English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Micro-video background music recommendation is a complicated task where thematching degree between videos and uploader-selected background music is amajor issue. However, the selection of the user-generated content (UGC) isbiased caused by knowledge limitations and historical preferences among musicof each uploader. In this paper, we propose a Debiased Cross-Modal (DebCM)matching model to alleviate the influence of such selection bias. Specifically,we design a teacher-student network to utilize the matching of segments ofmusic videos, which is professional-generated content (PGC) with specializedmusic-matching techniques, to better alleviate the bias caused by insufficientknowledge of users. The PGC data is captured by a teacher network to guide thematching of uploader-selected UGC data of the student network by KL-basedknowledge transfer. In addition, uploaders' personal preferences of musicgenres are identified as confounders that spuriously correlate music embeddingsand background music selections, resulting in the learned recommender system toover-recommend music from the majority groups. To resolve such confounders inthe UGC data of the student network, backdoor adjustment is utilized todeconfound the spurious correlation between music embeddings and predictionscores. We further utilize Monte Carlo (MC) estimator with batch-level averageas the approximations to avoid integrating the entire confounder spacecalculated by the adjustment. Extensive experiments on the TT-150k-genredataset demonstrate the effectiveness of the proposed method towards theselection bias. The code is publicly available on:\url{https://github.com/jing-1/DebCM}.

Debiased Cross-modal Matching for Content-based Micro-video Background Music Recommendation