A full-stack news intelligence web platform with text mining and NLP techniques

Abstract

This project aims to develop a news intelligence platform to deliver user-friendly visualization for easy reading in complicated news articles. NewsPoint, which is the full-stack web application developed with text mining and NLP techniques, is constructed as a client-server architecture with a database. The client-side supports automatic newspaper extraction by web scraping and allows users to input articles through different modules. The backend server, which is a Python Flask framework, can perform a variety of analytical approaches, including text summarization, key point extraction, lexical analysis, sentiment classification, and Twitter opinions collection. The backend engines are designed in unsupervised and supervised learning, thus the former aims to implement the NLP algorithm while the latter is used for text prediction and classification. The frontend and backend server is communicated through RESTful API. Once the client activates the text mining task through an HTTP request, With data fetching, the server will handle the API request and return the JSON file to the frontend. By the Jinja2 engine decoding and Bootstrap visualization, the analytical result can be displayed in graphical representation. For the UI/UX design pattern, the app allows users to easily control and view a dashboard with a computer and mobile phone vision. To conclude, NewsPoint is the application for the public to utilize the text mining tools to simplify or deep-understand the content of newspaper articles, to cultivate the customized reading strategy.