基于机器学习的抖音短视频用户分析（软件工程专业 ）.docxVIP

PAGE 26 内容摘要 随着社会的发展。抖音短视频APP的碎片化时间娱乐，进行抖音短视频APP的用户数据分析，有利于不同用户的选择以及创作者的内容创作。 针对短视频用户分析是建立在抖音App上的用户评论数据分析，进行抖音用户数据的爬虫，然后进行数据的分析。基于抖音短视频App的用户分析，使用机器学习的情感分析算法进行用户数据的区分以及分析。 在本次研究中，情感分析算法使用贝叶斯模型进行语句的情感分析预测。使用jieba分词系统用于对视频内容进行分类区分。使用机器学习算法中的snownlp库进行中文语句的判断预测，snownlp库可以为中文语句的情绪判断进行预测。当判断返回值越接近1的时候，表示为正面情绪，越接近0的时候表示负面情绪。 在本研究中，采取了百度AI情感分析以及本研究使用的基于贝叶斯模型的snownlp库情感分析的对比数据测试，从数据样本中随机抽取1000条数据进行准确度测试。经过测试分析，本研究使用的情感分析算法与百度AI得出的情感分析的准确率达到85%。通过数据测试，证明本研究所使用的情感分析算法准确率可以得到有效的保证以及可信度。 关键词：数据分析； 机器学习算法； 数据爬虫； 分词 Abstract With the development of society. The fragmented time entertAInment of Douyin Short Video APP, user data analysis of Douyin Short Video APP is beneficial to the choice of different users and the content creation of creators. The analysis of short video users is based on the analysis of user comment data on the Douyin App, crawling the user data of Douyin, and then analyzing the data. Based on user analysis of Douyin Short Video App, machine learning sentiment analysis algorithm is used to distinguish and analyze user data. In this study, the sentiment analysis algorithm uses Bayesian models to predict sentiment analysis of sentences. Use jieba word segmentation system to classify video content. Use the snownlp library in the machine learning algorithm to judge and predict Chinese sentences. The snownlp library can predict the sentiment judgment of Chinese sentences. When it is judged that the return value is closer to 1, it represents positive emotions, and the closer it is to 0, it represents negative emotions. In this study, BAIdu AI sentiment analysis and the comparative data test based on Bayesian model-based snownlp library sentiment analysis were used, and 1,000 data were randomly selected from the data samples for accuracy testing. After testing and analysis, the sentiment analysis algorithm used in this study and BAIdu AIs sentiment analysis accuracy rate reached 85%. The data test proves that the accuracy of the sentiment analysis algorithm us