Fake Social Media Account Detection Using ML
Pramit Karki**, Anurag Karki**, Shreya Karki**
#
Scope – Vellore Institute of Technology
Prof. Dr. Suresh A.
Introduction
"Developing an advanced system for the detection of fake social media accounts".
This project focuses on creating a sophisticated system that utilizes machine learning
techniques and data analysis to identify fraudulent accounts on social media
platforms. By addressing the challenge of fake accounts, the project aims to mitigate
issues such as misinformation, cyberbullying, and privacy breaches.
"Enhancing online security, trustworthiness, and user experience". The ultimate goal
of this project is to improve the overall online environment by enhancing security,
trustworthiness, and user experience within social media platforms. By accurately
differentiating between genuine and fake profiles, the system aims to promote a safer
and more trustworthy online space, ensuring users can interact with authentic
individuals and content.
Figure 2: Architecture of Adaboost Classifier
Motivation Adaptive Boosting (AdaBoost) is a popular ensemble learning algorithm that works
Counteracting fake accounts is vital for preserving trust in online interactions. By by combining multiple weak classifiers to form a strong classifier.
developing an advanced system to detect fraudulent profiles, this project aims to Each weak classifier is trained on a subset of the data and assigns weights to
maintain users' confidence in the authenticity of their online connections. This will misclassified samples to improve the performance of the next weak classifier.
create a more trustworthy and reliable online environment. Classifiers with more accuracy is assigned with high weight to get the final output.
The detection of fake accounts plays a crucial role in defending against
misinformation and mitigating cyber threats. By using state-of-the-art machine
learning algorithms, this project classifies accounts as genuine or fake, preventing the
spread of false information and minimizing the risks of cyberattacks. Additionally, it
ensures the protection of users' personal information from exploitation, enhancing
Result and Conclusion
privacy in the online space. Figure 3: Telegram bot output
Scope of the Project
To develop an advanced system to detect fake social media accounts. This includes
utilizing machine learning techniques to identify patterns and behaviors associated
with fraudulent profiles, enhancing online security and trustworthiness.
The project will focus on selecting relevant features to capture the characteristics of
fake accounts and creating a comprehensive training dataset. This dataset will be used
to train and validate the detection system, ensuring its accuracy in differentiating
between real and fake profiles.
Methodology
The provided information mentions that the model was able to correctly classify
fake and genuine social media accounts.
It is important to note that the model's ability to correctly classify social media
accounts demonstrates its effectiveness and robustness in present context.
Figure 1: Block Diagram Figure 4: Result
Load the current social media dataset using Instagram Scrapper and Kaggle.
Split the dataset into training, validation, and testing sets to train, tune, and evaluate
the model's performance.
Preprocess and cleanup the unnecessary data, filling in the empty rows and columns,
Reference
and fixing or eliminating inaccurate, incomplete, or duplicate data from a dataset. Singh, N., Sharma, T., Thakral, A., & Choudhury, T. (2021). Detection of Fake Profile
Extraction of relevant features from the preprocessed data and train our machine in Online Social Networks Using Machine Learning. In 2021 International Conference
learning algorithms on the extracted features.
on Advances in Computing and Communication Engineering (ICACCE).
Analyze the importance of each feature in the ML model and each model is
experimented with different hyper parameters to improve the model’s performance. Frunze, A. D., & Frolov, A. A. (2021). Methods for Detecting Fake Accounts on the
Develop adaptive thresholding strategies informed by anomaly detection and real Social Network VK. In 2021 IEEEConference of Russian Young Researchers in
time monitoring. Electrical and Electronic Engineering (ElConRus).
Deploy the trained model to detect fake accounts in real-time and the performance is Kirchknopf, A., Slijepčević, D., & Zeppelzauer, M. (2021). Multimodal Detection of
evaluated on a held-out test set. Information Disorder from Social Media. In 2021 International Conference on
Create a continuous learning system that updates with fresh data and detects model Content-Based Multimedia Indexing (CBMI)
drift.
