Building automatic mind map generator for natural disaster news in Bahasa Indonesia

摘要

The use of the internet today cannot be separated from people's lives. The available information is getting bigger and easier to obtain. Such information can be found in blog articles, news sites, and even statuses on social media. However, the available information sometimes cannot be utilized properly due to lack of a better understanding of the obtained information itself. The purpose of this research is to develop an automatic mind map generator application that can create a mind map from the input of news articles automatically. This is expected to help users understand the contents of the article. The study will take a case study of natural disaster news articles. In its application, researchers used the Support Vector Machine (SVM) method of multi-label classifier one vs rest with linear kernel to classify sentences in the news into the 5W+1H class (what, when, where, who, why, how). Beside classification, this research also includes summarization task as a preliminary task. To get the accurate model, we conducted some experimental study by examining combination of filtering features and candidate features. Our classification model raises F1-scores of 75%. The classification maps the word or phrase into each class, each class is determined as each node in mind map visualization with the root node is the image which shows the title of news article. The usability of our application was evaluated using The System Usability Scale and got the score of 78.5. This mind map generator also provides model evaluation by users, each user can review the classification result and if they agree with the result, they can update the model. By this scheme, the accuracy of our model is getting more accurate and lets us grab new data set automatically.