Informatics Publishing Limited

Feyza Altunbey ¹, Bilal Alatas ¹

Affiliations
1 Firat University, Department of Software Engineering, Elazig, TR

Abstract

Parallel to growth of the Internet, social networks have become more attractive as a research topic in many different disciplines and many real systems can be denoted as a complex network. Identifying major clusters and community structures allow us to expose organizational principles in complex network such as web graphs and biological networks. It has been shown that communities are usually overlapping. Overlap is one of the characteristics of social networks, in which a person may belong to more than one social group. In recent years, overlapping community detection has attracted a lot of attention in the area of social networks applications. Many methods have been developed to solve overlapping community detection problem, using different tools and techniques. In this paper, one of the most recent social-based metaheuristic algorithm, Parliamentary Optimization Algorithm (POA), has been firstly proposed to discover overlapping communities in social networks.

Keywords

Social Networks, Overlapping Community Detection, Parliamentary Optimization Algorithm.

Full Text

   

Feyza Altunbey Ozbay ¹, Bilal Alatas ²

Affiliations
1 Department of Software Engineering, Firat University, Elazig, TR
2 Department of Software Engineering, Firat University, Elazig, IN

Abstract

Frequently studied structural property of networks is community structure which is described as a group of users. User interactions inside the group are more than those outside the group. Communities in networks may be overlapped as users belong to multiple groups at once. This paper proposes a new socially inspired metaheuristic search and optimization algorithm, Parliamentary Optimization Algorithm (POA), to acquire promising solutions to overlapping community detection problems considering multiple objectives. The salient and unique feature of this work is that for the first time POA has been designed as a multi-objective search method for overlapping community detection. There is not any work about multi-objective overlapping community detection problem in the related literature. For this reason, simulation results of the proposed algorithm have not been compared with any results of works. The experimental studies on both artificial and real world social networks indicate that the POA ensures beneficial results for defining multi-objective overlapping community structure. A novel and interesting application area of POA has been introduced with this work. Parallel and distributed versions of social based POA with optimized parameters may also be efficiently designed and used for different social network problems.

Keywords

Complex Networks, Computational Intelligence, Evolutionary Computation, Heuristic Algorithms.

Full Text

   

Sule Kaya ¹, Bilal Alatas ¹

Affiliations
1 Department of Software Engineering, Firat University, Elazig-23000, TR

Abstract

Hate speech, which is a problem that affects everyone in the world, is taking on new dimensions and becoming more violent every day. The majority of people’s interest in social media has grown in recent years, particularly in the United States. Twitter placed 5th in social media usage figures in 2022, with an average of 340 million users globally, and human control of social media has become unfeasible as a result of this expansion. As a result, certain platforms leveraging deep learning approaches have been created for machine translation, word tagging, and language understanding. Different strategies are used to develop models that divide texts into categories in this way. The goal of this research is to create an effective a new hybrid prediction model that can recognize racist, xenophobic, and sexist comments published in English on Twitter, a popular social media platform, and provide efficient and accurate findings. 7.48 percent of the data were classified as racist, genderist, and xenophobic in the used dataset. A new hybrid LSTM Neural Network and Recurrent Neural Network based model was developed in this study and compared with the most popular supervised intelligent classification models such as Logistic Regression, Support Vector Machines, Naive Bayes, Random Forest, and K-Nearest Neighbors. The results of these several models were thoroughly examined, and the LSTM Neural Network model was found to have the best performance, with an accuracy rate of 95.20 percent, a recall value of 48.94 percent, a precision of 60.95 percent, and an F1 Score of 51.32 percent. The percentage of test data was then modified, and the comparison was made by attempting to get various findings. With a larger dataset, these deep learning models are believed to produce substantially better outcomes.

Keywords

Artificial Intelligence, Deep Learning, Genderism, Racism, Xenophobia

Full Text

   

Sule Kaya ¹, Bilal Alatas ²

Affiliations
1 Department of Software Engineering, Firat University, Elazig-23000, TR
2 Department of Software Engineering, Firat University, Elazig-23000., TR

Abstract

One of the most common effects in the use of social media today is thatpeople constantly make fun of each other or certain issues or do not take them seriously. Some comments made by sarcastic people in this widespread effect are misunderstood or taken seriously by other users. Some sarcastic comments, especially in the news headlines, create false effects on the readers and create some misunderstandings for people who donot have this sense of humor. Although there are numerous studies on the problem of sarcasm detection, even low performance increment in automatic sarcasm detection is very important and popular task. In this paper, a new hybrid deep neural model is proposed for more efficient automatic detection of sarcastic context. It is aimed to detect sarcasm using a hybrid neural network model CNN+BILSTM and BERT models with bidirectional language processing in a dataset consisting of headlines of The Onion News, which made such sarcastic headlines,and professionally prepared headlines without any sarcastic comments. When the results of this study were examined, it was seen that the model that gave the best results was BERT. In addition, accuracy, precision, recall and F1 score values were checked without using Glove embeddings in the CNN+BiLSTM model, and then the results were compared by applying Glove embeddings. In this comparison, the CNN+BiLSTM model without Glove embeddings gave relatively better results.

Keywords

Sarcasm Detection, BERT, CNN+BiLSTM, Deep Learning, Hybrid Model.

Full Text

 

Sinem Akyol ¹, Muhammed Yildirim ², Bilal Alatas ¹

Affiliations
1 Department of Software Engineering, Firat University, Elazig-23000, TR
2 Department of Computer Engineering, Malatya TurgutOzal University, Malatya-44000, TR

Abstract

Metaheuristic algorithms are widely used for problems in many fields such as security, health, engineering. No metaheuristic algorithm can achieve the optimum solution for all optimization problems. For this, new metaheuristic methods are constantly being proposed and existing ones are being developed. Dandelion Optimizer, one of the most recent metaheuristic algorithms, is biology-based. Inspired by the wind-dependent long-distance flight of the ripening seed of the dandelion plant. It consists of three phases: ascending phase, descending phase and landing phase. In this study, the chaos-based version of Chaotically Initialized Dandelion Optimizer is proposed for the first time in order to prevent Dandelion Optimizer from getting stuck in local solutions and to increase its success in global search. In this way, it is aimed to increase global convergence and to prevent sticking to a local solution. While creating the initial population of the algorithm, six different Chaotically Initialized Dandelion Optimizer algorithms were presented using the Circle, Singer, Chebyshev, Gauss/Mouse, Iterative and Logistic chaotic maps. Two unimodal (Sphere and Schwefel 2.22), two multimodal (Schwefel and Rastrigin) and two fixed-dimension multimodal (Foxholes and Kowalik) quality test functions were used to compare the performances of the algorithms. When the experimental results were analyzed, it was seen that the Chaotically Initialized Dandelion Optimizer algorithms gave successful results compared to the classical Dandelion Optimizer.

Keywords

Metaheuristic Algorithms, Dandelion Optimizer, Chaos, Global Optimization.

Full Text