E-Mail Spam Detection Using SVM And RBF
I.J. Modern Education and Computer Science, 2016, 4, 57-63Published Online April 2016 in MECS (http://www.mecs-press.org/)DOI: 10.5815/ijmecs.2016.04.07E-Mail Spam Detection Using SVM and RBFReena SharmaChandigarh University/Computer Science, Mohali, 160055, IndiaEmail: sharmareena551@gmail.comGurjot KaurChandigarh University/Computer Science, Mohali, 160055, IndiaEmail: randhawa789@gmail.comAbstract—In today‘s life internet is an important part.We spend most of our time on internet. One of theimportant features of internet is communication. Email isa mode of communication which is used for the personaland business purpose. Spam emails are the emailsrecipient does not wish to take delivery of; it is alsocalled unwanted bulk email. Emails are used each day bynumber of user to converse around the world. At presentlarge volumes of spam emails are reasoning serioustrouble for Internet user and Internet service. Such as itdegrade user investigate knowledge, it assiststransmission of virus in network, it increases load onnetwork traffic. It also misuses user time, and energy forlegal emails among the spam. For evade spam there areso many conventional anti-spam technique includesBayesian based sort, rule based system, IP blacklist,Heuristic based filter, White list and DNS black holes.These methods are based on satisfied of the post or linksof the mail. In this paper we proposed an efficient spamfiltering technique based on neural network. Thetechnique used is RBF a neural network technique inwhich neuron are trained. The results obtained by usingthis technique are compared with SVM. The parametermeter for comparison is precision and accuracy. On thebasis of these two parameters we compared the proposedtechnique with SVM.Index Terms—AdaBoost, Content Spam, Black andWhite Listing, Link Learning, RBF, Spam Filter, SVM.I. INTRODUCTIONSpam is termed as unwanted money oriented mail [4].It is needed to note down that certain individuals want tohave this type of messages. There are viewers for e-mailpublicity, regardless of the product that is being sold.Spammers try to reach these types of individuals. Theydo not know which people are made up of this group.They send spam to peoples, so that than can reach thepeople who comprise the group. This is because theydon‘t know who will respond to the message and whowill not. Spammers are the persons which are technicallyskilled and are hired by companies to send spam.Through a third party, the companies try to keepthemselves from take legal action [5]. Spamming can bevery profitable for a company if it done in right way.Copyright 2016 MECSLet‘s take an example company is selling defected toysfor 50 dollars a toy. If the company lets the spammersend out 10 million mails and the response rate is just0.1% it will make half a million dollars [3].E-mail addresses are get by the spammers throughwebsites, newsgroups etc. [6]. It can be turn this into abenefit, by fooling spammers with foggy e-mail addressesand thus collecting their spam.Spam is not restricted to e-mail. It exists in textmessaging services (SMS) [8], newspapers and othercommunication media. SMS spam can cost even morethan E-mail spam. For example, user has subscribed toreceive a notification via SMS when they receive e-mailat their mail account. They pay for every SMS receivedregardless if it is a spam or a ham.Cell phone spam is a type of junk message in the formof text message. This is defined as SMS spam or textmessage spam.As the vogue of Cell phones rushed in the late 1990s,consistent users of mobile began to see a large scalegrowth in the number of unwantedmarketableadvertisements being sent to their cell phones throughtext messaging. This is mostly irritating for the receiverbecause, unlike in email, some receivers may be chargeda fee for every spam message.Cell phone spam is usually less tenacious than mailspam, where in 2011 around 95% of email is spam. Thevolume of mobile spam differs generally from area toarea. In America, SMS spam has increase at large scalefrom 2007 through 2013, but remains below 1% as ofDecember 2012. In Asia approximately 35% of sms werespams in 2013[32].II. LITERATURE SURVRYRamachandran A et al .in their work they studied thenetwork level junk mails. The spams are detected fromthe network level. In their work they use DNS server forhosting. Spams are detected at IP level. BGP routeralgorithm is used to detect the spam mails [7].Krishnan et al. proposed an Anti- trust Rankalgorithm for the web spam detection. The algorithm isbased on the approximate isolation principle. Thresholdvalues are set for the set of spam web pages [13]. Theresults obtained from Anti Trust Rank and Trust Rankalgorithms are compared.I.J. Modern Education and Computer Science, 2016, 4, 57-63
