Personalizing Incremental Video Search with Hybrid Text and ID Embeddings

2026-07-15

AI Summary

This work addresses the challenge of ranking in incremental video search on Apple TV, where user queries are extremely short (1–3 characters) and intent is ambiguous. To enable real-time personalization, the authors propose a ranking method that jointly leverages textual semantics and collaborative filtering signals. Item representations are learned via a multilingual text encoder and ID-based collaborative embeddings, while user vectors are dynamically constructed from recent viewing history. Cosine similarities between user and item embeddings serve as features for an XGBoost pairwise ranker. The key innovation lies in the first-time integration of dual-path text and ID embeddings for real-time personalization in incremental search. Offline experiments show a 2.99% improvement in NDCG@10 (8.63% for ambiguous queries), and online metrics demonstrate a 1.14% increase in click-through rate, a 1.23% gain in conversion rate, and a 2.91% uplift in average conversion item ranking.

Abstract

Incremental video search requires high-quality ranking after each keystroke, where intent is often underspecified (e.g., 1-3 character prefixes). We present a personalization system for Apple TV search that combines complementary semantic and collaborative signals at ranking time. Our approach learns two item embedding spaces: (i) a text-based multilingual encoder (TextEmb) fine-tuned on co-engagement triplets via contrastive learning, and (ii) an ID-based collaborative embedding model (IdEmb) trained on interaction-derived positives. At serving time, we construct user representations from recent watch history and inject text- and ID-based user-item cosine similarities into a pairwise XGBoost ranker. We evaluate with temporally held-out offline datasets and a three-week online controlled experiment. Offline, for sessions with user history, the personalized ranker improves NDCG@10 by 2.99% and MRR by 3.30% over the non-personalized baseline. Slice analyses show that personalization is most needed in incremental search, where intent is still forming: on ambiguous prefix queries (1-3 characters), NDCG@10 lift is +8.63%, versus +1.46% on longer, fully specified queries. Longer-history users benefit more: NDCG lift rises from +2.13% for users with 1-5 history items to +4.37% for users with 51-100, even though baseline relevance is lower for these cohorts (NDCG@10 drops from 0.733 to 0.680), indicating that personalization adds the most value where default ranking underperforms. Online, treatment yields statistically significant gains of +1.14% tap-through rate and +1.23% conversion rate, with a 2.91% improvement in converted-item rank position. We further analyze coverage-precision trade-offs between semantic and collaborative embeddings via ablations isolating each signal, and evaluate embedding quality on a held-out corpus with LLM-judged similarity labels to reduce click/exposure bias.

Problem

• incremental video search
• query ambiguity
• personalization
• ranking
• underspecified intent

Innovation

• incremental video search
• hybrid embeddings
• personalized ranking
• contrastive learning
• collaborative filtering