mkr-recommendation's Introduction

mkr-recommendation

MKR is a Multi-task learning approach for Knowledge graph enhanced Recommendation. MKR consists of two parts: the recommender system (RS) module and the knowledge graph embedding (KGE) module. The two modules are bridged by cross&compress units, which can automatically learn high-order interactions of item and entity features and transfer knowledge between the two tasks.

Usage

For movie:

      python preprocess.py -d movie
      python main.py -dataset movie
      python predict_test.py -d movie  # Testing the .pd model

For book:

      python preprocess.py -d book
      python main.py -dataset book
      python predict_test.py -d book  # Testing the .pd model

For music:

      python preprocess.py -d music
      python main.py -dataset music
      python predict_test.py -d music  # Testing the .pd model

File structure

model/
- movie/, book/, music/
  - restore/: model save recovery save/restore method, use it to restore model weights
  - result/: save the .pd model, deploy model using tensorflow serving
  - vocab/: save the embedding, in order to transfer weight, use it for iterative training if new users or new movie/music/book join
data/
- book/
  - BX-Book-Ratings.csv: raw rating file of Book-Crossing dataset
  - item_index2entity_id.txt: the mapping from item indices in the raw rating file to entity IDs in the KG
  - kg.txt: knowledge graph file
- movie/
  - item_index2entity_id.txt: the mapping from item indices in the raw rating file to entity IDs in the KG
  - kg.txt: knowledge graph file
  - ratrings.dat: raw rating file of MovieLens-1M
- music/
  - item_index2entity_id.txt: the mapping from item indices in the raw rating file to entity IDs in the KG
  - kg.txt: knowledge graph file
  - user_artists.dat: raw rating file of Last.FM