UsingCross-MediaRelationstoReduceSPIT资源-CSDN文库

需积分: 1 164 浏览量 2012-11-19 20:26:35 上传评论收藏 536KB PDF 举报

使用跨媒体关系减少SPIT（互联网电话垃圾信息）：一种创新方法摘要部分明确指出，大多数合法的电话呼叫来自于与被叫方具有强社会联系的人或组织，如朋友。然而，也有一些合法的电话呼叫来自于与被叫方仅有弱社会联系的实体，例如在线预订过餐桌的餐厅。由于被叫方的联系人列表通常只包含与其有强社会联系的人或组织的地址，因此单纯依赖联系人列表来过滤未请求的电话容易产生误报。为减少这些误报，研究者首先分析了通话记录，发现合法的电话往往源自于通过网络交易或电子邮件交流与被叫方建立了弱社会联系的个人或组织。本文提出两种利用跨媒体关系来标记来电的方法，旨在基于之前接触的线索识别潜在的来电者。第一种方法是潜在的来电者向被叫方提供其联系方式，这些联系方式可能在未来电话中使用。另一种方法则是被叫方向潜在来电者提供一些弱保密信息，来电者在未来的通话中传递这些信息，以此证明自己是被叫方以前通过其他媒介接触过的个体。后一种方法使被叫方即使没有来电者的身份标识，也能对来电进行标记。减少在过滤过程中的误报将有助于减少SPIT（互联网电话垃圾信息）。关键词：未请求的电话；SPIT预防；跨媒体关系；电子邮件；万维网；网络；VoIP；SIP ### 引言未经请求的电话通常来自被叫方不熟悉其联系方式或未曾接触的个人或组织。由于基于IP的基础设施更容易受到未请求电话的侵扰，正如文献[1]中所描述的那样，人们近期遭遇的SPIT电话数量显著增加。相比之下，大多数合法的电话都有被叫方先前见过的来电者标识符（ID）。然而，有些合法的电话却没有明确的来电者ID，例如预约确认、预订或订单确认等电话。为了应对这一挑战，文章提出了一种基于跨媒体关系的新方法，即利用之前在不同媒体渠道（如电子邮件、网页、社交媒体等）上的互动来建立信任和识别。这种方法的核心在于，它不仅依赖于传统的电话通信记录，还结合了其他媒体渠道的信息，从而更准确地判断来电的合法性，避免了仅依赖电话簿或来电显示的局限性，有效减少了合法来电被误判为垃圾电话的可能性，进而降低了SPIT的干扰。 ### 方法论 #### 方法一：共享联系信息该方法鼓励潜在的来电者主动分享其联系方式，包括但不限于电子邮件地址、社交媒体账号等，以便于被叫方将来识别。这种预先共享联系信息的做法，能够建立起初步的信任，并使得未来通信时更容易被识别为已知来源。 #### 方法二：利用弱保密信息第二种方法则更为巧妙，它要求被叫方在初次接触时向潜在的来电者提供一些特定的、不完全保密但足以作为识别线索的信息。这些信息可以是一段对话的摘录、某个事件的细节或其他不易复制但具有独特性的内容。当来电者在未来的通话中正确引用这些信息时，就表明了其确实与被叫方有过前期接触，从而增强了其合法性的可信度。 ### 结论通过采用跨媒体关系，结合多种媒介渠道的信息，我们能够更有效地辨别合法来电与垃圾信息，减少误报，提升通信体验。这不仅对个人用户有益，也对整个通信行业的健康运行至关重要。随着技术的不断进步，未来有望开发出更多智能化、个性化的方法，进一步优化来电识别机制，为用户提供更加安全、便捷的通信服务。

资源推荐

资源详情

资源评论

Have I Met You Before?

