LingVarBench: Benchmarking LLMs on Entity Recognitions and Linguistic Verbalization Patterns in Phone-Call Transcripts

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Mohammadi, Seyedali, Paldhe, Manas, Chhabra, Amit, Son, Youngseo, Seshagiri, Vishal
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914254134706176
author Mohammadi, Seyedali
Paldhe, Manas
Chhabra, Amit
Son, Youngseo
Seshagiri, Vishal
author_facet Mohammadi, Seyedali
Paldhe, Manas
Chhabra, Amit
Son, Youngseo
Seshagiri, Vishal
contents We study structured entity extraction from phone-call transcripts in customer-support and healthcare settings, where annotation is costly, and data access is limited by privacy and consent. Existing methods degrade under disfluencies, interruptions, and speaker overlap, yet large real-call corpora are rarely shareable. We introduce LingVarBench, a benchmark and semantic synthetic data generation pipeline that generates linguistically varied training data via (1) LLM-sampled entity values, (2) curated linguistic verbalization patterns covering diverse disfluencies and entity-specific readout styles, and (3) a value-transcript consistency filter. Using this dataset, DSPy's SIMBA automatically synthesizes and optimizes extraction prompts, reducing manual prompt engineering and targeting robustness to verbal variation. On real customer transcripts, prompts optimized solely on LingVarBench outperform zero-shot baselines and match or closely approach human-tuned prompts for structured entities such as ZIP code, date of birth, and name (F1 approximately 94-95 percent). For subjective questionnaire items, optimized prompts substantially improve over zero-shot performance and approach human-tuned prompts. LingVarBench offers a practical and cost-efficient path to deployment in a direct-answer setting, with real annotations later enabling additional refinement.
format Preprint
id arxiv_https___arxiv_org_abs_2508_15801
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LingVarBench: Benchmarking LLMs on Entity Recognitions and Linguistic Verbalization Patterns in Phone-Call Transcripts
Mohammadi, Seyedali
Paldhe, Manas
Chhabra, Amit
Son, Youngseo
Seshagiri, Vishal
Computation and Language
Artificial Intelligence
Human-Computer Interaction
Machine Learning
We study structured entity extraction from phone-call transcripts in customer-support and healthcare settings, where annotation is costly, and data access is limited by privacy and consent. Existing methods degrade under disfluencies, interruptions, and speaker overlap, yet large real-call corpora are rarely shareable. We introduce LingVarBench, a benchmark and semantic synthetic data generation pipeline that generates linguistically varied training data via (1) LLM-sampled entity values, (2) curated linguistic verbalization patterns covering diverse disfluencies and entity-specific readout styles, and (3) a value-transcript consistency filter. Using this dataset, DSPy's SIMBA automatically synthesizes and optimizes extraction prompts, reducing manual prompt engineering and targeting robustness to verbal variation. On real customer transcripts, prompts optimized solely on LingVarBench outperform zero-shot baselines and match or closely approach human-tuned prompts for structured entities such as ZIP code, date of birth, and name (F1 approximately 94-95 percent). For subjective questionnaire items, optimized prompts substantially improve over zero-shot performance and approach human-tuned prompts. LingVarBench offers a practical and cost-efficient path to deployment in a direct-answer setting, with real annotations later enabling additional refinement.
title LingVarBench: Benchmarking LLMs on Entity Recognitions and Linguistic Verbalization Patterns in Phone-Call Transcripts
topic Computation and Language
Artificial Intelligence
Human-Computer Interaction
Machine Learning
url https://arxiv.org/abs/2508.15801