• Journal of KIISE:Software and Applications
• /
• v.32 no.5
• /
• pp.378-384
• /
• 2005

In this paper, we describe a novel method of spam-filtering to improve the performance of conventional spam-filtering systems. Conventional systems filter emails by investigating words distribution in email headers or bodies. Nowadays, spammers begin making email accounts in web-based email service sites and sending emails as if they are not spams. Investigating the email accounts of those spams, we notice that there is a large difference between the automatically generated accounts and ordinaries. Based on that difference, incoming emails are classified into spam/non-spam classes. To classify emails from only account strings, we used decision trees, which have been generally used for conventional pattern classification problems. We collected about 2.15 million account strings from email service sites, and our account checker resulted in the accuracy of $96.3\%$. The previous filter system with the checker yielded the improved filtering performance.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.04b
• /
• pp.223-225
• /
• 2002

인터넷의 급속한 성장과 함께 E-Mail은 대표적인 통신수단의 하나가 되어버렸다. 편리하다는 점을 이용해서 엄청난 양의 스팸메일이 매일같이 쏟아져 오고 , 그 문제점의 심각성에 정보통신부에서 정보통신망 이용촉진 및 정보보호 등에 관한 법률이라는 새로운 법률까지 생겨났다. 본 논문에서는 이 법률에서 요구하는 '광고'라는 문구를 걸러내는 등의 메시지 규칙을 갖는 시스템과 기존의 문서 분류에 널리 쓰이던 나이브 베이지안 분류자(Naive Baesian Classifier)를 결합한 스팸 메일 필터링 시스템(Spam-mail Fitering System)을 제안한다. 제안된 시스템에서는 사용자가 직접 규칙을 작성할 필요없이 학습한 데이터를 갖고 자동으로 스팸메일을 분류할 수가 있다. 들어온 메일은 메시지 규칙 기반 필터가 먼저 적용되고, 메세지 규칙 기반 필터에서 분류되지 않으면 나이브 베이지안 필터에서 분류된다. 실험에서는 제안된 시스템의 성능을 평가하기 위해서 메시지 규칙을 사용한 시스템 및 나이브 베이지만 분류자 시스템과 비교 평가하였다. 또한 임계치를 변경함으로써 제안된 시스템의 성능을 높일 수있도록 하였다.

• The KIPS Transactions:PartB
• /
• v.13B no.4 s.107
• /
• pp.409-416
• /
• 2006

In this paper, we propose an algorithm for extracting text regions from spam-mail images using color layer. The CLTE(color layer-based text extraction) divides the input image into eight planes as color layers. It extracts connected components on the eight images, and then classifies them into text regions and non-text regions based on the component sizes. We also propose an algorithm for recovering damaged text strokes from the extracted text image. In the binary image, there are two types of damaged strokes: (1) middle strokes such as 'ㅣ' or 'ㅡ' are deleted, and (2) the first and/or last strokes such as 'ㅇ' or 'ㅁ' are filled with black pixels. An experiment with 200 spam-mail images shows that the proposed approach is more accurate than conventional methods by over 10%.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.18 no.2
• /
• pp.129-137
• /
• 2008

In this paper, we propose an improved Bayesian Filtering mechanism to reduce the False Positives that occurs in the existing Bayesian Filtering mechanism. In the existing Bayesian Filtering mechanism, the same Bayesian Filtering DB trained at the e-mail server is applied to each e-mail user. Also, the training method using receiving e-mails only could not provide the high quality of ham DB. Due to these problems, the existing Bayesian Filtering mechanism can produce the False Positives which misclassify the ham e-mails into the spam e-mails. In the proposed mechanism, the sending e-mails of the user are treated as the high quality of ham information, and are trained to the Bayesian ham DB automatically. In addition, by providing a different Bayesian DB to each e-mail user respectively, more efficient e-mail filtering service is possible. Our experiments show the improvement of filtering accuracy by 3.13%, compared to the existing Bayesian Filtering mechanism.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.10c
• /
• pp.222-225
• /
• 2006

한국은 다른 나라에 비해 많은 인터넷 사용자를 가지고 있다. 이에 비례해서 한국의 인터넷 유저들은 Spam Mail에 대해 많은 불편함을 호소하고 있다. 이러한 문제를 해결하기 위해 본 논문은 다양한 Feature Weighting, Feature Selection 그리고 문서 분류 알고리즘들을 이용한 한국어 스팸 문서 Filtering연구에 대해 기술한다. 그리고 한국어 문서(Spam/Non-Spam 문서)로부터 영사를 추출하고 이를 각 분류 알고리즘의 Input Feature로써 이용한다. 그리고 우리는 Feature weighting 에 대해 기존의 전통적인 방법이 아니라 각 Feature에 대해 Variance 값을 구하고 Global Feature를 선택하기 위해 Max Value Selection 방법에 적용 후에 전통적인 Feature Selection 방법인 MI, IG, CHI 들을 적용하여 Feature들을 추출한다. 이렇게 추출된 Feature들을 Naive Bayes, Support Vector Machine과 같은 분류 알고리즘에 적용한다. Vector Space Model의 경우에는 전통적인 방법 그대로 사용한다. 그 결과 우리는 Support Vector Machine Classifier, TF-IDF Variance Weighting(Combined Max Value Selection), CHI Feature Selection 방법을 사용할 경우 Recall(99.4%), Precision(97.4%), F-Measure(98.39%)의 성능을 보였다.

