Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yi, Zhongchao, Feng, Kai, Ma, Xiaojian, Wang, Yalong, Liu, Yongqi, Li, Han, Zhou, Zhengyang, Wang, Yang
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2511.12518
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917222735151104
author Yi, Zhongchao
Feng, Kai
Ma, Xiaojian
Wang, Yalong
Liu, Yongqi
Li, Han
Zhou, Zhengyang
Wang, Yang
author_facet Yi, Zhongchao
Feng, Kai
Ma, Xiaojian
Wang, Yalong
Liu, Yongqi
Li, Han
Zhou, Zhengyang
Wang, Yang
contents In large-scale industrial recommendation systems, retrieval must produce high-quality candidates from massive corpora under strict latency. Recently, Generative Retrieval (GR) has emerged as a viable alternative to Embedding-Based Retrieval (EBR), which quantizes items into a finite token space and decodes candidates autoregressively, providing a scalable path that explicitly models target-history interactions via cross-attention. However, deploying GR in short-video feeds remains challenged by long-short interest interference, context-induced noise in hierarchical SID generation, and the lack of explicit learning from exposed-but-unclicked feedback. To address these challenges, we propose DualGR, which combines (i) a Dual-Branch Long/Short-Term Router (DBR) with selective activation, (ii) Search-based SID Decoding (S2D) that constrains fine-level decoding within the current coarse bucket for efficiency and noise control, and (iii) an Exposure-aware Next-Token Prediction Loss (ENTP-Loss) that treats unclicked exposures as coarse-level hard negatives to promote timely interest fade-out. On the large-scale Kuaishou short-video recommendation system, DualGR has achieved outstanding performance. Online A/B testing shows +0.527% video views and +0.432% watch time lifts, validating DualGR as a practical and effective paradigm for industrial generative retrieval.
format Preprint
id arxiv_https___arxiv_org_abs_2511_12518
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DualGR: Generative Retrieval with Long and Short-Term Interests Modeling
Yi, Zhongchao
Feng, Kai
Ma, Xiaojian
Wang, Yalong
Liu, Yongqi
Li, Han
Zhou, Zhengyang
Wang, Yang
Information Retrieval
In large-scale industrial recommendation systems, retrieval must produce high-quality candidates from massive corpora under strict latency. Recently, Generative Retrieval (GR) has emerged as a viable alternative to Embedding-Based Retrieval (EBR), which quantizes items into a finite token space and decodes candidates autoregressively, providing a scalable path that explicitly models target-history interactions via cross-attention. However, deploying GR in short-video feeds remains challenged by long-short interest interference, context-induced noise in hierarchical SID generation, and the lack of explicit learning from exposed-but-unclicked feedback. To address these challenges, we propose DualGR, which combines (i) a Dual-Branch Long/Short-Term Router (DBR) with selective activation, (ii) Search-based SID Decoding (S2D) that constrains fine-level decoding within the current coarse bucket for efficiency and noise control, and (iii) an Exposure-aware Next-Token Prediction Loss (ENTP-Loss) that treats unclicked exposures as coarse-level hard negatives to promote timely interest fade-out. On the large-scale Kuaishou short-video recommendation system, DualGR has achieved outstanding performance. Online A/B testing shows +0.527% video views and +0.432% watch time lifts, validating DualGR as a practical and effective paradigm for industrial generative retrieval.
title DualGR: Generative Retrieval with Long and Short-Term Interests Modeling
topic Information Retrieval
url https://arxiv.org/abs/2511.12518