【作者】 雷兵；

【导师】 盛步云；

【作者基本信息】 武汉理工大学 ， 工业工程， 2013， 博士

【摘要】 随着互联网的普及,越来越多的客车企业利用营销型企业网站开展商务活动。与传统营销相比,企业网站记录了客户的访问行为数据,包括点击流和用户产品评论。这些海量的、干净的访问行为数据,为企业获取客户需求提供了重要的数据源。然而,这些数据并不能直接利用。如何从访问行为数据中提取有用的客户需求信息,已成为客车企业网络营销需要解决的重要问题之一。本文旨在研究基于客车企业营销型网站访问行为数据的客户需求获取方法,具体研究了以下三个问题：(1)研究了客车产品客户需求知识的表示及构建方法。针对从客车企业营销型网站访问行为中提取客户需求效用的目标,提出了基于本体的客车产品客户需求知识表示及构建方法,包括概念模型、知识库模板和知识获取三部分。其中,针对概念模型,提出了从客户需求形式化表示到基于本体的知识表示的映射方法；针对知识库模板,提出了将其结构分为逻辑结构和采集点的两类表示方式；针对知识获取,提出了基于包装器归纳的营销型网站网页模板抽取方法。(2)研究了客车产品评论文本中客户需求特征的抽取方法。针对客车企业营销型网站产品评论文本的特点,提出了需求触发词的概念,并在此基础上提出了一种抽取方法,该方法包括产品特征词识别、基本需求检测以及基本需求内容识别等三部分。其中,对于基本需求,首先将其分为参数型需求、布尔型需求、情感型需求等三种类型,随后研究了三种基本需求的需求触发词特点及获取方法。(3)研究了基于营销型企业网站访问行为的客车产品客户需求聚类方法。针对客车企业营销型网站访问行为数据,对其进行了客户需求效用分析,并在此基础上提出了一种根据客户需求相似偏好的客户聚类方法,该方法包括数据建模、客户聚类及其特征归纳等三方面。其中,对于数据建模,设计了基于“访客-客户需求”矩阵的客户需求效用数据模型,并提出了两步转化的构建方法。综合上述研究,设计并实现了一套客车企业客户需求获取系统,并应用于国内某大型客车企业,取得了良好的效果。更多还原

【Abstract】 With the spread of the Internet, more and more bus enterprises are carrying out business activities via enterprise web site of marketing type. The mass, clean visiting behavior data provides an important data source for enterprises to gain customer needs. However, these data cannot be used directly. How to extract useful customer demand information from visiting behavior data is becoming one of the important problems that the bus enterprises need to be solved.This paper aims to research on customer demand data acquisition method of access behavior based on bus enterprise website of marketing-type. Specificly, the following three key problems are included:(1) Study the customer requirement of coach products on knowledge representation and construction method. Aiming at customer needs utility extracted from coach enterprise web marketing access, the paper proposes customers demand of coach products on knowledge representation and construction method based on ontology, including concept model, knowledge base template and knowledge acquisition.Among them, a mapping method is proposed in view of the conceptual model to convert customer needs formalization into knowledge representation based on ontology; two kinds of representation is proposed in view of the knowledge base template that it could be divided into logical structure and collection point; a web page templates extracting method of marketing website is proposed in view of the knowledge acquisition that the extracting methods should based on wrapper induction(2) Extraction method of customer requirement characteristics in coach product reviews text. In terms of characteristics of remarking texts in bus enterprise website of marketing type, this paper puts forward the concept of demand trigger and puts forward a kind of extraction method consisting of three steps based on the concept: product key word identification, basic demand detection and basic needs content identification. Among which, the basic demand detection can be divided into three types-parameter type needs, Boolean type needs, and emotional type needs, and studies needs trigger word characteristics and acquisition method for basic demand the of three kinds(3) Customer requirement clustering method of car product based on access behavior to enterprises website of marketing-type. This paper makes pragmatic analysis for customer needs based on access behavior data for bus enterprise website of marketing-type, and puts forward a customer clustering method of similar preference accordingly. The method includes three aspects-data modeling, customer clustering and characteristics summarizing. This paper designs "visitors-Pages" customer demand utility data model, and put forwards two-step conversion method.A set of requirements elicitation system for bus companies is designed and realized based on the above research. It has been applied one large-scale coach company and has achieved good effect.更多还原