summer-sea / aush

Paper

This is the code repository for AUSH: a GAN-based RS attacking framework

Please kindly cite our paper:

@inproceedings{Lin2020Attacking, author = {Lin, Chen and Chen, Si and Li, Hui and Xiao, Yanghua and Li, Lianyun and Yang, Qian}, title = {Attacking Recommender Systems with Augmented User Profiles}, year = {2020}, booktitle = {Proceedings of the 29th ACM International Conference on Information & Knowledge Management}, pages = {855–864}, location = {Virtual Event, Ireland}, series = {CIKM '20} }

step1:Pre-processing

test_main\data_preprocess.py transforms amazon 5cores ratings to tuples [userid,itemid, normalized float rating]

step2: Initialize

test_main\data_preprocess.py

• select attack target
• select attack number (default fix 50)
• select filler size
• selected items and target users
• settings for bandwagon attack

step3. Training

• baseline attack models

python main_baseline_attack.py --dataset filmTrust --attack_methods average,segment,random,bandwagon --targets 601,623,619,64,558 --filler_num 36 --bandwagon_selected 103,98,115 --sample_filler 1

• evaluation

python main_train_rec.py --dataset filmTrust --attack_method segment --model_name NMF_25 --target_ids 601,623,619,64,558 --filler_num 36

• RS performance before attack

python main_train_rec.py --dataset filmTrust --attack_method no --model_name NMF_25 --target_ids 601,623,619,64,558 --filler_num 36

• training AUSH

python main_gan_attack.py --dataset filmTrust --target_ids 601,623,619,64,558 --filler_num 36

• Evluation (AUSH)

python main_train_rec.py --dataset filmTrust --attack_method gan --model_name NMF_25 --target_ids 601,623,619,64,558 --filler_num 36

• Comparative Study

python main_eval_attack.py --dataset filmTrust --filler_num 36 --attack_methods gan,segment,average --rec_model_names NMF_25 --target_ids 601,623,619,64,558

python main_eval_similarity.py --dataset filmTrust --filler_num 36 --targets 601,623 --bandwagon_selected 103,98,115