竞争情报系统中一种自动文本分类策略——以民用航空客服行业为例  

A Strategy of Automatic Text Categorization in Competitive Intelligence System:Taking Civil Aviation Customer Service Industry as an Example

作　　者：秦宝宝[1] 宋继伟[1] 董尹[1] 牛青[1] 吕美香[1] 陈彬[1] 李骁[1] 

机构地区：[1]南京大学信息管理学院,南京210093

出　　处：《图书情报工作》2012年第24期39-43,119,共6页Library and Information Service

摘　　要：将自动文本分类引入竞争情报系统,并结合民航客服业的实际应用场景展开研究。在对文本分类的理论基础进行研究与分析的基础上,设计一种自动分类策略,其核心是针对传统机器学习方法依赖历史数据的局限,优化特征选择和学习样本获取方法。选用SVM算法,详细描述分类词表构建、分类规则提取、分类模型训练等过程,最终取得较为理想的实际效果。This paper focuses on the application of automatic text classification in the competence intelligence system （CIS） based on the practical application scenario of the civil aviation industry. After researching and analyzing the theoretical basis of automatic text classification, it designs a new strategy. Traditional machine learning method relies on the historical data heavily, which makes the process complex. This paper optimizes the method of feature selection and illustrates how to obtain learning samples. Using SVM as the core algorithm, it gives a detailed description of the process of classification vocabulary building, classification rules extracting, classification model training, and uhimately achieves a satisfactory practical effect.

关 键 词：自动文本分类 竞争情报系统 民用航空客户服务 支持向量机 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]