Abstract: The concept of association reduct is firstly introduced by Dominik，which adopts the notion of attribute reduction in rough sets.It aims to mine all association rules from the perspective of global dependencies between attributes in an information system.Since obtaining all association reducts or most informative association reducts has been proved to be NP-hard，two kinds of constructive algorithms for a specific attribute（or considered as a transaction） are put forward，namely，Multi-Single algorithm and Single-Multi algorithm.With these algorithms，some special association rules can be mined，which will be beneficial for evaluating transactions in a given information system.The validity of the proposed algorithms is illustrated by an example.

摘要： 关联约简由Dominik首次提出，其采用粗糙集理论属性约简思想，从全局属性依赖的角度，挖掘出信息系统中所隐含的关联规则。由于求取给定信息系统所有关联约简和最优关联约简已被证明为NP-难问题，针对特定属性（事务）给出了两类关联约简构造性算法：Multi-Single算法和Single-Multi算法，从而挖掘出针对特定事务的关联规则，有益于综合评价各事务在信息系统中的作用。实例分析表明了所提算法的有效性。