多范畴信息的自动分类方法

Automatic Classification Method of Multi-category Information

提出一种基于多范畴属性约简和复合相似度计算的自动分类方法。在分类中引入分类决策属性,然后计算各范畴的决策类和广义决策类,获得多范畴分类属性的约简集族,并依此计算多范畴信息分类对象的复合相似度,并依计算结果对分类对象进行排序和标引,实现自动分类。此方法有效地解决了多范畴不完备信息系统的自动分类问题,通过与Google自建分类系统的对比分析,验证了建立在此方法基础之上的多范畴信息分类系统在查全率和查准率方面明显优于传统的自动分类系统。

A method of automatic classification of multi-category information is proposed, which is based on the attribute reduction of multi-category and calculation of complex comparability. It imports the decision attribute into the classification, gets the reduction set group of multi-category classification attribute by calculating the value of decision class and broad-sense decision class of each category, then calculates respectively the complex comparability of each classification object based on the set group, in the end, fulfills the automatic classification of the object based on result of complex comparability calculating. The results of contrast emulate experimentation show that system based on this kind of method gets better obviously precision ratio and recall ratio than traditional system＇s.