Analysis of News Dissemination Characteristics under the Trend of Short Video Convergence Based on Multi-Priority Dynamic Threshold Algorithm

The rapid development of science and technology and the continuous iteration of mobile terminals have led to great changes in the news dissemination mode. After summarizing the characteristics of news publishing in the short video perspective, the study constructs a behavioral communication characteristics analysis model based on multi-priority dynamic thresholding. The model is able to automatically cluster according to the characteristics of the dataset by establishing a dynamic threshold selection model, which eliminates the difficulty of setting parameters manually. Meanwhile, the text similarity calculation based on the kernel method is used to realize the mining of short video news information. Conducting experimental analysis and practical application of the model's performance, it is found that the scores of the multi-priority dynamic threshold model on AR, PR, RE and F-score are 0.71, 0.67, 0.72 and 0.68, which are better than those of the comparison model, reflecting the model's good ability to mention information of short news videos. Who (97.21%) and What (100%) elements are the highest in news short video titles, and the proportion of monosyllabic and disyllabic words is 93.2%. It was found that short video news is characterized by fragmentation, concise content and interactivity.