联系我们

论文查询结果	返回搜索

论文编号：	14129
作者编号：	2120203004
上传时间：	2023/6/13 12:22:29
中文题目：	视频特征与弹幕之间的关系研究 ——以哔哩哔哩弹幕视频网站为例
英文题目：	Study on the Relationship Between Video Features and Danmaku --Taking Bilibili Danmaku Video Website as an Example
指导老师：	王芳教授
中文关键字：	弹幕；弹幕视频；哔哩哔哩；B站；视频特征；弹幕特征
英文关键字：	Danmaku; Danmaku Videos; Bilibili; Video Features; Danmaku Features
中文摘要：	近年来，弹幕视频发展迅速，越来越受到视频生产者和消费者的重视和喜爱。然而，视频平台之间和视频作者之间对观众和流量的竞争也愈发激烈。为了在竞争中取得优势，就需要充分发挥弹幕对于提升观众观看体验的重要作用，提高弹幕数量和质量。而要实现对弹幕的优化，研究视频的特征与弹幕的特征会呈现何种相关关系尤为重要。 基于以上背景，本文选取视频的主题分类、播放量、点赞率、评论率、收藏率、时长六个特征作为对象，研究它们与弹幕进度分布、形式、长度、词频、情感倾向之间的关系。首先利用爬虫获取B站视频与弹幕相关数据，然后对上述弹幕特征随视频特征的变化情况进行统计分析，并基于数据分析的结果进行理论解读和提出实践建议。 本研究获得的重要结论主要有：（1）不同主题分类的弹幕展现出了充分的特色。亚文化分区在弹幕形式方面更为多样，而大众化分区的弹幕形式则较为单调；在弹幕长度方面，娱乐区最高而时尚区最低；舞蹈区弹幕的词频集中度和情感积极性大幅领先其它分区；科技区弹幕的情感积极性显著低于其它分区。（2）视频播放量和点赞率越高，弹幕密度越大，形式越丰富，情感越积极。随着播放量的提高，弹幕还具有密度加速增长，文本逐渐变长，作者逐渐中心化的趋势。（3）在视频评论率、收藏率和视频时长与弹幕多项特征的相关关系中，都存在一到两个较窄的分界段，在分界段两侧，相关关系的幅度或方向会发生变化。 本文由六个章节组成：第一章为绪论，第二章介绍相关理论和研究现状，第三章提出具体研究设计并展示数据准备过程，第四章做数据分析并作小结讨论，第五章进行总结与展望。
英文摘要：	In recent years, danmaku videos have developed rapidly and become more and more popular among video producers and consumers. However, competition for viewers and traffic is intensifying among video platforms and among video authors. In order to gain an advantage in the competition, it is necessary to give full play to the important role of danmaku in improving the audience's viewing experience and improve the quantity and quality of danmaku. In order to realize the optimization of danmaku, it is particularly important to study the correlation between the features of video and those of danmaku. Based on the above background, this paper selects six features of the video: theme classification, video view, liking rate, comment rate, favorite rate and duration, as objects to study their relationship with schedule distribution, form, length, word frequency and emotion of danmaku. Firstly, crawler is used to acquire video and danmaku related data of Bilibili, and then statistical analysis is carried out on the changes of the above danmaku features with the video features, and theoretical interpretation and practical suggestions are put forward based on the results of data analysis. The important conclusions of this study are as follows: (1) Danmaku of different subject categories show sufficient characteristics and personalities. Subculture categories are more diverse in the form of danmaku, while popular categories are more monotonous. In terms of the length of danmaku, the entertainment area is the highest while the fashion area is the lowest. The word frequency concentration and emotion ofdanmaku in dance area are significantly higher than those in other areas. The emotion of danmaku in science and technology area is significantly lower than that in other areas. (2) The higher the number of videos played and the rate of likes, the higher the density of danmaku, the richer the form, the more positive the emotion, and the more concentrated the video head and tail. With the increase of video view, danmaku also has the trend of accelerated growth of density, gradually longer text, and gradually centralized author. (3) There are one or two boundary sections in the correlation between video comment rate, star rate and duration and danmaku. On both sides of the boundary section, the amplitude or direction of the correlation will change. This paper consists of six chapters: the first chapter is the introduction, the second chapter introduces relevant theories and research status, and the third chapter puts forward specific research design. The fourth chapter shows the data collection and processing process, the fifth chapter makes data analysis and summary discussion, and the sixth chapter summarizes and prospects.
查看全文：	预览  下载(下载需要进行登录)