Social media analysis for Indonesian language: case study flood in Jakarta

Social media had become worldwide phenomena. More than 80% of Internet's users are social media's users. When a disaster occurred, information needs will rise. Twitter is one of popular information resource especially in Indonesia. Because of that, twitter's information extraction system was needed. This research proposes a system that can detect topic in social media twitter by representing its content as a complex network graph using the implementation of natural language processing, graph concept, and complex network analysis. This system consists of 3 subsystems which are crawler, graph converter, and application for graph visualization. From testing result, we reach 89% success rate of keyword extraction using RIDF term weighting method and collecting messages by certain category. General topic about governor election and 13 subtopics was successfully extracted from set data flood in Jakarta.