qgyd2021/spam_detect|垃圾信息检测数据集|数据分析数据集

该数据集包含多个用于垃圾信息检测的子数据集，涵盖垃圾邮件、垃圾短信和广告识别等任务。每个子数据集的信息包括数据来源、语言、任务类型、原始数据/项目地址、样本个数、原始数据描述和替代数据下载地址。数据集中的样本示例展示了不同类型垃圾信息的内容和分类结果。

原始信息汇总

垃圾信息检测数据集概述

数据集概览

该数据集用于垃圾信息检测，包括垃圾邮件、垃圾短信息和广告识别。数据集包含多种语言和任务类型的样本，主要以英语和汉语为主。

数据来源

数据集从多个公开资源中收集整理，具体如下：

数据集名称	语言	任务类型	原始数据/项目地址	样本个数	原始数据描述	替代数据下载地址
enron_spam	英语	垃圾邮件分类	enron_spam_data; Enron-Spam; spam-mails-dataset	ham: 16545; spam: 17171	Enron-Spam 数据集是 V. Metsis、I. Androutsopoulos 和 G. Paliouras 收集的绝佳资源	SetFit/enron_spam; enron-spam
enron_spam_subset	英语	垃圾邮件分类	email-spam-dataset	ham: 5000; spam: 5000
ling_spam	英语	垃圾邮件分类	lingspam-dataset; email-spam-dataset	ham: 2172; spam: 433	Ling-Spam 数据集是从语言学家列表中整理的 2,893 条垃圾邮件和非垃圾邮件消息的集合。
sms_spam	英语	垃圾短信分类	SMS Spam Collection; SMS Spam Collection Dataset	ham: 4827; spam: 747	SMS 垃圾邮件集合是一组公开的 SMS 标记消息，为移动电话垃圾邮件研究而收集。	sms_spam
sms_spam_collection	英语	垃圾短信分类	spam-emails	ham: 4825; spam: 747	该数据集包含电子邮件的集合	email-spam-detection-dataset-classification; spam-identification; sms-spam-collection; spam-or-ham
spam_assassin	英语	垃圾邮件分类	datasets-spam-assassin; Apache SpamAssassin’s public datasets; Spam or Not Spam Dataset	ham: 4150; spam: 1896	数据集从email-spam-dataset的completeSpamAssassin.csv文件而来。	email-spam-dataset; talby/SpamAssassin; spamassassin-2002
spam_base	英语	垃圾邮件分类	spambase		将电子邮件分类为垃圾邮件或非垃圾邮件	spam-email-data-uci
spam_detection	英语	垃圾短信分类	Deysi/spam-detection-dataset	ham: 5400; spam: 5500
spam_message	汉语	垃圾短信分类	SpamMessage	ham: 720000; spam: 80000	其中spam的数据是正确的数据，但是做了脱敏处理（招生电话：xxxxxxxxxxx），这里的 x 可能会成为显著特征。而ham样本像是从普通文本中截断出来充作样本的，建议不要用这些数据。
spam_message_lr	汉语	垃圾短信分类	SpamMessagesLR	ham: 3983; spam: 6990
trec07p	英语	垃圾邮件分类	2007 TREC Public Spam Corpus; Spam Track	ham: 25220; spam: 50199	2007 TREC Public Spam Corpus	trec07p.tar.gz
trec06c	汉语	垃圾邮件分类	2006 TREC Public Spam Corpora		2006 TREC Public Spam Corpora
youtube_spam_collection	英语	垃圾评论分类	youtube+spam+collection; YouTube Spam Collection Data Set	ham: 951; spam: 1005	它是为垃圾邮件研究而收集的公共评论集。

样本示例

enron_spam 样本示例

wanted to try ci 4 lis but thought it was way too expensive for you ? <br> viagra at $ 1 . 12 per dose ready to boost your sex life ? positive ? time to do it right now . order viagra at incredibly low prices $ 1 . 12 per dose . unbelivable remove <br> spam

enron / hpl actuals for december 11 , 2000 <br> teco tap 30 . 000 / enron ; 120 . 000 / hpl gas daily ls hpl lsk ic 30 . 000 / enron ham

looking for cheap high - quality software ? rotated napoleonizes <br> water past also , burn , course . gave country , mass lot . act north good . from , learn form most brother vary . when more for . up stick , century put , song be . test , describe , plain , against wood star . began dress ever group . here oh , most world stay . <br> spam

ideabank website <br> please read the attached document for information about an exciting new website for ets employees ! ham

enron_spam_subset 样本示例

Subject: edrugs online viagra - proven step to start something all over again . nothing is more useful than silence . teachers open the door . you enter by yourself . how sharper than a serpent s tooth it isto have a thankless child ! spam

Subject: start date : 12 / 13 / 01 ; hourahead hour : 5 ; start date : 12 / 13 / 01 ; hourahead hour : 5 ; no ancillary schedules awarded . no variances detected . log messages : parsing file - - > > o : portland westdesk california scheduling iso final schedules 2001121305 . txt ham

Subject: cheapestt medss ! mediccationns at lowesst pricess everyy ! over 80 . % offf , pricess wontt get lowerr we selll vic od ( in v , ia . gra x , ana . x http : / / www . pr 3 sdlugs . com / ? refid = 87 <br> spam

