| At present,China is in the general trend of ”video”,uploader is slowly becoming an emerging profession,and more and more people are engaged in this industry.In the process of daily short videos,we often find that videos with the same content have obvious differences in the number of views.There are even times when the original creator uploads a video and gets less views than the mover.In order to explore the reason,this article takes this as a starting point to help up master increase the number of video playback as the goal,and takes the short video platform of B station as an example to generate a software that provides help and suggestions when making and publishing videos.In response to the above objectives,this paper carries out the following work:(1)Based on the current real-time hot spots,this paper selects Bilibili with unique operation mode as the research object,and analyzes the short videos in its platform.In the study,the title features were extracted,and the factors affecting popularity were analyzed and descriptively analyzed.To support research,a system called ”Operation Bao” was developed and the relevant functions of the system were demonstrated using the video data of the food zone as an example.(2)Construction of the operation treasure system.The ”Operation Bao” system consists of four parts,namely the user login interface,the data layer,the title processing parser and the propagation effect parser.The system uses common algorithms Chinese text analysis,stepwise regression,and random forest models to derive significant factors that affect the number of playbacks.After users enter keywords,they directly generate relevant information to provide help and suggestions for video production and publishing.In order to demonstrate some functions of the system,this article takes 4500 data from Bilibili as an example to analyze and display.(3)In order to expand the research scope of video titles,this paper attempts to extract the features of titles and study the influence of title features on the popularity of videos.The word segmentation tool and the supplementary custom word segmentation dictionary are used to improve the word segmentation,and the text analysis method is used to conduct detailed research from the three aspects of hot words,content and parts of speech.The results show that different video zones have different impacts on the number of views,taking the food zone as an example,the food production zone and the food detective zone have a promoting effect on the spread of the video,while the food evaluation zone has an inhibitory effect on the spread of the video,and the video duration should not be too long,and it is best to control it within 5 minutes.Of all the variables,the number of favorites,video duration,and number of comments had the greatest impact on the number of views,and the number of followers only partially moderated.In the video title and production,it is recommended to select the current hot topic as the main content of the video,according to the hot topic calculated by the system,the title is composed of numbers + hot words +strong descriptive words,and the title length is recommended to be 26-30 words. |
