No abstract

The social network has become an important part of our lives today which is prone to online impersonation as well as fake accounts. These online profiles are used by intruders to cause harm in online social networks (OSNs) in order to perform malicious activities in form of privacy intrusion like theft identity. So it becomes necessary in OSN, to identify the account whether it is real or fake. In this article we have proposed several algorithms for classification like deep neural network and SVM (Support Vector Machine) algorithm. It also studies comparisons of these methods for classification based on the spam account dataset that is used to select the best.

No abstract

An information retrieval system stores and indexes documents such that when users submit a query, the system gets relevant documents and assigns a score to each one. The higher the score, the more important the document is. IR systems typically yield vast result sets, and users must spend a significant amount of time sifting through them to identify the elements that are genuinely important. Different suggestions for applying evolutionary computing to the topic of information retrieval will be reviewed from the specialist literature. To do so, researchers looked at a variety of IR issues that were addressed using evolutionary algorithms. Some of the current ways will be detailed in detail; for example, when dealing with specialized domain knowledge, this challenge can be solved by embedding a knowledge base into existing information retrieval systems that illustrates the relationships between index words. The fuzzy set theory may be used to change the knowledge in the bases to cope with the ambiguity that is typical of human knowledge. In this work, a novel way for implementing a similarity measure utilizing fuzzy logic for IR is provided. A suggested similarity metric is based on many IR system attributes that boost IR system performance. This method's strength is that it can extract the majority of a document's characteristics. Fuzzy rules, which translated domain knowledge into fuzzy sets, were also designed to make this most effective. Our suggested similarity metric is validated using the CACM and CRAN benchmark datasets.