Spam Filtering Research Articles

Boyer Moore string-match framework for a hybrid short message service spam filtering technique

Advances in technology and the proliferation of mobile device have continued to advance the ubiquitous nature of computing alongside their many prowess and improved features it brings as a disruptive technology to aid information sharing amongst many online users. This popularity, usage and adoption ease, mobility, and portability of the mobile smartphone devices have allowed for its acceptability and popularity. Mobile smartphones continue to adopt the use of short messages services accompanied with a scenario for spamming to thrive. Spams are unsolicited message or inappropriate contents. An effective spam filter studies are limited as short-text message service (SMS) are 140bytes, 160-characters, and rippled with abbreviation and slangs that further inhibits the effective training of models. The study proposes a string match algorithm used as deep learning ensemble on a hybrid spam filtering technique to normalize noisy features, expand text and use semantic dictionaries of disambiguation to train underlying learning heuristics and effectively classify SMS into legitimate and spam classes. Study uses a profile hidden Markov network to select and train the network structure and employs the deep neural network as a classifier network structure. Model achieves an accuracy of 97% with an error rate of 1.2%.

Bayesian Spam Detection Framework on Mobile Device

Background: The spam message aimed at mobile devices has become a prevalent activity in recent years. About 76% of mobile users received spam messages of Ads, fraud messages and phishing links. It has led to a significant financial loss of 91.5 billion Yuan per year and this problem will be more severe in the coming future. Although the operators are making much effort against spam message, the number of spams are increasing day by day. Aim: Many schemes designed to detect and filter spam have been proposed in the research area. The Bayesian classification algorithm is one of the most popular schemes. Most of the spam detection schemes based on Bayesian are designed for communication providers. The scalability of filter policy is hard to control because the strict policy might filter the normal messages while some spam messages would not be detected due to the flexibility in the policy. Methods: The current mobile device has a robust computational capability so it can execute complex jobs. A spam detection framework is designed in the present study. Results: It can sniffer the coming messages in mobile device by hooking the Android SMS API, and send them to the filter module, -which is responsible to filter messages into normal or spam based on Bayesian classification. This is a light-weight framework of consuming low power, which is suitable for mobile devices. Extra experiments are conducted to prove their accuracy and efficiency. Conclusion: The results showed that it could filter the spam among receiving the messages in realtime and become more accurate by learning the users' feedback.

A comprehensive survey of imbalanced learning methods for bankruptcy prediction

AbstractIn practical datasets used for supervised learning, the uneven distribution of the amounts of data between classes is known as the class imbalance problem, and can reduce the performance of basic classifiers. The class imbalance problem arises in various areas such as medical diagnosis, spam filtering, and fraud detection. Bankruptcy prediction is of particular research interest due to the current economic upheaval. In a bankruptcy prediction dataset, there are two classes containing bankrupt and normal companies, and this problem can therefore be solved with binary classification methods. Several advanced models for handling the class imbalance problem in bankruptcy prediction have been developed to improve the predictability performance on a Korean bankruptcy (KB) dataset. To give an overview of imbalanced learning methods for bankruptcy prediction, this study first reviews several state‐of‐the‐art approaches for handling this problem in bankruptcy prediction, including an oversampling‐based framework, a cost‐sensitive method (the CBoost algorithm), a combination of resampling techniques and a cost‐sensitive framework, and an ensemble‐based model (the XGBS algorithm). We also conduct empirical experiments to evaluate the methods surveyed here in terms of two performance metrics, the area under the receiver operating characteristic (ROC) curve and the geometric mean. The results show that the ensemble‐based model outperforms other methods in terms of bankruptcy prediction on the KB dataset.

A comparative analysis on question classification task based on deep learning approaches.