58E-Mail Spam Detection Using SVM and RBFCarlos Castillo et al presented a learning algorithmWITCH (Web Spam Identification through Content andHyperlinks). This algorithm during learning phase usesthe hyperlink structure in addition to page features. Theway of graph regularization is used to utilize thehyperlink which yields a predictor that differs smoothlyamong interconnected pages [13].Guang Gang Geng et al. in their work they proposeda semi supervised learning link based algorithms. Thesealgorithms are used to speed up the performance of aclassifier. This classifier merges the old self-training withtopological dependency based on link learning [12]. l/webspam/datasets/uk2006/)benchmarkindicated that the algorithms are productive.Loredana Firte et al. presented a new approach forspam detection filter. The solution proposed is an offlineapplication that uses the K-Nearest Neighbor algorithmand pre- classified email data set for the learning process.KNN algorithm classified the messages which is based onfeatures extraction from the email‗s properties andcontent[18].Rafiqul Islam and Yang Xiang did sorting of workeremails form saturation of spam. In their paper, ―EmailClassification is done with the help of Data ReductionMethod‖ which is an effective email classificationmethod. This method is based on data purifying technique.A novel filtering technique using instance selectionmethod (ISM) is introduced. ISM reduces the useless datasamples from training model and then categorizes the testdata. ISM helps to recognize which samples (examples,patterns) in email should be selected as representatives ofthe all dataset, without any loss of facts. They have usedWEKA tool in our joined classification model and testeddiverse classification algorithms [23]. Their experimentalstudies illustrate significant performance in terms ofclassification accuracy with reduction of false positiveinstances.M.Basavaraju et.al, in this paper text clustering spamdetection approach which based on vector space model isproposed. By using this technique one can detect email asspam and non- spam. The Proposed method contains thedistance among all of the elements of an email [19].Saadat Nazirova performed a work,‖ Survey on SpamFiltering Techniques‖. In this paper the existing e-mailspam filtering methods are described. The grouping,judgment, and comparison of traditional and learningbased techniques are provided. Some private anti-spamproducts are verified and compared [26]. The declarationfor new method in spam filtering technique is considered.Faraz Ahmed et al. [9] Markov clustering basedapproach for the detection of spam profiles on OSN‘s ispresented in this paper. Work is based on a data set ofFacebook profiles, which include both benign and fakeprofiles. Three features are identified and used for tomodel public collaboration of OSN user using a weightedgraph. Markov clustering is applied to exploit thebehavior similarity of profiles and mine the clusterexisting in profile data set.Copyright 2016 MECSR. Kishore Kumar et al proposed the survey of emailspam filter over data mining techniques. In their work,―Comparative Study on Email Spam Classifier using DataMining Techniques‖ is proposed. TANAGRA datamining tool is used to analyze the spam data .It explorethe efficient classifier for email spam classification.Firstly, feature creation and feature selection is done todraw out the relevant features. Then numerous groupingalgorithms are applied on this dataset and cross validationis done for each of these classifiers [25]. In conclusion,best classifier for email spam is acknowledged on thebasis of error rate, precision and recall.Lourdes Araujo et al, present a work in which theytries to detect the tweets as spam in actual time bymeans of language as primary tool. Paper also introducedan general valuation method that has permitted showinghow the system is able to obtain an F-measure at the samelevel as the best state of the art system based on thedetection of spam accounts [16].Siddu.Pacingill. Algur et.al, proposed a system inwhich spam web pages are detected with the help of linkand content spam detector. System classifies the webpage as spam based on threshold which is set by algebraicmethod. Unsupervised web spam detection problem isstudied. For link spam detection the URL is taken astarget and the link spamicity is calculated. In contentspamicity the content of the web page is considered [29].The result obtained from both is average spam scorewhich is compare with the threshold value?Nosseir, Khaled Nagati and Islam Taj-Eddinperformed a work,‖ Intelligent Word-Based Spam FilterDetection Using Multi-Neural Networks‖. They proposeda character-based technique. A multi-neural networksclassifier is used by this approach. A normalized weightvalues derived from the ASCII value of the wordcharacters are used to train the neural network [3].Results obtained from experiment show high falsepositive and low true negative percentages.Sahil Puri et al, in this paper spam detection is doneon the basis of content and a rule based filtering. A newfilter has been introduced in suggested work by theinterfacing of rule based filtering followed by contentbased filtering for more efficient results [27].Mohammed Mikki et al [1] An improved filteringtechnique is presented which is based on the improveddigest algorithm and DBSCAN clustering algorithm.Vandana Jaswal in this paper an image spamdetection system is introduced. Hidden markov modelwas used to detect all the spam images.Asmeeta Mali [6] performed a work, ―Spam Detectionusing Bayesian with Pattern Discovery‖. In her paper sheproposed an operative procedure to recover the efficiencyof using and apprising revealed patterns for conclusionappropriate information using Bayesian filteringalgorithm and effective pattern. Discovery technique wecan detect the spam mails from the email dataset withgood correctness of term.Neha Singh performed a work, ―Dendritic Cellalgorithm and Dempster Belief Theory Using ImprovedIntrusion Detection System‖. To reduce the false alarmI.J. Modern Education and Computer Science, 2016, 4, 57-63
E-Mail Spam Detection Using SVM and RBFrate she proposed a new dual detection of IDS based onSimulated System that assimilating the Dendrite CellAlgorithm and Dempster Belief theory in her work [21].R.Malarvizhi et al.a summary for spam filtering, andthe techniques of evaluation and evaluation of differentfiltering methods is present in this paper. Fisher RobinsonInverse chi square, Ad boosts classifier, Bayesianclassifiers are discussed. Bayessian method is used tocreate the spam filter in this paper [24].59ProblemIdentificationCreate spam word dictionaryFeature Extraction of the spamwordsIII. PROPOSED SOLUTIONIntegration of Neural Network toproposed frameworkFrom literature survey we have studied the varioustechniques which are used for the spam detection. Thesetechniques have the problem with accuracy and precision.We proposed the RBF technique. It is a neural networktechnique. Proposed technique improves the accuracy andprecision. Results obtained from the RBF are comparedwith the SVM.Training of Neurons/SVMTesting with respect to training dataIV. WORK DONEStep1: Spam can be identified by various methods.From literature survey we studied the various techniqueswhich are used to detect the email as spam.Different spam detection techniques are considered.These techniques are found to have some limitations.Step2: Markov clustering, DBSCAN, List filtering areavailable techniques which are used to detect the spam.These techniques have less accuracy, precision values.After the spam detection accuracy precision andStep3. Pervious methods have less accuracy andprecision. We proposed the RBF which produces thebetter output. It is a neural network based technique.Another technique named SVM is used. We compare theresult which we obtained from both the techniques.Step4: All experiments are performed in MATLABframework. The framework is used for theimplementation of the SVM and RBF algorithms. TheRBF algorithm is implemented to detect the email as aspam or ham. We calculated the accuracy, precision. TheStep5: Implementation: The implementation ofproposed technique is described. Firstly we identify theproblem, then we start to find the solution of this problemwhich we describe step by step in the diagram. Matlab isa framework where we done the implementation.For spam detection firstly we collect the spam words.We create a spam word dictionary. These words are usedfor training and testing. After the creation of dictionarywe need to extract the feature of these words so that wecan use these words for training and testing.Feature extraction can be done by various ways. Inprevious feature extraction is done with the help ofclustering. We are doing the feature extraction on thebasis of weights of the alphabets. The process of featureextraction is done as follow.Copyright 2016 MECSResult ExtractionComparisonTable 1. Pseudo Code for feature ExtractionStep1Step2Load all input all wordsfor each word in