News4U
所属分类:金融证券系统
开发工具:Python
文件大小:5462KB
下载次数:0
上传日期:2023-01-25 23:17:20
上 传 者:
sh-1993
说明: 一个RT新闻推荐系统,使用新闻管道从CNN等各种资源中获取最新新闻...
(An RT news recommendation system that uses a news pipeline to scrape latest news from various of resources such CNN, BBC and Bloomberg etc.)
文件列表:
LICENSE (1210, 2021-08-02)
architecture.jpg (55370, 2021-08-02)
backend_server (0, 2021-08-02)
backend_server\demo_news.json (6787229, 2021-08-02)
backend_server\launcher.sh (462, 2021-08-02)
backend_server\operations.py (2675, 2021-08-02)
backend_server\operations_test.py (1002, 2021-08-02)
backend_server\service.py (1503, 2021-08-02)
client (0, 2021-08-02)
client\package-lock.json (413121, 2021-08-02)
client\package.json (440, 2021-08-02)
client\public (0, 2021-08-02)
client\public\favicon.ico (91862, 2021-08-02)
client\public\index.html (1665, 2021-08-02)
client\public\manifest.json (317, 2021-08-02)
client\src (0, 2021-08-02)
client\src\App (0, 2021-08-02)
client\src\App\App.css (139, 2021-08-02)
client\src\App\App.js (486, 2021-08-02)
client\src\App\logo.png (20579, 2021-08-02)
client\src\Auth (0, 2021-08-02)
client\src\Auth\Auth.js (506, 2021-08-02)
client\src\Base (0, 2021-08-02)
client\src\Base\Base.css (50, 2021-08-02)
client\src\Base\Base.js (1397, 2021-08-02)
client\src\Login (0, 2021-08-02)
client\src\Login\LoginForm.css (103, 2021-08-02)
client\src\Login\LoginForm.js (1555, 2021-08-02)
client\src\Login\LoginPage.js (1833, 2021-08-02)
client\src\NewsCard (0, 2021-08-02)
client\src\NewsCard\NewsCard.css (766, 2021-08-02)
client\src\NewsCard\NewsCard.js (1996, 2021-08-02)
client\src\NewsPanel (0, 2021-08-02)
client\src\NewsPanel\NewsPanel.js (2266, 2021-08-02)
client\src\SignUp (0, 2021-08-02)
client\src\SignUp\SignUpForm.css (104, 2021-08-02)
client\src\SignUp\SignUpForm.js (1872, 2021-08-02)
client\src\SignUp\SignUpPage.js (2392, 2021-08-02)
... ...
# News4U-News Manager
## Introduction
News4U-News Manager is a real time news scraping and recommendation system. This system uses a news pipeline to scrape latest news from various of resources such CNN, BBC and Bloomberg etc. To render the news, this system integrates with a single-page web application built by React. In addition, it generates a customized news list for each user based on news topics. To achieve this function, a click log processor collects users click logs to update a news preference model for each user, and an offline training pipeline models news topics.
![demo](https://github.com/rishavgiri6/News4U/blob/master/demo.png)
## Architecture
![architecture](https://github.com/rishavgiri6/News4U/blob/master/architecture.jpg)
__SOA__
* __Client__: a single-page web application built by React.
* __Web Server__: handles the sign-up and login functions with node.js and Express.
* __Backend Server__: requests news from database, labels specific news with "recommend" tag based on the response of news recommendation service and sends user's click events to click log processor
* __News Recommendation Service__: responses a list of prefered news topics for the current user.
* __Click Log Processor__: updates a user's preference model using a time decay method.
* __News Topic Modeling Service__: predicts news topics using a CNN model generated by an offline training pipeline.
* __News Monitor__: with the help of News API, finds the latest news from 20+ resource websites. It integrates with Redis to filter news with the same titles.
* __News Fetcher__: obtains a scraping task from the task queue and scrapes news using Newspaper3K library.
* __News Deduper__: utilizes NLP techniques to compare the content of the scraped news and existing news in mongoDB, calls the news topic modeling service and then stores the unique news into database.
## How to run it
To set up this system
```
./launcher.sh
```
To scrape more news
```
cd news_pipeline
./news_pipeline_launcher.sh
```
To update user'preference model
```
cd news_recommendation_service
python3 click_log_processor.py
```
近期下载者:
相关文件:
收藏者: