Hybrid Classification Model for Twitter Data - A Recursive Preprocessing Approach
Muhammad Bux Alvi, Naeem Ahmed Mahoto, Majdah Alvi, Mukhtiar Ali Unar, M. Akram Shaikh · 2018
Tremendous advancements in communication systems and computational power have ushered rapid unremitting increase in the size of data. Tweets, blogs, product and service reviews are a few among main sources of user generated text data. These sources depict public reaction and response about an entity. Sentiment analysis, a major application of Text Analytics, aims to know about collective sentiment of people about the entity under discussion. This work encompass a novel hybrid method involving recursive data munging module with machine learning techniques to glean classification of closed domain twitter dataset on the issue of global warming and climate changes. Experimental work proved that this hybrid model gave better results and achieved up to 86.18% accuracy for the given twitter dataset in comparison to base-line classification model.