load allValue(i) generator(word);Step3 EndStep4 Void generator()Step5 for each char in word S searchpos(g. file)If S! EmptyVar val Val S;After the feature extraction the training is done usingSVM and RBF. RBF is a neural network technique whichuses the hidden neurons to process the input and to givethe output.SVM create a hyper plane which separate the differenttypes of data. The training of SVM is done by followingthese stepsTable 2. Pseudo code for SVMStep1 Initialize training data (xi , yi ) for i 1 NStep2 Generate the weight vector and bias such thatf x W t bStep3 Train data using svmtrainStep4 Generate the groups for training set such thatsvmstuct svmtrain(training set, groups)Step5 Find the support vector by using svmclassify)such that svmclassify(svmstruct, testdata,groupsStep6 endI.J. Modern Education and Computer Science, 2016, 4, 57-63
60E-Mail Spam Detection Using SVM and RBFRBF training and testing is done. The Liebenbergalgorithm is used in this technique. Following is thepseudo code for RBFTable 3. Pseudo Code for RBFStep1 For each word set in all word generate weight Where a constant, x provided data, b randomweightStep2 Epoch. System 100;Step3 Hidden Neurons 10Step4 Initialization fn init p;Step5 Training function trainlnStep6 Type ― feed forward back propagation‖Method ―RBF‖;Algo ―Lavenberg‖;Step7 If processed hidden neuron true;Step8 Find epochsStep9 Error;Step10 Output layer sim(train set, test set)endTo compare the both techniques we calculate theaccuracy, precision, recall, frr and far so that we canidentify which technique is better. The values calculatedare stored in a table. The according to these values areplotted.Error is calculated firstly so that we can calculate theother values by using this. It can be calculated as:Error (training data – testing Data)2/ Length of testingsetNow with the help of Error we can calculate our othervalues which we are used for result.Far (False Acceptance Ratio): Number of spamclassified as non- spam. It also called false positive ratio.Far (error- test data) / Length of testing set.Frr (False Rejection ratio): Number of non-spamclassified as spam. It can be calculated as:Frr (Error – Far)/ Length of testing set.Precision: it is the percent of positive spam data that iscorrect. We can calculate it as:Precision (Test set-Error)/ Length of testing set.Recall: Its value should be low. It is percentage ofpositive labeled instance.Recall (Test set-precision)/ Length of testing set.Accuracy: It describes how close a measured value toactual value. The technique which has high accuracy isbetter.Accuracy (1-(far frr)/100)Table 4. RBF 7599.7846699.88790.0038950.004307Using RBF we calculated the above values. Thesevalues are used to plot the graphs.Fig.1. Accuracy using RBFFig.2. Recall using RBFRecall is plotted against diiferent testsets. Recall valueshuold be low.By using above formulas we calculate the accuracy,recall, precision, Frr and far. Now with the help of thesevalues we identify which technique is better for the spamdetection. Values are listed in tables. Below table 1.4contains the values which we obtained by using RBF andtable 5 contains the values which we obtained by usingSVM.Fig.3. Far using RBFCopyright 2016 MECSI.J. Modern Education and Computer Science, 2016, 4, 57-63
E-Mail Spam Detection Using SVM and RBFFig.4. Frr using RBF61Fig.7. Precision using SVMFig.8. Recall using SVMFig.5. Precision using RBFTable 5. SVM .300560.0001230.1174Using SVM we calculated the above values. Graphsare plotted using these values.Fig.9. Far using SVMFig.10. Frr SVMFig.6. Accuracy using SVMCopyright 2016 MECSI.J. Modern Education and Computer Science, 2016, 4, 57-63
62E-Mail Spam Detection Using SVM and RBF[2][3][4][5]Fig.11. Comparison of Accuracy[6][7][8][9]Fig.12. Comparison of PrecisionThe above graphs show the comparison of accuracyand precision. The accuracy and precision values of RBFare higher than the SVM. The proposed technique givesthe better results.