【作者】 李浩；

【导师】 李华；

【作者基本信息】 重庆大学 ， 工程硕士（计算机技术领域）（专业学位）， 2016， 硕士

【摘要】 博客又被称为网络日志,是一种个人信息的发布形式。从个人角度来说,博客是一种表达思想,以及与他人分享有价值资源的一种方式。随着博客用户的激增,越来越多的各行各业的人员利用博客进行广泛的交流、知识的共享。当今博客在人们获得资源的途径中占有重要位置。随着互联网的发展,各种应用都允许用户进行评论。用户的评论则直接体现了用户的情感倾向性。博客应用也不例外,用户总会在看到一篇高质量的博客时忍不住留下自己的敬佩之情。因此可以认为含有很多好评的博客是一篇优质的博客文章。本文通过分析博客文章评论的情感来优化博客搜索结果,使得用户在搜索相关博客时,搜索结果在相关度的基础上按照博客质量进行排序。本文主要工作如下:首先,需要准确的抓取博客的评论内容。由于一些博客网站的评论内容是由Ajax动态生成的,传统的网络爬虫不能完成动态页面的抓取工作,系统中采用phantom JS来改进传统的爬虫。为了从抓取到的网页中准确提取出来评论的内容,本文采用了一种基于特殊标识的最大DOM树算法。该算法可以准确地抽取出来网页中评论的内容。其次,准确提取到评论内容之后就可以进行情感倾向性分析。情感分析的目的是生成一个整体的评论得分,该得分作为搜索博客返回结果的排序依据。本文使用了文本分类的方法进行文本情感倾向性分析,并通过构建博客评论领域的情感词典和改进特征提取算法来提高文本分类的精度。最后,本文在Nutch开源软件的基础之上设计并实现了一个博客搜索引擎系统(CBlog)。Nutch系统本身只考虑了检索关键词因子以及链接分析因子来影响文档得分,CBlog系统增加了对评论内容的情感分析因子来进一步优化检索结果。CBlog通过为用户搜索时返回高质量的博客,使得用户有更好的搜索体验。更多还原

【Abstract】 Blog also known as web logs, is a way to release the personal information. From a personal point of view, blog is a way of expressing ideas and sharing valuable resources with others. With the proliferation of blog users, more and more people use blog to communicate with each other and share knowledge. Today,blog occupies an important position as the way of people access to the important resources.With the development of Internet, various applications allow users to comment.The comments of users directly reflect the bias of user’s emotional. Blog application neither exception, when we read a good blog, we’ll give a high evaluation. So we can think a blog that contain a lot of favourable comments is a good one. In this paper, we optimize the blog search results through the analysis of blog comments, so that when users in the search of related blogs, search results are sorted according to the quality of the blog on the basis of the correlation degree. The main works are as follows:First, the system need to grab blog comments accurately. Due to some of the comments in the blog site are dynamically generated by the Ajax, traditional web crawler can not complete the grasping job of dynamic pages, so we will use phantom JS to improve the traditional crawler. In order to extract the contents of the comments from the captured web pages, this paper uses a maximum DOM tree algorithm to solve this problem. This algorithm can accurately extract the comments of the web page.Second, after the completion of grasping the comments is to analyze the text orientation.The purpose of the analysis is to generate an overall review score,which will be used as a guide for the sort of the search blog results. In this paper, the text categorization method is used to analyze the emotional tendency of the text and improve the accuracy of text classification by constructing the sentiment dictionary and the improved feature extraction algorithm.Finally,this paper designs and implements a blog search engine system(CBlog) based on the Nutch open source software. Nutch itself only considers the key words and link analysis factor to calculate the score of the document,CBlog system increases the emotion analysis factor. By returning the high quality blogs, CBlog improve the user’s search experience.更多还原