idejie / recsys_web

A Flask Web for Recommendation System

A Flask Web for Recommendation System

It's just a homework for UCAS-091M5042H -网络数据挖掘.

Because the dataset is too small(the User-Iterm matrix only 0 or 1) , this rec system only run offline.

• For more details , you can visist my blog ：记又一次推荐系统。

Introduction

• DashBoard

  • Extracted the topic and tags of all news and Data reformated

  {
  	'100648598': { 
  		'title': '消失前的马航370',
  		'content': '【财新网】（实习记者葛菁）据新华社消息........',
  		'news_time': '2014年03月08日12:31',
  		'topic': '时政',
  		'tags': [
              ('马航', 3.9849225009666664),
              ('370', 3.9849225009666664), 
              ('消失', 2.074425740486667)
          ]
  	}
  }
  

  • The list of news on web

  

• User interface

  • Analysis the topics and the tags-cloud of users

  

  • Recommand news for users

  

  • Recommand similar users and provide the history of user

  

• News interface

  • Classified by topic

  

  • Provide the content /topic and details of news

  

  • Provide the similar news and Potential users

    

  • Provide the hot news and the users who read this

    

Environment

• Python3
• Some python packages , you can install them by pip install requirements
• The data for project provided in another repo: text-classification

Usage

• You can import this project using PyCharm

• or in terminal python3 app.py

TODO

Reference

• gaussic/text-classification-cnn-rnn
• jieba
• NicolasHug/Surprise
• ECharts: A Declarative Framework for Rapid Construction of Web-based Visualization Deqing Li, Honghui Mei, Yi Shen, Shuang Su, Wenli Zhang, Junting Wang, Ming Zu, Wei Chen. Visual Informatics, 2018 [PDF]