https://github.com/danyang-liu/KRED

This is the source code for paper: KRED: Knowledge-Aware Document Representation for NewsRecommendations
https://github.com/danyang-liu/KRED

Last synced: 3 months ago
JSON representation

This is the source code for paper: KRED: Knowledge-Aware Document Representation for NewsRecommendations

• Host: GitHub
• URL: https://github.com/danyang-liu/KRED
• Owner: danyang-liu
• Created: 2020-08-01T10:26:48.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2021-07-12T09:28:37.000Z (over 3 years ago)
• Last Synced: 2024-06-24T05:53:53.688Z (5 months ago)
• Language: Python
• Size: 64.3 MB
• Stars: 44
• Watchers: 1
• Forks: 11
• Open Issues: 7
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• StarryDivineSky - danyang-liu/KRED

README

        
## Introduction

This repository is the implementation of [KRED: Knowledge-Aware Document Representation for News Recommendations](https://arxiv.org/abs/1910.11494)

## Model description

![](./framework.PNG)

KRED is a knowledge enhanced framework which enhance a document embedding with knowledge information for multiple news recommendation tasks. The framework mainly contains two part: representation enhancement part(left) and multi-task training part(right).

##  Data description:

We use [MIND dataset](https://msnews.github.io) in this repo, due to MIND dataset doesn't contain location information, so in this experiments we do not use the local news detection task.

##  Running the code

```

$ python main.py  (this will run KRED for user2item single task training set by default parameters)

```

we also offer a quick example in this notebook: kred_example.ipynb

## Environment

The code has been tested running under Python 3.6.10, with the following packages installed (along with their dependencies):

- numpy==1.19.1

- pytorchtools==0.0.2

- scikit-learn==0.23.2

- scipy==1.5.2

- torch==1.4.0