, in which heuristic rules are constructed accorcding to experienced offset printing operators and appropriate paper records ; and the user interface, which use the expert system tool, Nexpert Object, to process knowledge and information.
In knowledge engineering, the knowledge acquisition and representations are based on the editors of Nexpert Object. For instance.several quality control devices, such as dot gain scale, are classified into objects and rules. The heuris- tic rules are not crystal, transparrant or unchanged. These rules combine to form the essential part of PQCES.
The purpose of PQCES is not only to serve as a Decision Support System (DSS) for offset printing operators, but also, to create a database by accumulating the trouble-shooting experiences porvided by printing operators and quality control personnel in related departments.