Using Cross-Media Relations to Reduce SPIT

Kumiko Ono

Columbia University

New York, USA

kumiko@cs.columbia.edu

Henning Schulzrinne

Columbia University

New York, USA

hgs@cs.columbia.edu

ABSTRACT

Most legitimate calls are from persons or organizations with

strong social ties such as friends. Some legitimate calls, how-

ever, are from those with weak social ties such as a restau-

rant the callee booked a table on-line. Since a callee’s con-

tact list usually contains only the addresses of persons or or-

ganizations with strong social ties, ﬁltering out unsolicited

calls using the contact list is prone to false positives. To

reduce these false positives, we ﬁrst analyzed call logs and

identiﬁed that legitimate calls are initiated from persons or

organizations with weak social ties through transactions over

the web or email exchanges. This paper proposes two ap-

proaches to label incoming calls by using cross-media rela-

tions to prior contact. One approach is that a potential caller

oﬀers the callee his contact addresses which might be used

in future calls. Another is that a callee provides a potential

caller with weakly-secret information. In order to be iden-

tiﬁed as someone the callee contacted before through other

means, the caller can convey the information in future calls.

The latter approach enables a callee to label incoming calls

even without caller identiﬁers. Reducing false positives dur-

ing ﬁltering using our proposed approaches will contribute

to the reduction in SPIT (SPam over Internet Telephony).

Keywords

Unsolicited calls; SPIT prevention; cross-media relations;

email; WWW; web; VoIP; SIP

1. INTRODUCTION

Unsolicited calls usually originate from persons or organiza-

tions, whom the callee does not know their contact addresses

nor met before. Since an IP-based infrastructure is more

vulnerable to unsolicited calls, as described in [1], people

have recently been experiencing more SPIT calls. Most le-

gitimate calls, by contrast, have caller identiﬁers (IDs) that

the callee has seen before. Some legitimate calls, however,

have unknown caller IDs. Examples of these legitimate calls

include conﬁrmations of appointments, reservations, or de-

Permission to make digital or hard copies of all or part of this work for

personal or classroom use is granted without fee provided that copies are

not made or distributed for proﬁt or commercial advantage and that copies

bear this notice and the full citation on the ﬁrst page. To copy otherwise, or

republish, to post on servers or to redistribute to lists, requires prior speciﬁc

permission and/or a fee.

IPTCOMM’09 July 7-8, 2009, Atlanta, Georgia, USA.

Figure 1: Existing SPIT ﬁlter

liveries, and recorded notiﬁcations of ﬂight delays or school

closing on a snowy day. These legitimate calls are often mis-

takenly labeled as unsolicited calls at a SPIT ﬁltering system

since their caller IDs are not found on the callee’s white list.

Figure 1 illustrates a typical SPIT ﬁltering system, which

authenticates caller IDs and looks them up on a black list

or reject-list and a white list or accept-list. For a VoIP

(Voice over IP) call using the SIP (Session Initiation Pro-

tocol) [2], the SIP Identity header [3] enables a callee to au-

thenticate the caller ID. Some legitimate calls, however, are

sent with “unavailable” caller IDs because the authentication

of the caller IDs fails. For example, most international calls

or calls through a VoIP–PSTN (Public Switched Telephone

Network) gateway have no authenticated caller ID. These

anonymous calls limit the eﬀectiveness of labeling incoming

calls based on the caller ID.

Generally, the callee’s black list contains the contact ad-

dresses of undesired callers or links to a reputation service

that gathers IDs of well-known malicious callers. Unfortu-

nately, however, callers can pick a new caller ID for each

call easily especially for VoIP calls. Thus, the eﬀectiveness

of the black list is limited.

