The term “Clickbait” refers to content that has the express intention of grabbing the reader’s attention. It has become an annoyance for social media users, because of the deception contained in the titles of Clickbait. Many studies detect Clickbait using Deep Learning and Machine Learning models. But detecting Clickbait in Arabic titles was addressed by a few studies, all of which used Machine Learning techniques, and here is where our turn came from. In our proposed work ”Wals” which is an expression of lying in Arabic, we will detect Clickbait in the Arabic titles using Deep Learning techniques which are Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNNs), Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU), and Bidirectional Gated Recurrent Unit (BiGRU). We have followed two tracks to train models to reach the Wals model to find an optimal model with the suitable optimizer and best case of different cases (with or without pre-processing and with or without Word2vec). The models were trained using an unbalanced dataset in the first track and using a balanced dataset in the second track. According to the best results obtained from the first track, we built the first Wals hybrid model consisting of LSTM+BiGRU using the Adam optimizer, and by applying Word2vec and pre-processing, the LSTM model obtained the highest Macro-F in the track equal to 0.79 using Adam and by applying both Word2vec and pre-processing, the BiGRU model also obtained the same LSTM value in the same path, but by applying pre-processing only. For the second track, we built another Wals hybrid model consisting of LSTM+CNN, the LSTM model got the highest accuracy in the track equal to 0.79 using Adam and by applying both Word2vec and pre-processing, the CNN model also got the same LSTM value with the same track but without applying either Word2vec and pre-processing. Its results did not show a clear difference between the two tracks, and for the use of Early Stop, some models showed better results than those not using it such as LSTM, BiGRU, and CNN. Also, all the two hybrid models of Wals gave close results, reaching an accuracy equal to 0.77 and with the Early Stop application as well.
12/04/2023
Due to the capability of generating data, generative modeling using deep learning methods has gained a lot of attention in the computer vision field. Medical deepfakes can cause serious resource drains in hospitals or even result in fatalities if they are not recognized. Injecting and removing tumors from medical scans is one method of deepfake generation used by the medical industry to generate data. This study intends to solve the problem of recognizing cancer manipulation and other relevant disease samples. In this study, multiple deep-learning models are trained to improve the classification performance of tampered lung cancer. We trained six CNN models, namely ResNet101, ResNet50, DensNet121, DenseNet201, MobileNetV2, and MobileNet. The models were trained using 2000 samples over three classes (Untampered, False-Benign, False-Malicious). We have enhanced the ResNet101 by adding more layers, achieving a training accuracy of 99%. The obtained results solved the problems that occurred in the first experiment considering Overfitting and Imbalanced dataset.
22/03/2023
In recent years, cybercrimes have been increasing tremendously and affecting victims around the world.Cybercrime analysis and detection help protect and prevent targeted unwary users and turning them into victims. Scams (money loss, identity theft, and malware installation) re- sult in losses of billions of dollars each year. It is crucial to quickly identify and respond to such dangers. Blacklists are used most often to detect criminals in general. However, blacklists are not thorough and do not include the ability to recognize freshly created malicious Websites.The generality of malicious URL detectors has to be improved, It has become necessary to build a reliable system to detect malicious websites by analyzing related data. Although, there are sev- eral cybercrime detection techniques such as Statistical Methods and Machine Learning. Thus, our project sets out to build a machine learning model and develop a website to detect malicious URLs. The used dataset had 651,191 URLs containing safe URLs, defacement, phishing, and malware URLs. We built a model using the Random Forest algorithm and were trained with our dataset. The model had 96.85% Accuracy. Therefore, we integrated the model with the website we designed which had a simple UI/UX.
05/03/2023
Nowadays, Twitter is identified as one of the largest social networking sites and has become a huge part of many people’s lives. Many activities, such as communication, promotion, adver- tisement, news, and agenda creation, have begun to be carried out through it. Twitter offers a primary feature called trending hashtags, where a name, phrase, or topic that is preceded with a hash sign (#) is mentioned at a greater rate than others. Twitter’s trending hashtags attract much attention; thus, they can affect the public agenda. With over 230 million users on Twitter, it comes as no surprise that this valuable feature has been abused by malicious campaigns. In the wrong hands, Twitter’s trends can be used to mislead people and disseminate fake news. One way to manipulate trending hashtags is through hashtag hijacking, where a trendy hashtag on social media is taken over to promote a different message than the originally intended one. Hashtag hijacking is normally done by fake accounts, especially by spammers or trolls. This type of hashtag manipulation can have several negative impacts on the original message or the reason for being promoted. It can confuse users, blur the original message, and lead to online harassment and trolling, in addition to damaging reputations. In this project, eleven pioneer- ing machine learning algorithms are used separately to build a detection system for trending hashtags, and their results are compared to find the best-performing one. Experimental results revealed that the best-performing model is SVM trained with TF-IDF using oversampling which achieved an accuracy of 0.980. However, the VotingClassifier model trained with TF-IDF has a slightly higher accuracy of 0.981, but it incurs an overhead of 23 minutes to run compared to SVM which only tooks seconds.