Question classification is one of the essential tasks for automatic question answering implementation in natural language processing (NLP). Recently, there have been several text-mining issues such as text classification, document categorization, web mining, sentiment analysis, and spam filtering that have been successfully achieved by deep learning approaches. In this study, we illustrated and investigated our work on certain deep learning approaches for question classification tasks in an extremely inflected Turkish language. In this study, we trained and tested the deep learning architectures on the questions dataset in Turkish. In addition to this, we used three main deep learning approaches (Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), Convolutional Neural Networks (CNN)) and we also applied two different deep learning combinations of CNN-GRU and CNN-LSTM architectures. Furthermore, we applied the Word2vec technique with both skip-gram and CBOW methods for word embedding with various vector sizes on a large corpus composed of user questions. By comparing analysis, we conducted an experiment on deep learning architectures based on test and 10-cross fold validation accuracy. Experiment results were obtained to illustrate the effectiveness of various Word2vec techniques that have a considerable impact on the accuracy rate using different deep learning approaches. We attained an accuracy of 93.7% by using these techniques on the question dataset.

Detecting spam e-mails using stop word TF-IDF and stemming algorithm with Naïve Bayes classifier on the multicore GPU

<span>A spam filter is a program which is used to identify unwanted emails and prevents those messages from getting into a user's mail. The study was focused on how the algorithms can be applied on a number of e-mails consisting of both ham and spam e-mails. First, the working principle and steps which are followed for implementation of stop words, TF-IDF and stemming algorithm on NVIDIA’s Tesla P100 GPU are discussed and to verify the findings by executing of Naïve Bayes algorithm. After complete training and testing of the spam e-mails dataset taken from Kaggle by using the proposed method, we got a high training accuracy of 99.67% and got a testing accuracy of about 99.03% on the multicore GPU that boosted the speed of execution of training time period and testing time period which is improved of training and testing accuracy around 0.22% and 0.18% respectively when compared to that after applying only Naïve Bayes i.e. conventional method to the same dataset where we found training and testing accuracy to be 99.45% and 98.85% respectively. Also, we found that training time taken on GPU is 1.361 seconds which was about 1.49X faster than that taken on CPU which is 2.029 seconds. And the testing time taken on GPU is 1.978 seconds which was about 1.15X faster than that taken on CPU which is 2.280 seconds.</span>

Junk Filtering through Naive Bayesian Algorithm

As the world is seamlessly developing at a very high pace, we have been seeing enormous growth in various sectors of Technology. Networking has played a crucial part in the exchange of technological culture around the globe, and the Internet being the sole medium of Network enhancement has taken over every aspect of our society. Today, most of the professional communications are done through emailing. As far as email has proven to be an efficient, professional and easy way of communication, it also comes with the disadvantage of unwanted bulk bombarding of spam content. This has been a critical concern for email users. Consequently, it has become very difficult for spam filters to efficiently filter the unwanted emails, since nowadays emails are written in such a manner that any existing algorithm cannot give 100% accuracy in predicting spam. This paper deals with Naive Bayesian Classifier that is a Machine Learning algorithm for antispam filtering, which gives satisfactory results by automatically constructing anti-spam filters with extended conduct. The review over the researched performance of Naive Bayes algorithm is done by the investigations of Spam ham csv datasets. The performance of the algorithm is evaluated based on the accuracy, recall and precision it shows on the mentioned datasets. This technique gives 96-97% accuracy and 89% precision on the investigated dataset. The result also highlights that the content of the email and the number of instances of the dataset has an apparent effect on the performance of the algorithm.

