Salvato in:
Dettagli Bibliografici
Autori principali: Sim, Ju Yong, Kim, Seong Hwan
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2506.06180
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910992370237440
author Sim, Ju Yong
Kim, Seong Hwan
author_facet Sim, Ju Yong
Kim, Seong Hwan
contents We develop a voice phishing (VP) detector by fine-tuning Llama3, a representative open-source, small language model (LM). In the prompt, we provide carefully-designed VP evaluation criteria and apply the Chain-of-Thought (CoT) technique. To evaluate the robustness of LMs and highlight differences in their performance, we construct an adversarial test dataset that places the models under challenging conditions. Moreover, to address the lack of VP transcripts, we create transcripts by referencing existing or new types of VP techniques. We compare cases where evaluation criteria are included, the CoT technique is applied, or both are used together. In the experiment, our results show that the Llama3-8B model, fine-tuned with a dataset that includes a prompt with VP evaluation criteria, yields the best performance among small LMs and is comparable to that of a GPT-4-based VP detector. These findings indicate that incorporating human expert knowledge into the prompt is more effective than using the CoT technique for small LMs in VP detection.
format Preprint
id arxiv_https___arxiv_org_abs_2506_06180
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Detecting Voice Phishing with Precision: Fine-Tuning Small Language Models
Sim, Ju Yong
Kim, Seong Hwan
Computation and Language
We develop a voice phishing (VP) detector by fine-tuning Llama3, a representative open-source, small language model (LM). In the prompt, we provide carefully-designed VP evaluation criteria and apply the Chain-of-Thought (CoT) technique. To evaluate the robustness of LMs and highlight differences in their performance, we construct an adversarial test dataset that places the models under challenging conditions. Moreover, to address the lack of VP transcripts, we create transcripts by referencing existing or new types of VP techniques. We compare cases where evaluation criteria are included, the CoT technique is applied, or both are used together. In the experiment, our results show that the Llama3-8B model, fine-tuned with a dataset that includes a prompt with VP evaluation criteria, yields the best performance among small LMs and is comparable to that of a GPT-4-based VP detector. These findings indicate that incorporating human expert knowledge into the prompt is more effective than using the CoT technique for small LMs in VP detection.
title Detecting Voice Phishing with Precision: Fine-Tuning Small Language Models
topic Computation and Language
url https://arxiv.org/abs/2506.06180