跳到正文
arXiv cs.CL· Filippos Ventirozos, Tom Armitage, Mahmoud Oshagh, Peter A. Appleby, Matthew Shardlow·· 15 小时前AI 评分26

有用多样性还是幻觉多样性?面向电商搜索的 persona 条件化合成查询生成

Useful Diversity or Hallucinated Variety? Persona-Conditioned Synthetic Query Generation for E-Commerce Search

AI 导读

研究在 47 字段英国汽车电商 schema 上,用 6 个 LLM、9 组 persona 条件化提示词生成 118,098 条合成查询对,对照 1,214 条人工标注与 15,304 条生产查询。

正文 · 原文

View PDF HTML (experimental)

Abstract:Synthetic supervision can bootstrap semantic parsers in new search domains, but prompt-induced diversity is useful only when generations are faithful to the structured intent. We explore this for cold-start e-commerce search: 118,098 synthetic pairs produced by 6 LLMs with 9 persona-conditioned prompt families on a 47-field UK automotive schema, against 1,214 human-annotated and 15,304 production queries. We show that the IPIP personality bundle improves parser F1 in all four matched contrasts (Holm-corrected $p<.001$), and all three bundles have a positive main effect over the complete $2^3$ grid. Syntax conditioning is the axis that moves generations towards real traffic. It also induces lexical diversity. Alignment sKL, the divergence between von Mises-Fisher distributions fitted to synthetic and human-annotated queries, is an unlabelled proxy for parser F1 ($\rho=-0.917$ against human queries written for simulated shopper scenarios, lower is closer). An LLM committee, checked against a blinded author pass, triages the 927 queries missed by every parser and labels 37% as hallucinated. Dropped constraints dominate. The four syntax-bearing families have the largest hallucinated shares, 38 to 42%, while IPIP alone has the lowest at 28%.
Comments: Accepted at IJCNLP-AACL 2026 (Findings). 20 pages, 2 figures, 16 tables
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR)
ACM classes: I.2.7; H.3.3
Cite as: arXiv:2610.35102 [cs.CL]
  (or arXiv:2610.35102v1 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2610.35102

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Filippos Ventirozos [view email]
[v1] Mon, 26 Oct 2026 09:43:29 UTC (121 KB)

来源:arXiv cs.CL · arxiv.org