傳統上,萃智理論(又稱為發明問題解決理論)是基於邏輯推理來辨識機會和解決問題。除此之外,不同的萃智工具有不同的問題表達模式,且沒有統一的方式來表達問題。故本研究提出問題特性陣列可以統一問題表達模式、以解答陣列充分表達在萃智理論背景的解答。依據萃智原理“相同類型問題有相類似的答案或可利用相似的流程解決”,故本研究以功能與屬性為基礎問題特性陣列與解答陣列的方式,可以模式化萃智問題的解題流程,此解題流程也可以被視為分類問題。經由本研究解題流程,藉由數理化的演算法計算,可自動搜尋萃智理論的通用解答,而不依賴於人類的知識和經驗,透過數理的計算方式可以成功辨識合適的發明原則與相關演化趨勢的解答。 本研究的貢獻含:1) 建立問題特性陣列和解答陣列,成功整合和統一的問題和解答表達方式。因此,允許平行地蒐集的問題解答的資料庫,與問題與問題解答之間相似程度;2) 啟用電腦輔助自動辨識相關的演化趨勢和發明原則;3) 導入數理的計算至傳統萃智理論中,在未來可以使用很多的分類工具來辨識萃智解答模式;4) 提供不需要依靠主觀經驗與知識的辨識解答模式。
Traditionally, TRIZ (Theory of Inventive Problem Solving) is based on logical reasoning to identify opportunity and solve problems. In addition, different TRIZ tools have different ways of problem modeling representations. There was no unified model to represent problems. This research proposed a form of unified problem model, in terms of Problem Characteristic Array (PCA), and a form of Solution Array(SA) to fully represent a problem with its solutions in the context of TRIZ problem solving characteristics. With the proposed function-attributed based Problem Characteristic Array and Solution Array, the research was able to model the TRIZ problem solving process from PCA to SA as similarity problems based on the TRIZ concept of “Like problem like solution”. The process of identifying model of solution can also be regarded as classification problems. In this way, searching of TRIZ generic solutions can be automatically done by computational algorithms without relying on human knowledge and experience. The approach was successfully tested with good performance by using mathematical computation to identify proper Inventive Principles and relevant trends for generic solutions. The contributions of this research include: 1) Establishing a Problem Characteristic Array and Solution Array enabling an integrated and unified representation of heterogeneous problems and solutions thus allowing parallel collections of problem-Solution database and similarity rating between problems and their solutions; 2) Enable the capability of computer-aided automatic identification of relevant Trends and Inventive Principles for problem solving; 3) Introducing mathematical computation into otherwise traditional logical TRIZ. This open up the possibility to use many classification tools to identify TRIZ model of solutions in the future; 4) Providing objective identification of model of solution without relying on subjective experience and knowledge.