Cross-social-platform picture recommendation algorithm based on content or emotion similarity

Abstract

The invention discloses a cross-social-platform picture recommendation algorithm based on content or emotion similarity, and relates to the technical field of smart media calculation and big data analysis. The algorithm comprises the following steps that: firstly, analyzing the contents of a text generated by a user in an image-text sharing social platform, extracting the keyword information of the contents, meanwhile, using the same method to carry out content analysis on pictures in a picture social platform, and carrying out matching according to the content consistency of the text and the picture to obtain a content-based initial recommendation list; then, carrying out emotion analysis on the text, and if the text contains emotions, carrying out matching on the basis of the emotion consistency of the text and the picture to obtain an initial recommendation list based on emotion matching; then, independently carrying out modeling on the user in two social network platforms to obtain user preference; and finally, carrying out fusion on the contents, the emotion and the user preference, and generating a final picture recommendation list. By use of the algorithm, the expression of the user is full and appropriate so as to be favorable for greatly improving user experience on a social network and improving user viscosity.

Description

technical field [0001] The invention relates to the technical field of intelligent media computing and big data analysis, in particular to a picture recommendation method. Background technique [0002] In the Web2.0 era, the development trend of social networks is the continuous enrichment of multimedia interactions and the continuous reduction of user interaction thresholds. From the initial long-text interactive social networks (news websites, blogs, etc.), to the later short-text interactive social networks (twitter, Facebook, Weibo, Renren, etc.), to the later visual social networks (Flickr, Snapchat, etc.) , Instagram, etc.), media types are constantly enriched (text-image-video), and the threshold for user participation is constantly lowered. Later added functions such as forwarding, favorites, and likes allow users to express their personal opinions with one click with the help of content generated / published / forwarded by others. Correspondingly, the expression with ...

AI Technical Summary

Problems solved by technology

As an important platform for people to obtain information, communicate, and communicate, on the one hand, social network platforms have become a rich source of data; on the other hand, due to the convenience of information release, the update speed of information is very fast, resulting in huge redundant information , a serious information overload has occurred. In this case, it is difficult for users to find the data they need from the massive data, which makes it difficult for users to find satisfactory images in a short period of time.

Images