@inproceedings{yuan-etal-2026-gram,
title = "{GRAM}-{V}2: Generative Reasoning and Ranking Alignment for Large-Scale {E}-commerce Search",
author = "Yuan, Chunyuan and
Pang, Ming and
Deng, Yiming and
He, Xiaoyu and
Fang, Zheng and
Xie, Donghao and
Qu, Fanyi and
Peng, Changping and
Law, Ching",
editor = "Atanasova, Pepa and
Haffari, Gholamreza and
Wang, Minghan",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-industry.106/",
pages = "1830--1839",
ISBN = "979-8-89176-460-6",
abstract = "Click-through rate (CTR) prediction is a critical task for e-commerce advertising systems. However, industrial ranking models still rely on sparse IDs and historical behavior signals, which limits their generalization to cold-start products, long-tail queries, and complex user intents. Although large language models (LLMs) provide strong semantic understanding and can generate behavioral rationales, directly using them for online ranking is impractical under strict latency constraints.\par We propose GRAM-V2, a rationale-supervised ranking framework that transfers LLM-generated behavioral rationales into an efficient CTR model. GRAM-V2 uses a Teacher LLM to generate click-oriented rationales, distills them into a lightweight Student LLM, and aligns a two-tower CTR model with the Student{'}s representations and rationale signals. During online serving, GRAM-V2 uses only distilled representations and cached item embeddings, without generating rationales autoregressively. Large-scale offline experiments and online A/B testing show that GRAM-V2 improves CTR prediction accuracy, CTR, and advertising revenue while satisfying production latency constraints."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yuan-etal-2026-gram">
<titleInfo>
<title>GRAM-V2: Generative Reasoning and Ranking Alignment for Large-Scale E-commerce Search</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chunyuan</namePart>
<namePart type="family">Yuan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ming</namePart>
<namePart type="family">Pang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yiming</namePart>
<namePart type="family">Deng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoyu</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zheng</namePart>
<namePart type="family">Fang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Donghao</namePart>
<namePart type="family">Xie</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fanyi</namePart>
<namePart type="family">Qu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Changping</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ching</namePart>
<namePart type="family">Law</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pepa</namePart>
<namePart type="family">Atanasova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gholamreza</namePart>
<namePart type="family">Haffari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minghan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-460-6</identifier>
</relatedItem>
<abstract>Click-through rate (CTR) prediction is a critical task for e-commerce advertising systems. However, industrial ranking models still rely on sparse IDs and historical behavior signals, which limits their generalization to cold-start products, long-tail queries, and complex user intents. Although large language models (LLMs) provide strong semantic understanding and can generate behavioral rationales, directly using them for online ranking is impractical under strict latency constraints.\par We propose GRAM-V2, a rationale-supervised ranking framework that transfers LLM-generated behavioral rationales into an efficient CTR model. GRAM-V2 uses a Teacher LLM to generate click-oriented rationales, distills them into a lightweight Student LLM, and aligns a two-tower CTR model with the Student’s representations and rationale signals. During online serving, GRAM-V2 uses only distilled representations and cached item embeddings, without generating rationales autoregressively. Large-scale offline experiments and online A/B testing show that GRAM-V2 improves CTR prediction accuracy, CTR, and advertising revenue while satisfying production latency constraints.</abstract>
<identifier type="citekey">yuan-etal-2026-gram</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-industry.106/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>1830</start>
<end>1839</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T GRAM-V2: Generative Reasoning and Ranking Alignment for Large-Scale E-commerce Search
%A Yuan, Chunyuan
%A Pang, Ming
%A Deng, Yiming
%A He, Xiaoyu
%A Fang, Zheng
%A Xie, Donghao
%A Qu, Fanyi
%A Peng, Changping
%A Law, Ching
%Y Atanasova, Pepa
%Y Haffari, Gholamreza
%Y Wang, Minghan
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-460-6
%F yuan-etal-2026-gram
%X Click-through rate (CTR) prediction is a critical task for e-commerce advertising systems. However, industrial ranking models still rely on sparse IDs and historical behavior signals, which limits their generalization to cold-start products, long-tail queries, and complex user intents. Although large language models (LLMs) provide strong semantic understanding and can generate behavioral rationales, directly using them for online ranking is impractical under strict latency constraints.\par We propose GRAM-V2, a rationale-supervised ranking framework that transfers LLM-generated behavioral rationales into an efficient CTR model. GRAM-V2 uses a Teacher LLM to generate click-oriented rationales, distills them into a lightweight Student LLM, and aligns a two-tower CTR model with the Student’s representations and rationale signals. During online serving, GRAM-V2 uses only distilled representations and cached item embeddings, without generating rationales autoregressively. Large-scale offline experiments and online A/B testing show that GRAM-V2 improves CTR prediction accuracy, CTR, and advertising revenue while satisfying production latency constraints.
%U https://aclanthology.org/2026.emnlp-industry.106/
%P 1830-1839
Markdown (Informal)
[GRAM-V2: Generative Reasoning and Ranking Alignment for Large-Scale E-commerce Search](https://aclanthology.org/2026.emnlp-industry.106/) (Yuan et al., EMNLP 2026)
ACL
- Chunyuan Yuan, Ming Pang, Yiming Deng, Xiaoyu He, Zheng Fang, Donghao Xie, Fanyi Qu, Changping Peng, and Ching Law. 2026. GRAM-V2: Generative Reasoning and Ranking Alignment for Large-Scale E-commerce Search. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 1830–1839, Budapest, Hungary. Association for Computational Linguistics.