[10]V. CONCLUSIN AND FUTURE SCOPE[12][11]During the study of this dissertation, it has been widelyobserved that there are numerous spam detectiontechniques available around us. Most of these techniqueseither lack in performance or level of accuracy. Theproposed methodology is adopted to enhance theprecision quotient of the existing spam detection methods.New mechanism using RBF is proposed. The proposedmechanism improves the accuracy, precision, recall Frrand Far. The proposed mechanism is compared withSVM and the results have been comparatively better. Weuse a database of approximately 1000 spam words in ourcurrent research work; in future we can use larger data setfor spam detection. The advanced neural networktechniques can be used in future for better results. Theproposed algorithms can be used with other algorithms tomake a hybrid algorithm which helps to improve theperformance of the spam detection system.REFERENCES[1]Gomes, L.H, Caztia , in Proceeding 4th ACM SIGCOMMConference on Internet Measurement ,ACM, pp.356-369,2014Copyright 2016 MECS[13][14][15][16][17][18]Megha Rathi and Vikas Pareek, ―Spam Mail DetectionThrough Data Mining –A Comparative PerformanceAnalysis‖, International Journal of Modern Educationand Computer Science, Vol.5, No.5, 12 December 2013.R.Malarvizhi and K. Saraswathi, ―Content – Based SpamFiltering and Detection Algorithms-An Efficient Analysisand Comparison‖, International Journal of EngineeringTrends and Technology, Vol.4, Issue 9,Septmber 2013N. Singh,‖ Dendritic Cell Algorithm and Dempster BeliefTheory Using Improved Intrusion Detection System―International Journal of Advanced Research inComputer Science and Software Engineering, Vol. 3,Issue 7, July 2013 ISSN: 2277 128XAsmeeta Mali, ―Spam Detection Using Bayesian withPattern Discovery‖, International Journal of RecentTechnology and Engineering, ISSN: 2277-3878, Vol.2,Issue-3, July 2013.J. Vandana and N. Sood, ―Spam Detection System UsingHidden Markov Model‖, International Journal ofAdvanced Research in Computer Science and SoftwareEngineering, Vol.3, Issue 7, July-2013Ahmed, H. Alaa, ―Improved Spam Detection onal Journal of Computer Applications, Vol. 6,May 2013.S. Puri, M. Ahuja, ―Comparison and Analysis of SpamDetection Algorithms‖, International Journal ofApplication or Innovation in Engineering andManagement, Vol. 2, Issue 4, April 2013A. Nossier, khaled kagati, and Islam Taj-Eddin,―Intelligent Word- Based Spam Filter Detection UsingMulti Neural Networks‖, International Journal ofComputer Science, Vol. 10, Issues 2, No. 1, March 2013ISSN(Print): 1694 -0814 ISSN(Online): 1664- 0784S. Roy, A. Patra, and S. Sau, ―An Efficient Spam FilteringTechniques for Email Account‖, American Journal ofResearch, Vol. 2, Issue 10, 2013F. Ahmed and M .Abulaish, ―An MCL Based Approachfor Spam Profile Detection in Social Network‖, inproceedings of 11th International Conference on Trust,Security and Privacy, IEEE, pp. 602-608, 2012T. Mahmoud and M.Mahfauz, ―SMS Spam filteringTechniques Based on Artificial Immune System‖,International Journal of Computer Science, Vol. 9,Issue 2,No.1, March 2012J. Martinez Romo and Lourdes Araujo, ―Detectingmalicious Tweets in Trending Topics Using a StatisticalAnalysis of Language‖, Elsevier, 2012R. Kishore Kumar, G. Poonkuzhali, P. Sudhakar,―Comparative Study on Email Spam Classifier using DataMining Techniques‖, in Proceedings of the InternationalMulti Conference of Engineering and Computer Science,Vol 1, 2012Siddu. P, Alger and N. tarannumPendari, ―HybridSpamicity Approach to web Spam Detection‖, inProceeding of Conference on Pattern Recognition,Informatics and Medical Engineering IEE, March 2012Tiago. A ―Contribution to the study of SMS SpamFiltering New Collections and Result‖, in Proceedings ofthe 11th ACM symposium on Document engineering ACM,pp. 259-262, Sept 2011S. Nazirova, ―Survey on Spam Filtering Techniques‖,Communication and Network, August 2011G. Kumari Tak and S. Taposwi, ―Query Based Approachtowards Spam Attacks Using Artificial Neural Network‖,International Journal of Artificial Intelligence andApplication, No.4,Oct 2010I.J. Modern Education and Computer Science, 2016, 4, 57-63
E-Mail Spam Detection Using SVM and RBF[19] M.Basavaraju and Dr.R.Prahakar, ―A Novel Method ofSpam Mail Detection using Text Based ClusteringApproach‖, International journal of ComputerApplications, Vol. 5, August 2010.[20] R. Islam and Yang Xiang, ―Email Classification UsingData Reduction Method‖, in Proceeding of 5thInternational ICST Conference on Communications andNetworking in China (CHINACOM), , pp. 1-5. IEEE,2010, June 16, 2010[21] L. Firte, C. Lemnaru and R. Potolea, ―Spam DetectionFilter Using KNN Algorithm and Resampling‖, inProceedings of the 2010 IEEE 6th InternationalConference on Intelligent Computer Communication andProcessing, pp. 27-33. IEEE, 2010.[22] M.Soranamageswari and C.Meena, ―Statistical FeatureExtraction for Classification of Image Spam UsingArtificial Neural Network‖, in Proceeding of 2ndInternational Conference on Machine Learning andComputing, February 2010[23] Guang-Gang Geng, Qiu-Dan Li and Xin-Chang Zhang,―Link Based Small Sample Learning for Web SpamDetection‖, ACM, April 2009[24] J. Abernethy, Oliver Chappelle and Carlos Castillo,―WITCH: A New Approach to Web Spam Detection‖, inProceedings of the 4th International Workshop onAdversarial Information Retrieval on the Web (AIR Web.2008)[25] Alex Brodsky (Canada) and Dmitry Brodsky (USA), ―ADistributed Content Independent Method for SpamDetection‖ International Journal of ComputerApplications, 2007.