Semi-supervised Event Message Identification System for Targeted Domain
San San Nwe, Nang Saing Moon Kham · 2018
Social media have become increasingly popular components of our everyday lives in today's globalizing society. They provide a context where people across the world can communicate, exchange messages, share knowledge, and interact with each other regardless of the distance that separates them. This research trend, extraction of events for specific domain from these social media is emerging speedily ranging from business intelligence to nation security field. The short length of Twitter messages and frequent use of informal and ungrammatical language challenge many long standing approaches for automatically detecting and categorizing events using streamed data in Event Message Identification system. A semi-supervised approach with Support Vector Machine (SVM) in combination with the corpus to identify the events from twitter for targeted domain in specific location is proposed in this paper. The experimental results show that the proposed semi-supervised SVM model is more efficient than a strong state-of-the-art semi-supervised classification model of Logic Regression, Naive Bayes and Decision Tree.