• Journal of Information Processing Systems
• /
• v.19 no.3
• /
• pp.289-301
• /
• 2023

Due to the development and dissemination of modern technology, anyone can easily communicate using services such as social network service (SNS) through a personal computer (PC) or smartphone. The development of these technologies has caused many beneficial effects. At the same time, bad effects also occurred, one of which was the spam problem. Spam refers to unwanted or rejected information received by unspecified users. The continuous exposure of such information to service users creates inconvenience in the user's use of the service, and if filtering is not performed correctly, the quality of service deteriorates. Recently, spammers are creating more malicious spam by distorting the image of spam text so that optical character recognition (OCR)-based spam filters cannot easily detect it. Fortunately, the level of transformation of image spam circulated on social media is not serious yet. However, in the mail system, spammers (the person who sends spam) showed various modifications to the spam image for neutralizing OCR, and therefore, the same situation can happen with spam images on social media. Spammers have been shown to interfere with OCR reading through geometric transformations such as image distortion, noise addition, and blurring. Various techniques have been studied to filter image spam, but at the same time, methods of interfering with image spam identification using obfuscated images are also continuously developing. In this paper, we propose a deep learning-based spam image detection model to improve the existing OCR-based spam image detection performance and compensate for vulnerabilities. The proposed model extracts text features and image features from the image using four sub-models. First, the OCR-based text model extracts the text-related features, whether the image contains spam words, and the word embedding vector from the input image. Then, the convolution neural network-based image model extracts image obfuscation and image feature vectors from the input image. The extracted feature is determined whether it is a spam image by the final spam image classifier. As a result of evaluating the F1-score of the proposed model, the performance was about 14 points higher than the OCR-based spam image detection performance.

• Journal of Information Processing Systems
• /
• v.13 no.5
• /
• pp.1397-1409
• /
• 2017

For text categorization task, distinctive text features selection is important due to feature space high dimensionality. It is important to decrease the feature space dimension to decrease processing time and increase accuracy. In the current study, for text categorization task, we introduce a novel statistical feature selection approach. This approach measures the term distribution in all collection documents, the term distribution in a certain category and the term distribution in a certain class relative to other classes. The proposed method results show its superiority over the traditional feature selection methods.

• The Journal of the Korea Contents Association
• /
• v.10 no.6
• /
• pp.43-48
• /
• 2010

Useful emails influence on consumers' purchase behavior and activate them to visit retail stores. Regular contact with consumers by e-mail has positive effects on brand loyalty. However, email marketing has a limitation. Spam now accounts for over half of all e-mail traffic. The increase of email users has resulted in the dramatic increase of spam emails during the past few years. In this paper, we proposed an ontology-based system offering personalized email services to overcome such limitation. Our method is not the ontology-driven spam filtering, but a personalized content service considering personal interests and relations among people by using FOAF and domain ontologies. Our system was successfully tested in email marketing domain.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2017.07a
• /
• pp.35-36
• /
• 2017

이메일 서비스 사용자들은 스패머가 무차별적으로 발송하는 스팸메일에 의한 정신적 경제적 피해를 입고 있다. 이러한 피해를 막기 위해 필터링, RBL (Real-time Blackhole List)과 같은 스팸차단 기법이 등장하였고 많은 메일서버에서 사용되고 있다. 그러나 이는 스팸메일의 근본적인 원인은 해결하지 못하며, 높은 차단율을 유지하기 위해서는 지속적인 관리 및 업데이트가 필요하다. 이러한 한계점을 극복하기 위한 기법으로 송신자 자가인증 기법이 있다. 본 논문에서는 송신자 자가인증 기법을 적용하여, 스팸메일을 근본적으로 차단하고 지속적인 업데이트가 필요 없는 스팸차단 서버 및 애플리케이션을 구현하였다.

• The KIPS Transactions:PartC
• /
• v.12C no.3 s.99
• /
• pp.361-368
• /
• 2005

Internet mail has become a common communication method around the world because of tremendous Internet service usage increment. In other respect, Most Internet users' mail addresses are exposed to spammer, and the damage of Junk mail is growing bigger and bigger. These days, Junk mail delivery problem is becoming more serious, because this is used for an attack or propagation scheme of malicious code. It's a most dangerous dominant cause for computer system accident. This paper shows the Junk mail filtering model and implementation which is based on FQDN (Fully Qualified Domain Name) check and evaluates it for proposing advanced scheme against Junk mail.