On the other hand, the callee’s white list contains the ad-

dresses from his contact list or address book, which is pop-

ulated by contact addresses of people with strong ties in his

social network [4] such as family members and friends. For

business use, the white list usually links to a directory service

located on an LDAP (Lightweight Directory Access Proto-

col) [5] server. For either use, however, the white list does

not usually include the addresses of persons or organizations

with weak social ties [4] such as friends of a friend in an SNS

(Social Network Service). When people who have with weak

social ties place calls for the ﬁrst time, their calls are ﬁltered

out since their caller IDs are not found in the callee’s white

list. This is called the “introduction problem.” To mitigate

this introduction problem, some systems forward these calls

to a voice mail box, rather than reject them. However, this

is not a desirable solution because it requires callee’s time to

check them and causes the delay of notiﬁcations. Thus, we

need a better approach to label incoming calls from persons

or organizations who have with weak social ties.

For this purpose, we analyze how legitimate calls from peo-

ple with weak social ties are triggered. We then propose two

mechanisms to label incoming calls by using cross-media re-

lations between calls and previous contacts. For our ﬁrst

mechanism, a potential caller oﬀers the callee his contact

addresses which he might use in future calls. If the callee

agrees, these contact addresses are added to his white list.

We describe this mechanism further in Section 4.1. For

our second mechanism, a callee provides a potential caller

with weakly-secret information that the caller can use in

future calls in order to be identiﬁed as someone the callee

has had prior contact through other means, as outlined in

Section 4.2. Section 5 describes a use case integrated with

an SNS and Section 6 describes implementation to achieve

these mechanisms. Finally, Section 7 concludes the paper.

2. RELATED WORK

Similar to preventing bulk unsolicited emails, spams, there

is no panacea for preventing unsolicited calls; thus, a col-

lection of solutions is needed. As described in [1], the so-

lution space can be divided into two categories: one places

procedural, computational, ﬁnancial, and/or legal burdens

on callers, and another labels incoming calls on the callee

side. Our mechanisms using cross-media relations are used

in conjunction with adding procedural burden on the caller

side and enhancing the labeling mechanism on the callee

side. Consent-based solutions in SIP [6] are also used in

conjunction with the two categories. To grant a permission,

whereas the consent-based solutions use additional SIP mes-

sages, our mechanisms reuse messages through other means

than the SIP.

Most well-known solutions for labeling incoming calls are

based on authenticated caller IDs as described in Section 1.

Since maintaining static lists of caller IDs as white and black

lists has the introduction problem, many approaches using

social graphs have been proposed. To expand white lists us-

ing social networks, Ceglowski and Schachter [7] introduced

address book sharing with privacy as an email attachment,

while we [8] oﬀered address book propagation within SIP

messages. To update white lists based on communication

history, Balasubramaniyan and his colleagues [9] introduced

call credentials based on the call history of a caller. Dantu

and Kolan [10] described learning systems based on unso-

licited call traﬃc patterns in order to update reputation and

black lists. Although their communication history limits to

calls, Shacham and Schulzrinne [11] addressed using alter-

native communication channels, web transactions to collect

potential caller IDs. This is the base work for our mecha-

nisms using cross-media relations, which we expand to use

email exchanges.

To label incoming calls without caller IDs, one of our label-

ing mechanisms is based on the destination address with sub-

addressing [12], which has already been deployed for emails.

For calls, subaddressing in the userinfo of the SIP-URI is

new, but the concept of extensions in the tel-URI is similar

to call distribution at a PBX (Private Branch Exchange).

Relying on the observation that many unsolicited calls play

prerecorded messages to decrease cost to the callers, Quit-

tek and his colleagues [13] proposed Turing tests to detect

human communication patterns. However, some legitimate

calls from government agencies, credit card companies, or

dealers among others are automated recorded messages. The

SPIT detection system proposed by Mathieu and his col-

leagues [14] relies on a SPIT characteristic that unsolicited

calls originate more error messages than legitimate calls.

However, legitimate automated calls also share this prop-

erty. Also, unless the originating carrier cooperates, it may

be diﬃcult to measure the outgoing call volume. This infor-

mation may also be considered privacy sensitive or a business

secret.

