Feature expansion using word embedding for tweet topic classification

One of Online Social Network (OSN) roles is a source of information, especially during an emergency. Twitter is an OSN service that enables users to send and read message but is limited to only 140 characters. Therefore, the tweet is written very short, not always using grammatically correct and using a lot of word variations. The use of word variations increase the likelihood of vocabulary mismatch and make the tweets difficult to understand without some kind of context. In this paper, we used word embeddings based on word2vec to reduce the vocabulary mismatch for tweet topic classification.