【摘要】 为解决在互联网文本信息爆炸性增长的前提下,在大规模文本数据中如何发现隐含的、有价值的潜在知识的问题,提出基于多层次文本聚类的文本知识挖掘方法,针对不同规模的文本数据进行不同粒度的聚类,实现不同层次知识的挖掘。针对最广义层次的文本知识挖掘可实现各主题事务划分,针对子级分类数据的文本知识挖掘可发现下一层次主题分类,针对自定义层次的文本知识挖掘可发现该事件中存在的具体细节。对诉求实际数据的分析结果表明,该方法可在所有诉求数据中挖掘出各种诉求主题,精确挖掘出其中的细节问题,为管理者提供数据和决策支持,提高服务效率。更多还原

【Abstract】 To solve the problem that how to find hidden and valuable potential knowledge in large-scale text data on the premise of the explosive growth of text information on the Internet,a multi-level text clustering-based text mining was proposed for different scales.The text data were clustered with different granularities to achieve different levels of knowledge mining.Text mining for the most general level realized the division of each topic transaction.Text mining for sub-classification data found the lower level of topic classification.Text mining for custom level found the specific details in the event.The analysis of actual data in the appeal shows that the proposed method can find the major appeal topics in all data,and accurately mine the details.It can provide results and strategies to managers so as to improve their service efficiency.更多还原