Subject: fw : picture

the following is an aerial photo of the wtc area . it kinda brings on vertigo , but is a phenomenal shot .

http : / / userwww . service . emory . edu / ~ rdgarr / wtc . htm ham

---

ling_spam 样本示例

Subject: internet specialist 007 - the spy <br> internet specialist 007 - the spy learn everything about your friends , neighbors , enemies , employees or anyone else ! - - even your boss ! - - even yourself ! this mammoth snoop collection of internet sites will provide you the newest and most current addresses available on the net today . . . = 20 * track down an old debt , or anyone else that has done you wrong ! it s incredible , and so many new data sites have come online in the past 90 days . . . * over 300 giant resources to look up people , credit , social security , current or past employment , mail order purchases , = 20 addresses , phone numbers , maps to city locations . . . * investigate your family history ! check birth , death , adoption or social security records check service records or army , navy , air force or = 20 marine corps . * locate an old friend ( or an enemy who is hiding ) or a lost = 20 love - - find e-mail , telephone or address information on anyone ! = 20 even look up * unlisted * phone numbers ! * find work by searching classified ads all over the world ! * screen prospective employees - - check credit , driving or criminal records verify income or educational accomplishments = 20 * check out your daughter s new boyfriend ! * find trial transcripts and court orders ! * enjoy the enchantment of finding out a juicy tid-bit about a co-worker . the internet is a powerful megasource of information , = 20 if you only know where to look . i tell you how to find = 20 out nearly anything about anybody , and tell you exactly where to find it ! you will be amazed to find out what personal information = 20 other people can find out about you ! check your credit = 20 report so you can correct wrong information that may be = 20 used to deny you credit . research yourself first ! you ll be horrified , as i was , = 20 at how much data has been accumulated about you . any my huge collection is only the beginning ! once you = 20 locate these free private , college and government web sites , you ll find even more links to even more = 20 information search engines ! = 20 if you believe ( like i do ) that the information that is stored about each one of us should be freely accessible , you ll want to see the snoop collection i ve compiled . verify your own records , or find out what you need to = 20 know about others . i m telling you , it s incredible what you can find out using the internet ! we will accept checks by fax at 813-269 - 9651 or > > > send $ 14 . 95 cash , check or money order to : > > > the coldwell group > > > p . o . box 3787 > > > dept 1007 > > > petersburg , va 23805 i will rush back to you my snoop information for fastest service include your * e-mail * address . = 20 * what information is available - - and exact url to get there ! * exactly where to look for - - and the clever way to use - - = 20 the above search engines , and tons more ! * my easy-to - browse categorized megacenter of information has my own description of how to use each site , and what you ll find when you get there - - and tricky tips on how to = 20 extract the best data ! you can know everything about everybody with this internet specialist collection ! * * soon to be available - - the most complete international internet spy = 20 sites available on the web today * * don t miss this one or you ll be sorry = 20 to be removed from our list please fax your address to 813-269 - 9651 . l = e3 = 01 @ u = 0b <br> spam

Subject: usage - based models - symposium <br> announcing the sixth biennial symposium of the rice university department of linguistics usage-based models of language rice university march 15-18 , 1995 invited speakers : mira ariel tel aviv university joan bybee university of new mexico john du bois university of california , santa barbara michael israel university of california , san diego sydney lamb rice university ronald langacker university of california , san diego tom givon university of oregon brian macwhinney carnegie - mellon university janet pierrehumbert northwestern university john sinclair university of birmingham ( u . k . ) arie verhagen university of utrecht description : the goal of this symposium is to explore approaches to linguistic theory that have in common the aim of accounting for linguistic usage . the empirical data for such theories is not restricted to linguistic intuitions about acceptibility , but comes from usage events of varied types . the focus is on the patterns found in the various sorts of usage data examined , and how those patterns can be extracted , represented , and used by the human mind . research from a variety of traditions will be represented , including corpus-based analyses , discourse studies , experimental studies of language processing and language acquisition , and instrumental phonetics . the approaches taken can be called data-driven , rather than model-driven , in that the fewest possible prior assumptions are made about what types of data are relevant , and that large sets of usage events are observed so that the detailed patterns found in actual usage can emerge . moreover , the various approaches taken show signs of converging toward a view of language as a dynamic system in which linguistic knowledge is not separate from its processing in language use . the linguistic models representing this view are usage-based by virtue of three factors : ( 1 ) the importance placed on usage data for theory construction ; ( 2 ) the direct incorporation of processing ( production and comprehension ) into linguistic theory ; and ( 3 ) the requirement that the models arrived at , whatever the direct source of evidence , must be testable with reference to language use . registration : no charge . symposium attendance on a space-available basis . for further information , contact suzanne kemmer ( kemmer @ ruf . rice . edu ) or michael barlow ( barlow @ ruf . rice . edu ) snailmail : dept . of linguistics , rice university , houston tx 77251-1892 . <br> ham

Subject: domani <br> new improved with free software , free bulk e mail system , free web site = to do what you wish , ongoing support ( optional ) , and a lot more ! all = included . . . . . . . . . . . this is a one time mailing . . . . . . . . . . . . . . . $ = you are about to make at least $ 50 , 000 in less than 90 days read the enclosed program . . . then read it again . . . / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /