家用電話詐欺預警機制之研究

 

A Study on Home Phone Fraud Prevention

 

顏志平

中央警察大學總務處

桃園市龜山區大崗里樹人路56

peter@mail.cpu.edu.tw

 

董正談

中央警察大學資訊管理所

桃園市龜山區大崗里樹人路56

tung@mail.cpu.edu.tw

 

陳芳逸

中央警察大學資訊管理所

桃園市龜山區大崗里樹人路56

im1033084@mail.cpu.edu.tw

 

 

摘要

隨著資訊通訊科技的發展,電信業者所能提供給消費者的電信服務較以往更多元及低成本,而電信詐欺的犯罪型態及手法亦與時俱進。犯罪集團亦透過各種通訊管道,包括電子郵件、通訊軟體、智慧型手機、傳統市話等來接觸被害人,使用電信業者所提供的電信服務來達到躲避偵查、節省通訊成本及竄改主叫號碼等目的,並以精心設計的詐騙劇本與情境詐騙,由詐騙人員在電話另一端以其逼真的聲音、語調與情境來取得被害對象的信任,典型的內容包括慈善詐欺、緊急救援詐欺、拍賣詐欺、恐嚇詐欺、退稅詐欺、台灣的篡改主叫號碼、猜猜我是誰詐欺等,這些詐欺手法往往利用人性弱點來實現。政府與民間為了減緩詐騙案件所造成的人民財產與社會經濟損失,除透過宣導與教育來提升民眾反詐騙意識外,也利用智慧型手機所提供的功能與服務發展詐騙防制與預警APP,如WhoscallAPP防駭通等,因此智慧型手機上的反詐騙預警機制可以說是發展相當成熟,包括來電黑名單過濾、簡訊詐騙關鍵詞偵測等。反觀傳統家用電話尚無相關防制機制,儘管傳統家用電話已逐漸式微,但大多數家庭仍有配置家用電話,且有相當大比例之使用者為年齡層偏高之民眾,其接觸反詐騙資訊的管道也相對較少,因此為詐騙被害者之高危險群。有鑑於此,本研究利用嵌入式系統 Raspberry Pi為例,結合語音辨識與關鍵詞偵測等相關技術提出有效的防制措施與解決方案,建構傳統電話詐騙預警系統,期能有效防制傳統家用電話之詐騙犯罪。

 

關鍵詞:電信詐欺、嵌入式系統、語音辨識、關鍵詞偵測。

 

 

 

 

 

 

Abstract

Due to the development of information and communication technology, telecom service providers can provide consumers telecom service with more variety and lower cost. But the criminal types and method of telecom fraud also advance with times. Criminal organizations also use all kinds of communication channels to contact with victims, including emails, communication softwares, smart phones and traditional home phones. They use the telecom service provided by telecom service providers to avoid investigation, save communication cost and tamper the caller ID. And they try to gain the victim’s trust through well-designed script and lifelike voice and intonation on the phone. Typical cases include charitable fraud, emergency fraud, online auction fraud, intimidation fraud, tax fraud, and so on, these fraud techniques often exploit human weaknesses to achieve. In order to mitigate the loss of civil property and socio-economic from fraud, the official and private try to elevate the civil anti-fraud awareness through propaganda and education, they also use functions and services provided by smartphones to develop the fraud prevention and warning app, such as whoscall, so the anti-fraud prevention schemes on smartphones are literally mature, including caller ID blacklist filter and SMS fraud keywords detection. Conversely, there are no any relevant prevention schemes on traditional home phone, even it has become gradually declined, a large proportion of families still have it, and most of the users are elderly, they are regarded as high risk population. Therefore, this study tries to combine relevant techniques, such as speech recognition and keywords detection, and using the Raspberry Pi, to propose an effective method and construct a home phone fraud prevention scheme, wishing to prevent the home phone fraud crime effectively.

 

Keywords: telecom fraud, embedded system, speech recognition, keywords detection.