基于贝叶斯模型的专利分类  

Patent categorization based on Bayes model

作　　者：郭炜强[1] 文军[1] 文贵华[1] 

机构地区：[1]华南理工大学计算机研究所,广东广州510641

出　　处：《计算机工程与设计》2005年第8期1986-1987,1996,共3页Computer Engineering and Design

基　　金：国家自然科学基金项目(60003019)

摘　　要：朴素贝叶斯分类器理论基础好,分类精度高。利用特征词权重函数修改朴素贝叶斯分类器,进而利用它实现专利文本的自动分类,不仅减少了专利人工分类的工作量和分类错误,而且为技术跟踪、竞争分析等提供了有效支持。实验与应用表明改进的朴素贝叶斯分类器用来解决专利分类是有效的。Based on naive bayes classifier having solid theory foundation and high accuracy rate of classification, the classical naive bayes classifier was firstly improved by using term weight function in text, and then the patent categorization was implemented. This approach not only reduced manual labor and the categorization error, but also supported for the technology tracing, competition intelligence etc.The experiments and applications illustrate that the improved naive bayes classifier can be utilized to classify patents efficiently.

关 键 词：专利 朴素贝叶斯分类器 专利分类 特征词权重 文本挖掘 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程] G306[自动化与计算机技术—控制科学与工程]