Survey on Spam Review Detection Using Spam Filtering Algorithms

Generally the people trust on product on the basis of that product reviews and rating. Reviews can affect an organization or profile of a brand. The corporation has to assess market reactions towards its goods. However, it is not straightforward to track and organize popular reviews. Many public views are hard to manually process in social media. A methodology is then required to categories positive or negative public assessments automatically. Online feedback will provide customers with an insight into the consistency, efficiency and advice of the product; this provides prospective buyers with a better understanding of the product. One such unrealized opportunity is the usability of web assessments from suppliers in order to fulfill client requirements by evaluating beneficial feedback. Good and negative reviews play a major role in assessing customer needs and in quicker collection of product input from consumers. Sentiment Analysis is a computer study that extracts contextual data from the text. In this study a vast number of online mobile telephone ratings are analyzed. We classify the text as positive and negative, but we also included feelings of frustration, expectation, disgust, apprehension, happiness, regret, surprise and confidence for spam review detection. This delimited grouping of feedback helps to holistically assess the product, allowing buyers to decide better.

Marginal Attacks of Generating Adversarial Examples for Spam Filtering

Digit information has been used in many areas and has been widely spread in the Internet era because of its convenience. However, many ill-disposed attackers, such as spammers take advantage of such convenience to send unsolicited information, such as advertisements, frauds, and pornographic messages to mislead users and this might cause severe consequences. Although many spam filters have been proposed in detecting spams, they are vulnerable and could be misled by some carefully crafted adversarial examples. In this paper, we propose the marginal attack methods of generating such adversarial examples to fool a naive Bayesian spam filter. Specifically, we propose three methods to select sensitive words from a sentence and add them at the end of the sentence. Through extensive experiments, we show that the generated adversarial examples could largely reduce the filter’s detecting accuracy, e.g. by adding only one word, the accuracy could be reduced from 93.6% to 55.8%. Furthermore, we evaluate the transferability of the generated adversarial examples against other traditional filters such as logic regression, decision tree and linear support vector machine based filters. The evaluation results show that these filters’ accuracy is also reduced dramatically; especially, the decision tree based filter’s accuracy drops from 100% to 1.51% by inserting only one word.

Spam E-Mail Filtering: A Review of Techniques

Every year, the number of uninvited email received by the common email user will increase dramatically. According to IDC, Spam has accounted for 38 percent of the 31 billion emails sent each day in North America in 2004, up from 24 percent in 2002. keeping pace with amount of spam is that the quantity of filtering solutions out there to assist eliminate it. This paper describes in detail how several of the most common spam filtering technologies work, how effective they are at stopping spam, their strengths and weaknesses, and techniques used by spammers to circumvent them.

PROTECTOR: An optimized deep learning-based framework for image spam detection and prevention

Image spam is a spamming technique that integrates spam text content into graphical images in order to bypass conventional text-based spam filters. In order to detect image spam efficiently, it is important to analyze the image data. The existing image spam detection techniques in literature focus on textual or graphic features of the image. None of the existing techniques considered the link information of the image which results in low accuracy and performance degradation. So, to fill these gaps, in this paper, we analyzed the link properties of image, for image spam detection and prevention. We propose an optimized framework called as PROTECTOR. In PROTECTOR, the rank score is generated by using the linkage information of the image, textual information of the image, and metadata information of the image. The computed rank score indicates the relevance of an image. This rank score is then used to train a deep neural network design of deep learning, which yields the accuracy of 96% with respect to various performance evaluation metrics. Also, the optimization algorithm, i.e., genetic algorithm is fitted in the results according to the defined fitting function. The proposed framework is validated with standard dataset of Image spam Hunter.

Prediction of Spam Email using Machine Learning Classification Algorithm

Unsolicited e-mail also known as Spam has become a huge concern for each e-mail user. In recent times, it is very difficult to filter spam emails as these emails are produced or created or written in a very special manner so that anti-spam filters cannot detect such emails. This paper compares and reviews performance metrics of certain categories of supervised machine learning techniques such as SVM (Support Vector Machine), Random Forest, Decision Tree, CNN, (Convolutional Neural Network), KNN(K Nearest Neighbor), MLP(Multi-Layer Perceptron), Adaboost (Adaptive Boosting) Naïve Bayes algorithm to predict or classify into spam emails. The objective of this study is to consider the details or content of the emails, learn a finite dataset available and to develop a classification model that will be able to predict or classify whether an e-mail is spam or not.

