微博热点话题抽取及其情感分类

发布时间：2018-04-18 17:00

本文选题：微博 + 话题抽取　；参考：《哈尔滨工业大学》2014年硕士论文

【摘要】：随着Web2.0时代的到来，尤其是社交媒体的飞速发展，越来越多的互联网用户开始在网上发布自己对一些话题的看法。如何从海量的数据中抽取出当前的热门话题，并就用户对该话题的看法进行分析成为一个非常有意义的研究点。本文就针对这个研究点进行了深入的分析。微博热点话题与普通话题的定义有所不同。在微博中，热点话题是指在短时间内被大量用户高度关注的话题。这些话题可以是真实的社会事件，也可以与社会事件无关，但仍在短时间内获得了很高的兴趣度，为大量微博用户所关注讨论。本文将该研究点分成两阶段任务：微博热点话题抽取与话题情感分类。关于微博热点话题抽取，本文提出了基于话题标签的微博热点话题抽取算法。该算法利用启发式规则和聚类算法来实时抽取微博热点话题。算法主要包括三个部分：话题发现、话题聚类以及话题排序。实验证明该方法抽取的微博热点话题具有很高的准确率。关于话题情感分类，，本文提出了基于篇章关系的微博情感分类算法。该算法共归纳出四种可能对情感分类的结果产生影响的篇章关系，并利用与这些篇章关系相对应的关系指示词来识别微博文本的主干成分。在此基础之上，探索了多种不同类别的微博情感分类特征。此外，本文还构建了一个面向微博领域的大规模情感词典。实验结果表明，本文提出的算法在微博情感分类上能够取得不错的效果。此外，我们还将上述的两个算法有机融合，构建了一个用户情绪在线分析平台。该平台能够实时的分析微博热点话题，并就广大用户针对这些话题的情感分布进行分析，具有十分重要的意义。
[Abstract]:With the advent of the Web2.0 era, especially the rapid development of social media, more and more Internet users begin to publish their views on some topics online.How to extract the current hot topic from the massive data and analyze the user's views on the topic has become a very meaningful research point.This article has carried on the thorough analysis to this research point.Weibo's definition of hot topic is different from that of ordinary topic.In Weibo, hot topic refers to the topic that is highly concerned by a large number of users in a short period of time.These topics can be real social events, or not related to social events, but in a short period of time has gained a high degree of interest, which has attracted a large number of Weibo users.This paper divides the research point into two tasks: Weibo hot topic extraction and topic emotion classification.In this paper, a topic extraction algorithm based on topic label is proposed for Weibo hot topic extraction.The algorithm uses heuristic rules and clustering algorithm to extract Weibo hot topics in real time.The algorithm includes three parts: topic discovery, topic clustering and topic sorting.Experimental results show that the proposed method has a high accuracy in extracting hot topics from Weibo.On topic emotion classification, this paper proposes Weibo emotion classification algorithm based on text relationship.The algorithm induces four kinds of text relations which may have an effect on the result of emotion classification, and identifies the main components of Weibo's text by using the relational deixis corresponding to these text relations.On the basis of this, this paper explores the characteristics of Weibo's emotional classification in different categories.In addition, this paper also constructs a large-scale emotion dictionary for Weibo.Experimental results show that the proposed algorithm can achieve good results in Weibo emotional classification.In addition, we combine the above two algorithms organically and construct an online analysis platform for user sentiment.The platform can analyze Weibo hot topics in real time and analyze the emotional distribution of these topics by the majority of users, which is of great significance.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092

【共引文献】