在過去的幾年中，有眾多的研究在探討購物籃分析(Market Basket Analysis)，購物籃分析是藉由粹取交易資料庫中的關連性，作為挖掘客戶購買模式之絕佳方法。現今的企業中，在世界不同區域擁有子公司、分公司以及代理商已經是相當普遍的經營模式，然而，只針對單一資料庫而設計之傳統關聯規則演算法已經不適用於這樣的多商店環境，為了解決傳統關聯規則演算法對於多商店環境的不適用性，一個多商店關聯規則演算法變由此而產生。 倘若我們就決策者的角度而言，規則不僅僅是需要被發掘，規則的易讀性以及其他可利用性更是重要，除此之外，不同時空組合之下的規則對於不同階級的決策者也會有著不同的意義，例如：一位跨國企業的總裁，他所感興趣的資訊是包含在較大範圍的時空組合之下，如2005年全球銷售的產品中隱藏著什麼樣的規則，而一位區域的決策者，他所在意的資訊將會包含在較小的時空組合之下，如在春天的日本其銷售的產品中，隱藏著什麼樣潛在的規則。在不同的時空組合之下，將會隱含著不同的零售知識，而由於時空因素的差異，不同階級的決策者需要在不同的情境中運用不同的零售知識。本論文研究的目標，即是滿足這樣的需求，藉由延伸在多商店環境下的關連規則方法，我們提出一個可以在不同時空維度組合之下，找出關連規則的方法，來滿足企業內不同的決策需求。實驗的結果得知，本論文提出的方式能達到運算上之效率。 Over the past few years, a considerable number of studies have been made on market basket analysis. Market basket analysis is a useful method for discovering customer purchasing patterns by extracting association from stores’ transaction database. In the business world today, it is common for a company to have subsidiaries, branches, or dealers in different geographical locations; hence, considering only the association rule of an individual store is not suitable for a multi-store environment. Therefore, a store-chain association rule is proposed in [30] to compensate this over-generalization. The rules discovered in [30] are represented by way of rule-by-rule, that is, the store-chain association rule is a rule-oriented method in multi-store environment and each discovered rule will be attached with a series of pairs of time-and-place in which the contexts each rule apply to. However, from the perspective of a business strategist, not only do the rules have to be discovered, but the rules also must be readily interpreted for easy reading and further usage. In addition, different executive personnel will require different interpretation of the rules for different scenarios because under different granularities of time-and-place, the retailing knowledge will be different and the goal of our work is to satisfy such dynamic needs. By extending of the existing techniques of mining association rules in a multi-store environment, we develop an algorithm that can find the rules under different granularities of time-and-place to satisfy the different demands of different decision makers within the company. Our empirical evaluation shows that the proposed method is computationally efficient.