A hybrid classification method for Twitter spam detection based on differential evolution and random forest

SummarySocial networking services are online platforms that are distributed across different computers over long distances. Twitter is the most popular microblogging site that allows users to share their opinions and real‐world events. Due to its popularity and ease of use, Twitter has also attracted spammers. As a result, spam detection is one of the most critical problems. In order to provide a spam‐free environment, it is necessary to identify and filter spam tweets as well as their owners. A hybrid method, which is based on Synthetic Minority Over‐sampling TEchnique (SMOTE) and Differential Evolution (DE) strategies, is presented to enhance the spam detection rate in real Twitter datasets. SMOTE is applied to tackle the imbalanced class distribution of datasets, while DE is used to tune Random Forest (RF) hyperparameters. Compared with related work and based on evaluation results, the presented method significantly enhances the classification performance in imbalanced datasets. The detection rate of optimized RF with excellent F1‐score and Area Under the Receiver Operating Characteristic Curve (AUROC), which are 98.97% and 0.999, respectively, demonstrates the high efficiency of the proposed method.

Eshort Message Service Spam Detection and Filtering Using Machine Learning Approach

A robust and efficient spam filtering and detection system is important in today’s ever changing and growing world. While basic filters and detector are there most of them are not efficient. There is a scope of further development in the systems used. Hence, we are doing our work in way to provide and design a system more efficient and user friendly according to user preferences. Most of the work that has been carried out in this field has been done on data-sets sourced from European countries and mostly on European language. We would be designing a SMS spam filter, which will be universal in nature.As well as our model would be smart enough to distinguish whether a SMS from an operator will be useful to the user or not and rank them accordingly.

A Machine Learning Approach to Prevent Malicious Calls over Telephony Networks

The paper is to present about identification of electronic junk mail called spam by LSTM layers in recurrent neural network. These methods are applied for detecting and filtrating of those junk messages in successful manner. This is to give high accuracy frequency as because it is used to simplify the text into word format. The most important factor in purchasing a product that customer looks for reviews in online, even producers and dealers are monitored on those reviews from customers because of the created collection. Development of a technique is important to detect and also to filter spam as for the profit spammers usually manipulate the reviews. In this, Artificial Intelligence model is proposed effectively to continuous recognition of those mail shot.

Memristor-based in-memory processor for high precision semantic text classification

Text classification is an important component of digital media such as natural language processing, image labeling, sentiment analysis, spam filtering, chatbots, and translators. In this work, effort was devoted to develop an in-memory processor for Bayesian text classification using memristive crossbar architecture, in which memristive switches were employed to store information required for the classification of text. The efficacy of the proposed circuit was tested on two distinct datasets consisting of a total of 55,575 texts. The circuit was found to be efficient to categorize the texts with an average accuracy of 91%. This work paves the way for hardware realization of cognitive systems using in-memory processors.

A New Bio-Inspired Method for Spam Image-Based Emails Filtering

Electronic mail has become one of the most popular and frequently used channels for personal and professional online communication. Despite its benefits, e-mail faces a major security problem, which is the daily reception of a large number of unsolicited electronic messages, known as “spam emails.” Today, most electronic mail systems have simple spam filtering mechanisms based on text spam filtering technologies. To circumvent these filters, spammers are introducing new techniques of embedding spam messages in the image attached to the mail. In this article, the authors propose a new method for spam image filtering. The proposed system can distinguish between legitimate images from spam images based on the texture characteristics of the image attached to an email. From each image, around 20 characteristics can be extracted from the gray level co-occurrence matrix (GLCM). Then, to classify the images as spam or ham, the authors use a new metaheuristic nature-inspired model for building classifiers based on the social worker bees and enhanced nearest-centroid classification method.

