USE OF SOCIAL POSTS FOR DISASTER DETECTION USING NATURAL LANGUAGE PROCESSING

Abstract

A person's attitude is reflected using his behavior. If we want to know a person's behavior, we can ask to his friend about him. With the growing importance of social media, researchers made social media as a business review machine. Using sentiment analysis on reviews, product's market value, lifespan of product etc. can be predicted. Social media can also be helpful to get reaction of public on some social issue. This will help to politicians for analyzing the impact of social issue on public's mood. Sentiment analysis of reviews from different social media such as short texts are insufficient for analysis. The main idea behind the proposed system is to make use of social media which is immensely active i.e. Facebook and use the posts which are posted. Using sentiment analysis and Natural Language Processing (NLP) on posts, disasters are extracted (riots, accidents, traffic issue, natural calamities etc.) and using Naive Bays classification technique disasters are classified. The challenges are processing of unstructured data and finding the annotated data. In this paper we find the solution for above challenges which will be beneficial for our system and provide solution to handle unstructured data easily.