04/03/2023
Steganography is one of the important methods that help in the process of hiding infor- mation in various digital objects. It is a science that involves the delivery of confidential information embedded in digital envelope objects suitable for multimedia such as images, audio and video files.This science aims to hide confidential data embedded within digital envelope objects. Steganography technology Through its development of computational power, it has improved and increased the security of existing data steganography techniques.Steganography aims at the inability to reveal hidden data contained within digital envelopes. The purpose of steganography is to maintain secret communication between two parties. This project provides an overview of image steganography, its applications,
04/03/2023
Recommendation systems, or also known as recommendation engines, have become an im- portant research area and are applied in various fields. In addition, the technologies behind these systems have been developed over time. In general, these systems help users to find products or services (such as books or music) through analyzing and aggregating other users’ activities, primarily in form of reviews, and then forming recommendations. Recommendations help facilitate the user’s decision-making process . In this project, we will develop a recommendation system of a pharmacy website to provide recommendations for customers to facilitate choosing items. Recommendation system predicts whether a user prefers a product or not, and gives new options based on the user’s profile and behavior in the past.
04/03/2023
The majority of conventional procedures for diagnosing plant diseases rely on human visual observation and inspection. This method, however, is time-consuming and involves considerable human work and specialist knowledge. Recent advancements in computer vision and deep learning offer a potential avenue for the development of a plant disease diagnosis system that enables quick disease identification across broad spatial areas with minimal human participation. In this paper, we developed a deep learning strategy for plant leaf disease classification problems and performed a variety of experiments to evaluate the performance of ResNet50, CNN, AlexNet, and DesNet169 state-of-the-art neural network architectures. The proposed models were trained using the colab platform with the PlantVillage dataset consisting of 54,305 photos across 38 plant disease classes. We assessed each architecture using four distinct performance metrics: accuracy, precision, recall, and F1-score. The DesNet169 neural network architecture surpassed all other Convolutional Neural Network architectures, producing an accuracy of 99.10% after 70 epochs of training.
03/03/2023
The bill is a time-stamped commercial document and a primary channel of communication between companies and their consumers. Managing time and expenses is crucial for our swiftly growing quotidian life. Hence, it is essential to have an efficient system for such purposes via an electronic platform. In our project, we aim to design an assistance billing system for public utilities (e.g. Electricity, Water and telecommunications). Our proposed system is a computer- ized Web- based approach to bill organizing with ease of use, simplicity of content, and clarity of classification, and for the security standards we encrypted the user password in DB by using MD5. In addition, the system collects all pre-set payment and billing notifications of an indi- vidual, and creates a reminder every month. The system is developed with Microsoft Visual Studio using PHP, web languages and (SQL) for creating database
03/03/2023
Tourism plays a critical role in the economy of Saudi Arabia hence it becomes extremely important that the correct information is delivered to the tourists. This research proposes a tourist guide system using the Internet of Things IoT, that provides explanations in different languages for tourists and visitors to tourist areas and museums through an application that works on smartphones. This system is linked to a set of RFID sensors for the current location of the user through which the system recognizes the current area that the explanation and talk should revolve around.
03/03/2023
Children’s use of technology and electronic devices is increasing rapidly, so it has become an important and necessary thing in some children’s lives in many ways, for fun or for education. Many parents are using electronic devices to teach their children some basics, such as letters or numbers, especially autistic children. It is no secret to us that the majority of autistic children have difficulty speaking or communicating and the ability to clarify their thoughts and feelings. From this perspective comes the Taalm application, an educational application for children with ASD that relies on several techniques such as PECS, which will help the child to communicate his/her ideas through pictures, in addition to many features that help the autistic child. Taalm app also will help the parents through the Awareness section to have a better understanding of ASD and it also allows them to take a diagnosis test for early ASD symptoms. It is a hope that with this app autistic children will be able to communicate, learn and express their feelings. Thus, this research discusses self-based learning applications and all the aspects that help in building this application and assist autistic children to improve their education and communication skills.
03/03/2023
The key to a student’s success is maintaining parent-teacher contact throughout the aca- demic year. Studies have indicated that when parents or other adult guardians are active, children perform better academically. Utilizing technology is one technique to inform parents about their children’s education and motivate them to get involved. We are creating a ”Follow Up” application, this project aims to make communication between parents and school easier. The application, which is characterized by simple learning via use, tries to eliminate complexity in existing apps. The application offers parents a simple means of contact with the school by alerting them of all topics pertaining to their child, including homework assignments, evalua- tions, behavioral levels, etc. Parents may schedule a meeting with their child’s teacher using the app, as well as submit the school concerns and ideas. The application interfaces is developed using cross-platforms (IOS, Android, windows) and supporting Arabic, which make it accessible for targeted users: students, parents and teachers.
03/03/2023
The emergence of social commerce has grown electronic commerce to both businesses and consumers. Social commerce is a merger between electronic commerce and web 2.0 technolo- gies. The number of publications on social commerce has grown exponentially in the past 10 years. Now, social commerce has become a significant emerging research area. We discussed in this project electronic commerce, social commerce, and social networks. We discussed social commerce types and benefits and the differences between social commerce and electronic com- merce. There are two types of social commerce, the first one is the social networks that provide E-commerce and the second one is the E-commerce sites that provide social network capabil- ities. We chose the second type of E-commerce site that provides social network capabilities. We built a platform to help users find carpentry easily, and choose the specifications required in the design. We used social network features on an e-commerce site for carpentry owners. We focused on three features. First is content sharing. Secondly, customers and carpenters will be able to have conversations. Thirdly, customers can follow their favorite stores on the platform. The price is agreed upon between the customer and the carpenter through chat. We made it easier for the user to implement their own design. By having such a system, the opportunity for carpentry companies to enter the world of e-commerce is supported by the characteristics of the social network. This may stand in the direction of Vision 2030 for the Kingdom of Saudi Arabia.
23/02/2023