TranSearch
A pytorch and tensorflow GPU implementation for our paper:
Yangyang Guo, Zhiyong Cheng, Liqiang Nie, Xin-Shun Xu, Mohan Kankanhalli (2018). Multi-modal Preference Modeling for Product Search. In Proceedings of ACM MM 2018.
Please cite our MM'18 paper if you use our codes. Thanks!
Please download the Amazon Dataset from http://jmcauley.ucsd.edu/data/amazon.
The requirements are as follows:
-
python 3.5
-
pytorch 0.4.0
-
tensorflow 1.7
Example to run:
-
Make sure the raw data, stopwords data are in the same direction.
-
Preprocessing data. Noted that the category >Clothing contains both the >Men's Clothing and >Women's Clothing. Therefore, before changed from other datasets to >Clothing dataset, remember to comment line 115 and uncomment line 116.
python script/extract.py --data_dir --count --is_clothing --img_feature_file
- We leverage the PV-DM model to convert queries and product representations to the same latent space.
python script/doc2vec_img2dict.py --dataset --img_feature_file
- Move the processed folder to the main model folder.
Pre-training:
python pytorch/pre_train.py --embed_size --neg_number=20
- Train and test the framework.
python pytorch/TranSearch.py --mode='text' --embed_size