基于内容和用户标识的混合型垃圾弹幕识别与过滤研究

发布时间：2018-05-27 17:03

本文选题：弹幕 + 贝叶斯　；参考：《杭州电子科技大学》2017年硕士论文

【摘要】：随着互联网的不断发展和推进,网络信息技术的快速发展为电影、电视剧、综艺节目等视频内容创作开启了一个崭新时代。网络在线视频已经在人们观看视频过程中占据了重要地位,成为大众视频消费的主流。“弹幕”因其高互动性和娱乐性开始进入人们的视野,弹幕是指用户在视频网站上观看视频时,可以直接将评论发送并显示在屏幕上方的一种显示方式。在国内最早使用弹幕技术的是动漫圈内的视频吐槽网站Acfun(A站)和bilibili(B站),随着弹幕的发展,弹幕也不再仅仅使用在弹幕视频网站中,如今,国内弹幕应用虽然火爆,但是国内弹幕文化和技术的整体发展时间较短,当前针对弹幕的研究还相对匮乏,对弹幕的有效管理与研究还较少。本文以提高用户的弹幕使用体验为目的,探讨对弹幕文本的识别与处理,提出一种新的基于内容和用户身份标识的混合型垃圾弹幕识别算法,对已有算法进行改进,以更好的引导我国弹幕文化的积极发展,改善用户在使用与观看弹幕时的用户体验,提高网站的用户使用度和黏性,绿化弹幕使用环境。本文针对弹幕文本特征进行分析,对比以往垃圾文本识别研究中较多涉及的邮件以及短信等文本类型,总结出弹幕文本的自身特征,结合其自身特性以进一步加深垃圾弹幕识别的研究。结合弹幕文本自身特点,本文以基于内容的文本分类算法和基于用户身份标识的分类算法为基础展开研究,在已有研究基础上提出一种新的基于以上两种文本分类算法相结合的混合型垃圾弹幕识别算法,并据此进行垃圾弹幕识别与过滤。本文利用从土豆网热播电视剧《琅琊榜》中爬取的弹幕数据作为数据来源进行试验分析,以验证本文提出算法模型的表现效果。通过实验研究可以看出,本文提出改进算法能够有效提高垃圾弹幕识别的准确性。
[Abstract]:With the continuous development and promotion of the Internet, the rapid development of network information technology has opened a new era for the creation of video content such as movies, TV dramas, variety shows and so on. Online video has played an important role in the process of watching video, and has become the mainstream of popular video consumption. Because of its high interactivity and entertainment, "projectile screen" begins to enter people's field of vision. It refers to a display mode in which users can directly send comments and display them at the top of the screen when they watch a video on a video website. In China, the earliest use of bullet screen technology is the animation video complaints website Acfun(A) and the bilibili(B station. With the development of the projectile screen, the projectile screen is no longer used only in the projectile video website. Nowadays, although the domestic projectile screen is popular, the application of the projectile screen is very popular. However, the overall development time of domestic projectile culture and technology is relatively short, the current research on projectile screen is relatively scarce, and the effective management and research of projectile screen is still less. In order to improve the user's experience of projectile screen usage, this paper discusses the recognition and processing of projectile text, and proposes a new mixed algorithm of garbage screen recognition based on content and user identification, which improves the existing algorithms. In order to better guide the positive development of the projectile culture in our country, improve the user's experience when using and watching the projectile screen, improve the user's using degree and stickiness of the website, and afforeze the projectile screen using environment. In this paper, we analyze the features of projectile text, and summarize the characteristics of project-screen text such as email and short message, which are more involved in the research of spam text recognition than in the past. Combined with its own characteristics to further deepen the research of garbage screen recognition. According to the characteristics of projectile text, this paper researches on content-based text classification algorithm and user identification based classification algorithm. On the basis of the previous research, a new hybrid algorithm of garbage screen recognition based on the above two text classification algorithms is proposed, and then the garbage screen recognition and filtering are carried out. In order to verify the performance of the algorithm proposed in this paper, the projectile data from the popular Tudou TV series "Langya Bang" is used as the data source. The experimental results show that the improved algorithm can effectively improve the accuracy of garbage screen recognition.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.1;F49

【参考文献】