A hybrid filtering for micro-video hashtag recommendation using graph-based deep neural network

Abstract

Due to the growing volume of user-generated content, hashtags are employed as topic indicators to manage content efficiently on social media platforms. However, finding these vital topics is challenging in micro-videos since they contain substantial information in a short duration. Existing methods that recommend hashtags for micro-videos primarily focus on content and personalization while disregarding user's modality-specific tagging preferences. Moreover, the cold-start user issue prevails in hashtag recommendation systems. Considering the above, we propose a hybrid filtering-based MIcro-video haSHtag recommendatiON (MISHON) system to recommend hashtags for micro-videos. We construct a heterogeneous graph to model user's modality-specific tagging patterns by establishing links with constituent modalities of previous micro-videos, further encompassing user-to-user and modality-to-modality interactions. We then refine modality-specific and user representations using message-passing strategy to recommend pertinent hashtags for micro-videos. The empirical results on three real-world datasets demonstrate that MISHON attains a comparative enhancement of 3.6%, 2.8%, and 6.5% concerning the F1-score, respectively. To address cold-start problem, we propose a content and social influence-based technique to recommend hashtags that are not only relevant to content but also popular, thereby empowering cold-start users to broaden their network and content visibility. The proposed solution shows a relative improvement of 15.8% in the F1-score over its content-only counterpart. © 2024 Elsevier Ltd