Skip to content

Code and data for the ACM CIKM 2022 paper "Implicit Session Contexts for Next-Item Recommendations"

License

Notifications You must be signed in to change notification settings

claws-lab/iscon

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Implicit Session Contexts for Next-Item Recommendations (ACM CIKM 2022 paper)

Overview

Implicit Session Contexts for Next-Item Recommendations
Sejoon Oh, Ankur Bhardwaj, Jongseok Han, Sungchul Kim, Ryan A. Rossi, and Srijan Kumar
ACM International Conference on Information and Knowledge Management (CIKM) short paper, 2022

Link to the paper PDF

ISCON is a novel session-based recommendation framework that contextualize sessions and utilizes the contextual embeddings for the next-item prediction.
Datasets used for in the paper are available at the following link.
https://drive.google.com/file/d/1Q8bCw3PqoPayP9RiKO7TaRHhKAotevkA/view?usp=sharing

Usage

The detailed execution procedure of ISCON is given as follows.

  1. Install all required libraries by "pip install -r requirements.txt" (Python 3.6 or higher version is required).
  2. "python src/main.py [arguments]" will execute ISCON with arguments, and specific information of the arguments are as follows.
--data_name: name of the input data (e.g., reddit). The corresponding data should be in the data directory with the file name "[data_name].tsv".
--epochs: number of epochs for training (default: 200)
--gpu: GPU number will be used for experiments (default: 0)
--output: name of the output log file (e.g., reddit_output)
--contexts: number of session contexts (default: 40)
--batch_size: mini-batch size for training (default: 1024)
--train_ratio: training/test split ratio (default: 0.9)
--session_emb_dim and --emb_dim: session embedding dimension (default: 128) and user&item embedding dimension (default: 256)
--context_dim: contextual embedding dimension (default: 32)
--topk: number of predicted contexts per session (default: 3)

Demo

To run the demo, please follow the following procedure. ISCON demo will be executed on the Reddit dataset.

1. Check permissions of files (if not, use the command "chmod 777 *")
2. Execute "./demo.sh"
3. Check "output/reddit_demo" for the demo result of ISCON on the Reddit dataset

Tested Environment

We tested our proposed method ISCON in NVIDIA DGXStation machines equipped with 8 NVIDIA Tesla V100 GPUs.

About

Code and data for the ACM CIKM 2022 paper "Implicit Session Contexts for Next-Item Recommendations"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published