[26] A. Ramachandran and Nick Feamster, ―UnderstandingNetwork – Level Behaviour of Spammers‖, ACM,September 2006.[27] Sender policy framework (SPF) for authorizing use ofdomains in e-mail, Version.1, No. RFC 4408. 2006.63[28] Erik D. Demaine, F.H. Meyer auf der Heide, U. Paderborn,Rasmus Pagh, Mihai Pˇatra scu, ―On DynamicDictionaries Using Little Space‖, ARVIX, 2005.[29] P.A.Chitira, J.Diederich and W.Nejdl, ―MailRank: usingRanking for Spam Detection‖, in Proceedings of 14thInternational Conference on Information and KnowledgeManagement ACM, October 2005.[30] Anti-Spam site, Claws and paws, Aug 2004 clawsandpaws. com/spam-l/tracking.html[31] J. Junod, ―Serves to Spam: Drop Dead‖, Computer andSecurity Elsevier, Vol.16, 1997[32] Anti - Spam abuse site – http://spam.abuse.net/Authors’ ProfilesReena Sharma student at ChandigarhUniversity. My area of interest is NeuralNetworks. I have completed my thesis workon Email spam detection.Er. Gurjot kaur Assistant Professor at UIE-CSEChandigarh University. I have completed myM.tech from Punjabi University Patiala. My areaof interest cloud computing, network security andCryptography.How to cite this paper: Reena Sharma, Gurjot Kaur,"E-Mail Spam Detection Using SVM and RBF", InternationalJournal of Modern Education and Computer Science(IJMECS), Vol.8, No.4, pp.57-63, 2016.DOI:10.5815/ijmecs.2016.04.07Copyright 2016 MECSI.J. Modern Education and Computer Science, 2016, 4, 57-63
send out 10 million mails and the response rate is just 0.1% it will make half a million dollars [3]. E-mail addresses are get by the spammers through websites, newsgroups etc. [6]. It can be turn this into a benefit, by fooling spammers with foggy e-mail addresses and thus collecting their spam. Spam is not restricted to e-mail.
Anti‐Spam 3 10 Anti‐Spam Email Security uses multiple methods of detecting spam and other unwanted email. This chapter reviews the configuration information for Anti‐Spam: Spam Management Anti‐Spam Aggressiveness Languages Anti‐Spam Aggressiveness Spam Management
To reduce the false detection rate. To classify between the spam and ham (non-spam) tweets. 2. Related Works [5] For detecting the spam existing in the social media platform of Twitter, a framework of semi-supervised spam detection (i.e., S3D) was proposed in the research work. Two different modules namely spam detection module
Anti-spam scanning relates to incoming mail only , and in volv es chec king whether a message needs to be categorised as spam or suspected spam (depending on the spam rating of the message) and taking appropr iate action. A spam digest email and w eb based spam quar antine enables end users to manage their quarantined spam email.
Spam related cyber crimes, including phishing, malware and online fraud, are a serious threat to society. Spam filtering has been the major weapon against spam for many years but failed to reduce the number of spam emails. To hinder spammers' capability of sending spam, their supporting infrastructure needs to be disrupted.
2 Spam detection accuracy is the industry -standard metric used to measure how accurate an anti spam filter is at correctly identifying spam. Generally, higher spam detection accuracy is obtained at the cost of a higher false positive rate. A good anti-spam filter will have an acceptable trade-off between the two metrics.
learn to identify spam e-mail after receiving training on messages that have been manually classified as spam or non-spam. A spam filter is a program that is mainlyemployed to detect unsolicited and unwanted email and prevent those messages from reaching a user's inbox. Just like other types of filtering programs, a spam filter looks for certain
Spam Filter User Guide Page 3 Getting to Know Your Spam Filter Features. Your spam filter consists of four tabs: Messages, Settings, Policies, and Status. The default setting is Messages, which displays all of the messages quarantined by the spam filter. Managing Your Quarantined Messages. The Inbound quarantine section will show the
Barracuda Spam Firewall: Login and logout activity: All logs generated by Barracuda spam virus firewall when login or logout is happened on barracuda spam firewall web interface. Barracuda Spam Filter: User login success: This category provides information related to user login success into barracuda spam filter.
