Skip to content

Latest commit

 

History

History
72 lines (46 loc) · 2.37 KB

README.md

File metadata and controls

72 lines (46 loc) · 2.37 KB

Fake Profile Detection on Social Media Platforms 🛡️

Detecting fake profiles using keystroke dynamics.

🎯 Project Goal

Develop a lightweight system to identify fake profiles on social media platforms using keystroke features.

🔍 Overview of the Key Parts of the Code Base

Workflow

1. Features

  • Word level features are extracted from the features/word_parser file.
  • Keystroke features (KHT & KIT) are extracted from the features/keystroke_features file.

2. Verifiers

  • Template Generator: Processes our compact CSV file containing keystroke data.
  • Verifier Library: A set of verifier algorithms implemented as class methods for easy usage. These include:
    • Absolute Verifier
    • Similarity Verifier
    • ITAD

3. Fusion

  • Score Level Fusion: Features several fusion algorithms:

    • Mean
    • Median
    • Min
    • Max

    Fusion scores are derived by iterating over the ITAD, Absolute, and Similarity matrices. These scores are then employed to determine the top_k_accuracy_score ranging from k=1 to k=5.

  • Decision Level Fusion: Fusion at the score level is determined using empirically set thresholds specific to each verifier. Profiles are labeled genuine or otherwise based on the outcome.

4. Heatmap Generation

Generates a heatmap using a matrix of the scores from a particular verifier algorithm for all user IDs. The scores are derived from keystroke features and, optionally, word-level features.

For example, this is an example heatmap with probe and enrollment IDs both for Facebook , combining KHT and KIT Flight 1 features, using all available session IDs, and the similarity verifier as the algorithm: HeatmapExample

5. Configuration File

Configure experimental conditions via the config file:

  • use_feature_selection: Opt for feature selection with true or skip with false.
  • use_word_holder: Decide to use word level features with true or avoid with false.
  • gender: Choose the gender data to incorporate:
    • all: All IDs
    • male: Male IDs only
    • other: Other IDs only

🚀 Getting Started

Prerequisites

All of the runnable scripts are in the repository root or in the plots directory To install dependencies, simply run:

pip3 install -r requirements.txt

Now you are ready to go and can run any of the runnable scripts