基于对话内容的交互型文本会话主题挖掘

doi:10.11959/j.issn.1000-0801.2016238

电信科学 ›› 2016, Vol. 32 ›› Issue (9): 139-145.doi: 10.11959/j.issn.1000-0801.2016238

基于对话内容的交互型文本会话主题挖掘

彭杰¹,石永革¹,高胜保²

¹ 南昌大学信息工程学院，江西南昌 330029
² 中国电信股份有限公司江西分公司，江西南昌 330029

出版日期:2016-09-15 发布日期:2016-10-20
基金资助:
国家自然科学基金资助项目;江西省科技计划基金资助项目

Session topic mining for interactive text based on conversational content

Jie PENG¹,Yongge SHI¹,Shengbao GAO²

¹ Information Engineering College,Nanchang University,Nanchang 330029,China
² Jiangxi Branch of China Telecom Co.,Ltd.,Nanchang 330029,China

Online:2016-09-15 Published:2016-10-20
Supported by:
The National Natural Science Foundation of China;Science and Technology Program of Jiangxi Province of China

摘要/Abstract

摘要：

传统的主题挖掘模型一般仅从交互型文本中挖掘出文档主题，为了能够从中挖掘出会话主题并提高挖掘模型的普适性，提出了一种基于对话内容的交互型文本会话主题生成模型。首先通过分析交互型文本的特征，基于主题树的概念，定义了一个5层结构的对话生成树。以此为基础，再基于LDA构建会话主题生成模型(ST-LDA)。最后采用吉布斯抽样法对ST-LDA进行推导，得到会话主题及其分布概率。使用实际数据进行验证，结果表明，ST-LDA模型可以从交互型文本中有效地挖掘出会话主题。此外，成果可以降低分类算法的复杂度，回溯主题—参与者关联关系，具有较好的普适性。

关键词: 交互型文本, 对话内容, 会话主题挖掘, 对话生成树, LDA

Abstract:

Traditional theme mining model generally digs out the document theme from the interactive text only.In order to explore the session topic and improve the universality of mining model,a kind of interactive text session topic generation model based on the content of the dialogue was put forward.Firstly,by analyzing the characteristics of interactive text and based on the concept of topic tree,a dialog spanning tree was defined with a five-layer structure.Based on this and LDA,the model of session topic generation(ST-LDA)was built.At last,Gibbs sampling method was adopted to deduce the ST-LDA and obtaining session topic and its distribution probability.The results show that the ST-LDA model can dig out a session topic effectively from the interactive text.Besides,the results can reduce the complexity of the classification algorithm and can be back to the theme—participants association.It also has a good universality.

Key words: interactivetext, conversationcontent, sessiontopicmining, dialogspanningtree, latentDirichletallocation

彭杰,石永革,高胜保. 基于对话内容的交互型文本会话主题挖掘[J]. 电信科学, 2016, 32(9): 139-145.

Jie PENG,Yongge SHI,Shengbao GAO. Session topic mining for interactive text based on conversational content[J]. Telecommunications Science, 2016, 32(9): 139-145.

图/表 6

图

图2

图3

表1

图4

表2

参考文献 14

[1]	林小俊, 张猛, 暴筱 ,等. 基于概念网络的短文本分类方法[J]. 计算机工程, 2010,21(11): 4-6. LIN X J , ZHANG M , BAO X . Short-text classification method based on concept network[J]. Computer Engineering, 2010,21(11): 4-6.
[2]	张志飞, 苗夺谦, 高灿 . 基于概念网络的短文本分类方法[J]. 计算机应用, 2013,33(6): 1587-1590. ZHANG Z F , MIAO D Q , GAO C . Short text classification using latent Dirichlet allocation[J]. Journal of Computer Applications, 2013,33(6): 1587-1590.
[3]	孙建旺, 吕学强, 张雷瀚 . 基于语义与最大匹配度的短文本分类研究[J]. 计算机工程与设计, 2013,34(10): 3613-3618. SUN J W , LU X Q , ZHANG L H . Short text classification based on semantics and maximum matching degree[J]. Computer Engineering and Designing, 2013,34(10): 3613-3618.
[4]	邬晓钧, 郑方, 徐明星 . 基于主题森林结构的对话管理模型[J]. 自动化学报, 2003,29(3): 275-283. WU X J , ZHENG F , XU M X . Topic forest based dialog management model[J]. Acta Automatica Sinica, 2003,29(3): 275-283.
[5]	ABDUR R , BASHER M A , BENJAMIN C M F . Analyzing topics and authors in chat logs for crime investigation[J]. Knowledge ＆Information Systems, 2014,39(2): 351-381.
[6]	蒋莹莹 . 基于对话的主题提取研究[D]. 武汉:华中科技大学, 2009. JIANG Y Y . Research on topic extraction based dialogue[D]. Wuhan:Huazhong University of Science＆Technology, 2009.
[7]	房冠南 . 面向对话语料的标签推荐[D]. 北京:北京邮电大学, 2012. FANG G N . Tag recommendation for dialogue corpus[D]. Beijing:Beijing University of Posts and Telecommunications, 2012.
[8]	张晨逸, 孙建伶, 丁轶群 . 基于MB-LDA模型的微博主题挖掘[J]. 计算机研究与发展, 2011,48(10): 1795-1802. ZHANG C Y , SUN J L , DING Y Q . Topic mining for micro-blog based on MB-LDA model[J]. Journal of Computer Research and Development, 2011,48(10): 1795-1802.
[9]	BLEI D , NG A , JORDAN M . Latent dirichlet allocation[J]. Journal of Machine Learning Research, 2003(3): 4-5.
[10]	吕超镇, 姬东鸿, 吴飞飞 . 基于LDA特征扩展的短文本分类[J]. 计算机工程与应用, 2015,51(4): 123-127. LV C Z , JI D H , WU F F . Short text classification based on expanding feature of LDA[J]. Computer Engineering and Applications, 2015,51(4): 123-127.
[11]	QUAN X , LIU G , LU Z ,et al. Short text similarity based on probabilistic topics[J]. Knowledge and Information Systems, 2010,25(3): 473-491.
[12]	GRIFFITHS T L , STEYVERS M . Finding scientific topics[J]. Procof the National Academy of Sciences of the United States of America, 2004(101): 5228-5235.
[13]	GRIFFITHS T L , STEYVERS M . Finding scientific topics[J]. PNAS, 2004,101(1): 5228-5235.
[14]	樊兴华, 孙茂松 . 一种高性能的两类中文文本分类方法[J]. 计算机学报, 2006(1): 124-131. FAN X H , SUN M S . A High performance two-class chinese text categorization method[J]. Chinese Journal of Computers, 2006(1): 124-131.

参数	购物	数码产品	股票	旅游	招聘	健康医疗
正确率P	0.82	0.838	0.896	0.814	0.859	0.866
召回率R	0.731	0.83	0.862	0.798	0.926	0.975
F1	0.772	0.834	0.878	0.802	0.889	0.915

主题	正确率P平均值	召回率R平均值	F1平均值
文档主题	0.752	0.579	0.654
对话主题	0.832	0.849	0.84
会话主题	0.849	0.853	0.848

基于对话内容的交互型文本会话主题挖掘

Session topic mining for interactive text based on conversational content

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 6

参考文献 14

相关文章 0

Metrics

推荐阅读 0