Abstract:

Multi source all media information comes from different platforms, channels, and formats, including text, images, videos, audio, and other forms. The data has high dimensions and complexity, making information representation difficult. Therefore, a multi-source all media interaction information similarity search algorithm based on double-layer attention is proposed. A dual layer attention model is applied to extract features from all media information, effectively capturing key features of information at different levels. The fuzzy C-means algorithm is used to cluster the entire media information database and classify similar media information together.Using similarity search algorithms, by calculating the similarity between sample information and other information in the entire media information database. The most similar information content can be quickly and accurately searched, providing users with a better search experience. Experimental results have shown that the proposed method can achieve similarity search for all media information with accurate search results.

Key words: