Crisis Events Information Extraction

Unfortunately, crises occur quite frequently throughout the world. In an increasingly digital age, where most news outlets post articles about events online, there are often tens or even hundreds of articles about the same event. Although the information found in each article is often similar, some information may be specific to a certain article or news outlet. And, as each news outlet usually writes a lengthy article for each crisis event that happens, it can be hard to quickly locate and learn the basic, important information about a given crisis event.

This web app project aims to expedite this lengthy process by consolidating any number of articles about a crisis event into who, what, where, when, and how (WWWWH). This information extraction is accomplished using machine learning for named entity recognition and dependency parsing. The extracted WWWWH info is displayed to the user in an easily digestible table, which allows for users to quickly learn the essential information regarding any given crisis event. Both the user’s input and the output data will be saved to a database, so that users can see their previous usages of the program again at any time. While users must manually input web articles into the program, whether as links or .txt files, there is potential in the future to use a web crawler to automate this initial article gathering.

The stack for this applications utilizes the MERN Stack. MongoDB was chosen due to its flexible document structure. For the back-end features such as natural language processing and our server we utilized Python and Express/Node.js. The front-end consists of React which is used to fetch our data and utilizes component libraries such as MUI for consistent design language.

The deliverables for this project include our Final Presentation and Final Report which show our progress throughout the development stages, and finally our code for the application which are submitted to our professor and client, Mohamed Farag.