Aspect-level sentiment capsule network for micro-video click-through rate prediction

Micro-videos, a new form of videos that are constrained in duration, gain significant popularity recent years. The volume and rate online micro-videos urgently calls for effective recommendation algorithms to help users find their interested ones. Although some previous works have investigated how model users’ historical behaviors predict the click-through micro-videos, they generally based on positive feedback only but overlook negative which can understand user preference at finer granularity. jointly imply user’s different sentiments aspects, where each aspect is one component micro-video such as video_scene video_subject. To this end, we propose an spect-level s entiment cap sule network(ASCap) prediction by aggregating both feedback, with attempt make more explainable. More specifically, aspect-specific gating mechanism firstly utilized extract aspect-level features from target feedback. Then, following sentiment capsule network, paired those respectively identify capsules. Finally, layer employed calculate overall click probability Experimental results two real-world datasets demonstrate proposed method significantly outperforms state-of-the-art methods.