Generating Actionable Interpretations from Ensembles of Decision Trees

Machine-learned models are often perceived as “black boxes”: they are given inputs and hopefully produce desired outputs. There are many circumstances, however, where human-interpretability is crucial to understand (i) why a model outputs a certain prediction on a given instance, (ii) which adjustable features of that instance should be modified, and finally (iii) how to alter a prediction when the mutated instance is input back to the model. In this paper, we present a technique that exploits the feedback loop originated from the internals of any ensemble of decision trees to offer recommendations for transforming a k-labelled predicted instance into a k'-labelled one (for any possible pair of class labels k, k'). Our proposed algorithm perturbs individual feature values of an instance, so as to change the original prediction output by the ensemble on the so-transformed instance. This is also achieved under two constraints: the cost- and tolerance of transformation. Finally, we evaluate our approach on four distinct application domains: online advertising, healthcare, spam filtering, and handwritten digit recognition. Experiments confirm that our solution is able to suggest changes to feature values that help interpreting the rationale of model predictions, making it indeed useful in practice especially if implemented efficiently.

Hybrid categorical expert system for use in content aggregation

The subject of this research is the development of the architecture of an expert system for distributed content aggregation system, the main purpose of which is the categorization of aggregated data. The author examines the advantages and disadvantages of expert systems, a toolset for the development of expert systems, classification of expert systems, as well as application of expert systems for categorization of data. Special attention is given to the description of the architecture of the proposed expert system, which consists of a spam filter, a component for determination of the main category for each type of the processed content, and components for the determination of subcategories, one of which is based on the domain rules, and the other uses the methods of machine learning methods and complements the first one. The conclusion is made that an expert system can be effectively applied for the solution of the problems of categorization of data in the content aggregation systems. The author establishes that hybrid solutions, which combine an approach based on the use of knowledge base and rules with the implementation of neural networks allow reducing the cost of the expert system. The novelty of this research lies in the proposed architecture of the system, which is easily extensible and adaptable to workloads by scaling existing modules or adding new ones.

Weighted k-Nearest Neighbour for Image Spam Classification

E-mail is an efficient and reliable data exchange service. Spams are undesired e-mail messages which are randomly sent in bulk usually for commercial aims. Obfuscated image spamming is one of the new tricks to bypass text-based and Optical Character Recognition (OCR)-based spam filters. Image spam detection based on image visual features has the advantage of efficiency in terms of reducing the computational cost and improving the performance. In this paper, an image spam detection schema is presented. Suitable image processing techniques were used to capture the image features that can differentiate spam images from non-spam ones. Weighted k-nearest neighbor, which is a simple, yet powerful, machine learning algorithm, was used as a classifier. The results confirm the effectiveness of the proposed schema as it is evaluated over two datasets. The first dataset is a real and benchmark dataset while the other is a real-like, modern, and more challenging dataset collected from social media and many public available image spam datasets. The obtained accuracy was 99.36% and 91% on benchmark and the proposed dataset, respectively.

2 Years in the anti-phishing group of a large company

The email threat landscape is constantly evolving and hence difficult to counteract even by carrier-grade spam filters. Dangerous spam emails may thus reach the users and then result in damaging attacks spreading through the corporate network. This paper describes a collaborative approach for early detection of malicious spam emails and its application in the context of large companies. By the joint effort of the employees and the security analysts during the last two years, a large dataset of potentially malicious spam emails has been collected with each email being labeled as critical or irrelevant spam. By analyzing the main distinguishing characteristics of dangerous emails, a set of both traditional and novel features was identified and then tested and optimized by applying common supervised machine learning classifiers. The obtained massive experimental results show that Support Vector Machine and Random Forest classifiers achieve the best performance, with the optimized feature set of only 36 features achieving 91.6% Recall and 95.2% Precision. These results, confirmed by a large empirical experiment conducted on 40,000+ company employees, led to the re-engineering of the email threat management process to ensure a high level of security in the company, as well as an increased security awareness of all company employees.