基于语义分析的网络服务智能交互系统研究

发布时间：2018-11-21 09:33

【摘要】：互联网的日益普及使得人们对网络的依赖程度越来越高,通过搜索引擎、官方技术支持网站、提问回答互动平台等各种途径获取自己所需要的信息,已经成为一种时尚一种习惯。目前网络环境下,淘宝、京东这样的C2C或者B2C网上交易平台已经推出了“店小二”或者售前咨询服务,中国工商银行等各大银行也推出了常见问题汇总区,惠普售后支持论坛和中国移动都已经相继推出了在线咨询平台,就是为了应对用户获取信息的需求。但是目前的这些交互方式都很局限,交易平台一般采用人工回复来应对问题,而服务支持网站则以常见问题连接的方式来处理大多数问题,汇总以上两种典型的回复方式可以看到两个明显缺陷：一是时效性差,用户在进行信息咨询的过程中很可能还在同时浏览其他的网页和商品,当回复不及时则用户会失去兴趣；二是服务质量无法保证,特别是用户找不到自己想要咨询的主题时同样会失去耐心。基于以上网络服务提供的现状,本文提出了智能交互系统这样一个思路,并对最基本的基于文本的智能交互过程做了一些研究。本文倡导的理想化的智能交互系统是基于网络环境的,主要是功能是用户描述自己的信息需求,系统通过一系列算法来匹配用户需求,从而给出一个用户满意的回复。本文的主要内容就是整理前人在问题和回复方向的相关研究成果,概括出详细的基于概率潜在语义分析的智能交互过程,并通过程序实现。在此基础上结合蚁群算法,提出了基于蚁群算法结合概率潜在语义的智能交互算法,同样编程实现并统计程序运行时间,与基于概率潜在语义分析的交互过程进行对比,效率有显著提高。在此算法的应用的基础上,本文总结了系统设计方面的一些问题以及系统自主更新和扩展的一些设想和方法。在总结和展望部分主要概括文章研究的结论,提出系统处理文本的核心算法选择、系统融入自主学习型算法思想等几条思路,智能交互系统是一个很庞大的工程,需要更多的人来做更多的工作。
[Abstract]:With the increasing popularity of the Internet, people rely more and more on the Internet. They can obtain the information they need through search engines, official technical support websites, interactive platforms for answering questions and so on. Has become a fashion a habit. At present, under the network environment, C2C or B2C online trading platforms such as Taobao, JingDong have launched "shop first class" or pre-sale consulting services, and various major banks, such as Industrial and Commercial Bank of China, have also launched common problem gathering areas. Hewlett-Packard Aftermarket support Forum and China Mobile have launched online consulting platforms to respond to users' demand for information. However, these interactions are very limited at present. Trading platforms generally use manual responses to deal with problems, while service support sites handle most of the problems in the form of frequent problem connections. Summarizing the above two typical response methods can be seen two obvious defects: first, poor timeliness, users in the process of information consultation is likely to also browse other web pages and products, when the reply is not timely, the user will lose interest; Second, the quality of service can not be guaranteed, especially if users can not find their own topics to consult the same impatience. Based on the status quo of the above network services, this paper puts forward the idea of intelligent interactive system, and does some research on the most basic text-based intelligent interaction process. The idealized intelligent interactive system advocated in this paper is based on the network environment. The main function is that the user describes his information requirements. The system matches the user's needs through a series of algorithms, and then gives a satisfactory response. The main content of this paper is to sort out the related research results of the former people in the direction of question and reply and generalize the detailed intelligent interaction process based on probabilistic latent semantic analysis and realize it through the program. Combined with ant colony algorithm, an intelligent interaction algorithm based on ant colony algorithm and probabilistic latent semantics is proposed. It is also programmed to realize and calculate the running time of the program, which is compared with the interactive process based on probabilistic latent semantic analysis. Efficiency has been significantly improved. On the basis of the application of this algorithm, this paper summarizes some problems in system design and some assumptions and methods of independent updating and extending of the system. In the part of summing up and looking forward to the conclusion of the paper, the author puts forward some ideas, such as the selection of the core algorithm for the system to deal with the text, the integration of the system into the autonomous learning algorithm, and so on. The intelligent interactive system is a huge project. More people are needed to do more work.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.1

【参考文献】