在日常生活當中總是遇到許多決策問題，目前已有許多解決決策問題的方法，例如：作業研究、統計分析、數學方法、統計分析等等，其中最重要的方法之一為多目標決策，傳統的多目標決策包含：輸入、輸出、解決方法，其輸入可能包含多個方案，每一個方案的多個屬性都有一個值，代表有利程度，而問題的輸出往往是要決定每個屬性的重要性，然後透過決策模式，決定最後的單一最佳解或是所有方案的排序，本研究不以產生最佳解與方案排序為目標，而是希望能提出一個摘要化的方法，把各方案的各目標以一個摘要表格表示，此一表格可更進一步轉換成一個雷達圖，用以呈現方案對於各屬性值的分布狀況，我們的整個問題可以定義為當把整個資訊摘要化之後，和原來完整資訊的差異程度大小（資訊遺失程度），而所謂資訊遺失程度＝所有m*n的值和其摘要化取代值的距離之總和，假設使用者指定要把原來m*n的表格摘要為s*t的表格，則我們的目標是要尋找一個s*t的表格，且它的資訊遺失程度最小，這個問題是一個NP hard問題，因此可以用Genetic Algorithms來決定該如何來群集目標及方案。;We always face with various decision-making problems in our daily life. To solve the decision-making problem, more recently there are various approaches have been applied in decision-making problem, such as operation research, statistical analysis, mathematics, sensitivity analysis, and so on. One of the most important approaches is Multi-Criteria decision making (MCDM). Generally, typical forms of MCDM may include three parts, i.e., input, output and solution approach. In general, the input can be expressed as: m alternatives with n criteria. It usually assumes that the underlying input data can be represented as a decision table. Besides, output usually is a single optimal solution and a set of solution preferred by decision maker. In this research, we propose summarization techniques which can summarize decision table and represent it as a summarization table for decision maker to support their making decision. Furthermore, the summarization table can be represented as radar chart to display each alternative for the distribution of the criteria values. The objective of our research is to discover summarization table, and its Degree of Information Lost (DIL) is minimum. Where Degree of Information Lost (DIL) means the sum of distances between original decision table ra×b and table vi×j. Finding a minimum output result is an NP-hard problem. Therefore, we use generic algorithms to improve summarization result.
