/ISMA

Primary LanguagePython

Attacking Social Media via Influential Interaction Poisoning

This project is for the paper "Attacking Social Media via Influential Interaction Poisoning".

The code was developed on Python 3.6 and tensorflow 1.15.0.

Usage

prepare Weibo data

mkdir data/weibo/fast

Download data and put them in data/weibo/fast.

run main.py

Generate poisoning data of IPRO.

python main.py --data weibo --model mlp1 --data_size 0.4 --poison 0 --train 1 --method inf --cal_inf --cal_cand

Evaluate IPRO

python main.py --data weibo --model mlp --data_size 0.4 --poison 1 --train 1 --method inf