3. LEGITIMATE CALLS FROM WEAK SO-

CIAL TIES

Our quick survey gives a rough sense of how often people

are experiencing unsolicited calls, how well-maintained con-

tact lists are eﬀective in labeling legitimate calls, and how

legitimate calls from weak ties are initiated. In this survey,

we gathered call records of 246 calls from eight cell phones

and 136 calls from four landline phones from our colleagues

at our lab. We also asked the participants about their rela-

tionship to legitimate callers whose IDs were not found on

their contact lists.

Figure 2 indicates a signiﬁcant diﬀerence in the propor-

tions of unsolicited calls between cell and landline phones.

Whereas only six percent of the incoming calls on cell phones

were unsolicited calls, 52 percent of those on landline phones

were unsolicited. We suspect that this diﬀerence was caused

by the FTC (Federal Trade Commission) regulations that

prohibit telemarketing calls to cell phones [15]. Even though

we can reduce unsolicited telemarketing calls using national

“Do Not Call”registry service [15], the eﬀect is unfortunately

limited. This is because their jurisdiction is limited over do-

mestic telemarketers, not over international ones nor calls

using VoIP. Also, some telemarketers appear to be ﬂouting

the law. Thus, we still need a technical mechanism to help

a callee decide whether to accept incoming calls.

Figure 2 also illustrates a diﬀerence in the proportions of

legitimate calls with known caller IDs. A larger proportion,

78 percent, of the calls for cell phones carried known caller

IDs, which were found on the contact list, compared to 18

percent for landline phones. Since people usually maintain

their contact lists on cell phones better than landlines, the

result shows how well-maintained contact lists are useful to

label incoming calls.

剩余6页未读，继续阅读

评论收藏

内容反馈

ooOO005

粉丝: 0
资源: 4

Using Cross-Media Relations to Reduce SPIT

Charles.River.Media.Cross-Platform.Web.Services.Using.CSharp.and.Java

DisCrasher-2_spit_

Alpops-Pizza:为SPIT Web-It-Up构建的简单Pizza Maker网站2021

json-schema:一个根据Clojure规范生成JSON模式的库

福建省南安市石井镇厚德中学九年级英语上册 Unit 2 Topic 2 All these problems are very

论文研究-基于社会网络分析的IMS网络SPIT识别方法 .pdf

动词不规则变化.doc

ble-sniffer_win_1.0.1_1111_Sniffer (1).zip_BLE win_ble-sniffer_b

React Spit-crx插件

Spit-开源

中考英语作文常考典型句子归纳.doc

NeuralResearch:实习研究的神经网络实施和未来目标

Camel-Spit-Run-game:带有骆驼英雄的跑步者 HTML5 游戏

高中英语必修五Unit15 Learning作业题及答案解析6份2精选.docx

高中常用不规则动词表.doc

初中英语句子成分.ppt

Natural Language Processing in Python 【2016】

一&#8226;三&#8226;六导学案湖北省咸宁市嘉鱼县城北中学九年级英语上册Unit2Topic2Alltheseprobl

最新英语四级高频词汇

浙江高考英语--600高频词汇.doc

自考英语（二）大纲单词.docx

promise-child:从ChildProcess创建承诺

2012新四级700高频词汇

卫星移动通信业务介绍专题培训课件.ppt

Apress.Pro.C.Sharp.2008.and.the.dot.NET.3.5.Platform.4th.Edition.Nov.2007.part2

2022届高考英语一轮复习第一部分教材基础回扣练选修八课时作业含解析牛津译林版202107061150

08年四级高频词汇详解

2023.10.21 雷蛇+鼠标宏+PUBG+绝地求生 步枪通用

115转存助手ui优化版3.9.1网友魔改-转存提取全修复-user

最新资源

高中英语必修五Unit15　Learning作业题及答案解析6份2精选.docx

一•三•六导学案湖北省咸宁市嘉鱼县城北中学九年级英语上册Unit2Topic2Alltheseprobl

2023.10.21 雷蛇+鼠标宏+PUBG+绝地求生步枪通用