這篇論文嘗試以企業資源規劃系統中的交易資料來產生更多的績效評估指標。在企業資源規劃系統中，資料以實體-關係模型的形式存放其中，但是在查詢系統中資料則以星狀綱要的形式來表示其查詢結果，如何減少這之中的差異?第一，我們透過詞頻逆向文件頻率的方法去計算關鍵字的權重，以分類實體-關係模型中的實體，然後我們可以將星狀綱要中的關鍵數字引進實體分類的類別裡。第二，我們透過已存在查詢系統中的公式，為屬性附加新的公式結構。第三，結合屬性與關鍵數字所產生的公式，我們可以透過系統產生更多績效評估指標。最後，將此結果轉換成星狀綱要的形式，以落實在系統，並藉此改善查詢系統。 This paper tries to generate more business performance indices from transaction data in ERP (Enterprise Resource Planning) system. The data is designed in ERP system as an ER-Model (Entity Relationship Model) but the query system is used Star Schema to represents the results. How to reduce this gap? The first of all we use the TFIDF (Term Frequency Inverse Document Frequency) to count the number of Key Words for the purpose of classifying Entity in ER-Model. Then we can include the Key Figure from the Star Schema by the Key Word classes. Second, we take the formulas that have designed in query system for the basis to bring the Attribute into this formula structure. Third, we generate more KPIs (Key Performance Indices) from including Attribute and Key Figure. Final, we transfer the results into the Star Schema for performing our results and improving the query system.
