【作者】 王德兴；

【导师】 胡学钢；

【作者基本信息】 合肥工业大学 ， 计算机软件与理论， 2003， 硕士

【摘要】 数据库中的知识发现是当前涉及人工智能、数据库等学科的一门相当活跃的研究领域。数据挖掘是从数据中提取人们感兴趣的、潜在的、可用的知识，并表示成用户可理解的形式。关联规则挖掘是数据挖掘的一个重要分支，是描述数据库中数据项(属性、变量)间存在的潜在关系。 概念格通过概念的内涵和外延及泛化和例化之间的关系来表示知识，因而适用于从数据库中挖掘规则的问题描述。在概念格的内涵中引入等价关系并将其外延量化，得到量化概念格。本文是基于量化概念格的关联规则挖掘为中心而展开的。 本文中创新性的主要内容如下： ①提出了基于量化概念格的关联规则及基于兴趣度加权的量化概念格的关联规则挖掘的思想、算法以及性能分析。基于兴趣度加权的量化概念格关联规则挖掘选择大于兴趣度加权阈值的项目构造量化概念格，在此基础上交互地挖掘感兴趣的关联规则。 与Apriori算法相比，两种方法所挖掘出的规则结果完全吻合，前者具有较好的时间性能，规则表示更直观，减少了算法的搜索空间和计算量，提高了挖掘的效率和准确性。 ②改进了传统的购物篮分析，由于传统的购物篮分析只关心顾客是否购买商品，忽略其购买的数量，因而在实际应用中，有很大的局限性，在本文中，不仅要关心顾客是否购买商品，而且考虑顾客购买的数量，在传统的购物篮分析中，引入兴趣度加权思想，并提出了如何获取兴趣度加权阈值的方法，因此在改进了传统的购物篮分析基础上，基于量化概念格所挖掘出的关联规则有更贴近于实际和应用价值。更多还原

【Abstract】 Knowledge discovery in databases (KDD) is a rapidly emerging research field relevant to artificial intelligence and database system. Data Mining is the process of mining the interesting, potentially useful, valid and understandable knowledge in data. Association rule mining is an important sub-branch of Data Mining, which describes the potential relationships between attributes and variables in databases.Concept Lattice represents knowledge with the relationships between the intension and the extension of concepts, and the relationships between the generalization and the specialization of concepts, thus it is applied to the description of association rules mining in databases. The Quantitative Extended Concept Lattice (QECL) evolves from concept lattice by introducing equivalence relationships to its intension and quantity to its extension. The paper is presented by the main ideas, the research on the model of association rules mining based on quantitative extended concept lattice.There are original main ideas in the paper zs follows:(1) The main ideas, algorithm and capability performance analysls of the model of association rules mining based on quantitative extended concept lattice and that of association rule mining by interest-weighted have been proposed, Association rule mining by interest-weighted on quantitative extended concept lattice is an algorithm that we choose those items whose interest-weighted are bigger than the interest-weighted threshold, generate QECL, then mine mutually interest-weighted association rules according to user’s interests.Compared with Apriori algorithm, the uniform results of association rules have been obtained by the two methods, but association rules mining by interest-weighted on quantitative extended concept lattice has high quality of time complexity, shows association rules more brief and visual, reduces much searching space and computation of the algorithm, then improves the efficiency and veracityof association rules mining.(2) Traditional marked-basket analysis has been improved, Since it only cares for that the customer have bought something or not, ignores the quantity of those bought, There are some more limitations in practical application. In the paper, I am concerned about both cases, then introduce the idea of interest-weighted to marked-basket analysis, put forward the algorithm how to acquire the interest-weighted threshold, therefore, The association rules mining by interest-weighted on quantitative extended concept lattice is more practical.更多还原