摘要： 传统的关联规则数据挖掘的支持度-置信度框架存在着弊端。一方面，它不能发现同时满足支持度和置信度而其前提和结论却相互独立的规则；另一方面，也不能区分正负关联规则。该文提出了一种评价关联规则的可量化的标准，进一步提出一种同时挖掘正负关联规则的框架，用此框架来分析调研问卷。

关键词: 负关联规则, 相关性, 数据挖掘, 问卷

Abstract: The conventional framework for mining association rules is the support-confidence framework which has some limitations. For one thing, it can not prune such useless rules as satisfy both minimum support and minimum confidence with their antecedents and consequents independent. For another thing, it can neither separate the negative association rules from the positive ones. The purposes of this paper are to find a quantitative criterion for mining both positive and negative association rules, and further put forward a novel framework whose efficiency is tested in analyzing a questionnaire.

Key words: Negative association rules, Correlation, Data mining, Questionnaire