Generating Behavior Features for Cold-Start Spam Review Detection（2019DASFAA）（2020Information Sciences） http://link.springer.com/chapter/10.1007/978-3-030-18590-9_38（DASFAA） https://www.sciencedirect.com/science/article/pii/S0020025520302437?dgcid=author（Information Sciences）
The implementation of the model in the paper.
1）review_shuffle_w2v_c1w8-i20h0n5s100.txt is a pre training model for text information trained by word2vec.
- trainEmb are indexs of all examples while train、test are indexs of examples with labels.
3）bf_embedding are six RBFs of users'.
4）rd_embedding/time_embbding are EAFs of users'.
5）textConv are text features trained by textcnn.
3/4/5 only give an example.
In pinjie, there is the final result of spliced RBFs.
If the codes help you, please cite our paper:
[1]Xiaoya Tang,Tieyun Qian,Zhenni You. Generating Behavior Features for Cold-Start Spam Review Detection[J]. 2019DASFAA.
[1]Xiaoya Tang,Tieyun Qian,Zhenni You. Generating Behavior Features for Cold-Start Spam Review Detection with Adversarial Learning[J]. Information Sciences,2020.